. .. ... .. . .. .. .. . . . . .

. . . . ..
.
.
....
...
..

..
....
..

............
..
.....
..
..
..

.......

....
........
...
....
...-.... . ..... ... . . . .. .. .. ... ... .....
..........
.......
....
. . ...........
. . .. . ............
. . . ......
.. ............
.. . . . . . . . .. . .:-.
. . ..........
. ..... ......
......
. . . ...~-
. . . ...
. .... .
.. .........
. . ... .....
. . . ..... ..
.. . . .. ...
. .
. . .... .... .
. ......... ....
. ...
. .........,.....
...
. ..
. ...
.................................
.....
..........:-.~::.
..
.. .............. . ..-... .......: ..:: : ..:.....
..
..
........................................
....
....
: :;:. .:: :..::.::...:..:.::.::: :::.::..:.:..:::.:.:.:::.::.::.::.: :.:..::..: :::.::.::.:.::::::::
~-.::.:. :::.::.::.-:.::.::.: :.::::.:.::.::...::...:::
. .::.:::.::::.::.::.::::::::::::.:::::::.::::.:.
.. . . . . . . . . . . .::.:::::::
. . :'.::.:::::
.. . . .::.:.:::.:.:::...:::.:::..::.::.: .:.:.:.:.::::
..
...
..

..
.. ..-- .... .... ... ..... :. .............
..
-- ........ .. .. ............. . . ...... ....... -.. _. .. .. ... . . . ...... .. .
..::::::
....:::-::.:
...... ........ .. . .. :... : ..: ...... ... .. . ......... ....~ ....... .. : -..:: : ....... ....~.. .. .
:: .:: ..
::::::.::::
::.:::::::::::::
..
::::::.:::.:::::.:::::
.. ~::.:::::.:::::::::.: :.... : .::..:.:.: .. : ::::::.: ::.::: .':'":::::: .:::
.
..
..
-..
...

..
.:
.
.

..

....
.
-
..
..
..
..........
....
.
. . . . . . . . . . . . . . . . . . . . . . . , _ . . . . . . . .. . . . ... . . ..
. . ..
. ...
. .. . :-. .....
0-. .
. . ... ... ..
. . . ......
. . . . . ....... ...
. ..
. .......
... .. . .
...................
.... .,.. . . . . . .. . . .....--.. ..... ... .. .. ........... ~.
......
,_. .
.
..
.
.. ..
.
~ . ....
.. . ... . ..-...
.. . .

..
.
..
...
..
...~
....
..
..
................................,...
..
..
..

..
.
.......
...
..................
. ...
. ..
21
..

.........
.-. ... ..~ .. .. .... ........
.........
. . .
..
..-......... ..................
....... .... .
.. ... ....
......
...........
......
...... . ..
. .. ... . .. .. . . *.
:
..
..

....

"

..

..................:

- -

..

: . : . : . : . : . : . : . . . . .. . . . . . . . . . . . . . . . . . . .

...............

..

. . . .,._

..

..

0-

. . . . . .. . . . .. . :

........:

,_

0-

..

..

- _ ........

...... .

. .... -..... ...~...... -:....:::::-..... "'.... .....-i .... .......:. : ~. ...:..--.... ...... .. .......,......
:.::...
:.:.::.:.::-.;:":.. ::.:.::..:..:...::..;::::..........
:."::.::::;:;.:::::::..::.: -:::.:::..:::::.:.:...
::::::..:...
:::::::......
::;;;:::'::::::.::.:::::;::
:::....::.:.::
"::::.......
::.:.::: ::.:.::::.:::":::::::::.:'.:::::::":::: ::::::::::::::::::
.
.
.
..
..
.....
..... . ....::.:-::-,.~=::::::.;:.::::::
. . . . . .......::.
:.::: .::.::: :::.::.::::::::,:::..:::: :::: == :-: :: :: :-~ :::.::::::::::.:.::::.::::::: .:-...... :-:.=:::::
::.:::::::::.::-: ::::.............
:.:.:.:.:.: .:.::..: ::::.. ::.:.. .:............
::.::::.: :::::;:: : ::::::.::.................
:: ::::::::: : :::.::::::::::: :...
:::..-....
::.::: .:.:.::::::.
:.:: ....
::::............................
::: :::.:.:: ..::::::: ::.:: : .:::::::.. :: .:-::.: :: :::: : .::.,.:...
::: :.:-:.
............
......
. ...+
..- .. .......
. .....--..
..........
. ...
'"'"
.......
.... .......
. ..
...."
....
. .. ...
..
....
.-- . ...
- .
-..... .-
- . ......
"'"' - - .. ..'4.,.
, - ..
.. ,... __ ......
..;,.. .......
.-..
.. ..
.....-.--..
..-. ...
..

....... ...... .. ........ .. ....... . ',...;,:' .........'" ... . ... . . . .... . . - .. ~ . . .... . ....... .......... .. . ..........
....
.
... .........
......
...
-
........
.....":..
.............
.
.....
.........
.. ..
. ..
....
.. ....
....
...
....--.-.

.
.

.
.:.:.:
..:
....
..

..
.
..
...
.
.
...
...
....
.
...

.
.
..--.....
.
..
....
..

.
..
...
...........
...
...
.. . ..
..
..
.
... ...
-.
.. .
. .....:...
. ....
.~
.....
..............
..
.
...
..
....
......
. ....
..
.....
.....
.........
... . .... .:
.
...
.
,,,_.......
....

.....
..
..
..
~ ....
.
..
....
.
..
..
..
.
..
...

.. .....
.
..."4.......
.
.
.
. ...
.
.
.
..
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...... .....
... ..~.. ... ....
:-:-......
. .. ....
. ..
. .. ..-.~.
-. ......... ........... ..:..
.. ...
. ..:
. ......,:.
.~
. . .:.:--=
.............:...!
... ..
. . ..,. .....
....,_ .....
. ...........::
.: ....
. ..=.. . ..: .:.:.:-.
. .:. ....
....~........
. . .....
.. ............
. ......... .-.
.. .

..:
...
.
::
:::

.
.

....

.....

...
.
..

.....
...
.....;-.
.- .-. :
:: : .: .:
.. .......
... . . . . . ... .:.:::
::::
:::: : ::. : .
.. .... .-:
. ........::
. . ..........
"'........
1;...... . .
-~. .....~
........ ,,..._~-. . . . -
. ....
.-. ). . ,. ... , ......, ... \. "..ii: .
. .....
-
r .r.

- . . .. . . . .. . . . ..

.. .# . . . .

..

:-...... - ... .. . . . . ....

..-

,._

.
...

\ ~

-

.........
0

..

..

.... ,

._ ,

...~--..
- 1, ___.

.-..- . . . . . . .

... . . .

...-*

0-

"-<- "

,, .. ,,,,,,.... . . ...~...... .~ ... . .... ~r ... .-.. .. ......-. ~ ......

.. ;

,\
.. . , ..
..
... .. ..,..

..

- -,- '-' - .:-...:'..-'.'.

.. ...-

.,,

,i

r-;-, :-:-:-:-:"" . . . . . . . . . .
,""-

...

.....
.. .

. .. .
..

. .. ,,,

,...,l'l"f' . - . ...... ,.... .... , ......... ' '."'\ ..,..

~, -.
.. . ......
... ) ;
. ., . ;. ....1.. , { . ".. . ..r . .. . . . " .

. . . ""
.. .....' ,
, . .. . . .. . .

::: : ..:" :: : ':: :::: .p,. . :>"" ~ ::: f.:: .' "..,:: ~=.::i
;.,.:
...:.. - : :'': ..: '" ::."::.r,
-:.1 f "~~:::~:.::::::: : . : :.::::
"'
"
"'

M
._.!....-~" - " ... . .... ...... . . .. . . ..
=--::: ::: ::.::.: :: : :::::::-::.''""..--..:...--:~ ..: ::... :: -:::::::::::.: . ..:: :.:. : :::: :: : :: . .. : ::
....
........:::
...... - .
...
.......
. .....
...........
.. . ..........0-.......
..... .. .. . .:::::
. . . . . . . ..
... .... . .. ......
.... . . .. ....
.... .
. .....
.. ....
. . ...:::
. .. . :..::-..
. . . . . . .......:....
.. ..
.. ...
... ..
....
.....
.... .. ....
..........
:.
::.:.:::::.::::::.:
: .::: : . :;::,::::
:: .:::::.::.::..:.:.: .., :;.\:'t-::\::.:
:V.
~:~f'- :-. -. ::~~
:::::""'./"\~-::-~
::.::[.::" r: : ::\ .i<::::::
:: ::::::.::::..:.:::
~::::::::.:.:.::::.::::
.;: :::::: ::.
. .
. ..... . ... .. .
r. .. ..
.....
. ....
.. ...
- .
. .. . . . . . .. . . ... ~ ... \..1
,,

.
.
.
.
.
.
.

.
.
.
.
.
.
.
..
.
.
.
.
_
,.,
...
"'
l,0
'
.
..
.
...

....
.
.
.
.
.
.
.
.
.
.
.
.
. ....
... ....".',- ..
.,.

........ .
. "..t , .
.. .

.
..

~ .
..
,,..._ .
.
..
..
~..
-.
.............
...
.....
.
.. ~ ....
....
,,_. ..
.. ..
. ........-. ........ ...............
.
.
.
. .. ... ..... .......
.
......
.
.....
..........
.
.......
. ...
.
......
.
.
.
.
.. . ..
....... .
.
..... .
.. ..
.. ...
. . .. . -....... .:
. ...:
.
.
.. ... .......
.
. .. .... . .... .....
..... -.....
.
. ... ...
....
.:-.:
~ .. .. .. . . .. ... :: . .. ... ... .... .... .. ...... .. : ..... ... . ... . .. ... ..... . ...
::
.
:
:-:. ...
: .~. :......
::.:
.. :
:.-.
:.. "':'..... -.
~.
.:. ..---.-..
:. .. ...
:::::.
..:
....: ,.::
. :~.
:::.
: : :..
::
......
: :-: : :.. .
. .:.. :
...
..
..
~
..

. J~-~~
~- ..... .,. ,
:.~
~...., . .,._,r
........
~~
..
.
.
.,....,,
.-:
.
.
.
..
.
.
.
~.-:
.
.
...
.

'''

..

..,
....
.
.
,
.
.
~
...

""o

.
i':(
j
'
'.._
,
.
~

~
i

.

. . . . . . . . .. .. - ' . . . . ... ~ ~. ' ' '. . . . . .. . . ,.. .
.. .... , , I. ' ','4-!- f .......... ~. . \.
.. . . . . . . . . ....
-~. ~- . .~""'
1 1

..
\: . ~ ... .
~
'"\..'~
.-:.....
M. .'.::-,..,.~'!--:._-..~
..:--
....:..'. . ...
. .......-.:.:.
.........
. ."" ......
. .. .....
.
.
.
.
.
. ....... .... ....;..,.I'
..., ..
......
.._:--~--:.~
. . ... ........
.
.
.
.. .::: ....; .,.:.,...,:,;..;~:
.
.
.. .........
... ~.. ....:....::....-........ .....
....~....
.. .. . .-:.
.
.
. ..:::...::
.
. .. . . . . . .:::.::.
.. . . . . .
............
..
..
.
..
..
..

..

..

..~
.
...

..
.
........
.
...
v
'
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....... ..:...
.
.
.
.
.
.
..
.
.
.
.
...
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.....
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...................................-............................................ ............................................................... ..................................................................................................................... .................................................................................................. ................................................................... ................................
..... ...
....................
...:0-...:-::
...
... . ....
.
....... .. .........
..................
.... .
.................
. ...
... .......
...
.....
... .......
......
.. .............
.. '. .......
.......:...
........:
............
.. .... ...
. ....
........
..... .... ..........
..........
...........
.....- .....
... ......,....
.... ... .
.......
.
.
.
.
.
.
.. .....
.................... ....
.
:.:
.: ::.:: :.::::::.::
:::::::::::.:::.
... ..::::.=.::
. ::
:::.:.:::::-:
..:::-::
:.:.:..:
-:-:-:.:::
.-:.. ....
: .::
: .:
:
...

.
...
...
..
..
..
.:
:-
::.
:-
:-:
..:..;
.
..
....
.. .....
...
.
..
...::::.
::...::::
::
.
.::::
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
...
.
,
_
....
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
..
..
.
.
.
.
.
.
.
.
.
.
.
.
.
..........
......
. ..... ........ ... ... . ... . ... . ..... .... .... .. ..-... ......... ..... ...... .... .. .. . . .. . .... ... . . .. ... ... ............. ... ... . . . . ..... .. . .. .... ... ........ . .
:: : .. .:..........
.....
...........::.:
:-: -: ...
.... ..
, ...- . .. . ... . . ... .. . ::
: : .....
........
.... ......:
:: : . ::: : ..
.....
.. . ..... .. ....
....
.::.
: :-. ... . . ..... . . ........ ... ...

. -

..

..

..:... :\,;

_,,,,

"

' <r"'TI

. . , .

. . - .....

,_. __ ... -

0-

. .

..

- .

.. ... .. . . .v., ._ . . "" ... . . ... -...... . . . " .. - .. ,. . ...-- . -... .". . .. :- . . . . ... . ................. .
.
.
.
...... ............ .......... .... .. ............ ........... ................... ...... .......
........... .......... ................ .. ........ .... ...' ........................... ............ .. ..... . ..................
....... ...
.
.
.
.
.
.
.
.
.
.
.
.
.. .::: ..... . . .. ..... .. ...: ...... . ... . . .. .... .... .. .. .. . : .. : : . ::: :'":. ... ... .... ... . .. .
- :""')

. . . . . . . . . . ' \._.

.. .

... . , . \! . 7 . J

., , ..

111 ,

~...

..

. . ,

..

:::~ . .::-::: :::::.: :::.:::::: :::.::.: :: .:::::::::= .. ::::::::: :: :::-:::: ::':': :.::::::: ::::::-::::-:: ::::.:::::.::.:::::::: .. :::::::::::
.::
::.
;.. .. . ... .....::: .....:::: :-.:.:: .. ..;..
......
.. .. ....
: .. .. :: :-: ; ..::.:: ... :......
.... . ....
.,,
..
.. ..
..........-.
...............
.
.
.
.
..........
.. ... ....,................~
.
.
.
.
.
.
.
.......
"";.. ....... . ...... ... ........ ...... . . . . ....... ....
.
.
.
.
.
.
.
.
.................... ......
.
.
.. . ... . . . ... : ................. .. ..... ...
.
.
. .. .......... . ....... ... . ..... . .
.... ..
.. ... .. ...
...
: . .....-... .........................................
... ................ .. ........................:::
. .-:.. --. .. ..,, .. . ...-- ....:
...
.
. .......
, ......
........
..----

*
....
......
...........
. . . ..... - . ..
..' ...
.........
... .. ... .. ... . . .. .........,,.. . . . ......
.. . . ...
.
...
...
.
..
....
..i..........~......
.
..
....
.
.::: :-: :.:. :_:.:. .. : : :::::::::: :: ... ::::: :: .::::::::::: ,; :: : .:.: :: : :: : :::. ::::::::::: ::: :: .:: .. ::::.:::: ::::: :: :: .:: ::..::. .. :.:: .... :::: .... :.:: : :: :.:-: ::: .. ::::.:: :: :::: ::: .
. ... . . ... .... .... ...... ....... ... ... .... . .. ..... . . ....... ....... . ... .. . .............. ..... ...... ..-. ............ ............ . .. . . . ............ . ...... ................ .. .
::-::: :::::: ::::.: :::. :::.:: ::::. :-:.. ::-: .::.:.: :.. : . .. : .::: . : .::: .:.::.:: ::: :::: :::: ::: ::: :: .::.:.:.:. :-:-:.:.:. . :-:::::::::::::: :: ::.: : .... -: ::.: :.:.:.:: :::
.:::::
. . . .. ...
,, . ... ...... .. ... . .. ... .. . . . . . . . . . . . . . . . . . .. ........ . . . . . . . . . . . . . .. . . . . . . .. . . . . . . . . . ... .
...... ..... ........ ...
...........
............!.,',
. ....................
. .........
.......:
..........:_..........:.::
.... ..::
......:::
... ............. ..........
..................
... ........... ...... ....::.:
...... .. ........:::::
.. ........:
. . ..:::::
. ................. ....
.... ..........
................. ..... .
.
.
.
.
.
.: :::::.: ::: :-;-: ::.::::::::.: : . .:::::.:::::: : :::: ::: :: :: .. . : ..:.::::::: :. :: :: : :::.
:::: .. t: ::;:-: :. :-:..: :::.:-:::::..::.::.:: : :-.:: :.::-:~:::::::::::.::::::. :-:::: ::::::. : :. .. ..::: :
. ..::: .: .:... . . . ...... .. ..---.. ::. :::::: ::::.: : .: : . .. ....... ....... .. .: . ::~ : : :::::: : : : : :::.: :-
.
...
....
.. ..
.....
.....:..
...
. ...::~.:.:
.. . ...........--......... ... . ..-.. -.. .. ... ,.... .. ....... .......... ..
:
:
..
-:
...
...
.
.
....-......
:
:-.::
........,*"-'
:.::::
...:::" . ... ----........ . .. .. ..
...
....: .. .~.:::.: .....::: :-:: ;. . . ...:.: ..........
::::::::: :-::::
:::::::,:::::.:
.........::::::::::::::::::
.:.:....:. .:..:...:..:...::.::;...::...:.::......:..:..::..:..:::.:.:..:.:..:.:..:...:.:....:..:...::....:....:.:::...:..:....:.:...:...:..:..:...::.: ..::.... ..:.:..::..:....:......:.:.:..:...:.:..:::...:...:..:..::.. :.:.:......::...:.:.:.:....:.::.....:......:.:..:.:..:..::..:.:..:.:..:...:...::..:.:...::..:.:..:..:..:.:..... :...:::......::...:..:..:..::...::..:......:..::...::......:..:.:..::..::.:..::..0-:::..........:::..:.
........... .. .. ...
.. . .....:
.
.~.. . .... ...........
........ .........
.
.
.
.. .........
..........
.
.
........
... ....:.
. . . . ..........
.. ...
........
-... ....
.
.
....
... . ....~
. .. ........ ...
.. ....-.- -.. .......
... .. --... .......
. . ..........
..... ......
..
.. .
.
.
..
..
.
.::
:
:-:.
:.:::-............
..
..:
::::::::
:.::
..
:
........
...
...

..
....:.
:-:
..
:::
..
..
..
. ... . ... ... . . . . . . . . . . . . .. . . . . . ..... ... .... . . . . . . . . . . . . . . . . . . . ... ..... . . .. ... . . . . . . . . . . . . . . . . . . . . . . ... ...
:::.:::::-:: ::::: :-:::-: :-::::::-: :-:::::=:::~:: =::-::: ::::!::: :-:::: :-:::::::-:.::::.::::.::::,.::::::::. : :.
:: :-:: .-::.::
:::. =:: ::.:.:::::::
:: :. :: :-::..
: -: -:! ...
.:-. :: ..:-:::= ::.--......
:::::.:::. :: --::-r:-:"::.:
: .
: :.:.. ::.
.: :. : : : :: :.::: ::. ::.:::
:'::.:: : ::: ::.: :: .;:."...
:.:. :::..:
.....
.......:.......
........
.... .. .... . .. ...
.
..
-....
.
.. ... .............
.
.
.
.
.
.
.
..... ...
.
... ... ................
.
.
. ....... . ......:,.
. . . . . .......
.
.
.
......
..................
...... ..... .. .........
.
.......... ..,......
....... . . ... . .. .. . ...
. . ...... ..
.. . .....
...
..
.
..
...::
:-::-:-::
...
::::::
.
...
.....-:.
..:::
:--:~-:
.::::
.....
.
.
..
......::.
.::::
:-.
.
.
.
.
.
.
.
.
..~ ...... . ......... ........ ...... ............. . . . . . . ... . . ... ................. .......... . .......... ........................ ............... ...... .. .. .. ... ... .:...
.....
..;.. .. .. .. .. .... ...........::........ ...............:. .... .. .. ....-.... .. . ..
::: . ::: : .... :::: : :: :: :.: ::.::: ::::::::::::-:::.: ::::::::: : :: ::: : :: :: .:: :::: .::.::::::: ::::::::..:::: :::: :: .::.:: :: ::: :.: :: ::.::::: :: :: :: ..:::::.: : ::: :: :::: "::::: .. ::: :: ...: :. : : .. : .:
......:-..:::.-::.:::.-.
....
...+ :::::
... .
............
. ... . ... ........
.
.. .....::
.... . ......
., . . . . . .. . .
.................
. . . . .. ........
. .....
. . . . .
.... ...
. ....~
.. ..::
. . ....:.,:::::::::::::
. ... .................
.... . . . . . ........:.
.. ".....
.. . ... ..... ...........
..... ~
.. .
. .......
. . . .......... ......::......:.:....:.........: .:;::
. .
::: :::: : :::: : : . :: : : . : . : : : . :: : : : :;: : :; : : ! : . :!;;: :::.::::::::::::::::::::::: .:::::::::::::.::::~: =~
:i:::::.:::::.:::::: ~:: ::~:.:::::.: ::. :::::::::::::::::::::::.::::::::::::::.:.:::.::.
. . . .. . .... . . , ... . .. . .. ... . . ,., ., P : ,.1 . . .. . . . . . . . ... .. . .. ....... . . . ....

' o o '
'

'

..

' '' '

'

'

'..

'
'o' ' o

..

'

''

1' ..

' '

'

'

'

'

'

'

' ' o o

.. .. ... .. ..

'

'

..

... o

o
o

o
o

o " o

'

' '
'

o o ._ ' '

..

----

'

' '
' ' 'o '
o

'

o '

'

..

..

..

..

..

'

..

'

' I
' o '

'

. . . . . . . . . . . .

0---

'

'

'

'

.,

"

,_,

'

'

'

'

' ' '

.:.::::::::

";.:"".

.-

.: : :. : ::
........
..:: ..--= ::.:::::::::
: :: ).:...c-- ..::::..:: .:::::

...... ".

. . . . .. . - ... . . . . . ~ .....

"

"

'

'

'

'

'

'

..::::::::
.. .. .:.: .: ::.:::
..
.....

..: ::: ::::::: :: :::.. :. ::::. :: :':::.:::.::::.:::.:::.::.::::::::.. :: .. ~~,~: : :-.::.. ... :: .. :.. :.::::::::::::::::::.:-.::: ::: :-:::::::: ::::::::::.
: ;.... . .. . . ..:: :: . :.::.: . .=:: : ...... .. ..........';". -.:..;:::::: ..:: ....... ...:. ..::.::: ..::::::.:::::::
~: ~: ~ ~~ :~.:: ~::.:::: ~.:: ~:~: ~::: ~:::::: :~:::~:::::: :~:::::~::
~ :~~~ :~;
:~::.: :~::::: ~: ~ :~:::: :~: ~: ~::: ~=~ ~: ~: ~::: ~=~: ~. ~-:: ~::::::
::
,; : '\:
:--::::: :::=-:::::::~:
.::.::::::::.:::
.. .:.
. .. :.. :.. . ::::--:::-::::::,.:-::::::::::
... . ....... :::::::~.,:
- t'r.
"\.:...'.;.:::;..:.:::.:.
"-"' ...
............

::.:: ::::::::::::::::.::::.::.::::::::~::: ~:::::::::::::::::.:::::::: ::::::::::::::::::::~. r,:... :::\: 3:~:::::::::::: :: :::::: :::::::::::::.: ::::::::.::::::::::::i:::::.:::::::.:::.:::::::::::::::
-.. : : :::.-... . ......~::..:.:::. . . . .. ' . '-: "\ 'f.: .. .. . ... .~-.: : :: :-: .
.. ........
. ..:-::
. ...
....---..
....
<'.........--.... ........... . ...........~ . .. ... . ... .. ....
..:.......
:::-:
..
.::-:..
.::::::
.:
-:
:..:::
..
:-:
:-:
:-;-;:
...
: ::.....:...
...:::. . ...:: ::. : : :: : : : ::: ..:: .. . :::::.: ..:.::: :::--: :.::. :: ::: :.: ...
..
...:::
.. :. ..:
...: ....
........ ........: ..::::.::: . . .............. ::.::.:: .....:-....:..... ..- ...........
.
..

...
.
.
'." .... ........... ... .. .. .. . .... .. ......:::- ...: ... . .. .. ........ .~ ..... ....... ....... ... ..
... .
... :-:.:-::::: .: : :-::
.. . ...:
...::--:.<:
.......
. ....
. ....
.. . " ..
. . . . ......
... . . .... ...
..... ...--........
. . . . ... ::
. ... . ....:.:
. ........
. . . . ...........
.....
. . ......
. .....
... .. . ......
.. .. . .~-. . ..... .: .. : :::.:
. . .- :~ ..
....
...
..
.
-
......................
...
........ ..
........
. . . .... ... ...............
......
......
...... .....
. .. .... .....
. . . . . ...............
. . . . .............. ......... ...... ..... ... ............ ........
... . ...
............
. .......
. ...
. .,,_.....
... ..
...... ................ .
............ .. .....
....
.
.
.
.
.
.:
..
.
.
.
.........................
.
..

.
..
.
.
.
.
.
.
.
.
........
:
..-....
.... ... . - ... ..
. .. . . . . .. ... ..
...... .. .. ...,._, . .............:. .-..
........ ...... ... .. --. .. .. ..... .. ..
. - ..... '.* . ....,...... .
..:..::. ::: : :
... .....-... .. ... :
. .................... .
......:. .... . . . ...... . .... ~ ... .. . .......-......--......... .................... ..... .,, .............. ... .. ... . :-::""'::.. .. .

::: : :;:

. . . .

::

..

..

.. ...

"

t . . . . .

: :::: : ::;. :-: _

::::::::::;

...... -.. .. .....,

..

..

.,_

..

" 1.-;\

..

..

..

..

..

::

..

0-

..

..

..

: :::::::::::::

::: :

-. . .

..

..

...

;o

--..

.,...

-~"-;i*j..~~~::~:~~:~:
~~ :~: ~:~ ~ ~~~~~:~~~:; ~~~~;:j~~~: ~~ ~\~ j~~~:~Ft~~~~~~~~:~:~~~~~~~:~::::~~;~~~~~~~~~~;\~~ j~j ~~~~:::~~ ~; ~ \:~~ j~~~f: ~~~~;:;~ ;~;~~~~;~;~~~ :~: \:.~:~:~:~~ \: j-~~~ ~~~~;:.~: ~:: ~::
.......-:_-..:. .~ .. .. .. :: i-~ :: ...-..... .... ::.-:. .:.::: :::: : : .:::..:::: '._!,::.:.:::. : :.: .. :. :: :: :.:
. ..

....
.

..
. ...
~.. ..
..
..... ...
.. ........::.
::: ........;o.....
. ..
...
....
...
. .....:..
.....:.
... ..
.:
. ...............
....
~... ......
\. ' ..
. . ...' ~..~
. ...
. ................
.....
. ....
. . . ....
... ..........
.. ..
.. ..
. . ..
. . . ...
. . .. --..........
. .............

.-:_ :-: :: ..
..
.0- .. ........
.. . .. .......
. ..
....... ...
.......:
. ....
. . :..

.....
..
, . . .. .... . ....... . .... . . . . . . . . . . . ... ..... .. ... . .~ . .

.. 0-. :~......".--.. .. ... . . .. .......

.

:.:
... . . . . . :-.--.... ... . . .. .. .,:::::::......... ...........:.:::.::.::::: .. .......--..~.....=--.::... ::. : ;.~. . .
.. :....;.
~.. :-::::.--! .... . . .........:::::.. ... .... .. . .""-.......... .... ....::::----..:.-....;>.. . ... ....... . .... . ..:::.
....: ~ .:::::.:: :::=-.: ::::.:..:.:.::::-- :::.:: ..~. :--;:::.:::::::::::""::::::::::===-::-..:...;::.: ..: !:.:..: ... ::.:.:.:_..: ...:. ::: .. ::: ::::: ::::. : ::::: :::::.:::: ...:.: ..: . :. :. : ... ..
... .. r.:. . ....:: .. . .:.... .. . ::-:-... . :: . : . . .... -. ! ~ .. ..- .. -- - ..... ..: :-:. ---:: .. -.".. ......... .... :: : .. ~
.. --:...::.:.:::::: ;. ...... ..... . . ... . .. : :::: :::-..... ... . .. :. ::::::::--:..: . .. . .. .... ...... :.- ......:. :::
.... .'r~~ ......
. .. :-:,. ..:.-..
...
..
.....
.........
: . : .......::: .. .... ... :: :-.:: . :.::::: .: .. :-: ...... .. .
..;......
.....
.
.:.:
:.
:
.:
..
..
Q....
.... . .~~ . ~:. ...--~
.....
.......
.. ........ .
.. .
.......~....
.... ..~- .. .. ..
. .. ..
. ..
~..
..
..
..
"---
..
...
~-
.~.:.: .:: :\:::~:: : .:::::..:::.::::.::.:..:::-r..:.:.:.:.: ...: :.:.::::::::.:::::::::::.:::::-=::::::.:.:::.:. ~ :.: . :.:.:.:.:.:.:..:...:.:.:.::.:::.:;:.:.::.. :;:::: :. :::.::::.. .:.::.:.:.... :.;
.....

.....!.,. ......... .. ..

-
......
.......'":"...,._
..........
.......
:z
......--
....
.......-.. ............
...... . ...
.. .. ~ . ......
. .....'"
....
...
."': :."-..
. . .. .. - . .~-:
. : . ...
.....~....
....
. . ...
......"-..... ..:.--..:.= -.. . .:..... :
. .. ...-*. ... ..
.: ..............
= ..............
:-:.. .=
..........
. .... ....:
: ... --
0- . . .

: - - : : .. : : : :

AN INTRODUCTION TO THE STUDY

OF CLASSICAL ARMENIAN
by

Robert GODEL

WIESBADEN 1975
DR. LUDWIG REI CHERT VERLAG

CIP-Kurztitelaufnahme der Deutschen Bibliothek

Godel, Robert
An introduction to the study of classical Armenian.
ISBN 3-920153-37-5

Dr. Ludwig Reichert Verlag Wiesbaden 1975

Alie Rechte vorbehalten
Photomechanische und photographische '\Viedergaben nur
mit ausdriicklicher Genehmigung des Verlages
Gesamtherstellung: Imprimerie Orientaliste, Leuven
Printed in Belgium

TABLE OF CONTENTS

FOREWORD

ABBREVIATIONS.
BIBLIOGRAPHY

THE ARMENIAN ALPHABET

VII

VIII

IX

XI

FIRST PART

A SYNCHRONIC APPROACH
1. Classical Armenian (1-1.2) . .
.
The Alphabet (1.3-1.42)
. . . .
Text Sample and Comments (1.5-1.642)

2. Phonology
. . . . . . .
. . .
The Classical Armenian Sound Pattern (2.1-2.12) .
. .

Syllable division : hiatus (2.2-2.213) . .

. . . . .
Vowel and consonant alternation (2.22-2.232) .
.
.

Syllable division [Link] : the neutral vowel (2.3). Consonant clusters

(2.31-2.345)
. . . . .
. . . . . . . .

The traditional pronunciation of Classical Armenian (2.4-2.43)

3. Morphology .
. . . . . . . .
. .
Noun I nflection. [Link] Remarks (3.1-3.12)
Declension Patterns (3.13-3.152) . . . .
. . .
The Armenian Grammarians' doctrine (3.16) . . . . .
Anomalous Nouns (3.17)
.
. .
Case endings (3.18-3.183) .
.
. . .
.
Verb Inflection. The Armenian Verb System (3.2-3.222). The
(3.223) . . . . .
.
. . . . .
Regular Inflection (3.23-3.236)
.
. . . . .
Diathesis (Voice} (3.24-3.243). Causative Verbs (3.244) . .
. .
A Classification of Armenian Verbs (3.25-3.255)
Compound Tenses (3.26) . . . . . . . . . .
Word Formation. On Derivation and Composition (3.3-3.32)
. . .
Regular Derivatives and Compounds (3.33-3.34)

.
.
.
.

. .
. .
.
.
. . . .
Verb 'to be'
. . . .
. . . .
. . . .
.
.
. . . .
. . .
. .

1
2
4
9
9
10
12
15
23
26
26
28
34
34
35
37
41
46
50
54
54
57

VI

Table of Contents

SECOND PART
A DIACHRONIC APPROACH
4. The Historical Framework (4.-4.122) . .
.

The Problems of Etymology (4.2-4.314) .

. .

Sound Change. The PIE Sound Pattern (4.32-4.323)

. .
The PA Stress Shift (4.33) . . .
. .
. .
Armenian Reflexes of PIE Phonemes (4.331-4.335) and Clusters (4.34. .
.
4.343). Developments of *y and *w (4.35-4.354) .
Comments (4.4-4.43). .
. . . . .
.
Vowel Prosthesis (4.44). Epenthesis (4.45-4.452). Vowel Contraction
(4.46). The Transformation of the PIE Diphthong Pattern (4.47). .
The Origin of the Classical Armenian Phonemes (4.5)
5. Historical Morphology . . .
. .
. .
. .
Noun Inflection. The PIE Background (5.1-5.146) .
.
Case Endings (5.2-5.236)
. . .
.
. .
Pronouns (5.24-5.245) . . .
.
.
Verb Inflection. The PIE Background (5.3-5.323)
.
A Tentative Reconstruction of the PA Verb System (5.33-5.333)
Thematic and Athematic Inflection (5.34-5.341)
. .
.
The Development of the (Medio)passive Conjugation (5.35-5.353)
Present and Aorist Stems (5.4-5.433) .
.
The Infinitive and the Participle (5.44-5.442) .
6. Conclusory Remarks
INDEX

A LIST OF CLASSICAL ARMENIAN WoRDS AND MORPHEMES

.
.
.
.
.

61
64

69
72
72
85
86
90

92
92
99
107
112
115
118
119
122
128
130

133
135

FOREWORD

This book does not contain anything more than what its title promises. Its
scope is limited to a description of the Classical Armenian language, and to ah
inquiry on its origin and growth. Both parts have been planned so as to balance
each other; and, as a historical study on syntax, in the present state of our
knowledge, would unavoidably be full of gaps, syntax has not been included
even in the descriptive part.
This latter, an outline of Classical Armenian phonology and morphology, must
not be mistaken for a grammar. Several items have been purposely set aside, in order
not to obscure the main features of the language. These deliberate omissions are
perhaps compensated by a more systematic treatment of noun and verb inflection.
Besides, more attention has been paid to phonology than is the case in recent
grammars or text-books.
As for the second part, it aims at supplying reliable information on etymology,
sound change and historical morphology, without concealing the difficulties that
attend the comparative study of such a language as Classical Armenian. Meillet's
fundamental work, Esquisse d'une grammaire comparee de l'armenien classique,
was first published in 1903. A second, revised edition appeared in 1936. Since
that date, many phonological [Link] have been discussed again. Meillet's
views on historical morphology, though they have been less controverted, are in
need of revision. This, I believe, is enough to justify a new approach.
The list of Armenian words and morphemes at the end of this volume would
have increased exceedingly, should it have included all the examples quoted in
the descriptive part. It therefore contains only those on which some comment
is made in connection with etymology, sound change, or morphological pecularities.
I express my gratitude to Professor Giorgio Raimondo Cardona, from the
University of Rome, and to Professor Calvert Watkins, of H arvard University,
who took the trouble of reading the manuscript of my Introduction. To both I
am indebted for many a helpful observation.
I also have to thank Mrs. Lilliam Hurst for carefully revising my English text
and removing the stylistic infelicities.

ABBREVIATIONS

Consonant
Vowel

N.
Acc.
G.
D.
L.
Ahl.
I.

Nominative
Accusative
Genitive
Dative
Locative
Ablative
I nstrumental

(P)IE
PA
Cl Arm.
ModArm.

(Proto)-lndo-European
Proto-Armenian
Classical Armenian
Modern Armenian

Alb.
Av.
Gk.
Goth.
Hitt.
Lat.
Li th.

Albanian
[Link]
Avestic
Ancient Greek
Gothic
Hittite
[Link]
Lithuanian

OCSI.
OHG.
Oicel.
Oir.
OP.
Skt.
Umbr.
Ved.

Old Church Slavonic

Old High German
Old Icelandic
Old Irish
Old Persian
Sanskrit
Umbrian
Vedic

AS.

Periodicals :
BSL
Bulletin de la Societe de linguistique de Paris
HA
Handes Amaoreay (Vienna)
IF
lndogermanische Forschungen (Berlin)
JAOS
Journal of the American Oriental Society (New Haven, Conn.)
KZ
Zeitschrift fiir vergleichende Sprachforschung (Gottingen)
[Usually called : Kiihn's Zeitschrift)
Lg
Language (Baltimore)
MSL
Memoires de la Societe de linguistique de Paris
NTS
Norsk Tidskrift for Sprogvidenskap (Oslo)
REArm
Revue des Etudes Armeniennes (Paris)
RicLing
Ricerche Linguistiche (Rome)
TPhS
Transactions of the Philological Society (Oxford)

BIBLIOGRAPHY

The following publications [Link] referred to in the text with the author's [Link], and the [Link],
if [Link] :
A. A. [Link]
1964 [Link] [Link]~ [A Classica.l Armenian Text-book]. 3rd ed., Erevan.
E. Benveniste
1945 Etudes iraniennes. III. Emprunts iraniens en armenien. TPhS, p. 68-78.
1952 La construction pass-ive du verbe traMitif. BSL 48, p. 52-62.
1957-58 Mots d'emprunt iraniens en armenien. BSL 53, p. 55-71.
1959 Sur la pkonitique et la syntaxe de l'armenien classique. BSL 54, p. 46-68.
1964 Elements parthes en armenien. REArm. n.s., vol. 1, p. 1-39.
G. Bolognesi
1951 Sul vocalismo degli imprestiti iranici in armeno. RicLing 2, p. 141-162.
1954 Ricerche sulla fonetica armena. RicLing 3, p . 123-154.
1960 Le fonti dialettali degli imprestiti iranici in armeno. Milano.
F. Feydit
1964 Considerations BUr l'a:tphabet de Saint Mesrop. Wien (Studien zur [Link]
Geschichte, vol. 11).
R. Godel
1965 Les origines de la conjugaison armenienne. REArm, n.s., .vol. 2, p. 21-41.
1970a. Questions de pkonetique et de morpkologie armeniennes. REArm. n.s., vol. 7, p. 1-7.
1970b Diachranic Armenian. Current Trends in Linguistics, vol. 6, Mouton, The Hague,
p. 139-159.
M. Grammont
1918 Notes de phonetique generale. VI. Armenien classique. MSL 20, p. 213-259.
H. Hiibschmann
1897 Armenische Grammatik. I. Teil, Armenische Etymologie. Leipzig. Repr. Darmstadt
1962.
H. Jensen
1959 Altarmenische Grammatik. C. Winter, Heidelberg.
St. E. Mann
1963 Armenian and lndo-European. Histories.I Phonology. Luzac and Co., London.
1968 An Armenian Historical Grammar in Latin Characters. Luzac and Co., London.
A. Meillet
1913 Altarmenisches Elementarbuch. C. Winter, Heidelberg.
1936 Esquisse d'une grammaire camparee de l'armenien classique. 2nd ed., Vienna.
1962 Recherchu BUr la syntaa;e camparee de l'armenien class-ique. Lisboa (Bibi. de la Fondation
C. Gulbenkia.n).
H. Pedersen
1905 Zur aTmenischen Sprachgeschichte. KZ 38, p. 194-240.
1906 Armenisch und die Nachbaraprachen. KZ 39, p. 334-485.

Bibliography

V. Pisani
1950 Studt aulla fonetioo. dell' armeno. RicLing 1, p. 165193.
1951 Studt aulla fonetioo. dell' armeno. RicLing 2, p. 47-74.

H . Vogt
1938
1958
1961
\V. Winter
1955

Armenien et Caucasique du Sud. NTS 9, p. 321-338.

Lu ocdusivea de l'armenien. NTS 18, p. 143-159.
Armenien et georgien. HA 75, p. 531-540.

Problem4 of Armenian Phonology. II. TM. Repreaentation of IE p, t, k (kw) in Annenian.

Lg 21, p. 4-8.
1962 Problema of Annenian Phonology. III. Consonant Cliutera. Lg 38, p. 254-262.

Other references will be found in the foot-notes. A Selected bibliography is appended to my

report on Diachronic Armenian (Godel, 1970b p. 155-159). As it does not include purely
descriptive studies, nor any 'vork on Classical Armenian literature, I think it appropriate to
quote here the following :
V. Inglisian
1963 Die Arnuni&che Literatur
G. R. Solta
Die Armeniache Sprache. Handbuch der Orientalistik. I . Abteil, Band 7. E. J. Brill,
Leiden.
H. Jensen
1964 Altarmeniacht ChreatomaJhie. C. Winter, Heidelberg.
P. Jungmnnn
1964 L'article defini en armenien claasique. REArm n.s., vol. l , p . 47-99.
S. Lyonnet
1933 Le parfait en armenien cla88iqu.e. P aris (Collection linguistique publiee par la Sooiete
de linguistique de Paris).

H. Vogt
1937

Lufonnea nominalu du. verbe armenien. NTS 8, p. 5-70.

After the completion of the manuscript of this Introduction, there appeared a detailed report on
Cla88ical Armenian studies, followed by an alm011t ex.h austive bibliography :
Rudiger Schmitt
Die Erforachung du

Klaaaiach-ANntn~chen

aeit Meillet (1936). Kratyloe XVII, 1972 [1974), p. 1-68.

THE ARMENIAN ALPHABET

Transliteration

[Link]
of the
letters

WJP

p p
q. If
'I- 'I

pbL
If frJ

/;

'l

b
g
d
e
z
e

I!.

t!.P

t'
z

{Jn
df
pfip
LfLb

fa. fd
J. d

,.,
l

hJ

r,

"Ju
L

(;

1.

!,

'l

'l

'fW

/;l.
qw

x
c
k
h

Jut

.lw

l .
l

(;w

*/;fi

f.n
'lWtn

(ayb)
(ben)
(gim)
(da)
(ec')
(za)
(e)
(et')
(t'o)
(ze)
(ini)
(liwn)
(xe)
(ca)
(ken)
(ho)
(ja)
(lat)

[Link] on

c
m

6f
J/;L

J/r

y
n
s

l.
"I

c'
p

l_W

Ii 6
lf d
8 J
t L
(;

n
9

"'

Names
of the
letters

bnL

lw

"{f

2t

nw

uf

c/..

1"-

t
r
c'
w
p'
k'

mpLb
pf
9n
f. pLb
PPL('
.[!f

fl.

s
r
8

""'
~

I'
9
L

'"

.[!

(ce)
(men)
(yi)
(nu)
(sa)
(o)
(c'a)
(pe)
()e)
(ia)
(se)
(vew)
(tiwn)
(re)
(c'o)
(hiwn)
(p'iwr)
(k'e)

(1-1.2]

FIRST PART

A SYNCHRONIC APPROACH

1. A systematic description of an obsolete language like ClArm. has to be made

on the ground of a 'corpus' affording sufficient and homogeneous data about its
phonemic structure, grammar, and basic vocabulary. Such a corpus, of course,
consists of texts only. The earliest Armenian texts have been written down in
the 5th century A.D. , the so-called 'golden century' (oskedar). Many are translations from the Greek, e.g. the Armenian Bible, and possibly P'[Link] Biwzant's
History, or from the Syriac, e.g. St. Ep'rem's Hymns. Others have been composed
or compiled by Armenian clerics : Eznik from Kolb, Koriwn, Lazar from P 'arpi,
and the anonymous author known as Agat'angelos.
1.1. In t his connection, it must be pointed out that IO) all the works of the
oskedarean authors and translators have come down to us in manuscripts from
later centuries. The most ancient one, the Moscow Gospel, was copied in 887;
and the only extant copy of Eznik's treatise against the heresies (Elc alandoc'},
preserved in the famous Library ([Link]) at Erivan, is dated from 1280,
while the original treatise had been written down in the years 441-448. The
reliability of the text and its orthography is therefore not quite above suspicion.
2) The dating of certain important literary works, traditionally referred to
the 5th century, has t urned out to be controversial. This is the case of Eli$e's
narrative of the first religious war, i.e. Vardan Mamikonean's unsuccessful
campaign against the Sassanian monarch Yazgerd II, in 451 A.D. (Vasn Vardanay
ew Hayoc' paterazmin}, and of Movses Xorenac'i's celebrated History of Armenia
(Patmut'iwn Hayoc'). Even those scholars who cling to the traditional dating
pave to admit that the extant versions of both works display interpolations or
other alterations from the following (6th to 9th) centuries.
1.2. In view of this situation, it is advisable not to include in the corpus such
texts as are likely to belong to, or bear the stamp of, later times, and to limit the
field to non controversial oskedarean literature. Even within these limits, a
description of ClArm. is not very easy; but it would become more complicated
were we to take into account the literature of the subsequent period. Besides, a
corpus covering several centuries would not suit the purpose of a synchronic
approach.

The Alphabet

(l.2-1.311

One must not be deceived by the long survival of ClArm. as a literary language
(grabar) : no literary language, however carefully cultivated and handed down, is
safe from gradual alteration. From the early Middle Age down to the beginning
of the 19th century, Armenian writers and poets, whatever dialect they spoke
in everyday life, kept on using the old language, and trying to maintain its
purity. To what extent they actually succeeded is a difficult question to answer.
Doubtless, those writers, as learned men and clerics, would mostly remain aloof
from the developments of the popular language, and avoid 'vulgar' words,
inflections, or idioms. But they would also yield to the temptation of improving
the grabar by artificial embellishments. This trend is perceptible in the so-called
Siwni school (6th to 8th century), where Armenian literates were trained to use
every kind of Greek idioms, even against the rules of Armenian syntax and
inflection. On the other hand, toward the same time, traces of 'vulgarity' appear
in the Armenian version of Zenob Glak's History of Tarawn, by Yovhan
Mamikonean (7th century).
1.3.

THE ALPHABET

The Armenian alphabet, like the Gothic and the Cyrillic (OCSl.), was created for
religious and cultural purposes. Christian preaching had reaclled Armenia toward the
end of the 3rd century, starting from the most active centers. in the neighbouring
countries: Caesarea in cappadocia1 and Edessa (Urfa) in Mesopotamia. Christianity soon became the state religion .of the Armenian kingdom (301 ). Later, in
king .P ap's days (367-374), the Armenian church released itself from the authority
of Caesarea's archbishop. Yet, many, years still elapsed before the Armenian
faithful could have the Gospel preached in their own language : Armenian clerics
had to learn Greek and Syriac, as well as make oral translations and comments.
1.31. The alphabet was c.r eated in 406 or 407 by a learned cleric, Mesrop
Maatoc (Koriwri and Lazar [Link] the latter name only). The translation of
the Bible ensued; Mesrop started with Solomon's Proverbs. It appears that
Mesrop had to invent an alphabet because Armenian h~d never been written
before that time. At least, there is no positive evidence to the contrary. Armenian
historians never refer to documents or chronicles written in the national language,
prior to the Mesropian period. They would hardly have ignored them, had there
been any. What they sometimes quote, besides foreign historical works, are
fragments of old native poetry. Movses has t\velve extracts from the 'Songs of
Golt'n' (Golt'an ergk'), mostly pertaining to the deeds and adventures of Armenian
kings or heroes of the past (180 B.C. to 72 _A.D.). These old epic lines are very
fine, and [Link] like to have more of them. The poets were very likely wandering

The Alphabet

[I.31-1.4)

rhapsodes (the profession has continued up to recent times), performing their

songs not only at the kings' and lords' courts, but also among the country folk.
All people enjoyed them, and in spite of the clergy's opposition (see P'awstos,
III. 13), these poetic memories of pagan times remained alive for centuries. But
the fragments quoted by Movses do not display archaic word forms, as would
probably be the case if they had been preserved through a written tradition.
Apparently, the stylistic features of the poems did not alter; but the language
is no longer the original one : it is Classical Armenian, very much like the
historians's own language.
1.32. The historical significance of Mesrop's "discovery of the letters" must
therefore not be underrated. Did he really invent his alphabet 1 1 The shapes of
the 36 letters [Link] rather unlike any known characters. But in spite of several
attempts to connect the Mesropian alphabet with other Eastern systems of
writing, the Greek alphabet can still be safely mentioned as a model, in view
of the succession of the letters :
Gk.
Arm.

[Link]{TJ 8 '
K
~ . v~o 1T pa ,,.
vcf>x(ifiw)
w p 'l 'I /; 11. f I!. p cf fo L /u b ~ {, a 11. 6 d J & l n l "I 2 fl u t/. Ill I' !J L "' .[!

and of the use of nL ( = Gk. ov) for /u/ (see below, 1.643). Furthermore, the shapes
of certain letters seem derived from a variety of cursive Greek :

P<

p; ~ < {}

etc.

Anyway, the Greek alphabet did not supply all the characters needed, al'.!d
Mesrop would have had to supplement it either by borrowing letters from some
other system, or by coining new signs. Coining letters, after all, is easier than
setting down the phonemic system of one's own language; and since he perfectly
succeeded in ~oing this, he may be credited with the invention of the 'Armenian
letters' (/!. <f L fu b l. a 6 J l 2 t/. p g).
1.4.

TRANSLITERATION OR TRANSCRIPTION 1

Transliteration means substituting Latin letters (with the addition of diacritic

marks, or exceptionally, of Greek characters) for the original signs of written
texts, regardless of whether or not the spelling faithfully reflects the 'pronunciation', i.e. the phonological shape of the language. It simply aims at showing
how a language is written, not how it exactly sounds. This device is the only
On a similar problem concerning the origin of the Avestio alphabet, see W. B. Henning's study
in Handbuch der Orientalistik I. 4/1 (1958}, p. 52.
2 Feydit, p. 31-36.
1

The Alphabet

[l.4-1.5)

scientific one in the case of defective or awkward writing systems, such as the
Middle Iranian alphabets and spellings.
Transcription, on the contrary, is devised so as to represent the oral word forms,
i.e. to give a phonological picture of a language. Phonetic alphabets have been
created to this very purpose, though they are also used for transliterating.
Transliteration, indeed, however tentative, is never totally haphazard : e.g. if
there are reasons to believe that a certain letter represents a dental stop (t, or
d, or th), it will not be transliterated by p ors. Consequently, a transliteration will
sometimes approximate a transcription, insofar as the phonological values of the
written signs can be conjectured with some certainty.
1.41. Regarding ClArm., the former method has proved more adequate. The
only scholar who deliberately departed from the usual practice is Stuart E. Mann :
his system of transcription reflects the phonological situation in the 11th century
(Mann, 1963; Preface, p. ii). He devised it with the purpose of sparing his readers
the difficulties that arise from the conflict between the 5th century spelling and
the traditional pronunciation. From a philological point of view, however, this
device is objectionable.
1.42. The current system of transliteration is mainly H iibschmann's, into
which Meillet's improvements have been incorporated (Meillet, 1936, p. 13). A
slight modification (c' and c' instead of y and 9) appears in the new series of the
REArm s. The following text sample has been transliterated accordingly. The
only particular for which I am responsible is the hyphen dividing prepositions
and articles from the words to which they are attached: y-anapat (1wuw"lwm)
'to the desert'; hogu.1ov-n (4nq1.11ifu) 'by the Spirit', etc. In the manuscripts, the
name of God, AstU<UJ, is written llb and the GDAbl. form, written llj , is usually
restated as Astucoy, according to the traditional reading. But word internal -[Link] not drop before the Middle Armenian period (12th to 14th century), so that
the classical form was doubtlessly AstU<UJoy (llumntwbnJ).
1.5. TEXT

SAMPLE :

the Temptation of Jesus (Luke, 4.1-13)

Ct 8fountu Lfo {nqtnt{ uppnt{ '[Link] fo (Jnp'fwuwuf b. t{wpfp {nqtnt{u JU1UU1U[U1U1 w1.11tpu
I Ew Yisus li hogwov srbov darjaw i Yordanane ew varer hogwov-n y-anapat (2) awurs

.f!WnwunLU pnpJ./;wL fo Uwmwuw1f b. nl liqlip b. nl wpp JWLntpuu JW]Unufof( b. fo qwmwp/;Lf,

k'arasun p'orjeal i Satanaye, ew oo' eker ew oCS' arb y-awurs-n y-aynosik, ew i katarel-n

ungw J!W'l!J/;WL : Ct wuf gfiw UwmwUWJ

cfJ/;

np'f fo /;u llumntwbn1

UIUUI

.pwpfo'f WJ'f"folf

noc'a k'alc'eaw. 3 Ew ase c'-na [Link] : Et'e ordi es Astuacoy, asa k'ari-d aydmik
s Jensen (1959, 31) already pointed out that 9 and l being actually aspirated b and
be more consistently transliterated as c' and c'. Nevertheless, he stuck to the usual practice.

6, would

The Alphabet

( l.5-1.61]

'l./' l.w9 L/19/i: fTJw111wu/uwfiJi h111 fi,Jw 8/1untu b. wuf. <[Link] f h{Jh n~ l.w9Jit Jf1w1fi 4t99f
zi hac' lic'i. 4 Patasxani et nma Yisus ew [Link] : Gree.I ii et'e oo' hao'iw miayn keo'c'ii

i5wl''I WJL wJ/;fiw1fi pwfiJi1. U.u111n1.U1l>n1 : b1. {.wfihwL '[Link]

/1

J_hwnfi Jfi pwpap h9n19 &i5w

mard, ayl amenayn baniw Astuacoy. 5 Ew haneal z-oa i learn mi barjr, ec'oyc' nma

'[Link];fiw1fi [Link] w2Juwpl./1 Ji 1[w1p4hwfi dwow&w4 /1

: b1.

wuf 9fiw Uwu1wfiw1-

[Link] t'[Link] [Link] i vayrkea.n ![Link]. 6 Ew ase c'-na. [Link] :

~/;'I. 111w9 '[Link] wJ/;&w1fi Jiifuw&nt{JJit& b. 'lifrwnu un9w, 'I./' Jifia 111nthwL
K 'ez ta.c' z-ays amenayo ilixanut'iwo ew z-p'ars soc'a, zi ioj tueal ii, ew um kamim

f b. nto l1wJfio

111wi5 '[Link] : U.p'I 'lnL bfJb wfi4bwL bp4 fip "fW'fwfiJi9bu wnw2/1 Jio, l!b'l. bqfig/i wJ/;fiw1fi :
tam z-sa. 7 Ard du et'c ankeal erkir paganic'es [Link] im, k'ez elic' i [Link].

fTJw111wuJuwfiJi b111 fiow 8Jiunw b. wuf. bpfJ jbwu Jio, Uw111w&w1. 'lfi 'll'bwL

f , hp4 fip

8 Patasxani et oma Yisus ew asii : Ert' y-ets im, Satanay, zi greal ii : Erkir
"{Ullf 9/;u

wbwnfi U.u111ntwl>n1 .j!nLO b. '[Link] Jfiw1fi "f Wl10bu9bu : b1. wl> '[Link] [Link]

pago'es teafn Astuaooy k'um ew z-na. miayn pateso'es. 9 Ew ac z-na. y-Erusaliim

b. 4wgn19 Ji i[bpwj w210wpw4 Ji 10w6wp/1fi, b. wuf 9fiw, bfJb np'lfi bu [Link]>nj wp4 'l../!b'l.
ow kac'oyc' i veray [Link] tallari-n, ew ase c'-na.: Et'e ordi es Astuacoy, ark z-k'ez

wuwp Ji i[wJI' 'l./1 'll'bwL

f bfJb l.pb2ww4w9 Jitpn9 "{U/[Link] f i[wufi .j!n "lw{.bL 'll!b'l.

asti i vayr, (10) zi greal ii et'e hre8takac' iwroc' patuireal ii vasn k'o [Link] z-k'ez,

'lfi Ji if.bpw1 atnw9 pwpa9b& 'l.l!t'l. .

JJi

bppb.j! l.wp9bu 'l.f!wpfi 'lnwri l!n : fTJw111wu/uwriJi t111

(11) zi i veray jerac' barjc'en z-k'ez, mi erbek' harc'es z-k'ari z-otn k'o. 12 Patasxani et

fio1u 8Jiun1.u b. wuf. U.uw9bwL

f fJb

n~

ifrnpabu9bu '[Link] [Link]> .j!n : b1. 4wwwpbwL

oma. Yisus ew ase : Asac'ea.I ii t'e oc' p'orjeso'es z-tiir Astuac k'o. 13 Ew katareal

'[Link]/;fiw1fi [Link] Uw111wfiw1Ji' Ji pw9 t4w9 Ji fiowfif wn dwi5wfiw4 1/fi :

z-amenayo p'orjut'iwns Satanayi, i bac' ekao' i nmaoii af Z&manak mi.

1.6.

COJ\iMENTS

c = ts (in Engl. Tsar); j = dz (in adze); x =Germ. ch (in Bach,

doch, etc.). Diacritic marks are used for hushing consonants: 8 = Engl. sh;
= Engl. z in azure;
= Engl. ch; f = Engl. j (joy), and for aspirates :
CONSONANTS:

p' t' k' c'

c'.

On land i, see below (1.62).

1.61. Data. [Link] the phonological values of Armenian consonants are afforded
by loan words from Middle Iranian, Syriac, or Greek :
"lwLww palat < Gk. [Link] ( < Lat. palatium)
4w1up kaysr < Gk. Kaisar (Lat. Caesar)
l>fil>'lwl cnclay < Syr. t?it?la, f?et?lii ( = f?et?l?')la) 'cymbal'
62owpfi111 csmarit 'true' < Mlr. casm~it 'visible, obvious'
pw11Jifi bagin < Mlr. bagin 'altar'
'lwl' dar 'slope' < Mlr. dar 'valley'
6pw'f crag 'lamp' < Mlr. cirag
ifrwn(.j!) p'ar(k') 'glory' < l'fIIr. *farr ( < OP. farnah- 'glory')

The Alphabet

[1.61 -1.631J

ifrwufrwu p'asian < Gk. [Link].n6s 'pheasant'

fJbJ t'em 'diocese' < Gk. thema
J/;.f!buw1 mek'ena.y < Gk. mekhane 'engine'
Notice the occurrence of aspirates before s :
ifrufrw{J p'sia.t' < Gk. psfathos 'mat'
J/;mw.[!u meta.k's < Gk. [Link]. 'raw silk'
just a.s in ./!uwu k'san '20' (see below, 2.31).
1.62. The transliteration of 'las a kind of l (l) is by no means arbitrary. Though
it is traditionally pronounced as a. fricative (the voiced counterpart of x), it
certainly represented in early ClArm. (5th to 8th century) a dark l, not unlike
the American l in 'life, [Link]', etc., while L was used for the clear l, as in French
'lac, bal'. Evidence for this phonological value is not wanting : beside the position
of 'l in the alphabet, there are loan words, e.g. mfrmqnu titlos < Gk. titlos ( < Lat.
titulus 'inscription, title'); 'f/111qnu Pawlos < Gk. Paalos, etc., and etymological
data, e.g. :
wq a.l 'salt' : Gk. hal-s; Lat. sal
bqu eln 'hind' : OCSl. jeleni
qbqfru delin 'yellow' and qwlwp dalar 'fresh, green' are derived from the same
root (Cf. Gk. th<illo 'I bloom', thaler6s 'blooming, sturdy').
ClArm. had also two kinds of r: n (r) and p (r). The difference can be approximately figured out by contrasting the Italian strongly trilled r with the American
intervocalic r (in 'berry, tomorrow') 4
1.63.

VOWELS

The transliteration of f by e is somewhat misleading : it seems to point to a

long vowel, while ClArm. presumably had no contrast of long vs. short. But it
reflects the diachronic background : f, as a development of a former diphthong
(ei/ey : see below 2.211), must have once been a long vowel. Furthermore, in
vowel alternation, it does not behave like either a, e, o or i, u (2.224). In Mesrop's
days, it probably stood in contrast to /; as a close e to an open one; but about
the 10th century, e and e eventually merged, except in word initial (1.632).
1.631. A similar development took place later on. In the early 10th century,
the aw 'diphthong' (1.641), before a radical consonant, changed to a long, open
4

See G. Bolognesi, Sul valore di armeno f, RioLlng 5 (1962), p. 141-145 (f = rr).

I 1.631-1.641)

The Alphabet

o, entailing the later addition of a new character to the alphabet : since the 12th
century, !.wtm 'herd', wtp 'day', f[w tlflf 'shoe' are written !,om, op, f[olfaf[. In Modern
Arm., the new vowel has merged with the original o (n), except in word initial.
1.632. In the course of the post-classical period, initial b developed an on-glide
y- (e- >ye-), but f did not. Similarly, n developed an on-glide w- (o- > wo-), but
o did not. The resulting situation in ModArm. can be illustrated as follows :
ba.p /yemk / 'we are' : f.p /ek' / 'you are'
np /vor / 'who' (rel. pronoun) op /or/ 'day'

\Vord internally b = f;

= o:

ubp /ser / 'progeny' : ufp /ser / 'love'

!,nm /hot/ 'odour' : !.om /hot/ 'herd'.

1.64. q, L and J never occur between consonants: consequently, they do not

represent vowels. The same preposition occurs as fa i before consonants, as J
before vowels :
fa qbpwJ i veray 'on top of'; fa LbwnL i [Link] 'toward the mountain'
1wLw"lwm y-anapat 'into the desert'; 1wtntpu y-awurs 'in the days'.

As to 1 and t, they stand in complementary distribution, the former occurring

in word initial (fa qwJp i vayr 'down'; qwuL .pn vasn k'o 'for thy sake' ; fa qbpw1
i veray, etc.) and after n (!.11qtnq hogwov 'by the Spirit'; q11qba govem 'I praise',
etc.); the latter, in every other environment. It thus appears that t , in spite of
its position in the alphabet, is not equivalent to Gk. v; otherwise, it would have
become superfluous, since Gk. /ii/ (v) had long ago merged with /i/ (,). Mesrop
seems to have taken it over with regard to the quasi diphthongs av, Ev, 11v, which
probably sounded approximately like Arm. wt aw, bt ew, pt iw before consonants.
1.641. All three letters occur either between vowels (!.w1faJ hayim ' I look at';
b'lbtpL elewin 'fir-tree'; !.11qwLp hovani 'shadow, shelter'), or after a vowel in word
(or syllable) final position (qlfun1 dsxoy 'queen'; wpb. arew 'sun'; q11pnq korov 'vigor';
w1q/1 aygi 'vineyard'; lfnqufu Movses ' Moses'), i.e. in such positions where any
consonant is allowed to occur. But 1 is never found between a vowel and an
implosive consonant 6 , while both L and J frequently are : w1q ayg 'dawn'; w1u
ays 'this'; qwjpqbwL vayrkean 'moment' ; L"Ju loys 'light' ; qn1L goyn 'colour'; Jwtm
mawt 'near'; wtcl awj 'snake'; wLbtp akun 'flour'; upt.L siw-n 'column', etc. Very

Except in some plural forms, on which see below (2.323).

The [Link]

(l.641-1.643]

likely, ay, oy, aw, ew, iw in this last case did not sound quite the same as in other.
environments. Armenian grammarians term them 'diphthongs'.
1.642. According to the transliteration principle, each Armenian letter has
to be substituted by one Latin letter, e.g. 1by v, L by w, regardless of the phonemic
problems involved. Htibschmann's system is therefore objectionable. He steadily
uses v for 1, but transliterates L by v or u, J by y or i, in turns, e.g. :
fd/1L t'iv 'number, figure'; fdpL& t'ivn

'the number'; &mL nav 'ship'; wp.f!mJ

ark'ay 'king' - but :

WLf aUr 'day'; wp/1W ariUn 'blood'; [Link] jain 'sound, VOiCe', etc.
thus anticipating the solution of t'vo problems : 1) Do 1 and L represent the
same phoneme 1 2) Did ClArm. have diphthongs 1 And if this was the case, how
would they have to be defined in phonemic terms? Meillet's transliteration is
more consistent: he uses v for 1, w for L, and y for J in all environments (except
for the particular case of nL : see below), and writes, accordingly : t'iw; t'iwn;
naw; ark'ay; awr; ariwn; jayn, etc.
1.643. Doubtlessly, n1 and nL do not reflect the same phonological reality:
ov and ow would of course serve the purpose, whatever value is assigned to the
latter. But this value can be safely stated : Mesrop simply borrowed from the
Greek the pseudo-diphthong ov, which had been for a long time, throughout the
hellenized world, the normal sign for u. The decisive proof lies in the fact that,
in vowel alternation, nL (u) behaves like p (i) (2.221), which would not be the
case if it had been a diphthong. Thus, the transliteration by u, though not quite
consistent with the principle, does not raise serious objections.

'

I would write: t'i11J-n (see above, 1.42 and below, 2.324).

(2-2.11]

2.

Phonology

Phonology.

The accuracy of Mesrop's alphabet has already been emphasized (1.32). It

certainly provides a sound basis for a classification of ClArm. phonemes.
Diphthongs as defined by the Armenian grammarians (1.641) need not be
included in the description : wL'I /awd/ 'air' displays the same phonological
pattern a8 W[''/ /ard/ 'now', WU'/ /and/ 'there'; qfoLlll /giwt/ 'diSCOVery' Can be
paralleled with p/Ip111 /birt/ 'coarse'; w1p /ayb / 'the letter w with w'lp /alb/ 'dirt', etc.
2.1.

TABLE OF

ClArm.

PHONEMES

Consonants
Stops:

/pf

"{

/t/

Ill

fkl
Affricates : /c/

/c/
Fricatives : /s/

*"'
6

/p'/

if

ft' I {J
/k' I ./!
/c'/ !/
/c'/ l

/sf l
/x/ Ju
/hi !.
Resonants:

/bf

/d/
/g/

'I

/z/

11.

/z/

cl

/1/

L
l'

Nasals: /m/ d
/n/ u

'l

/j/ a
/j/ 2

/r/

/l/

ft/

'l
n

{/v/ if
/w/ L
/y/ 1

Vowels
/a /
/e/

/of

/;

/e/
Ji/

fo

/u/

nL

(a [!)

2.11. As a result of recent studies (H. Vogt, 1958; Benveniste, 1959, p. 46-56),
the stops and affricates in the 3rd column prove to have been voiced aspirates
(b' d' g' etc.), as they still are in some modem Eastern dialects 1. Yet the usual
7

Thia accounts for their divergent developments in Eastern and Western dialects (Bologneai,
1960, p. 20-21).

10

Phonology

[2.11-2.211]

transliteration (b d g etc.) can be maintained without inconvenience, since [Link],m.

had no contrast of simple voiced vs. voiced aspirate.
2.12. From certain obsolete spellings like lfWJ'[, fil."l'l (fi2"'l) beside 9w1L 'wolf',
fil."JL 'ray of light', etc. it appears that l and l have formerly been phoneme

variants. They still partially stand in complementary distribution, l occurring in

word initial (LwJ 'I cry', Lhwnfi 'mountain', LuhiJ 'I hear', etc.) and after J, and l
before consonants (w'lp 'dirt', "'l2"Jfi 'greeting', .pw'[ghwL 'he felt hungry', etc.).
In other positions (intervocalic or final), however, they behave as phonemes :
wLf'.f! /alik' / 'waves, billows'
mwl /tal/ 'husband's sister'
'l"L /golf 'heat'

Similarly r and

: w'lfr.f! /alik' / 'bowels'

: mw'l /tal/ 'verse, [Link]'
: 'l"'l /golf 'thief', etc.

r are phonemes, although such minimal pairs as

i[wphJ /varem/ 'I lead'

wpwm /arat/ 'stain'

: i[wnhiJ /varem/ 'I burn'

: w11.1um /arat/ 'plentiful'

are not numerous. Before radical n, the opposition is mostly suppressed (2.231).
As to v and w, they have to be regarded rather as phoneme alternants (cf. 1.64,
and Pisani, 1950, p. 186).
SYLLABLE STRUCTURE. The usual patterns are (C)V, (C)VC: Ew Yi-sus
li Jwg-wov ... dar-jaw i Yor-da-na-ne ew va-rer ... ya-na-pat ... k'a-ra-sun p'or-jeal
i Sa-ta-na-ye ew oc' e-ker ... Ew a-se ... Sa-ta-nay: E-t'e or-di es As-tu-a-coy a-sa
k'a-rid etc.
However, (C)VCC too has to be reckoned with, in view of aydmik, vayrkean,
barjc'en. On the problems involved, see below 2.342.
2.2.

2.21 Hiatus, as a rule, is allowed in morpheme juncture, except after a or o,

which require the insertion of y:

fuw'lwJfr /xalayi/ 'I moved, walked', impf. of fuw'lwiJ, as against
[Link] /berei/ 'I brought' ; w11.fit1Lp /arnui/ 'I took', impf. of [Link], w11.fit1LiJ;
[Link]{JfLU /ark'ayut'iwn/ 'kingdom', from wp,pw] 'king'
'l"J"L{JpLfi /goyut'iwn/ 'being, existence', from 'f"J 'being, extant', as against
Lf"L{JpLfi /liut'iwn/ 'fullness', from LP 'full'
Jwpqwpht1L{JpLL /margareut'iwn/ 'prophecy', from Jwp'fwpf 'prophet'.

Conversely, there is no instance of y after e, i or u.

2.211.

In the most ancient manuscripts, e.g. the Moscow Gospel,

is

[2.211-2.213)

Phonology

11

substituted by /; when immediately followed by a vo,vel : Jwpqwpf 'prophet',

GDL. Jwpq1upbfo, I. JwplfwpbpL; derivates: JwplfwpbwfiwJ'I prophetise', JwplfwpbnLfJpLfi
'prophecy'. Similarly in the e- and i- conjugations, the imperfect ends in -bfo
2nd sg. -bfop 1st. pl. -bw.p, etc. (but 3rd sg. -tp). The paralleling of these endings
'vith those of the a- conjugation at once reveals the origin of the e/e alternation:
pbpbfo /berei/
pbpb/1p /bereir /
pbpfp /berer /
pbpbwl! /bereak' /

fuwqw1fo /xalayi/
fuwqw1fop /xalayir/
fuwqw1p /xalayr/
fuwqw1wl! /xalayak'/ etc.

PA. *ey yielded e before consonants (pbptp < *bereyr) and in word final while
-y- dropped between e and a vowel (pbpbfo < *bere(y)i). This explanation holds
good for all similar cases, e.g. Jwpqwpt < *margarey; JwpqwpbwfiwJ < *margare(y)anam.
But as soon as the 10th century t instead of /; appears in inscriptions. The
new spelling (pbptfo-JwplfwpbwfiwJ), involving the merger of e and e (1.63), has
become traditional. It must be noted that the former one was partially ambiguous :
in pbpbwJ!, JwplfwpfwfiwJ, ea does not make one syllable only as it does in other
instances (2.213).
2.212.

Hiatus occurs after u without qualification :

finL 'bride', GDAbl. finLllJ /nuoy/

Lb'lnL 'tongue', GDL. Lb'lnLp /lezui/ ; I . Lb'lnLwL /lezuaw/
burnL 'J gave' (aor.), 1st pl. urnLWJ! /tuak'/ 'we gave'; part. urnWW[ /tueal/ 'given'
wnfinLp /atnui/ ' I took' (impf.), 1st pl. wnfinLWJ! /atnuak' /, etc.,

but after i only when the radical morpheme (root or stem) is monosyllabic :
'I fo-bJ /diem/ 'I suckle' ; ( fo-wfiwJ /hiana.m/ 'I wonder, admire'
[Link] 'horse', GDAbl. J.fon1 /jioy/ ; [Link] /jiawor/ 'rider'
Jfi 'one', G. Jfrn1 /mioy/ or J/in2 /mioj/; D. J/inLd /mium/;
JfrwfiwJ /mianam/ 'I unite'; Jfrw1L /miayn/ 'only'; p Jfrwupfi /i miasin/ 'together';
Jfrw&a& /mianjn/ 'monk'.

2.213. The addition of a morpheme beginning with a to a polysyllable in -i

results in a compound phoneme /ea/, which counts as one syllable only, and can
therefore be rightly termed a diphthong :
"l''IP /ordi/ 'son'

>

np11bw~

+ -w~ /ak/ (a diminutive suffix)

/ordeak/ 'son' (with a connotation of endearment);

12

Phonology

(2.213-2.221)

mh'lfi /teli/ 'place'

-wL (I. ending)
> mh'lhwL /teleaw / (cf. L/;'lnL 'tongue', I . L/;'lnLwL).

The aorists .pw'l!/hw1 fk'alc'eay / 'I became hungry', p111fuhw1 /p'[Link].y/ 'I fled',
etc. [Link] derived from i-stems (k'ak'i-, p'axi-), and can be paralleled with the
u-stem aorists LnL -1111 /luay / 'I heard', hpqnL-111) /erduay / ' I swore'.
The peculiar nature of /ea/ as a contraction of i + a is proved by the aor.
3rd. sg. form h4hwg /ekeac'/, from 4hwa /kea.m/ 'I live' ; for, thee- prefix (augment)
is added to such aorist forms only as would otherwise be monosyllables, e.g.
h111nL /etu/ ' I gave'; h4hp /eker/ 'he ate'; h9n19 /ec'oyc'/ 'he showed', etc. (3.233).
NOTB.

Sfp /ter/ 'lord', GDL. mh111nL /tAlain/, is admittedly derived from 111jf /&yr/ 'man', and ha.a

been traced ba-0k to a PA. compound ti-ayr 'house man, landlord' (Pisani, 1951, p. 64-65). The
contrast with J/i1111fi /miayn/ 'only' might be explained on chronological grounds : 111fp probably
belongs to an earlier lexical layer than Jf1111fi. But the parallel word mfi* {ifi /tikin/ 'lady' does not
bear evidence to PA. ti-, because i should have dropped (see below, 2.221). Only teiJ- \Vould account
fo1 both urfp < te(y)-ayr and ur/14 {ifi < *tekin < tey -kin (2.222). See Godel, 1970a, p. 2-4.

2.22. REGULAR VOWEL ALTERNATION. Vowel alternation, as a rule, is the

result of diachronic developments. At an early stage of PA., the word stress
settled upon the last syllable but one, entailing firstly t he dropping of final vowels,
t ogether with the following consonants (except n, r and l) 8 Later on, certain
vowels and diphthongs in unstressed (non final) syllables were lost or reduced.
The effect of the latter change can be formulated in synchronic terms, as rules
of vowel alternation.
2.221.

Final i or u : non final (a)

qfrp /gir/ 'letter': GDAbl. 'lPnl /groy/; 'fp/;J /grem/ 'I write'
hp4 {ifi.p /erkink'/ 'heaven' : GDAbl. hp4fi{i9 /erknic'/
fr'la /ilj / 'longing, wish' : GDL. l!.'lafr /alji/ ; l!.'lawa /alja.m/ ' I wish'
/;Lfr.P /elik' / 'he left' ([Link].): 1st sg. L.f!P (lk'i/, 2nd sg. L.f!hp /lk'er/, etc.
L.f!P!J /lk'ic/ 'I shall leave' ([Link]. subj.) : 2nd sg. L.f!!J/;u /lk'c'es/, 3rd sg. L.f!!Jf

/lk'c'e/, etc.
.pnLfi /k'un/ 'sleep' : GDAbl. .pfin1 /k'noy /
wfiwunLfi /[Link]/ 'animal' : GDAbl. 111fi111ufin1 /anasnoy /
2nLp /Jur / 'water' : GD Abl. 2Pnl /jroy / ; 2Pw'lw!J / [Link].c' / 'watermill'
/;JnLur /emut/ 'he entered' ([Link].) : 1st sg. Jurfr /mti/, 2nd sg. Jurhp /mter/, etc.
A similar change took place much later in a few words and phrases, a.a the consequence of an
emphatic stress on the first syllable: 1111up /aysr/ < ayaor; 1111ua /[Link]/ < ayaum (G. and DL. of
111JU ' this') ; WJUd /ayzm/ < ay8 lam '(at) this hour, now'; PPP /ibr/ 'a.a' < . , bar 'which way, how? '
9 'Final' and 'non final' refer to the syllable, not to the vowel only.
8

(2.221-2.223)

13

Phonology

VO words with initial i or u display no alternation :

foci /iz/ 'viper' : GDL. fr<lfr /izi/

nLL /ul/ 'kid' : GD Abl. nLL''l /uloy /, etc.
Final -iw reduces to -u- :
"lwmpL /patiw/ 'honour': GDAbl. "(tumnw1 /patuoy/; "lwmnuu*wu /patuakan/
'honorable, valuable'

2.222.

Final e : non final i

Final oy : non final u

atq /meg/ 'mist, fog' : GDL. Jfq fr /migi/ ; Jf11waw<> /migamac/ 'misty'
wLpfu./! /awrenk'/ 'law, rule' : GDAbl. wLppfiw9 /awrinac'/
t2 /ej / 'he came down' (aor.) : 1st sg. fr2/r /iji/, 2nd sg. fr2bp /iJer/, etc.
Ln1u /loys/ 'light' : GDAbl. LnLun1 /lusoy/; LnLuwwp /lusawor/ 'luminous'
*w9n19 /kac'oyc'/ 'he set, placed' (aor.): 1st sg. *w9nL9p /kac'uc'i/, 2nd sg.
*w9nL9bp /kac'uc'er/, etc.

2.223.

Final ea : non final e

awmbwu /matean/ 'book': GDL. awmbufr /mateni/ ; awmbuwqfrp /matenagir/

'writer'
LbwnL /learn/ 'mountain' : GDL. L/;l'fru /lerin/ (Cf. 2.231)
ubwL /seaw / 'black' : GD Abl. ubL11J /sewoy /
wmbw9 /ateac'/ 'he hated' (aor.): lat sg. wmb9fr /atec'i/, 2nd sg. wmb9bp
/atec'er/, etc.

Consequently, eay in non final syllably was to yield *ey, which regularly developed
to e before vowels, according to 2.2,:I, and to i (through e) before consonants,
according to 2.222 :
2pbw1 /Hreay/ 'Jew': GDL. 2pbfr /Hrei/ (from *Hreyi); 2pbwumwu /Hreastan/
'[Link]'.

From wmbwa /ateam/ ' I hate', the impf. is regularly

1st sg. wmbfr /atei/ (from *at,eyi < at,eayi. Cf. fuwqw1fr)
2nd wmbfrp /ateir/
3rd wmbw1p /ateayr/ (2 syllables)
lat pl. wmbw./! /ateak'/ (3 syllables!), etc.
and the pres. subj.
1st sg. wmfr9ba /atic'em/ (from *at,eyc'em
2nd sg. wmp9bu /atic'es/, etc. 10.

< at,eayc'em.

Cf. fuw1iw19ba)
.

The pres. subjunctive of */;wJ 'I live' : */;9/;J, */;9/;u

opinion (1936, p. 21).
10

... is therefore irregular, against Meillet'e

14

Phonology

(2.223-2.231]

On the later spellings lpf Ji, lpf111uur111L, 1111nfp, see above (2.211).
2.224. It thus appears that /e/, /oy/ and /ea/ make up a peculiar set of
phonemic entities, occurring in final (stressed) syllables only, and standing in
contrast to the 'stable' vowels a, e, o on the one hand, and to the 'unstable' ones
(i, u} on the other.
2.23. While the rules of vowel alternation spread all over the morphological
system of [Link]., with scarce exceptions, only two instances of CONSONANT
ALTERNATION have to be mentioned.
2.231. f (before n) : r (in all other environments)
This alternation is found in the paradigm of a few irregular verbs, e.g. :
111nLbiJ /arnem/ 'I do, make' : aor. wpwpfr /arari/
rpunLwiJ /darna.m/ 'I turn, return' : aor. 1wpJ.1111 /[Link] /
[Link]'JbnLnw /ant'ernum/ 'I read' : aor. euf'Jbp91111 /ant'erc'a.y/
and in the inflection of several nouns with variable stems (3.14) :
LbwnL /learn/ 'mountain': GDL. Lbl'fru /lerin/, [Link]. Lbl'fru~ flerink'/
'lnLnfi /durn/ 'door' : GDL. 'Jpwu /dran/, [Link]. 'll'n1..u~ /drunk'/
'lnL"'"lwu /[Link].n/ 'door-keeper, porter'
wnu /arn/ GDL. of "'11' /[Link]/ 'man', [Link]. wp~ /ark'/, etc.
~ut

it is no longer predictable, because in other instances f has been propagated

through the whole paradigm, e.g. :
111nLnLd /[Link]/ 'I take' : aor. wnfr /ari/
2bnunw /Jei:num/ 'I warm (myself)' : aor. 2bnw1 /Jera.y /
pbnL /bern/ 'burden' (Cf. pbpbiJ /berem/ 'I bring'): GDL. pbnfru /berin/, [Link].
pbnfru~ /befink' /
Likewise : pnLnL /burn} 'fist' (GDL. pnfru /brin/); 'lwnL /[Link]/ 'lamb', 11wnu /darn/
'bitter', abnu /jern/ 'hand', etc.
Conversely, r is preserved before n in the oblique cases and derivates of nouns
or adjectives ending in -in or -un :
?l'nw /[Link]/ 'spring': GDAbl. 'l'lll'uu'Jun1 /[Link]/ (from a. derived
adjective qmpfiw1Jiu /[Link]/)
t{bpfiu /verin/ 'upper, supreme' : GD Abl. t{bpun1 /vernoy /, etc.

[2.2322.3)

2.232.

15

Phonology

c' (before a vowel) : s (before c',

f)

This alternation is peculiar to the subjunctive of -c'- aorists, provided the stem
is polysyllabic :
"lw2111h9fi /pastec'i/, aor. of "lw2111hJ /pastern/ 'I worship' :

Subj. 1st sg. "lw2111h9fi9 /pastec'ic' / 'I shall worship'

2nd sg. "lw2111hu9hu /pastesc'es/ (from *pastec'c'es < *pastec'w'es)
3rd sg. "lw2111hu9f /pastesc' e/, etc.
.
Jnnw9w1 /mofac'ay/, aor. of Jnnw&wJ /mofanam/ 'I forget' :
Subj. 1st sg. Jnnw9w19 /m?fac'ayc' / 'I shall forget'
2nd sg. Jnnwugfiu /mofasc'is/, etc.
On ftw9 pg /ka.c'ic' / 'I shall stand', 2nd sg. "w99hu (stem qw9- /ka.c' /) see below (2.341).
8C' is [Link].1 in wpwughu /[Link]'es/ ... , hfthu9hu ([Link] with hfth9hu) /ekeso'es/ ... , from
wpwpp9 /[Link]'/ 'I shall do', lift p!J /ekio'/ 'I shall come'.
NOTE.

2.3.

SYLLABLE STRUCTURE AGAIN: THE NEUTRAL VOWEL (<1)

The letter/!.. is written in word initial only, before m, n, or l

+ C:

'

e.J"IJ;J /ampem/ 'I drink', e.&111frp /antir/ 'select', e.&fthp /anker/ 'companion',
/!..'lawJ /aljam/ 'I wish; long for', etc. Notice also the prepositions e.&'l /and/,
e.u111 fast/ . .
In this respect, the orthography does not exactly reflect the phonological
reality : d doubtlessly occurred in many other positions, where Mesrop did not deem
relevant to have it written. One is thus confronted with two problems, a theoretical
one : Did d function as a phoneme, like any full vowel 1 And a practical one : In
which environments did it regularly occur 1
The phonemic character of d can be seriously questioned in view of the very
fact that it is not consistently written 11. Its occurrence seems to have depended
on certain phonological and morphological rules, so that th~ 5th century reade~
would be spared any hesitation. But regarding those rules; we are left with our
knowledge of the situation in Modern Armenian, and with partial information
from orthographical treatises, the earliest of which is dated from the 2nd half
of the 13th century. To apply the modern pronunciation to the classical language,
as many Armenologists do, is only a makeshift solution, bringing no real light on
the problem. The neutral vowel was probably apt to drop or to shift its place
under certain circumstances, e.g. .in word juncture. Even an optional use of .,,
.

u In such minimal pairs a.s php/;J'I bring': pphJ 'I dig'; pw&wft '[Link]': p&wft 'inhabitant, native';

ftwpbJ 'I can' : ftphJ 'I bear', etc., the phonemic contrast is better expressed a.s V /{} (/berem/: /brem/;
/banak/ : /bnak/, etc.) than a.a

Vt~ (/berem/

: /oorem/).

16

Phonology

[2.3-2.311]

in certain environments, has to be reckoned with. Thus, the difficulty for unskilled
readers, would have been directly opposite to that which French spelling involves:
the so-called 'e muet' is written four times in "Je ne te le dira.i pas", while the
usual pronunciation is "Je n' te l' dira.i pas". The rules of ClArm. spelling would
have required : zntldirepa.
2.31. Apparent initial clusters. In PA, the dropping of unstressed i or u might
result in clusters, sometimes entailing the partial assimilation of the first consonant
to the following one, as in :
.f!uw& /k'san/ '20' < *gsan < *gisan
fJ2nLwn /t'suai/ (and lnLwn /c'uai/) < *dswar < Mir. *duswarr 'unfortunate'
(Hiibschmann, p. 504; Grammont, p. 233; Bolognesi, 1954, p. 134-135).
In ClArm., however, all initial clusters were resolved by inserting a, as in .f!uw&
[k'asan], fJ2nLwn [t'asuai), or by prefixing it to sp-, st-, sk-, zb- etc. This is proved
by the active inflection of the root aorist: all the forms that lack the e- prefix
are actually dissyllables (Cf. 2.213) :
*LP [kali] 'I swallowed', 2nd sg. qlbl' [kaler], 1st pl. *LW.f! [kalak'], etc.
as against 3rd sg. bqnLL [ekul],
u"lw& [aspan] 'he killed'
as against bqbp [eker] 'he ate', bgn]g [ec'oyc'] 'he showed', etc.

Likewise, the [Link]. subj. of uuiw&wJ 'I get, obtain' is inflected like Jnnwgw1g
(2.232) : uuiwgw1g, uuiwug pu, uuiwugp .. The substitution of c' by s before c' proves
that the aorist stem uuiwg- /stac'-/ did not count as one syllable only; therefore
uinw&wJ has to be read [astanam].
2.311. According to this principle, any apparent cluster (CC-) in word initial
involves a non written a, either inserted (CaC-) or prefixed (aCC-) :
a) [JfJnL [t'at'u] 'sour'; pdfr2q [bazisk] 'physician'; ufuwLfra [saxalim] 'I err';
&Jw [nama] 'to him'; q&p.f! [kanik'] 'seal'; 'll'wfuui [daraxt] 'paradise'; !.pb2uiwq
[harestak] 'angel', etc.
The rule applies to prefixed particles, e.g. the preposition g- to, up to" :
g&w [c'a-na] 'to him' (as against gw1dJ (c'-ayzam] 'up to now').
b) u"l pmw* [aspitak] 'white'; uuif"l [astep) 'frequent'; uq P'lP& [askizban] 'beginning'; 'lPw'lbwl [azbaleal] 'occupied, busy', etc.
The preposition 'l- which is .chiefly used a.s a mark of the definite direct object,
is read [az-] before any consonant:
'l&w [az-na] 'him'; 'l.f!/;'l [az-k'ez] 'thee' (as against 'lw1u [z-ays] 'this', 'lnui&
[z-otan] 'the foot').

[2.312-2.321)

Phonology

17

2.312. In most instances, the initial cluster stands in contrast to a full stem
form with either i or u, according to 2.221. The examples given in that paragraph
can now be rewritten in phonetic transcription, by substituting CC- by CaC- :
q/1p 'letter' : GDAbl. qpnJ [garoy]; qpbi> [garem] 'I write';
blfi.P 'he left' : 1st sg. L.P fr [lak'i], 2nd L./!/;l' [lak'er], etc. ;
./!nJi 'sleep' : GDAbl. ./!Ln1 [k'anoy];
2nLf' ' water' : GDAbl. 2rn1 [Jaroy]; 2rwqw9 [jaralac'] 'water-mill';
bi5'nLU1 'he entered' : 1st sg. i5'ur/r [ma ti], 2nd i>urbp [mater], etc.

The morphological connection between the reduced stem form and the full one
accounts for such seemingly exceptional cases as uurbi> [satem] 'I lie' (as against
uurwLwi>, uur/"f, etc.): the full stem is found in unLur 'lie', GDAbl. uurn1 [satoy].
2.32. Word final. Non inflected word forms, i.e. mostly nouns or adjectives
in the sg. NAcc., display various final clusters (-CC). The following can be
considered regular 12 :
a) Fricative + stop : -st, -sk (and -sp, -st, -zd, -xt in loan words only);
b) Nasal or r, l + stop or affricate (m before p, b only);
c) r, f, l +nasal;
d) r + fricative : -rs (and -rz, -r z, -rh in Ir. loans only);
e) y, w + any consonant (Cf. 1.641).
Class c) calls for discussion. Phonologically, -fn, -ln are on a level with -rm, -lm.
The traditional pronunciation, however, is not consistent : on the one hand fiJwpi>
[t'arm] 'fresh', 2bpi5' [jerm] 'warm', .pn'f'i> [k'urm] 'heathen priest', (.nqi> [holm]
'wind', Jbqi> [melm] 'mild, soft'; on the other 11nurfi [dutan] 'door', LbwnL [leai'an]
'mountain' bqL [ elan] 'hind', wubqL [ aselan] 'needle', etc. But the substitution of
f for r in qntnfi, LbwnL (2.231) clearly points to an earlier pronunciation [dutn],
[learn], and similarly [eln), [aseln].
2.321. All other final clusters have to be resolved by inserting a before t he
last consonant (-n, -r or -l; seldom -m) :
[Link] [ewt'an) 'seven'; anL~L (jukan] 'fish'; [Link] [anjan] 'soul, self';
urwuL [tasan] 'ten'; [Link] [erduman] 'oath' ; frLL [inan] 'nine';
[Link] [p'ok'ar) 'little, small'; [Link] [barjar] 'high' ; (.wi5'p [[Link]] 'dumb'; Jbqp
[melar] 'honey'; urwpp [tarar] 'element';
'
11 Irregular clusters (e.g. -lc'a or fricative + fricative) are tolerated in loanwords: i5'[Link] [mak'e)
'toll, tax' (Syr.); Jburw./!u [metak's) 'silk' (Gk.); fuwpfrufu [xarisx] 'b&ae, anchor' ( ?) ; pqbwlfu [OOdeaex)
governor of a border province' (Iran).

18

Phonology

(2.321-2.322)

n6q [kocal] 'log' ; wumq [astal] 'star';

'1Pn2a [darosam] 'stamp, mark'.
See also the words quoted in f. n. 8 (~pp, w1up, w1ua, w1da).
2.322.

I nflected word forms

The addition of the plural endings N. -j!, Acc-L. -u very often brings forth
irregular clusters, e.g. .j!Ul'LlllJ!.j!, .[Link] 'cities' ; awq.p, awqu 'chicks'; J/1111.p, J{1mu
(pl. only) 'mind'; wll! wlu 'eyes', etc., not to speak of three consonant clusters :
n&.qm.p, nuzmu 'camels' ; [Link]!, anpau 'clothes'; nJpj!, n1pu 'blind'. etc.
The issue can be solved in terms of morpheme variation :

-.p [-k'/-ak']; -u [-s/-0s] (Cf. the English plural ending-a in 'hats, boys; houses').
The rule of distribution may have been a purely phonological one: -k' and -s
would then have occurred not only after vowels (i.e. practically i ~r u, seldom e),
but also after single consonants, whenever regular final clusters resulted; -ak', -as
automatically in all other instances :
[Link].p
Lb'lnw
Jbq,/!
Jbqu
"'l./!
UllU

(lezuk'] 'tongues', C>wnw1.p [ca:fayk'] 'servants' t/. wp I! [ vark'] 'behaviour'

' t/_wpu [vars]
[lezus]
C>wnw1u [ca:fays]
[melk'] 'sins',
pwfil! [bank'] 'words'
[mel0s]
pwfiu [banas]
[ac'ak] 'eyes',
n1p./! [koyrak'] 'blind'
[ac'0s]
n1pu [koyras]

A number of final clusters in the .pl.N., though without parallels in non inflected
word forms, apparently fit into some of the regular patter~ :
a) bpbu./! [eresk'] 'faces', 4pw2./! [harask'] 'miracles', awpwfu.p [maraxk'] 'grasshoppers';
b) wanll! [amolk'] 'pairs' 13, wal! [kaink'] 'will' 1, ifrwnl! [p'a:fk'] 'glory';
e) fi111LJ! [nawk'] 'ships', C>nt/../! [covk'] 'seas', etc.
They may, of course, have been felt regular. Yet one has to reckon with another
possibility: in all these forms, indeed, morpheme juncture is involved, so that
the a variants may have had a wider range of occurrence than can be assumed
on merely phonological grounds. The [Link]!,. ending, at least, is traditionally
pronou_n ced -as even after r and w, while the corresponding clusters -rs, -ws
actually occur in non inflected forms. Thus, e.g. :
[Link]. t/_wpu

[var-as]

'behaviour' but [Link]. t/_wpu [vars]

u Notice, however, that l does not elsewhere occur before a consonant (2.12).
14

Cf. the 1st pl. ending -mk' in

ba./! 'we are', fuwqwa./! 'we move', etc.

'hair'

(2.322-2.324)

19

Phonology

wqww [akaws] 'furrow'

"''lPLu [aliws] 'brick',

&ww
[naw-as) 'ships'
(w?_ptu [hasiw-as] 'accounts'

etc.
~

But we cannot decide to what extent this rule reflects the pronunciation of 5th
century Armenian, nor do we know 'vhether a similar rule held of the pl.N. ending
(see below, 2.323, note).

2.323.

The pl.I. of all nouns and adjectives is formed by adding -[! to the
sg.I. form:
pl.I.
Sg.I. wnmqwt (from wnmq 'parable')

/JJ RU/" /JJL[!

pwbpt (from pw& 'word; reason'
pwbpL[!
= Gk. logos)

(from If Pl' 'letter')

'I l'ni/.

If pniJ./!
.
(1upp
(from (u11p 'father')
(wpp[!

qwuwap (from q fiu 'woman')

"/JJ f, /JJ,,p/!

The last two examples, displaying the abnormal cluster -Obk', have doubtless
to be read [harbak'], [kanambak'). As to the other three, there is a clue in the
later spelling wnwqo/! (on which see above, 1.631): the change of aw to o points
to an original final cluster in wnmq111t[! [arakawk'], and analogically in pwflpt[!
[baniwk'], q pnifl! [garovk'].
The change of aw to {j did not take place in pl.N. and AccL. forms. The divergent developments
'vould be explained by starting from, e.g., UWL[![naw0k'], flww [nawos], as against wnwqwL[! [afakawk'].
NOTB.

The demonstrative (or deictic) particles -u, -11, -fl are used either in
connection with the demonstrative pronouns w1u, "'l'l w1u, or as definite articles :

2.324.

/tun-Says/ (Or UljU UJnW) 'this house'

urnwu 'the house (here)'
/!"'l'i''l "'J'IJ/iq /k'ari-d aydmik/ 'to that stone' (Luke, 4.3)
11utntpuu 1w1unupq /y-awurs-n y-aynosik/ 'in those days' (Luke, 4.2).
urnWU llljU

Notice, in Luke 4.1, the use of the -fl particle with reference to a previous
occurrence of the same noun : LP (nqwiJ. uppnif 'full of (the) Holy Spirit' ...
i[wpfp (nqtniJ.u 'he was led by the (said) Spirit'.
As in the case of the pl.N. and AccL. endings, morpheme variation must be
here a"8Sumed, but the rule of distribution can be stated more confidently :
-u after vowels : [-s] ; after consonants : [-as]

-q

,,

: [-d]

"

: [-ad]

20

Phonology

(2.324-2.331]

-u

after vowels : (-n] after consonants : [-an]

mnwu [tun-as] (NAcc.)
fr mwufu [i tane-s] (Abl.) 'from the house'
.f!wpfr'l [k'ari-d] (D.) 'to the stone'
: .f!WP'I [k'ar-ad] (NAcc.)
!.n11w~u [hogwov-an] (I.)
l.n11 fru [hogi-n] (NAccL.) 'the spirit'

Thus the addition of a demonstrative particle never generates a cluster.

Cf. !.npu [hor-as] 'the well (here)' with npu [ors]

'hunt; game'

with wu11 [and] 'there~

with mwLfi [tawn] 'feast'
with 11n1u [goyn] 'color', etc.

pwu'I [ban-ad] 'the word'

fiwLfi [naw-an] 'the ship'
fun1u [xoy-an] 'the ram'

However, in view of the developments in ModArm., polysyllabic word forms

ending in -ay, -oy may have admitted the postvocalic variants : wp.f!wJ'I [ark'ay-d]
'the king'; !.n11LnJU [hogwoy-n] (GDAbl.), from !.n11fr 'spirit'.

2.33. Since the only function of a is, it seems, to preclude forbidden consonant
clusters, one will safely assume that it dropped, or rather failed to be pronounced,
whenever its insertion became superfluous, i.e. whenever a modification in the
environment allowed for a normal syllable djvision. This remark applies to a
in final syllables, and to initial a before st, sk, etc. (2.31).
2.331. The addition of a demonstrative particle to such word forms as have
been described above, 2.321-323, would entail the dropping of the preceding a:
anL*U (ju/kan) 15
,//;'LP [me/lar]
wum'l [as/tal]
pl.N.
wll! [a/c'ak']
~njp.[! [koy/rak']
[Link]. wlu [a/c'as]
*wuwtp.f! [ka/nam/bak']
pl.I.

+ -u >

+
+
+

+
+
+

anL*Uu (juk-nas) 'the fish (here)'

-'I > J/;'lP'I [mel/rad] 'the honey (there)'
-Ii > wum'lu [ast/lan] 'the star'
-u > [Link] [ac'/k'an] 'the eyes'
-Ii> *nJ{'.[!U [koyr/k'an] 'the blind'
-11 > wlu'l [ac'/sad] 'the eyes'
-u > ~wuwiJp.f!u [ka/namb/k'as] 'by the
women (here)'

16

The clash of k' lvith k' (in .[!W'[W.[!.[!U [k'alak'/k'an] 'the cities') and the more frequent clash
of a with a (wn a*finpuuii [at jaknors/san] 'at (or by) the fishers', Eznik IV 12) does not seem to have
been avoided: such a phrase as 1wJfruuu [y-amis/S<>S]JWJUnufr* 'in these months' would quite naturally occui:.
NOTE.

The slant lines mark the syllable boundary.

16 Cf. [Link] [awuras) .f!Wnwunw 'forty days' (Acc.), but [Link] [y-a-wurs-an]
those days' (L.) in Luke 4.2.
15

1w1finufr*

'in

(2.332-2.342)

21

Phonology

2.332. Similarly 9, either initial or in final syllable, might drop in derived

or compound words :
wuuuz [as/tal] 'star'
pwpJ.p [barjar] 'high'
[Link]~fi

[ju/kan] 'fish'

: wum'lfrY [ast/lik] 'little star'

: [Link] [barj /ranam] 'I rise'
: a~finpu [jak/nors] 'fisher' (Cf. npuwo 'I hunt,

captt1re')
u"lwfiwfibJ [ aspananem] 'I kill' : Jwp'Jwu"lwfi [mardas/pan] 'man-killer, murderer'
2.333. The last instance to consider is close word juncture. So, presumably
wqfi nLfipJ [ak/nunim 'I expect' (w~fi 'eye'; nLfi/10 'I have '); J/;'ll' nLmbfi [mel/rutenl
'they eat honey'; p u~'[Link] [is/kazbane] 'from the beginning'; uppbgp '[Link]'l.
[sirec'iz/jez] 'I loved you', etc.

.'

2.34. Word internal consonant clusters mostly consist of implosive

in agreement with the ndrmal syllable division :

+ releasing,

"fwmqtp [pat/ker] 'portrait'; nlfuwp [ oc' /xar] 'sheep'; OwmfinLO [ matnum) 'I
approach'; "lwmpwum [pat/rast] 'ready';
wfiwufin1 [a/nas/noy], qbmfin1 [get/noy], wfi'lfi'ln!J [an/dan/doc'], GDAbl. of wfiwunLfi
'animal', qbmpfi 'ground, floor', wfiqnLfi'JJ? 'abyss'.

Sim1larly, but with 9 in the first syllable: fJnlnLfi [t"ar/c'un] 'bird', 4'l'l.P [kal/zi]
'island'; [Link] [bazas/kem] 'I cure' (from [Link] 'physician'), ufifi1Jb1ufi [sa/nan/dean], GDL. of ufinLfilJ 'food', uq'[Link] [ as/kaz/ban], GDL. of uq P'[Link] 'beginning';
i1zmfi2bfiwqwfi [mas/tan/jenakan] 'perpetual'. As a rule, 9 is inserted, when
necessary, before the laxest consonant: w"luurwJp [apastamb] (not *apsatamb)
'rebel'; 6wqfi'lb'l [cakandel] 'beet'; llmp"{wmwqwfi [Atarpatakan] 'Atropateile
(modern Azerbayjan)'.
'

2.341. ClArm. had no 'double consonants', except -nn- (pfifibpnpq 'ninth')

and -rr- (Jpppq [marrik] 'tempest', GDL. Jppq p [marraki]). Consequently, the
apparent cluster -c'c'- in
qwg9hu, qw9gf ... {1st sg. qwg pg 'I shall stand')
L!J!Jhu, L!J!Jf
... (1st sg. L!JP!J 'I shall fill'), etc.

has to be resolved : [kac'ac'es], [lac'ac'e]. Otherwise, -c'c'- would have reduced to

-sc'-, as in "lwzurhu9hu, "lwzurhu9f ... (2.232).

2.342. Regular implosive clusters (2.32) are expected to occur word internally
as they do in word final, though perhaps less frequently. Thus, e.g. in simple
'

22

Phonology

[2.342-2.343)

words: wL11p111Lfi4 [and/ranik] 'first born'; wLJ.p/;L [anj /rew] 'rain'; ifw1p4bwL
[vayr /kean] 'moment' ; wtpCfi/;J [a wr /hanem] 'I bless' ; and in derivates or compounds
from words ending in -n, -r, -l, -m:
wLMlwufp [anj /naser] 'self-loving, selfish'
4n'lflUw4wL [kolm/[Link]] 'governor' (4n'lflU 'side'
- 4wL 'holder')
"'1'1Jfi4 [ayd/mik] < w111a [aydam] DL. of w111 'that'
111uur'lfi4, [Link] (2.332).

However, no general rule of occurrence for the (C)VCC pattern can be formulated,
for want of unambiguous evidence, as will appear from the following discussions.
2.343. In [Link] 'I lift, raise' (aor. [Link]), 11wnL111J 'I turn, return' (aor. 11wpJ.w1),
-rj/ has reduced to r before n: bafnam < *barjnam. But there are contradictory
instances: in the -nu- presents, at least, similar clusters seem to have been
preserved : u111p111Lnta 'I startle'; J!w'l9Lnta 'I feel hungry'; bpqfinta ' I swear'.
However, since the related aorists, as mentioned above (2.213) are built upon
-i or -u stems, this must hold true of the presents too, which can be confidently
traced back to *sarti-num, *k'alc'i-num, *erdu-num. But this diachronic statement does not solve the issue; for, the contradiction can be explained away in
two ways : either by assuming that the unstressed vowel, in this environment
(CCVC) did not drop, but reduced to a, so that we had to parallel :
1. *barj-nam
2. bainam

*sarti-num, *erdu-num
sartanum, erdanum,

or by assigning the conflicting results to different phases of the development.

Indeed, the change of *barjnam to [Link] doubtlessly took place at an earlier
stage of PA. than the dropping of unstressed i and u. Therefore, the preservation
of implosive clusters in the same environment at a later stage would be no wonder,
and the developments would have been:
1. *barj-nam
2. bainam
3. barnam

*sarti-num, erdu-num
*sartinum, erdunum
[Link], erdnum

Thus, there is no deciding whether u111purlinta has to be read [sartanum] or [sartnum].

One is confronted with the same ambiguity wherever apparent clusters prove to
have come about as the result of the dropping (or reducing) of i and u:
'l"'l'Pfili 'smith'
a"'l''l P4 'men, people'
w'l2fi4 'girl'

'l"'f'plin9 [darb(a)noc'] 'forge'

: GDL. Jwp'14wli [mard(a)kan]
: GDL. w'l24wli [alj(a)kan]

[2.343-2.4)

23

Phonology

.54 pmpl_ 'Baptist'

: GDL. a4pml_p [makart(a)c'i]

UIL'fnLm 'use, advantage' : GDL. UIL'ftnp [awg(a)ti]
wiiantq 'narrow'
: GDL. wiiaq p [anj(a)ki]
: GDL. hp'lawii [erd(a)man]
[Link] 'oath'
: 2nd sg. pwpa!Jhu [barj(a)c'es], etc.
pwpap!J 'I shall lift'
2.344. The oblique cases of polysyllables ending in -i display the rather
unusual sequence Ow (l.n11fo '.spirit' : GDAbl. l.n11LnJ). This spelling can be taken
at its face value, in view of the contrast of l.n'fLnJ [hogwoy] and "lwuinw1 [patuoy],
GDAbl. of "lwmpt 'honor' (2.221). Consequently, implosive clusters can be checked
in the oblique cases of WJ'lf 'viney~rd', w'lwLiip 'dove', np'Jp 'son' : UIJ'fWJ [ayg/woy],
w'lwLiiw1 [alawn/woy], np'}tnJ [ord/woy], etc. As to 'lw'lmiifo 'secret', 1w1uiiifo 'visible,
manifest', the GDAbl. forms ('fw'lmiiwJ. 1w1uiiiw1) certainly have to be read
[galtanwoy], [yaytanwoy) and the NAcc. may have been, accordingly: [galtani],
[yaytani]. But the alternative reading (without a) is by no means ruled out :
for, the insertion of an inorganic a in the oblique cases must be assumed in iiwfuiip
'prior, ancestor', GDAbl. iiwfuiitnJ); 4'l11.P 'island', GDAbl. 4'[Link], since neither
-xn- nor -lz- are regular implosive clusters. The only conceivable reading is
therefore:
iiwfuiifo [nax/ni] : iiwfuiitnj [naxan/woy]
4'1.11.P [kal/zi]
: 4'[Link] [kalaz/woy]
[Link], presumably :
'fUl'lmiip [galt/ni] : 11w'lui&w1 [galtan/woy).

2.345. Etymology sometimes allows for a decision: e.g. Av. afrinami 'I
praise' gives a clue as to the position of a in w'l'(.&/;J [awr/hanem] 'I bless'
(< PA. *awhrinem), later [Link] [orhanem] (1.631). Such cases however are
exceptional.
We are thus brought back to our first survey of the whole issue (2.3) : the
rules of occurrence of the neutral vowel partially escape our ~owledge. Yet,
Mesrop's spelling, puzzling though it appears to present day readers, is by no
means objectionable : it is phonemic, not phonetic. And this is a strong arg!lment
against substituting the usual transliteration by a tentative, and perhaps misleading, phonological transcription.
2.4. THE

TRADITIONAL PRONUNCIATION OF

CIArm.

CIArm., as a written language, fairly preserved its grammatical shape through

centuries, while the spoken dialects gradually altered (Cf. 1.2). But if it was
possible for writers to keep close to the morphology and syntax of the classical

24

Phonology

[2.4-2.42J

models, they would have hardly been able, and certainly did not trouble, to
preserve or restore the original phonology : they would instead conform to the
pronunciation of their own days. However, in view of the variety of modern
dialectal developments, it must be assumed that a tradition was established at
a certain stage of the evolution. The rules of pronunciation, as they are
formulated in Armenian grammars, seem to go as far back as the 11th or 12th
century.
2.41. These rules, equally valid for ClArm. (qpwpwp) and the modern literary
dialects (Eastern and Western w2fuwp(wpwp), can be summed up as follows :
.

1) On h and t, n and o (for earlier wL), see above, 1.632 : hyhp [yeker] 'he ate';
1wrtr [varer] 'he led (or was led)'; "l [voe'] 'not'; if1npahl [p'orjel] 'to try (or be
tried)'. Notice n1 [ov], even in word initial : fl1y pwunu [ovkianos] 'Ocean'.
2) hw = [ya] : q phwl [garyal] 'written'.
3) "J (before a consonant) = [uy]: 11n1u [guyn] 'color'; hgn1g [yec'uyc'] 'he
showed'.
4) fL (before a radical consonant) = [yu]: wppw [aryun] 'blood'; fLf'"!J [yuroc'],
pl. GDAbl. of PLf' [yur] 'his, her'.
5) On 'b see above, 1.62: wqpw (ayyus] 'brick'; ~wqghwL [k'ayc'yav] 'he felt
hungry'; hqpgp [yeyic'i] 'it shall be'.
6) Initial J = [h] : 8/1untu [Hisus] 'Jesus'; JWUWUJIUlll [h-anapat] 'to the desert';
J'lP [hayi] 'pregnant'. Both J"JL 'lazy' and (n1L 'troop, swarm' are pronounced
[huyl].
7) Final J is dropped: C;wnw1 [cara] 'servant'; U.u111nLC;n1 [Astuco], GDAbl. of
[Link](; 'God' (see above, 1.42); Jwphw1 [harya] 'I got up, rose'; 11w1 [ga] 'he comes
- except in monosyllabic nouns (l.w1 [Hay] 'Armenian'; fun1 [xoy] 'ram') and
interjections.
8) L = [ v] : C.n1 fL [hoviv] 'shepherd'; (nqw1 [hogvov], I. of (nq fi 'spirit'.
Besides, nL, when followed by a vowel, is also pronounced [v]: [Link](; [Ast(a)vac] 'God'; C.n1nL/i [hovvi], GDL. of C.n1f!L; unLf[' [naver] 'present'; uinLhwl [tavyal]
'given'.
2.42. Notice that rules 4) and 7) do not apply to nouns ending in -/iL, -w1
or -n1, either inflected in the pl.N. and Acc-L. (2.322) or followed by a
demonstrative particle (2.324) :
C.n1/iL [hoviv]

: pl.N. C.n1/i~ [hoviv(a)k']

Acc-L. C.n1fiLu [hovivas]
+ -u : C.n1fo1.u [hoviv-as]

(2.42-2.43)

Phonology

25

+ -fi : (nq/[Link] [hoviv-an]

pl.N. + -fi : (nq/[Link] [hovivk'-an], etc.
~wnw1

[ca:i:a]

: pl.N. C>wnw1.p [ca:i:ayk']

Acc-L. C>wnw1u [ ca:i:ays]
-11 : C>wnw111 [ ca:i:ay-d]

Likewise
awp11n1 [mardo], GDAbl. of awp11 'man'
+ -fi : au1p11n1fi [marduy-n]

2.43. Text sample (1.5), transcribed according to the traditional pronunciation i 1:

Yev H isus li hogvov sarbov darjav i Hordanane yev varer hogvov-an h-anapat
(2) avuras k'a:i:asun p'orjyal i Satanaye, yev voe' yeker yev voe' arb h-avurs-an
h-aynosik, yev i katarel-an noc'a k'ayc'yav. 3. Yev ase c'a-na Satana : Yet'e
vordi yes Astuco, asa k'ari-d aydmik zi hac' lic'i. 4. P atasxani yet nama Hisus
yev ase: Garyal e yet'e voe' hac'iv mi~yn kec'(a)c'e mard, ayl ame.n~yn baniv
Ast uco. ? Yev hanyal az-na. i lya:i:(a)n mi barjar yec'uyc' nama z-amenayn
t'agavorut'yunas asxarhi i vayrkyan zamanaki. 6. Yev ase c'a-na Satana : K'ez
tac' z-ays amenayn isxanut'yun
yev az-p'a:i:as .soc'a, zi inj tavyal e yev um kamim
.
tam az-na. 7. Ard du yet'e ankyal yerkir paganic'es a:i:aji im, k'ez yeyic'i amenayn.
8: Patasxani yet nama Hisus yev ase : Yert' h-etas im, Satana, zi garyal e : Y erkir
pagc'es tya:i:(a)n Astuco k'um yev az-na miayn pastesc'es. 9. Yev ac az-na
h-Erusayem yev kac'uyc' i vera astaraki tacari-n, yev ase c'a-na: Yet'e vordi
yes Astuco, ark az-k'ez asti i vayr, (10) zi garyal e yet'e harestakac' yuroc'
patviryal e vasan k'o pahel az-k'ez, (11) zi i vera je:i:ac' barj(a)c'en az-k'ez, mi
yerbek' harc'es az-k'ari z-otan k'o. 12. Patasxani yet nama Hisus yev ase :
Asac'yal e t'e voe' p'orjesc'es az-ter Ast(a)vac k'o. 13. Yev kataryal z-amenayn
p'orjut'yunas Satanayi, i bac' yekac' i namane ar zamanak mi.

17

The stops and affricates are transliterated according to the pronunciation in the [Link] dialect...

26

Morphology

3.

Morphology

[[Link] l)

18

In ClArm., as in most IE languages, inflection and word formation have to

be dealt with separately. Inflection will be given the precedence because it
pertains to both simple and derived (or compound) words, while word formation
is concerned with the latter only. The reverse order, however, is more usual.
Detailed tables of noun and verb inflection, such as are found in grammar
books, would be out of place in this Introduction. What is needed here is rather
a survey of the inflection system, together with a discussion of the theoretical
and practical problems involved.
3.1.

NOUN INFLECTION.

Nouns, adjectives (including numerals) and pronouns take different forms

according to number (sg. or pl.) and case. ClArm. has no gender distinction and,
consequently, no instance of such grammatical agreement as prevails in Latin
(magnus uir, magna urbs, magnum nomen) and in many other cognate languages.
On the morphological level, adjectives do not differ from nouns : [Link] 'wise',
nor 'new', ferm 'warm' belong to the -a- declension exactly like [Link] 'God',
gore 'work', beran 'mouth' etc. ; erkayn 'long, csmarit' 'true', to the -i- declension
like matean 'book', Birt 'heart', ef'kiwl 'fear', awe etc. On the syntactic level,
however, they behave differently, in that polysyllabic adjectives, when placed
immediately before a, noun usually remain uninfleoted :

csmarit [Link] 'the true God'

(as against : GD. [Link]-oy clmart-i
GDAbl. csmarit [Link]-oy
Abl. [Link]-oy clmart-e).
3.11.

Sg.N.
Acc.
L.
G.
D.
Ahl.
I.

The case system

es 'I'

Z-18

y-1s
.
im

lnJ

y-1nen
.
inew

1S

best represented by pronominal paradigms, e.g. :

du 'thou'
Z- k'ez
i k'ez
k'o
k'ez
i k'en
k'ew

na 'he, she'
z-na

inma
nora
nma

i nmane
novaw

nok'a 'they'
z-nosa
.
I nosa
noc'a
noc'a
i noc'ane
nok'awk'.

3.111. Even in these paradigms, there are examples of 'syncretism', i.e. of

one form covering two or three different cases: is, AccL.; k'ez AccLD.; na NAcc.;
In this and the following chapters. CIArm, words and morphemes, as a rule, will be quoted in
transliterated form.
18

[[Link] l-3.112]

Morphology

27

nma LD.; nosa AccL.; noc'a GD. Yet, in view of the contrasting distribution of
the case forms in na, nok'a, the existence of a seven case system can hardly be
denied. The L ., it is true, mostly coincides with the D. in the sg. (though not
in is !), and with the Acc. in the pl. at large, so that many Armenian grammarians
do not acknowledge it as a case in its own right. This opinion is not shared by
Western Armenologists. At any rate, some specific L . forms do occur in the sg.
(3.133; 3.181).
3.112. Ancient grammarians held the monophonemic prepositions (Iiwfu'IPI')
z- and i/y- (on this variation, see above, 1.64) as part of the Acc. and Ahl. forms,
a view t hat is partially supported by CIArm. orthography : z-is, y-is, y-inen,
z-k'ez etc. are written as single words (qpu, 1/iu, 1/ififfi, [Link]). Abrahamyan departs
from the tradition by bracketing q-, but not 1-. Western Armenologists slightly
disagree: Meillet and Jensen drop the prepositions altogether, while Mann maintains z- (with a hyphen) and drops i/y-. The old practice, though involving
an encroachment of syntax upon morphology, may be advocated on pedagogical
grounds, provided the prepositions [Link] bracketed or separated (as in the above
paradigms. Cf. 1.42).
The actual blending of a preposition with a case form is ascertainable in the
interrogative pronoun NAcc., zi, zinc' 'what 1' (G. er, D. (h)im, I. iw). I n any
other instance, the prepositions, however frequently joined to the Acc. (z-) or to
the L. and Ahl. (i /y-), never come down to mere 'case prefixes', but stand on a
level with the other prepositions that occur with the same cases :
Acc.

L.

Ahl.

ekn af is (ar na) 'he came to me (to him or her)'

ase c'-na 'he says to him'
i na yusamk' 'we hope unto him'
and is (and nma, and nosa) 'with me (with him, with them)'
af nma 'near by him (or her)'
ast is (ast k'ez, ast nma) 'according to me (to thee, to him)'
[Link] z-inen (z-nmane) 'they say about me (about him)'.

Besides, nouns are used in the Acc. without z- as indefinite direct object :
tan ptul 'they bear (give) fruit' Mark 4.20 (as against: tay z-ptul 'he bears
(the) fruit' Matth. 13.23 19);
ekn kin mi or uner Sis iwloy nardean 'there came a woman who had a
flask of nard oil' Mark 14.3 ;
nok'a arkin i na jefs 'they laid hands upon him' Mark 14.46, etc.
lt

In both places, the Greek original has a compound verb (karpophorei).

28

Morphology

[3.112-3.131)

or to denote a .space of time: awurs k'afasun '(during) forty days' (Luke 4.2).
Notice the occurrence of the bare Acc. (without z-) in formular phrases: patasxani
tam 'I answer (give answer)'; erkir paganem 'I prostrate myself (kiss the earth)';
aknunim ' I expect , look forward (have an eye to something)', etc.

3.12. R egular declension patterns are numerous. In classifying them, it seems

appropriate to contrast, first of all, variable and invariable stems (Meillet, 1913,
43). Mixed inflection and isolated anomalous words have to be recorded
separately.
Stem variation, an archaic feature reflecting PIE vowel alternation, is preserved in quite a number of nouns, including derivatives; and there is even
~tymological evidence for a propagation of the -n- inflection in PA beyond its
original range (see 5.14; 5.146). Notice that stem variation, as a morphological
item, has to .be kept apart from the regular vowel alternation described above
(2.22-2.223). The latter occurs automatically whenever the addition of a morpheme
entailed a stress shift; consequently, it has no bearing on the declension : such
words as aiafin 'first' (GDAbl. aiaf-noy), surb 'holy' (srboy), loys 'light' (lusoy),
mef 'middle' (mifoy) range among invariable stems, together with [Link] 'God'
(Astuacoy), <JOf'C 'work' (gOt"ooy), ker 'food' (keroy), ferm 'warm' (fermoy), etc. Both
stem variation and regular alternation operate simultaneously in
jukn 'fish' : GDL. jkan ( < *jukan)
dustr 'daughter' : GDL. dster ( < *duster)
learn 'mountain' : GDL. lerin ( < *learin), etc.

3.13. I NVARIABLE STEM INFLECTION. The \VOrd stem underlying .the whole
paradigm is identical to the [Link]. form. I t ends in
-C(C)
-a, -i, -u or (seldom)

-e

(On -CC, see 2.32 and fn. 12)

(On polysyllables in -i, see below, 3.133):

The oblique case endings consist in -V(C), exceptionally -VCC (in the pl.I ., 2.323).
On the pl.N. and AccL., see 2.322; 3.183.
Taking as a. criterion the .inflectional vowel, we get the following
.
. 3.131.
paradigms:

[Link].
GD.
Ahl.

I (-o-)
erg
'song'
erg-oy
erg-oy

I l a (-a-) .
Hayk
pr.n.
Hayk-ay
Hayk-ay

l ib (-a-)
lezu
'tongue'
lezu-i
lezu-e

I I I (-i-)

[Link]
'servant'
.

caray-1

caray-e

. -IV (-u-)

gah
'th.r one'
gah-u
gahe-

I.
[Link].

29

Morphology

(3.131-3.133]

erg-ov
erg-oc'

[Link]-a.w

lezu-aw
lezu-ac'

ca.i:a.y-iw
ca.i:a.y-ic'

ga.h-u
gah-uc'

The sg.L. is formally identical to the GD. in lib', III, IV,' and to the NAcc.
in I , Ila, which also display similar forms in the GD. and the Abl. In spite of
this parallelism, Ivleillet (1913, 49) regards Ila as a variety of lib, while Mann
only mentions it as an instance of 'minor deviation of declension (1968, p. 100).
In fact, the [Link] of the Hayk and kzu patterns is limited to the sg. : in the
pl. there is only one -a- declension. Furthermore, Ila. is a. marginal paradigm,
insofar as it does not suit genuine Arm.e nian words, and is mostly substantiated
by Iranian, Syriac or Greek proper names, e.g. Tigran, Anahit (GDAbl. Anahtay),
Bel, Asia (GDAbl. Asiay), Homeros (Homeray), etc: For more examples, see
Jensen, 135; Abrahamyan, 54. Nouns and adjectives are exceptional: Meillet
(loc. cit.) quotes hiwl 'matter', from Gk. ltyle, L. i hiwl-n; Abl. z-hiwleay (Eznik I 5);
Mann has dabir 'sanctuary', an indirect loan from Hebre'\v (through [Link]. dabir)
and bosor 'dark red', of unknown origin.
3.132. As a rule, every noun or adjective is bound to one declension pattern.
This holds true of derivatives, in particular, the paradigm being predictable from
the suffix : [Link] nouns in -ic', e.g., follow the normal a declension (IIb): araric'
'creator', GDL. ararc'i, I . ararc'aw. Infinitives and participles, too, never deviate
from the original o declension : asel 'to say', GDAbl. aseloy; asac'eal 'said ', GDAbl.

asac'ewy.

Simple words, however, may be observed to be inflected according to more

than one pattern : only in limited instances can the right paradigm be predicted
from the bare stem. Monosyllables in -i'- or u, e.g., belong to the o declension :
di 'corpse', GDAbl. di-oy; likewise li 'full', ji 'horse', bu 'owl', ju 'egg', etc. Most
polysyllables in ~ay (cafay), -e (margare '.prophet',) -oy (dsxoy 'queen'), tO 'the i
declension. But arbitrariness largely prevails, and as soon as the 5th century
some fluctuation is perceptible: Yovsep' 'J oseph' is inflected after Ila as .well as
after IV; . both Yovsep'-u and Yovsep'-ay occur as D. forips in quite . similar
environments (Matth. 2.13 and 19). Such examples are very rare in early Jiterature.
But later authors are less careful and use different declensions in t he sa;me word :
afu 'brook' (I or III) ; k01.1s 'virgin' (l ib or III); arf 'bear' (I , III cir IV) ; her 'hair'
(I , lib or III), etc., so that we sometimes are at a loss as to the original dectension .

The a and i declensions (li b and Ill) seem particularly exposed to mutual overlapping,
ein_ce the endings in the sg. GDL. and Abl: are the same. In fact, the difference in the other inflected
cases has been presenred to a larger extent than might have been expected.

NOTE.

.
3.133.

. .

Mixed [Link]. Among polysyllables ending in -i, some follow the o

30

Morphology

[3.133-3.14)

declension (I), while the others conform to a mixed paradigm with o and a endings
and a specific L. form in the sg. In both sets, final -i changes to -w- before -o(2.344). The paradigms run as follows :

[Link].

L.
GD Ahl.

I.
[Link].

I (-o-)
hogi 'spirit'
hogi
hogw-oy
hogw-ov
hogw-oc

V (-o/a-)
aygi 'vineyard'
aygw-oJ
aygw-oy
aygeaw ( < *aygi-aw : 2.213)
aygeac'.

In the [Link]., a couple of nouns of the -o/a- classe have a specific form with a
complex ending: tarw-of-e, telw-of-e, from tari 'year', teli 'place'. Mann (1968,
p. 71) mistakenly assigns tari to the -o- class. The sg.I. is tareaw, not *tarwov,
which would conflict with the L. form.
The aygi type is doubtless a variety of the normal -a- declension (IIb). The -i
ending of the GDL. (lezu-i) would have amalgamated with the final vowel; hence,
presumably, its substitution by -oy. But neither aygwoy could do for the L.
(Meillet, 1913, 46), nor would a similar form in the L. and the NAcc. fit in with
the normal -a- paradigm. On other occurrences of -of, -ofe, see 3.17 (kin); 3.181
Note (mi).
3.14. VARIABLE
(Stem I) ends in :

STEM [Link]

includes -r, -l and -n stems. The sg. NAcc.

ezr 'border', dustr 'daughter', hamr 'dumb'

astl 'star', kocl 'log' ao
kolmm 'side', jukn 'fish', jiwn 'snow', befn 'burden'
-Cn
tun 'house', anun 'name' 21
-un
-ik,. -uk calik 'flower', nuinuk 'child'.
-Cr
-Cl

The inflected forms are built up on modified stems, according to the following
distribution table (3.141-143). Notice that in all paradigms the [Link]. form is
identical to stem II :
ezr 'border': GDL. ezer (=stem II ezer-)
jukn 'fish' : GDL. jkan (= stem II jkan-)
ao Nouns in .wi range among invariable stems: ewl (I) 'oil'; enawl (llb) 'parent'; erkiwl (Ill)
'fear', etc.
21 Derived adjectives in Un (imaatun 'wise', from imaal 'intelligence') do not belong here, but
follow the O declension.

[3.141-3.142]

3.141.
a)

31

Morphology

Two stem declension

Stem I : [Link].

Nouns in -r, -l :
ezr 'border'
dustr 'daughter'
alewr 'flour'
.awr 'day'
astl 'star'

Stem II : oblique cases

pl.N. and AccL.
ezerdsteralerawur- (-ur instead of -er in this word only)
astel-, etc.

Exceptionally a few nouns in -n :

akn 'gem'
akankolmn 'side'
kolmananuananun 'name'
b)

Stem I : [Link].
pl.N. and AccL.
sun 'dog'
tun 'house'

and most nouns in -iwn :

ankiwn 'corner'
p'orjut'iwn 'temptation'
3.142.
c)

Stem II : oblique cases

v
santan-

ankeanp'orjut'ean-, etc.

Three stem dec'lension

Stem I : [Link].

So most nouns in -n :
akn 'source'
jukn 'fish'
skizbn 'beginning'
durn

Stem II : oblique cases Stem III : pl.N. and

AccL.
akanjkanskzbandran-

akunjkunskzbundrun- (2.231)

including all action nouns in -umn :

erdumn 'oath'
erdman-

erdmun-

and words lacking -n in the [Link]. :

calkancalik 'flower'
manuk 'child'
mankan-

calkunmankun-, etc.

32

(3.142-3.15]

Morphology

d) Stem I : [Link].

azn 'nation, race'

anjn 'soul, self'
learn 'mountain'
and some other -n stems.

Stem II : [Link]. _
(Ahl.) pl.N.
and AccL.
.
az1n..
anJinlerin-

Stem III : other oblique

cases
azan
anJanleran- (2.231)

The [Link]. is built up on stem II : azn-e ( < *azin-e), an(jn-e, kine-; seldom,
on stem I II: ezan-e, from ezn 'ox', GDL. ezin; befan-e (also befn-e), from befn
'burden', GDL. befin.

Some -n nouns or adjectives share the characteristics of both d) (in

the sg.) and c) (in the pl.), thus yielding the following paradigm :
3.143.

[Link].
GDL.
Ahl.

I.
Pl.N.
AccL.
GDAbl.

I.

t'oin 'grandson, grand'daughter'

t'oiin
t'otn-e (or t'oian-e)
t'oiam-b
t'otun-k'
t'oiun-s
t'oian-c'
t'oiam-bk'

'

Besides simple words, as ap'n 'bank, shore', masn 'part', matn 'finger', this
pattern also fits compounds of azn, anjn. Notice the resulting contrast in the
pl.N. and AccL. :
*Mazde-azn 'Mazdean' : pl.N. mazdezun-k' as against azin-k'
AccL. mazdezun-s as against azin~s
Likewise : vehanjun-k', from veh-anjn 'generous', as against anjin-k'.

Mixed infledion includes several nouns and a small group of adjectives,

which do not make up a homogeneous formal class, except for one common
feature: the contrast between the sg. and the pl. declension. Leaving aside some
particulars, we find two main varieties of mixed inflection, according to whether
an -n stem is assigned to the pl. or to the sg.
3.15.

(3.151-3.152)

3.151.

33

Morphology

Nouns with an -n stem in the pl. only

k'ar 'stone' (III)
k'ar-i
k'ar-iw
erec' 'elder, priest' (IV)

er1c
-u

GDL.
I.
GDL.

pl. stem : k'arin-/k'aran- (3.142 d)

J pl. stem : eric'un-/eric'an- (3.142 c)

Likewise: pafaw 'old woman'.

Adjectives :
GDL.
Ahl.

canr 'heavy'; barjr 'high' (IV)

can-u
barj~u
can-ue
barj-ue 22

pl. stem : canun- ; barjun-/barjan- (3.142 c)

Like canr are inflected : manr 'small'; p'ok'r 'little'; k'alc'r ' sweet', etc. (Meillet,
1913, 5lc; J ensen, 147).
At first sight, adjectives in -Cr (canr, p'ok'r) stand in contrast to nouns (ezr, d'U8tr, alewr),
the adjectives losing -r in all inflected forms, while the nouns preserve it througho~t :
canr
ezr
GDL.
can-u
ezer
pl.N.
canun-k'
ezer-k'
There are contradictory instances, howeve~, among adjectives (hamr 'dumb', GDL. hamer) as well
as among nouns : asr 'fleece', GDL. as.u; likewise calr 'laughter', melr' 'honey'. On the other h~nd,
artaww 'tear' and cunr 'knee' arenot inflected. in the sg.; the pl. belong to the -a- declension (IIb):
arkuu-k', cung-k', GDAbl. artaBu-ac', cng-ac'.
It thus appears that PA had two [Link] of words in -Cr: one with .,. as a [Link] of the stem (ezr,.
dwtr, hamr, etc.), [Link] another with -r as a morpheme marking the sg. NAcc. 23(asr, melr, canr, p'ok'r,
etc.). In.C!Arm. the former remained unaltered, except for a. [Link].I shift to the -a- declension in the
pl. (GDAbl. ezer-ac', beside ezer-c'), while the latter underwent various alterations, either in the ag.,
as the Joas of the inflection in artawr, cunr and some adjectives (goltr 'tender, lllild' is invariable), or
in the pl., as the expanded stem in canun-, p'ok'un-, and the shift. to the -a- declension in artasu-k',
r;ung-k'. The decay of the original inflection plainly appears in such later forms as GDL. meler,. I.
melerb (instead of mel-u); GDL. t'anjru, t'anjroy, from t'anjr 'thick', etc. (Abrahamyan, 59, p. 45).
NOTE.

3.152.

Nouns with an -n stem in the sg. only

..

akn 'eye'
GDL.

I.

akan
akam-b

pl. stem :

ac'- (II)

'

As against gah:e (3.131). The .same complex Ab!. ending also occurs in some nouns, e.g. dasu-e
(beside das-e), from das 'ord~r, [Link]', xrat-u-e, from xrat 'advice', etc.

23 Though also occurring in derivates and compounds: canr-agoyn 'very heavy', canr-a-airt"'hard
hearted, obstinate'; melr-a-hoa 'flowing with honey' (as against aau-i 'woollen').
22

34

Morphology
'

[3.1523.l 71

GDL.
Ahl.

unkn 'ear', GDL. unkan; pl. stem : akanf- (II)

dufn 'door', GDL. dran; pl. stem : dttr- (II), beside drun-/dran(3.142 c)
jern 'hand'
jeiin
pl. stem : jef -(II)
.. ieran-e

So also:

otn 'foot', GDL. otin; pl. stem : ot- (Il l)

GDL.
I.

serund 'breed, race'

serndean
serndeam-b

So also :

pl. stem : serund- (I)

and other derivatives in -und, -urd (zolovurd 'assembly' people,), -ist-, -ust (Meillet,
1913, 59a; Jensen, 148). On the historical development of -ean, see below, 5.237.
3.16. The Armenian grammarians 238 favour a somewhat different classification, based on what they call 'declension markers'. A declension marker
(!.nLnif p~) is the characteristic vowel of the oblique cases t in the sg., whether
t his vowel appears in the case endings or within the second stem. There are as
many declensions as markers: a, e, i, o, u, i/a, o/a. But t he position of the marker
entails a cross division : internal inflection (i.e. stem variation) vs. final
inflection. Thus, the same -a- marker will be found in : Hayk (Ila), GDAbl.
Hayk-ay, I . Hayk-aw (final -a- declension), as well as in: jukn, GDL. jkan, I.
jkam-b (internal -a- declension). Likewise, the i/a marker in lezu (l ib), GDL.
lezu-i, I . lezu-aw and in anjn, GDL. anjin, I . anjam-b. Notice that -e- occurs in
internal inflection only (3.141 a), while o, i and o/a {I , III, V) never do. This
classification is defendable from a descriptive (synchronic) point of view. In a
diachronic and comparative study, it would prove less adequate.
3.17. Anomalous words display irregular stem variations and, exceptionally,
uncommon case endings. The following, as notable reflexes of PIE noun inflection,
are worth recording in full paradigms :
[Link].
GDL.
Ahl.

ayr
kin
'man, husband' 'woman, wife'

arn
kn-oj

arn-e
kn-oj-e

hayr
'father'
hawr
hawr-e

k'oyr
'sister'
k'er
k'er-e

Following A. Bagratuni, whoae [Link] Grammar for Advanwl Sludenta (1.w1bpf & j!/;pw~w&n
"(f111u 'f.f''l'u!//;Ln!J) was published in 1852.
'' Remember that the Locative ie not taken into account.
Illa

t/Jpi.L p

[3.17-3.181)

I.
Pl.N.
AccL.
GDAbl.

I.

35

Morphology

[Link]-b
or kn-aw
ar-k'
kanay-k'
ar-s
kanay-s
aran-c'
kanan-c'
kanam-bk'
[Link]-bk'
aram-b

har-b

k'er-b

har-k'
har-s
har-c'
ha.r-bk'

k'or-k'
k'or-s
k'er-c'
k'er-bk'

On tir 'lord', tikin 'lady', see 2.21., Note. Notice the unexpected forms : Ahl.
teafn-e (with -ea- preserved, as against I. teram-b); I . tiknam-b, pl. [Link]-k',
AccL. tiknay-s, etc. (as against kanam-b, kanay-k', kanay-s, etc.). Like hayr are
inflected : mayr 'mother', elbayr 'brother'.
Gewl (also written geawl, giwl) 'village' belongs to the -i- declension (III), except
for the [Link]. gel-f and the Ahl. gel-f-e. The GDL. tumfean, from tiw 'day
time', is a parallel to serndean, wlovrdean, etc. (3.152), save that -ean is not added
to the bare stem, but to an original PA case form *[Link]. An obsolete Ahl. form
is preserved in the phrase i tue (< *tiw-e) 'by day' (lit. 'from day break'), beside
the analogical form tutJnfen-e, from tumfean.
3.18. CASE ENDINGS. In no paradigm is [Link] of the seven cases matched by
a specific form. But the distribution of the case forms is not quite the same in
the sg. and in the pl. Moreover, there are differences between nominal and
pronominal inflection.
3.181. On the L., see above (3.111). Aside from the mixed -o/a- paradigm
(3.133), sg.L. forms sporadically occur in the -o- declension (I) : i mif-i (beside
i mef) 'in the middle; i giser-i 'in the night', etc. Notice the isolated form y-amsean 'in the JUOnth', from amis, GDAbl. ams-oy (for more details, see Meillet, 1913,
50 and 74; Jensen, 136.1-3, 140.l, 159).
As to the G. and D., all pronouns have different forms in the sg., which is never
the case in nominal inflection. Here are some examples, in addition to the paradigms
quoted above (3.11) :

G.
DL.

ayn 'that'

o(v) 'who 1'

aynr
aynm

oyr
um

zi, zinc' 'what 1' or 'who, which'

(relative pr.)
er
or-oy
(h)im
or-um

On aynr, aynm (as against nora, nma), cf. 2.22 fn 8. The last example (or) points
to a connection of the DL. ending -um with the -o- declension: indeed, -um has
regularly developed from *-om, as can be shown by paralleling nma < *num-a
with nor-a, both from an original stem no-. The same pronominal ending is also

36

Morphology

[3.181-3.183)

found in some adjectives of the -o- declension, namely: mi 'one', ayl 'other',
mews 'another', nor 'new', kin 'old', arafin 'first', which [Link] inflected like the
relative pronoun : DL. mi-um, ayl-um, nor-um.
Besides 111wy, mium, there is also a by-form 111i-o/, whioh is chiefly used as a G. The Ab!.
.
is, expectedly, mw/-.
NOTE.

3.182. In the pl., the G. and the D. always coincide, except in the personal
pronou.n s of 1st and 2nd [Link]. Moreover, the Ahl. is formally identical to the.
GD., as in the sg. of I and Ila (3.131), except in the same pronouns and the
demonstratives : sok'a 'these', dok'a 'those (near you)', nok'a 'those, they';
aysok'ik, aydok'ik, aynok'ik. The latter three very often function as appositions :
k'arink'-s
. aysok'ik. 'the stones, these ones = these stones' (Matth. 3.3). The situation
is summed up in the 'following table :
.

G.
D.
Abl.

mek' 'we'
mer
mez
men]
v

duk' 'you'

1er

1ez
.1en3
v

nok'a 'those, they' aynok'ik 'those'

} noc'a
noc'ane

aynoc'ik
aync'ane

In spite of the constant occurrence of noc'ant, ayn.c'anl, etc. in ClArm. literature, these
[Link]!. forms &re missing in Meillet's paradigms: he &SSumes that noc'a, aynoc'ik, etc. aot as GDAbh
NOTE.

(1913, 64). Jensen quotes soc'ant e.s a by-form_( 215) and omits ayae'ant ( 217).

3.183. There is a striking contrast between the variety of declension patterns

and the uniformity of most ca~e ma?'kers ((nLn1wlfbpuiLkp), especially in the pl.
The [Link]. ending -e is common to all inflection classes, except I and I la. It
is added either to the bare stem (lezu-e, cafay-e, gah-e), or to the GDL. form,
as in canu-e, xralu-e (3.15 and fn 2Z), and in variable stems at large: ezer-e,
astel-e, tan-e, anken-e, p'orjut'en-e, jkan-e, azn-e, afn-e, hawr-e, etc. It also
shows a peculiar connection with such endings as -(o)f (tarwof-e, telwof-e, 3.133;
knof-e, .gelf-e, 3.17; miof-e, 3.181, ~ote) and -um: orm,-e, aylm-e (cf. DL. orum,
aylum, 3.181). In the demonstrative pronouns, the expanded ending -ane (nmane,
aynmane) has even been propagated to the pl. (noc'ane, aync'ane).
The sg.I. markers ar~ :
( V)v /10 (except in IV : -w is regularly dropped after -u)
{ (O)b '

In the pl., the case markers [Link], uniformly :

N.
_-k' (2.322)
-s
AccL.
GD A bl.
-c
I.
sg.I. marker + -k' (2.323).

(3.183-3.21)

Morphology

37

The combination of stem and case markers (or endings) in various declensions is
evidenced in the paradigms quoted above (3.131 and 133; 3.143; 3.17).
3.2.

VERB INFLECTION

The verb system, or conjugation, consists of the following moods and tenses
Indicative : present, imperfect, aorist;
subjunctive : present, aorist;
imperative (never used negatively) and prohibitive (with the negative particle
. mi 'do not', which is also used with the subjunctive and the infinitive).
Besides, it includes an infinitive (ending in -l), with derived verbal adjectives,
and a participle (ending in -eal). Both the infinitive and the participle belong to
the -o- declension (3.132). Verbal adjectives in -loc', being mostly used as
predicates, do not occur in the oblique cases. Those in -li (sireli 'lovable, beloved';
zarmanali 'admirable') follow the mixed -o/a- paradigm (3.133).
I have purposely omitted the so-called present participle in -ol, -awl, inflected after the
normal -a- paradigm, though Jensen( 267) and Mann (1968, p. 138-139) put it on a level with the
participle in -eal. A regular contrast is thus assumed between
Present (,or active} participle (e.g. lin-ol 'building' ; inawl 'builder')
and Past participle (e.g. lin-eal 'built, having built').
This concept ion is rooted in Armenian tradition. In Modern Arm., indeed, the pres. participle in -oy
(< -ol, -awl) is an integral part of the verb system. The Armenian grammarians, therefore, do not
hesitate to assign the same function to its classical antecedent (Abrahamyan, 275-278). This
assumption, however, is by no means cogent, since the modem verb system does not exactly reflect
the classical one.
The question therefore arises, whether the presumed participles really belong to the conjugation,
or whether they have to be classified among the verb derivatives (act ion nouns, agent nouns, adjectives,
etc.). A reliable criterion lies in the relative frequency of -eal and -ol (-awl) in [Link] texts. To take
an example at random, in ?.lark 1-4 there are only four occurrences of -ol (-awl) : karol 'able' (1.40
and 45; 4.33); [Link] 'sower' (4.3), as against fifty occurrences of -eal (not including the adver
bialized part. darjeal 'turning back > again'}. ?.ieillet's view, therefore, proves right: 5th century
Armenian had only one genuine participle (Meillet, 1913, 128). As to -ol, -awl, they were still [Link]
noun morphemes, like -ic' (Cf. 3.132). The further development, resulting in the }lodArm. present
participle, can be paralleled with that of -ac (?.fann, 1968, p. 139-140), though it began earlier: in
Movses Xorena.c'i's work, an agent noun in -ol occurs twice with a direct object (III, ch. 35 and 67,
p. 469 and 566 of the Venice ed., 1881).
NOTE.

3.21. The indicative and the subjunctive stand in free contrast to each other
in dependent and independent clauses, fairly as they do in Latin. Roughly speaking,
the indicative is used to mean what does or did happen; the subjunctive, to denote
what may happen or is expected to take place, as a purpose, a wish, a demand
or a prospect. The latter mood, therefore, is somehow related to the future ; an~
since the indicative lacks a future tense, the subjunctive - especially the aorist

38

Morphology

[3.21-3.22)

subj. - is frequently used to translate the Greek future. Hence the denomination
of w"lwnfip (future), applied by most Armenian grammarians to the aor. subj., and
their assigning it to the indicative. Abrahamya.n does not go so far. In his terms,
the subjunctive has no present and no aorist, but only two future tenses : 1st
and 2nd w"lwniip ( 191). To this terminology it can be objected that the term
'future' does not cover the whole range of occurrence of the subj., as [Link]
himself acknowledges; nor is the subj. the only expression of futurity : the Greek
future may also be translated by the pres. ind., e.g. when emphasis is involved,
as in: oc' hawatam 'I will not believe' (Joh. 20.25). The verbal adjectives in -loc',
on the other hand, provide periphrastic forms with a connotation of necessity :
Du es or galoc'-n es 1 'Art thou He who is to come 1' (Luke 7.20). Therefore we do
well to keep the harmless, but not misleading, term 'subjunctive' 25 , and to [Link]
'future', insofar at least as we [Link] concerned with morphology.
3.22. In ClArm., as in several IE languages ([Link], Germanic, Slavic, Persian),
the verb system has evolved towards a two stem paradigm : every verb form
pertains to either the present or the aorist stem. As a rule, one stem at least,
or even both, contain a suffix which is inserted between the root and the inflectional
vowel (V) in the present, between the root and the ending (E) in the aorist. Thus,
E) there are
[Link] with simple presents (Root V) and root aorists (Root
presents with various stem extensions (-an-, -n-, -c'-, -nc'-), and -(V)c'- aorists 2e
Examples:

Aorist stem ext,ended :

p'orj-em 'I tempt'
hay-im 'I look'
as-em 'I say'
ors-am 'I hunt'
yus-am 'I hope'

a)

Present stem extended :

pag-an-em 'I kiss'
hec-an-im 'I ride'
p'ax-c'-i1n 'I flee'

b)

aor. p'orj-ec'-i
hay-ec'-ay
as-ac <-1
<
ors-ac -1
yus-ac'-ay, etc.
pag-1
hec-ay
p'axe-ay (2.213), etc.

Mann (1948, p. 47) prefers 'potential'.

26 The morpheme division in the examples is somewhat arbitrary : the conjugation vowel in the
present ought to be separated from the personal ending (Cf. 5.4). In some extended present stems, it
proves to have been part of the suffix (daf-na-m, l-nu-m), as still appeal'll in such doublets as
mat-nu-mfmat-l'i-m 'I approach' ; and in monosyllabic presents (gam 'I come'. kam 'I stand', lam 'I
cry', tam 'I give'), a is the radical vowel.
25

[ 3.22-3.221)

c)

Both stems extended :

mot-an-am 'I forget'
1-n-um 'I fill'

Morphology

39

. ' -ay
mor-ac
1-c'-i, 3rd sg. elic' (root li-), etc.

The stem contrast is neutralized when a simple present is matched by a root

aorist:
han-em 'I draw'
nst-im 'I sit down'
t'ol-um 'I let, remit'

aor. han-i
nst-ay (root nist-)
t'ol-i 27

Only in the last instance (-um presents) is the root aorist predictable (3.253 e).
Simple -em or -im presents, as a rule, are paired with -c'- aorists.
Some anomalous verbs display a root contrast : fJmpem 'I drink', aor. arb-i;
ert'am 'I go', aor. c'ogay (in the indicative only : 3.221).
3.221. The paradigm, therefore, consists of two sets of forms in the following
distribution :

Present stem
Indicative present
Indicative imperfect
Subjunctive present
Prohibitive
Infinitive

Aorist stem
Indicative aorist
Subjunctive aorist
Imperative
(Participle).

The unique exception is the anomalous verb ert'am 'I go' : aside from the aor. ind.
c'og-ay (or c'ok'-ay), all the verb forms in the second column are built up on the
' ... ; imper.
.
root erl.- : aor. sub.J erl-aye,' ert.
-ices
ert' , p l . ert-ayk' ; part. ert-ea.l
The brackets mean that the participle, though mostly formed from the [Link].
stem, may lack the -ec'- suffix when it is related to a simple -em or -im present:
: part. gr-eal
gr-em 'I write' : (aor. gr-ec'-i)
bazm-im 'I sit' : (aor. bazm-ec'-ay) : [Link]. bazm-eal, etc.
as against:
hay-im 'I look' : aor. hay-ec'-ay : part.
koc'-em 'I call' : aor. koc'-ec'-i

hay-ec'-eal
koc'-ec'-eal (Luke 19.13)
t
: par { koc'eal (Luke 19.15)

For details, see Meillet, 1913, 96-97; Jensen, 272-273.

Hence, regularly, the imperative t'ol, pl. t'[Link]', which Jeruien mistakenly assigns to the present
( 245).
27

40

Morphology

(3.221-3.223)

From hanem, nstim, t'olum, etc. (3.22), the participle is, expectedly: han-eal,
nst-eal, t' ol-eal, etc.
A& pointed out above (2.213, cf. 2.24$), root [Link] in -eay [Link] built up on -i- stems: p'axeay
'I fled'< *p'axi-ay; sarteay 'I startled'< *aarti-ay, etc. Participles [Link] not derivable from such
stems, as they are from [Link]. stems in -u- (tu-eal 'given'; erdueal 'having sworn'). \[Link] the reason
of this discrepancy, the [Link] participles [Link] supplemented by those of the corresponding causative
verbs (3.244), e.g. :
p'ax-w;'.eal (from p'ax-uc'an-em 'I put to flight, chase') 'having fled' (as if from p'axc'im 'I flee').
NOTE.

The contrast between present and aorist is not explainable in terms

of tense. In the indicative, both the imperfect and the aorist are past tenses,
alternately used in narratives; and in the subjunctive, the present and the aorist
occur in similar environments. A striking instance is Luke 13.30, where the same
Greek future (esontai 'they shall be') is translated by elic'in (aor.) and linic'in
(pres.). Yet the contrast is by no means a merely formal one: it involves what
is currently termed 'aspect', i.e. the expression of an event with or without regard
to its duration. In English, e.g., the contrast of simple vs. continuous tenses
pertains to this category, and so very likely does the ClArm. contrast, although
no definitive statement can be ~ade, as yet, about its exa~t bearing. According
to Meillet, the present verb forms, including the impf. ind., denote an action as
developing ,while the aorist forms point to its completion (Meillet, 1913, 120;
cf. 1962, p. 93-104. But see also Jensen, 296). At any rate, the solution of the
problem depends on more comprehensive studies than have been made so far 2 8.
Incidentally, it is worth remarking that ClArn1. has hardly any trace of prefixes
used as aspect markers, as there are in Slavic and other IE languages. In view
of the sparse compound verbs that occur in classical literature (gathered in Meillet,
1962, p. 114-120), there is no deciding \vhether .this formation failed to develop
in PA., or whether the extant examples are survivals.
3.222.

The verb 'to be' has no aorist, and its paradigm is restricted to the
indicative and subjunctive forms, namely :
3.223.

Ind. pres.
em 'I am'
es

e
emk'

impf.
ei 'I was'

e1r
er
eak'

(2.211)

Subj. pres.
ic'em ( < *eyc'em, cf. 2.222) ' I may or
shall be'
ic'es

. e-

IC

ic'emk'

Armenian historians often use the participle a.s a. main verb, e.g. arareal 'having done > he did'.
instead of arar ([Link].) or afner (impf.). To my knowledge, nobody has studied the interplay of these
forms in narratives.
28

41

Morphology

[3.223-3.231)

ek'
en

ic'ek'
ic'en

eik'
.
e1n

Obsolete imper. forms are preserved in the greeting phrase : olf er (pL ek') 'may
you be healthy, hail'.
There is also a defective verb, of which the only usual forms are :
3rd sg.
3rd pl.

Ind. pres.
goy 'he is,
there is'
gon

impf.
goyr 'he was,
there was'

goyin

Infinitive
gol 'to be, to exist'

As to guo' e (< *goyc' e), originally a pres. subj. 3rd sg. 29, it has gone through the
same semantic development as its English equivalent 'maybe'. These defective
paradigms are supplemented by the verb linim 'I become, happen to be', aor. ele
(See below 3.255 c).
3.23. REGULAR INFLECTION. In agreement with the binary structure of the
verb system (3.221), the personal endings in the present group and the aorist
group are different. The stem form, however, is not relevant, so that the same
paradigms will do for simple and extended stems as well ~
pag-an-em is inflected like

hec-an-im
<
gr-ec -1

mor-ac'-ay

In each group,
are derivable.
3.231.

on~

gr-em
hay-im

pag-1
hec-ay, etc.
'

tense can be viewed as the primary one, from which the others

Present group

The primary tense is the pre_s. ind. The inflection vowel (a, e, i or u) combines
with the following endings :

a)

Sg. l
2
3

-m
-s

-(y)

Pl. -mk'
-(y)k'
-n

Concerning the 3rd sg. and the 2nd pL, it must be remembered that:

a, but never after i or u (2.21)

before Corin word end (2.211).

-y- occurs after

-ey- > -e-

As in the following sentence: ayl ew oc' i'e hogi aurb guc'e lueal e mer 'we have never even [Link]
tba.t there may be a Holy Spirit', Act. 19.2.
29

42

Morphology

(3.231)

We thus get four paradigms :

orsam 'I hunt'
orsas
orsay
orsamk'
orsayk'
orsan
b)

grem 'I write'

gres
gre
gremk'
grek'
gren

hayim 'I look'

hayis
hayi
hayimk'
hayik'
hayin

argelum 'I hinder'

argelus
argelu
argelumk'
argeluk'
argelun

The impf. ind. has vocalic endings, except in the 3rd sg. :

Sg. 1
2
3

-i
-ir

Pl. -ak'
-ik'

-(y)r

The inflectional vowel i changes to e, so that the 2nd and the 3rd paradigms
merge: hayei, like grei. On the morpheme juncture (-y- or hiatus), see 2.21-2.212;
notice, however, the peculiar form of the 3rd sg. in the -u- paradigm :
orsayi
orsayir
orsayr
orsayak'
orsayik'
orsayin
c)

grei
greir
grer
greak'
greik'
grein

(hayei)

argelui
argeluir
argeloyr
argeluak'
argeluik'
argeluin

The pres. subj. marker -ye'- (cf. ic'em), 3.223 is followed by the inflection
vowel e or i (u in the fourth paradigm). The endings are those of the pres. ind. :
orsayc'em
orsayc'es
orsayc'e

gric'em
(< *greyc'em)
gric'es
gric'e

hayic'im

argeluc'um

hayic'is
hayic'i

argeluc'us
argeluc'u, etc.

On the pres. subj. in -ayc'-im from intransitive verbs in -am, [Link]. -ac'ay (yu-sayc'im,
from yusam 'I hope'), see Meillet, 1913, 112; Jensen, 289.
The preservation of unstressed i, u (in hayic'im, argeluc'um), if not due to analogy, may
result from the regular development of -iy-. -uy- (Meillet, 1913, p. 22E).
NOTE.

d)

The prohibitive ends in -r in the 2nd sg.; the 2nd pl. coincides with the pres. ind.:
mi orsar
mf orsayk'

e)

mi grer
mf grek'

mi hayir
mf hayik'

mi argelur
mi argeluk'

The infinitive, as pointed out above (3.2) ends in -l. Here again, as in the
imperfect, i changes to e : hayel, like grel; later also hayil; but the oblique
cases always display e :

(3.231-3.233)

GD Abl.

43

Morphology

orsal
orsal-oy

gr el
grel-oy

hayel (-il)
hayel-oy

argelul
argell-oy.

3.232. Aorist group. There [Link] two sets of endings, which can conveniently be
termed active (A) and passive, or rather mediopassive (MP) in view of their
function in most transitive verbs (3.242). Three endings are common to both
sets, namely :
Aor. ind. 1st pl. -ak'
Aor. subj. 1st pl. -c'uk'
Aor. subj. 2nd pl. -fik'.
3.233.
Sg. 1
2
3

The primary tense is the aor. indicative. The endings are :

-i
-er

A.
pl. -ak'

-ek' (-ik')

-in

MP.

-ay
-ar
-aw

-ak'
-ayk' (-aruk')
-an

The paradigms show the perfect parallelism of root aorists (argel-i, nst-ay) and
-(V)c'- aorists (orsac'-i, hayec'-ay) :
argeli
'I hindered'
argeler
argel
argelak'
[Link]'
(-ik')
[Link]

orsac'i
'I hunted'
orsac'er
orsac'
orsa.c'a.k'
orsac'ek'
(-ik')
orsa.c'in

nstay
'I sat'
nstar
nstaw
nsta.k'
[Link]'
(-aruk')
nstan

[Link]'a.y
'I looked'
hayec'ar
[Link]'aw
hayec'a.k'
hayec'[Link]'
(-[Link]')
[Link]'a.n

The 3rd sg. A. often stands in contrast to the other forms :

a)

If the stem is monosyllabic, it takes the e- prefix (augment : 2.213, 2.31) :

(Root aor.)
hani 'I drew'
t'oli 'I let'
haner
t'oler
et'ol
ehan

(-c'- aor.)

kac'i 'I stood'

bac'i 'I opened'
kac'er
bac'er
eka.c'
ebac'
(From kam 'I stand', banam 'I open').

In 5th century literature, the augment never occurs before a vowel : el 'he went
out' ; af 'he took' (later : eaf); arb 'he drank' (later : earb).
b) If the stem is subject to vowel alternation (2.22-2.223), the 3rd sg. displays
the full grade vowel :

44

Morphology

kl-i 'I swallowed'

lc'-i 'I filled'
i]-i ' I came down'
anic-i 'I cursed'
luc-i 'I loosened'
p'axuc'-i 'I chased'
kec'-i 'I lived'
atec'-i 'I hated'

2nd sg. kl-er

lc'-er
v
lJ-er
.
aruc-er
luc-er
p'axuc'-er
kec'-er
atec'-er

(3.233-3.234)

3rd sg. ekul

elic'
-v
eJ
anec
eloyc
p'axoyc' (3.244)
ekeac'
atea.c'

As these last examples belong to the regular -a- conjugation (pres. ind. keani,
ateam), the e/ea contrast needs no particular explanation. But it. also occurs,
unexpectedly enough, in -(V)e'- [Link] related to simple -em presents:
grem ' I write'
: aor. grec'-i
katarem ' I achieve'
katarec'i

3rd sg. greae'

katareae', etc.

The common opinion is that -eae'-, reducing to -ec'- in non final syllables, is the
original form of the suffix (Meillet, 1913, 26; 1936, p. 115-116; J ensen, 43;
Abrahamyan, 229). Yet, Mann may well be right in suspecting that "the 3rd
sing. form of -e- stems (-eae') contains a non historical -a- of uncertain origin"
(1968, p. 44): see below, 5.433.
3.234. The subjunctive aorist has complex ending~, all of which, except the
2nd pl., contain the subj. marker -ie'- (-ye'- after a: cf. 3.231 c). In unstressed
syllables, i is regularly lost :

A.
Sg. 1
2

-ie'
-e'es
-e e-

pl. -e'uk'
-fik'
-e'en

MP.
-aye'
-e'uk'
-e'is (-e'es)
-fik'

-e'i (-e' e)

-e'in (-e'en)

It must be remembered that extended aorist stems, if polysyllabic, change e' to s

before C (e', j: 2.232, 2.341) :
argelic'
argelc'es
argelc'e
argelc'uk'

orsac'ic'
orsasc'es
orsasc'e
orsasc'uk'

nstayc'
nstc'is
nstc'i
nstc'uk'

hayec'ayc'
hayesc'is
hayesc'i
hayesc'uk', etc.

The bracketed A. endings in the MP. paradigm are peculiar to some anomalous
verbs (3.255 b), and to the subj. of i- stem aorists (3.221, Note). So, from p'auay
' I fled', the subj. is : p'auaye', p'axic'es, p'axic' e, 3rd pl. p'axic'en.
The aor. ind. obviou11ly contains an -a- morpheme of l\fP. inflection: -a-y, -a-r, -a-w, etc.
In the subj., this morpheme occurs in the 1st sg. only: -a-ye', so that the paradigm looks somewhat
NOTB.

(3.234-3.235)

45

Morphology

unbalanced. This situation seems to result from a partial remodelling of the MP. inflection. Such
passive forms as argelc'u 'thou willst be hindered', argelc'i, argelc'in may have been coined after
argelc'e8, argelc'e, argelc'en : here as elsewhere (3.241) the e/i alternation is used to contrast active
and mediopassive.. If this holds true, some peculiarities in the subj. of i- stem aorists turn out to be
archaic features. The paradigm of p'aa:eayc' displays active endings (c'e8, -c'e, -e'en), together with
an unstressed i (p'=ic'e8, p'=ic'e) reflecting earlier -ey- (2.222). Now, in this position. (i.e. word
internally before a consonant), -ey- is likely to have issued from -eay- (2.223). So the original inflection
appears to have been :
(
(.
(.
c

c
, _
p axeayc , p axeayc e8, p axeayc e ...
thus preserving the a morpheme t_h roughout the paradigm, while the creation of the new forms
entailed its e1imination from the normal inflection (argelc'u, argelc'i ... ). (Godel, 1970a, p. 4-6)'.

3.235. The imperative endings are :

.

A.
2nd sg.

-f:J

pl. -ek'

-ir /-f:J

MP.
-aruk' (--ayk')

Let us begin with the mediopassive -ir ending. It occurs

1) constantly in verbs with extended present stem (3.22 b) :
hecanim 'I ride' aor. hec-ay
darnam ' I turn'
darj-ay
p'axc'im 'I .flee'
p'axeay
.zgenum ' I dress'
zgec'-ay
2)

imper. hec-ir
darj-ir

p ax-ir

zgec-Ir

pl. hec-aruk'
darj-aruk'
p'axeruk'
zgec'-aruk'

optionally in verbs of the -anam ([Link]. -ac'ay) class (3.251 d) :

imanam 'I understand'
moranam 'I forget'

imac'-ay

-ay
morac

imac'-ir/ima
morac'ir/mora, etc.

In all other instances, the 2nd sg. is identical to the aorist stem, except that in
polysyllabic -c' - aorists (3.22 a and c) :
a)

-ac'

reduces to -a :

asem 'I say'

aor. asac'-i
hawatam 'I believe' hawatac'-i
yusam ' I hope'
yusac'-ay

Imper. asa
hawata
yusa

pl. asac'ek'
hawatac'ek'
yusac'-aruk'
etc.

..

b) -ec' changes to -ea in the active, to -eac'" in the mediopassive inflection

(cf. 3.233. last pa:ragraph) :
grem 'I write'
hayim 'I look'

grec-I
hayec'-ay

gr ea
hayeac'

pl. grec'ek'
hayec'aruk',
etc.

46

Morphology

(3.235-3.24)

Monosyllabic -c'- stems remain unaltered :

lam 'I cry'
keam 'I live'
banam 'I open'
lnum 'I fill'

lac'-i
kec'-i
bac'-i
lc'-i

lac'
keac'
bac'
lie'

lac'ek'
kec'ek'
bac'ek'
le' ek', etc.

and so do all root aorist stems :

argelum 'I hinder' [Link]. argel-i
han-i
hanem 'I draw'
barnam 'I lift'
barj-i
ark-i
arkanem ' I throw'
nstim 'I sit down'
nst-ay
NOTE.

imper. argel
ban
barj
ark
nist 30

pl. argelek'
hanek'
barjek'
arkek'
nstaruk', etc.

On the participle, see [Link], 3.221.

3.236. On the cohortative forms : 2nd sg. -ifir, pl. -ifik' (from -em or -im
presents}, and -fir, -fik' (from aorist stems), see Jensen, 243 and 261. Meillet
quotes the former (-ifir, -ifik') in connection with the pres. subjunctive (1913,
105 d, 122), while Abrahamyan describes the latter only under the heading :
cohortative, or future of the imperative ( 267-272).
These forms hold a rather marginal place in the verb system, except -fik', the
normal ending of the 2nd pl. in the aor. subjunctive (3.234). Presumably,- fir.
was first coined to match -fik' : the imperative of some anomalous verbs (e.g.
tur 'give', pl. tuk'; dir 'lay', pl. dik') could serve as a model. Hence, analogically :
[Link]-jir 'hinder' (cf. Lat. prohibeas) : pl. argel-jik'
besides the usual imperative : [Link] (Lat. prohibe) pl. argelek'.
The partial expansion of these endings to the pres. subjunctive bears additional
evidence to our [Link] statement (3.21), namely that the pres. subj. and the
[Link] 'future' belong to the same mood.
3.24.

DIATHESIS (VOICE)

Transitive verbs occur in active and passive clauses. In most inflectional

languages, the voice contrast is expressed by the verb forms, and the passive
conjugation runs parallel, and stands in [Link], to the active. However, one
can conceive of a language in which the same verb forms would convey the active
This example, though unique, is in agreement with the rule formulated above: in view of the
simple -im present mtim, the -ir ending is not expected.
80

47

Morphology

[3.24-3.241)

as well as the passive meaning, the contrast depending on the syntactic environment. This is actually the case of a small group of deponent verbs in Latin, the
so-called 'verba communia' : criminor consulem 'I accuse the consul' /criminor a
consule 'I am accused by the consul'. But in Latin grammar the case is marginal,
while in ClArm. not only do 'common verbs' exist as a particular class of verbs
(3.242 b), but contrastive (i.e. passive vs. active) inflection is balanced by common
inflection in the verb system at large. In the imperfect of all transitive verbs,
e.g., there is only one paradigm for both voices :

patmein z-ban-n Astuacoy i zolowurds Hreic'-n 'they proclaimed the word of

God in the synagogues of the Je,vs', Acts 13.5;
ew and amenayn le:[Link]-n Hreastani patmein bank'-s aysok'ik '!!>nd through
all the hill country of [Link] were all these things told', Luke 1.65.
3.241. In the present group, the expression of the voice contrast depends on
the inflection vowel, in that passive forms are freely derived from -em presents
by substituting i for e (or e). This device is closely connected with the general
distribution of -em and -im presents, the former being mostly transitive (grem,
asem, paganem, hanem, koc'em, kat,arem, etc.), the latter, intransitive (hayim,
bazmim, nstim, p'axc'im, ankanim 'I fall', etc.). The e/i alternation works in the
pres. indicative :
varem (-es, -e ... )'I lead' : varim (-is, -i ... )'I am led, I behave'
ase 'he says'
asi 'it is said'
[Link] 'they tell'
patmin '(things) are told'
eanac'em z'ims-n ew canac'im y-imoc'-n 'I know my own and I am known by
my own', Joh. 10.14
and in the prohibitive :
mi varer (-ek') 'do not lead'

mi varir (-ik') 'do not behave'

It has also been expanded to the subjunctive :

varic'em (-es,

-e ... ) 'I

banayc'em (-es,

shall lead'

s1

-e ... ) 'I shall open'

varic'im (-is, -i ... )'I shall be led,

behave'
banayc'im (-is, -i ...) 'I shall be
opened'.

But in the imperfect, the change of i to e (3.231 b) entails the suppression of the
formal contrast :
ai In translating the CIArm. subjunct ive, I use the future for the sake of simplicity.

48

Morphology

(3.241-3.242.1

A. varem - - -- - - - - .
- -impf. A/P. varei (-eir, -er ... ) 'I led/was led'
P . varim -------~
The other vowels (a, i, u) do not admit substitution, so that common inflection
is the only possible for all transitive presents in -am, -im, -um, including the
pres. subj. in -ic'im, -uc'um :
banam 'I open/am opened'
argelum 'I hinder/am hindered'
hamarim 'I regard/am regarded'
motanam 'I forget/am forgotten''
ant'etnu 'he reads/(something) is read'

(Aor. A. : bac'-i; P. : bac'-ay)

(Aor. A. : argel-i; P. : argel-ay)
(Aor. [Link]. : hamarec'-ay)
(Aor. [Link]. : mofac'-ay)
(Aor. A./P. : ant'erc'-aw)

(l mpf.)
banayi 'I opened/was opened'
argelui 'I hindered/was hindered', etc.
(Subj.)
argeluc'um 'I shall hinder/be hindered'
hamaric'im 'I shall regard/be regarded'
ant'etnuc'u 'he will read/(something) will be read', etc.
3.242. In the aorist group, the voice contrast is expressed by the personal
endings. Passive forms are regularly derived from the corresponding active :
(Ind.)
p'orjec'-i 'I tempted'
p'otjec'-ay 'I was tempted'
patmec'-in 'they told'
patmec'-an '(things) were told'
argel 'he hindered'
argel-aw 'he was hindered'
ziard ebac' z-ac's k'o? 'how did he open thy eyes 1 J oh. 9.26
ziard bac'an ac'k' k'o 1 'how were thy eyes opened 1 Joh. 9.10
(Subj.)
p'orjec'-ic' 'I shall tempt'
argel-c'en 'they will hinder'

p'orjec'-ayc' ' I shall be tempted'

argel-c'in 'they will be hindered'

(I mper.)
srbea 'make clean'
srbeac' 'be cleaned', Math. 8.3
mkrtea 'baptize'
mkrteac' 'be baptized', Acts 22.16
luac'-ek' 'wash'
luac'-aruk' 'wash yourselves', etc.
Exceptions, however, are not wanting.

(3.242-3.244)

a)

llforphology

49

Common endings in the pl. have been pointed out above (3.232) :

(Ind.)
p'orjec'-ak' 'we tempted/were tempted'
(Subj.)
argel-c'uk' 'we shall hinder/be hindered'
p'orjes-Jik' 'you will tempt/be tempted'
b) Besides, there are 'common verbs' (!.wuwpwq pwjbp, Abrahamyan, p. 98), i.e.
transitive verbs following the mediopassive inflection :
hamarec'-ay 'I regarded/was regarded' (from hamarim)
morac'-an 'they forgot/(things) were forgotten' (from mofanam)
ant'erc'-aw 'he read/(something) \Vas read' (from mt'efnum)
As appears from the examples, the corresponding presents end in -im, -anam, or
(seldom) -num.
3.243. To sum up : the ClArm. verb system does not provide a morphological
expression for the voice contrast through all moods and tenses. Contrastive
inflection prevails in the aorist group; common inflection, in the present group.
The impersonal verb forms (infinitive, verbal adjectives, participle) are also
common:

p'orjeal i Satanaye 'tempted by Satan', Luke 4.2/lianeal z-na 'having drawn him',
Luke 4.5
karek' ampel z-bazak-n z-or es ~mpeloc' em, kam z-mkrtut'iwn-n mkrtel z-or 32
es mkrteloc' -n em 1 'Are you able to drink the cup I am to drink, or to be baptized
with the baptism with which I am to be baptized 1', Math. 10.38.
Common inflection, indeed, does not automatically result in ambiguity; and if
necessary, ambiguity can be precluded either by stem substitution (e.g. hel-u 'he
pours' /hel-ani'(something) is poured'), or, more conveniently, by the use of
compound tenses. The latter device is especially resorted to in the imperfect :
bereal linei (lineir, liner ... ) 'I was carried' (instead of berei (-eir, -er .. .) 'I
carried/was carried'). See Jensen, 294; Abrahamyan, p. 222.
3.244. CAUSATIVE VERBS. The incomplete development of the mediopassive
conjugation is somehow compensated by the regular derivation of causative
verbs from intransitive - or less frequently, transitive - verbs. So from

bnakem 'I dwell' : bnakec'uc'anem 'I make to dwell, I establish'.

3 2 z-ml.."Ttut'iun~-n

... z-or : a. [Link] [Link], in imitation of the Greek original.

50

Morphology

[[Link])

Causative verbs are built up on the aor. stem by adding to it the characteristic
morpheme -oyc'- (> -uc'- word internally : 2.222) 33 The resulting stem, recurring
through the whole inflection, can be regarded as the 'basis' of the causative verb :
bnakem 'I dwell',
molorim 'I err'
kam 'I stand'
usanim 'I study, learn'
da:i:nam 'I turn'
p'axc'im 'I flee'
ya:i:nem 'I rise'

aor. bnakec'-i :
molorec'-ay
kac'-i
us-ay
darj-ay
p'axe-ay
yare-ay

(The last two examples show that -ecausative morpheme).

<

caus. vb. basis

bnakec'oyc'molorec'oyc'kac'oyc'usoyc'darjoyc'
p'axoyc'yaroyc'-, etc.

-i- in -eay aorists is lost before the

The present stem has the -an- extension: the aorist stem coincides with the basil'! :
molorec'uc'-anem 'I lead astray'
aor. molorec'uc'i (3rd sg. mokFrec'oyc')
kac'uc'-anem 'I set up, place'
kac'uc'-i (3rd sg. kac'oyc')
usuc'-anem 'I teach'
usuc'i (3rd sg. usoyc')
p'axuc'-anem 'I chase'
p'axuc'-i (3rd sg. p'axoyc')
Causative verbs, therefore, display the same paradigm as, e.g. paganem 'I kiss',
aor. pag-i (3.22 b), and belong to the same verb class (3.253 f). In the 2nd sg. of
the imperative, -oyc' is shortened to -o : bnakec'o 'establish', uso 'teach', etc.
Notice that causative verb bases are never monosyllabic: this accounts, at least,
for the dropping of -c' in bnakec'o, uso as in asa, hawata, grea (3.235).
3.25. A CLASSIFICATION OF ClArm. VERBS. The main division holds between
regular and irregular (or anomalous) verbs; but criteria of regularity have to be
defined. Regular inflection, as described above (3.23-3.236) is certainly one. Yet
it does not suffice : the formal relation of the aorist to the present, too, must be
considered. In this respect, regular verbs are those in which, given the present
form, the aorist stem is predictable. In classifying these verbs, grammarians
rightly start from the different present types.
Presents in which n is the last radical coll80nant must not be mistaken for extended -(a)npresente. Such are, e.g.
serman-em 'I sow', from aermn 'seed' (GDL. serman)
nman-im 'I resemble', from nman 'like, similar'
as against : pag-an-em 'I kiss', hec-an-im 'I ride', etc; or
Jan-am 'I strive', cf. fan 'effort, endeavour' (I. janiw)
spai:n-am 'I threaten', of. [Link] 'menacing'
as against : ba-n-am 'I open', dar-nam 'I turn, return'.
NOTE.

as Exceptionally -oyz- (UZ) or -oya- (-UB-): Meillet, 1913, p. 26; Jensen, 74..

51

Morphology

(3.251 -3.253)

3.251. Regular verb classes with -Vc' - aorist

Pres.
a)

-em

b)

-i1n

c) -am

Aor. stem
p'orjem 'I tempt'
-ec' pastern 'I worship'
bnakem 'I dwell'
-ec' hayim 'I look'
nmanim 'I resemble'
hamarim 'I regard'
-achawatam 'I believe'
yusa.m 'I hope'
Janam 'I strive'

t.

t.

p or3ec -1
pastec'-i
bnakec'-i, etc.
hayec'-ay
nmanec'-ay
hamarec'-ay, etc.
hawatac'-i
yusac'-ay
Janac'-i/-ay, etc.

The aorist has either the active or the mediopassive inflection, or both without
distinction: fanac'i, fanac'ay ' I strove' (Meillet, 1913, 112; Jensen, 289;
Abrahamyan, 243).
d) -a-nam -a-c'-

[Link].m 'I understand' : imac'-a.y

ant'a.c'-a.y
ant'ana.m 'I run'
moia.c'-ay, etc.
[Link] 'I forget'

[Link],
The aorist, as a. rule, follows the [Link] inflection. The voice
however, is sometimes expressed, as in

Zuac'i 'I washed'

luanam 'I wash' /wash myself { luac,ay ,1 wash ed myself'
Presents in -enam (< -i-anam) also belong here, e.g. merjenam 'I come nearer',
[Link]. merjec'ay. Notice that the aorist would fit an -im present as well, which
accounts for such doublets as : yagenam/yagim 'I get satiated'.

3.252. Some exceptions have to be mentioned in connection with a) and b) :

izcem 'I bring', berem 'I bear, bring', [Link] 'I draw', helusem 'I nail, fix', nstim
'I sit down' have root aorists: aci, beri, helusi, nstay (cf. 3.22). On the other hand,
e is substituted by a in asem ' I say', asac'i; gitem ' I know', gitac'i; karem 'I can',
karac'i.
3.253.
e)

-um

Regular verb classes with root aorist

-0-

argelum 'I hinder'

t'olum 'I let, remit'
zenum 'I immolate'

: [Link]-i
t'ol-i
zen-i, etc.

52

Morphology

-anem

-0-

paganem 'I kiss'

tesanem 'I see'
lk'anem 'I leave'
usuc'anem 'I teach'
.
ankanim 'I fall'
g) -anim -0hecanim 'I ride'
usanim 'I learn'
h) -(n)c'im (-e- : 2.213) p'axc'im 'I flee'
hangc'im 'I rest'
martnc'im 'I fight'
f)

(3.253-3.255]

pag-1
tes-i
lk'-i (3rd sg. elik')
usuc'-i (3.244), etc.
ank-ay
hec-ay
us-ay, etc.
p'axe-ay
hange-ay
marte-ay, etc.

On the participles (p'axuc'eal, hanguc'eal, etc.), see 3.221 Note. On the aor.
subjunctive : 3.234.
3.254.

-num

The -num verb class. The aor. stem is not predictable :

-c-

-0-

lnum 'I fill'

ant'efnum 'I read'
arnum 'I take'
jernum 'I warm'
erdnum ' I swear'
(< *erdunum : 2.213,

lc'-i (3rd sg. elic')

ant'erc'-ay
..
ar-1
jer-ay
erdu-ay
2:343).

The prevailing paradigm, however, is :

-num

(-e-)

k'alc'num 'I hunger'

aytnum 'I swell'

k'alc'e-ay
ayt.e-ay . ,

Since -eay aorists also correspond to -c'im presents, doublets do not fail to occur,
e.g. t'ak'num, t'ak'c'im 'I hide', aor. t'ak'eay ; matnuni, matc'im . ' I approach',
mateay, etc.
. 3.255. IRREGULAR VERBS. Not only is the aorist stem u11predictable, but
many .of these verbs partially deviate from the normal inflection pattern. Such
deviations, however, occur in t~e aorist group only : there is no instance of
irregular inflection in the present group. These criteria allow for the following
classification, in which root alternation is regarded as an utmost case of irregular
stem contrast.
a) Irregular stem contrast without inflectional peculiarities
:
.

harkanem 'I strike'

yanc'anem 'I trespass'

aor. har-i
yanc'e-ay

[3.255)

53

Morphology

barj-i
bafnam 'I raise, lift'
dafnam 'I turn, return'
darj-ay
C!l-nac'em ( < *canac'em) 'I know' cane-ay
empem 'I drink'
arb-i
unim 'I take hold, have'
kal-ay

..

Irregular stem contrast with peculiar forms in the imperative and/or the aor.
subju?J,ctive (3.2~4) :
b)

arar-i; 8 4 imper. sg. ara

tar-ay j imper. tar, tarayk'
yare-ay; imper. ari, arik'
c'og-ay; .aor. subj. ert'ayc', ert'ic'es .. .
mel-ay; aor. subj. melayc', melic'es .. .
lu-ay; aor. subj. luayc', luw'es ...
imper. lur, luaruk'
ker-ay (but 3rd sg. eker); imper. ker, kerayk'
aor. subj. kerayc', ker-w'es ...

afnem 'I do, make'

tanim 'I carry'
yafnem 'I rise, resurge'
ert'am ' I go'
melanc'em 'I sin'
lsem 'I hear, listen'
utem 'I eat'
.

c)

Irregular inflection in the aorist group generally :

gam 'I 9ome' tam 'I give'
dnem 'I lay, put' linim
'I become'
.
. .
'

Aor. ind.
eki
ekir
ekn
ekak'
ekik'.
ekin

etu
etur
et
tuak'
etuk'
etun

edi
edir
ed
edak'
edik'
edin

ele
eler
elew
eleak'
elek'
elen

ekic'
eke(s)c'es
eke(s)"c'e

tac'
ta.c'es
tac'e

edic'
dic'es
dic'e

elec'
lic'is (elic'is)
lic'i (elic'i), etc.

ek
[Link]'

tur
tuk'

.dir
dik'

ler (elijir : 3.236)

leruk'. (eleruk')

edeal

leal.(eleal)

'

Aor. subj.

Imper.

Participle

ekeal

tueal

.
The occurrence of the augment (3.233 a) here is not limited to the 3rd sg., but spreads all
over the aorist indicative {save tuak'), and even beyond (edic', edeal, ellc', etc.). In eki, ekir ... it has
coalesced 'vith the root, so that its original prefix character ~n be proved by comparative evidence only.
NOTE.

84

On the substitution of r by 8 in the aor. subj. (araric', arasc'es), see above, 2..232, Note.

..

54

Morphology

(3.265-3.3)

In the above table, the parallel inflection of dnem (root di-)" and linim (root *ky-) appeara at a
glance. On the other hand, the bracketed forms (elic'i8, elic'i, etc.) throw further light on the expansion
of the augment in tkt(a)c'u, tkt(a)c't, etc. These analogous developments, however, did not take place
at the same stage, and the later one (elic'ia, elic'i, along which lic'ia, lic'i) went on so far as to bring
forth a new present : elanim, whioh does not seem to oocur in 5th century literature ([Link],
306); besides, it is not reflected in Modern Armenian. Unfortunately, this obscure competitor has
been exalted by grammarians and armenologists. The result is that the original paradigm, as given
above, is generally misrepreeented in CIArm. grammara: linim is ranged among defective verbs,
while elanim is made the head of a full conjugation (Meillet, 1913, .117; Jensen, 251, 256.3; Mann,
1968, p. 120).

3.26. COMPOUND TENSES. The verb 'to be' (3.223), as an auxiliary verb,
combines with the verbal adjectives in -loc' (3.21) or with the participle.
The meaning of these compound tenses is active or passive (cf. 3.243). The
participle yields 'resultative' tenses, denoting a situation as the result of an
action:
zarmac'eal ein 'they were in a state of amazement', Luke 2.33 (as against:
zarmanayin 'they wondered', Luke 2.47)
t'oleal lic'in k'ez melk' k'o 'thy sins be (definitively) forgiven', Math. 9.2.
If the verb has a direct object, the agent is put in the genitive :
er nora hraman areal i hogwoy srboy 'he had received a promise from the Holy
Spirit', Luke 2.26
oc' ic'e ant'erc'eal jer z-or arar Dawit' 1' 'Have you possibly not read what
David did 1' Math. 12.3.
This construction has been rightly explained by Benveniste (1952) as an application
of the possessive genitive: er nora hraman areal 'he had received a promise' has
the same construction as: afn miof ein erku ordik' 'one man had two sons',
Luke 15.11.
3.3.

WORD FORMATION

Meillet (1913, 28-40) and Jensen ( 57-129, besides, 315) have devoted
extensive chapters to this part of morphology 38, thus sparing us a detailed
enumeration of derivational suffixes and composition patterns. Our description,
more sketchy in this respect, includes some historical comments : word formation,
indeed, is the morphological level that best reflects the background of a language,
so that it can serve as a link between the synchronic and the diachronic approach.
To substantiate our description, we draw upon a small corpus of derived and
compound words from the first chapters of Luke (1-8) and John (1-7).
The preservation of unetressed -i in dic'u, dic'e . . . is perhaps due to analogy. flee however
Meillet, 1913, p. 23 (top) and 105 (i < *-iy-).
35

(3.31-3.312)

Morphology

55

3.31. Both derived and compound words are built up on nouns or verbs, or
rather, on noun or verb stems. This statement implies that we leave aside the
compound verbs, i.e. those to which a preposition is prefixed, as in English forgive,
understand, etc. Indeed, though such verbs do occur in ClArm. - e.g. and-unim
'I receive' (unim 'I take hold, I possess'); z-gnam 'I ramble' (gnam 'I walk, go') they hold a quite marginal place in word formation, and can be neglected without
inconvenience, as mere lexical items (cf. 3.222, last).
Noun stems are mostly identical to the sg. NAcc. form (2.32; 3.13). Verb stem.a
are either roots, or extended -c aorist stems (3.251 c-d; 3.254). Stem vowels are
liable to regular vowel alternation (2.22-2.223).
There are some instances of derivation from phrases :
erkrpag-u 'worshipper', from erkir paganem 'I kiss the earth, worship' ;
zgetnem 'I throw down', from z-getni 'to the ground' (cf. z-k'ari 'against a
stone', Luke 4.11) ;
[Link]-ayn 'presently', from noynfum 'the same hour'.
These derivatives must not be mistaken for compounds : in the ClArm. lexicon,
there are no such words as *pagu, *getnem, *fumayn.
3.311. The members of a compound word, as a rule, are linked together by
the connective vowel -a-, unless the second member begins with a vowel:

c'ar-a-xaws 'slanderer' (c'ar 'malicious'; xawsim 'I speak');

lern-a-kolmn 'hill country' (learn 'mountain'; kolmn 'side, country');
barekam ( < *bari-a-kam) 'benevolent, friend' ( bari 'good'; kamim 'I will, wish'),
etc.
as against : cov-ezr 'seaside'; jkn-Of's 'fisher' (2.332).
3.312. Derivation and composition, as devices of word formation, stand in
contrast to each other. Certain composition patterns, however, have been more
productive than others, owing to the wider, more abstract meaning of the second
member, e.g. -arar 'doing' (aor. stem of afnem, 3.255 a): kendqnarar 'vivifying',
xalalarar 'pacific', barerar 'benefactor', etc.; or -kic' 'associate, fellow, co-' (kc'em
'I add, join') : orsakic' 'fellow hunter'; ulekic' 'fellow traveller' (uli 'way, road'),
etc. Such productivity may result in a morphological shift: the 2nd member,
together with the connective vowel, comes to be interpreted as a derivational
suffix. The most conspicuous example is -aw0'1' (lib) a1 in: zawrawor 'mighty'
See also G. Bolognesi, [Link] armeni I, RicLing 5 (1962), p. 105-127, and on compound words
Meillet's thoroughgoing study (1962, p. 159-184).
87 Remember that the declension of derivative nouns and adjectives is predictable (3.132). The
symbols I, lib, Ill, etc. refer to the paradigms (3.131, 133; 3.141-142).
86

56

Morphology

(3.312-3.32)

(zawr 'might, strength'), p'arawor 'glorious' (p'ark' 'glory') 38, marmnawor 'bodily'
(marmin 'body'), etc.; -wor (< *bhorii- 'bearing') is no more related to berem 'I
bear, bring', as it originally was (4.323).
.A similar, though more recent development accounts for :
,

-astan (IIb) (nouns of lands or places) : Hreastan '[Link]';

-agoyn (III) (comparative or elative) : zawragoyn 'mightier, mightiest' 39;
-(a)pes } (adverbs) : ayspes 'in this way, thus' (ays 'this'); p'ut'apes 'promptly'
-abar
(p'oyt' 'haste, zeal'); brnabar 'forcibly, violently' (burn
'fist, violence').
Neither *stan nor *p ea occur as single words. As to -abar, -agoyn, the original
connection with bark' 'manners, behavior' 3 8, goyn 'colour, kind' seems to have
been no longer perceptible to 5th century speakers. A decisive statement is. of
course not possible: one is confronted here with borderline cases: orsakic', ulekic',
etc. have doubtless originated as compound words (Meillet, 1962, p. 180); yet
the assumption of a derivational suffix -akic' (Meillet, 1913, 34 e; Jensen, 81)
is not objectionable.
3.32. DERIVATIVES would be most conveniently classified under the headings
'verbal' and 'nominal', according to whether the stem is a verb or a noun. This
division can be extended to compound words too, with regard to the 2nd member,
i.e., in Bloomfield's terms, the 'head' of the compound.
One suffix straddles both classes: -ut'iwn (3.141 b) yields quality as well as
action nouns: bazmut'iwn 'multitude' (bazum 'many'); csmartut'iwn 'truth'
(csmarit 'true'); k'ahanayut'iwn 'priesthood' (k'ahanay 'priest'), etc. On the other
hand : p'orjut'ium 'temptation' (p'orjem 'I tempt'); mkrtut'iwn -'baptism' (mkrtem
' I baptize'); yarut'iivn 'resurrection' (yarnem, 3.255 b), etc. The occurrence of a
denominative verb along with a radical noun makes the derivation ambiguous :
e.g. vkayut'iwn 'testimony' may be derived from vkayem 'I testify' as well as
from vkay 'witness'.
Exceptionally, a nominal suffix occurs with a verb stem, e.g. -akan (IIb) in
bazmakan '(one) who is seated' (bazmim 'I sit, am seated'), as against azgakan
'relative', from azg 'kin, nation'; tarekan 'yearly, year, age' (Luke 4.19), from
tari 'year', etc. Conversely, the verbal suffix -un (I) is found in imastun ',vise'
.
l\fany abstract nouns are 'pluralia tantum', i.e. lack tho ag. declension.
3 9 Notice. that zawragoyn, though derived from zawr (I V) 'strength, power', is semantically related
to zawrawor 'mighty'. Cf. mecawor 'possessing greatness, superior' : -awor being normally added to
[Link], not to adjectives, mu stands for mecut'iwn 'greatness'. In such cases the accumulation of
suffixes has been avoided. A special study on this phenomenon would be helpful.
38

[3.32-3.331]

Morphology

57

(imast 'intelligence, sense'), as against sarzun 'movable, moving', from sarzim 'I
move'; zelun 'running over', from zelum 'I pour/overflow'; p'aylun 'bright', from
p'aylem 'I shine', etc. Such deviations from the normal line of derivation do not
really bring into question the principle of classification : the original function of
the suffixes can be determined On historical and [Link]:ve evidence. For want
of such evidence, however, some derivatives have to be left unclassified, so e.g.
those in -ac : harac (I) ',vound' (harkanem, 3,255 a); ararac (I) 'creature, creation'
(afnem, 3.255 a); erkiivlac (Ilb) 'reverent' (erkiwl 'awe, reverence'); gt'ac (Ilb)
'compassionate' (gut' 'compassion'; gt'am 'I pity'). The examples are not
nu1nerous. In Modern Arm., -ac has evolved to a past participle morpheme (Mann,
1968, p. 139-140), but this late development does not allo'v for any decision as
to the origins.~ function of the suffix 4o.
NOTE.

In this connection, a problem worth discussing is that which arises from such word pairs as

grem ' I write'

gorcem 'I perform'
sirem 'I love'
nstim 'I sit down'
p 'aylem 'I shine'
anicanem 'I curse'
harc'anem 'I ask'

gir (I) 'writing, letter'

gore (I) 'work'
ser (I) 'love'
nist (I. III) 'sitting, seat, place'
p 'ayl (III) 'shine, brightness'
aneck' (ill) 'cursing, curse' 38
hare' (III) 'question', etc.

In each pair, the relation between the verb and the vadical action noun 'should be defined in terms
of derivation; but the issue is ambiguous, because the noun stem does not differ from the verb root
(or from the aorist stem, in verbs with a root aorist). A decision is often possible on historical grounds :
thus, gir, aer are found to be prior to the corresponding verbs; grem, airem are denominative, like
vkayem, eto. (3.332 1), whereas the action nouns belonging to the -i- declension are derived from the
verbs, especially from the aorist stem (anic-, hare'-; cf. 'ilnt'ac'k' (ill) 'running, run', which contains
the -e' stem of 'ilnt'ac'ay, 3.251 d). In some instances, however, the antecedent development is not
so clear: on nJJtim, nist, see Godel, 1965, p. 22; on gorcem, gore. Meillet, 1936, p. 105. Anyhow, gore
must have its O from the verb, either by analogical levelling or by actual derivation; for, the PIE
prototype had -e-, as is shown by Gk. (w)ergon, AS. weore, were.

3.33.
3.331.

a)

A survey of usual derivation and composition patterns.

DERIVATIVES FROM VERBS

Causative verbs : 3.244.

b) Action nouns in -umn (3.142 b): katarumn 'accomplishment' (katarem ' I

achieve'); korcanumn 'ruin, destruction' (korcanem ' I destroy' 41); sarzumn 'motion'
(sarzim 'I move'), etc. On ~ut'iwn, see above 3.32,. Less important suffixes are:
40

41

Meillet (1962, p. 181} and Mann (1968, p. 139} are inclined to derive -ae from acem 'I bring, handle'.
Cf. 3.25, Note !

58

Morphology

[3.331-3.332)

-st: imast (Ill) intelligence, sense" (imanam 'I understand'); govest (Ill) 'praise'
(govem 'I praise'). Nouns in -ist, -ust do not follow the -i- declension in the sg.
(3.152) : hangist, GDL. hangstean 'rest, repose' (hangc'im 'I rest'); korust 'loss,
perdition' (kornc'im 'I get lost, perish');
-iwn (3.141 b) : k'rt'mnfiwn 'grumbling' (k'rt'mnfem 'I grumble');
-an(k') (lib) : patuiran 'order' (patuirem 'I command');
-uac (I) : xndruac 'request' (xndrem 'I seek, require'); yaweluac 'abundance'
(yawelum 'I increase'); herjuac 'division, schism' (herjanem 'I cleave');
-urd, -und (3.152) : xorhurd 'thought' (xorhim 'I meditate, think'); cnund 'birth'
progeny' (cnanim 'I bring forth/am born' - a common verb.).
As appears from some of the examples, these derivatives may also denote the
means or the result of an action : p'rkank' 'ransom', beside p'rkut'iwn 'salvation'
(p'rkem 'I save'); kapank' 'fetters' (kapem 'I bind'); zolovurd 'assembly, people',
beside zolovumn 'gathering' (wlovem 'I gather').

c) Agent nouns in -ic', -ol/-awl (lib) (see 3.2, Note): p'rkic' 'saviour'; matnic'
'delator, traitor' (matnem 'I denounce, betray'); anicic' 'curser' (anicanem 'I
curse'), etc.; nmanol 'resembling, like' (nmanim 'I resemble'); arbec'ol 'drunkard'
(arbenam 'I get drunk'); cnawl 'parent' (cnanim), etc.
d) Adjectives in -un (I) : see above, 3.32.
3.332.

DERIVATIVES FROM NOUNS

a) Denominative verbs in -em/-im, seldom -am: vkayem 'I testify' (vkay

'witness'); bzskem 'I cure' (bzisk 'physician'); anuanem 'I name' (anun, 3.141 a);
nawem 'I sail' (naw 'ship'); srbem 'I clean' (surb 'clean, pure; holy'), etc.; yusam
'I hope' (yoys 'hope').
Inchoative verbs in -anam (3.251 d) are mostly derived from adjectives:

olfanam 'I recover' (olf 'sound, healthy'); c'oranam 'I become dry' (c'or 'dry');
nsemanam 'I become dim' (nsem 'dim'); tkaranatn 'I weaken' (tkar 'weak'), etc.
But also : k'ahanayanam 'I become a priest, serve as a priest'; veranam 'I rise,
am extolled' (i ver 'upwards').
b)

Nouns of state or quality in -ut'iwn : see above, 3.32.

c) Other derived nouns: collective nouns, inflected in the sg. only: mardik
(3.152) 'men, people' (mard 'man'); mankti (V) 'children' (manuk 'child'); xozean
'pigs' (xoz). Diminutive nouns in -ak (lib): covak 'lake' (cov 'sea'); eramak 'small
herd' (eram); ordeak '.son' (2.213). Nouns of places or containers in -anoc', -astan,
-aran (lib): zolowrdanoc' 'place of assembly, synagogue' (zolowurd); ganjaran

[3.332-3.334)

Morphology

59

'treasure-house' (ganj 'treasure'). Nouns of trees in' -eni (V) : moreni 'mulberry.
tree'; t'zeni 'fig-tree' (t'uz).
d) Adjectives. Among numerous suffixes, the following at least deserve
mention:
-akan (li b) 'pertaining or belonging to -'. See 3.32: yawitenakan 'eternal'
(yawiteank' 'eternity'); korstakan 'perishable, ruinous' (korust 'perdition').
With fairly the same meaning, but a more or less specialized application :
-ayin (I ), -ean (lib); -eay (III); -in (I). [Link] in -ac'i (I) refer to persons only,
and ar.e often substantivized: I srayelac'i 'Israelite'; Samarac'i 'Samaritan';
p'arisec'i 'Pharisee', etc. Likewise, many derivatives in -ord ((li b): orsord 'hunter'
(ors 'game'); arafnord 'leader' (afajin 'first').
-awor (lib) 'possessing, related to - '. See 3.312: melawor 'sinful, sinner' (melk'
'sin(s)'); erkra1vor 'earthly' (erkir 'earth'), etc.
-elen (li b), -eay (III) 'made of-' : garelen nkanak 'barley loaf', John 6.9 and 13;
xarazan c'uaneay' 'whip of cords', J ohn 2.15.
-agoyn (III) 'more, very'. See 3.312: diwragoyn 'easier, very easy' (diwr 'easy');
k'alc'ragoyn 'sweeter, very sweet' (k'alc'r 'sweet~); yarafagoyn ~previously' (yaraf
'before').
e) Adverbs in -abar, -apes (3.312): -aki; yankarcaki 'suddenly' ( = yankarc);
valvalaki 'at once' (val 'early', reduplicated).
3.333. COMPOUND WORDS with a verb stem as their 2nd member usually
follow the normal -a: declension (lib). Most are agent nouns: akanates 'eyewitness' (tesanem 'I see') 42 ; jknors 'fisher' (orsam 'I hunt'; cf. 2.332); c'araxaws
'slanderer' (xawsim 'I speak') ; sermanac'an 'sower' (c'ane?n ' I scatter, sow') 4 2;
p'oxatu 'changer' (tu-, aor. stem of tam 'I give', 3.255 c); barerar 'benefactor'
(3.312), etc. Some have t he passive meaning : andamal,oyc 'with loosened members,
paralytic' (lucanem 'I loosen'); diwahar 'struck by a demon, demoniac' (harkanem,
3.255 a). Action nouns, such as asxarhagir. 'world-enrollment' (grem 'I write')
are exceptional.
3.334. Compound nouns with a noun as their second member are either
determinative, as : lefnakolmn 'hill country', covezr 'sea-side' (quoted above,
3.311); k'ahanayapet 'chief priest' lit. 'priest master', c'orrordapet 'tetrarch'
(pet 'master, ruler'), etc., or possessive : mecatun 'rich', lit. 'having a large house' ;
aylakerp 'having another shape, transformed'; kisamah 'half dead', lit. 'befallen
42

Here, the let member coincides with stem II (3.141).

60

Morphology

[3.334-3.34)

by a half death'; mecajayn ' having a loud voice' (ase mecajayn 'he says with a
loud voice', Luke 8.28).
Compound~ of the latter class automatically turn privative when the first
member is a negative prefix : anarat 'speckless, pure' (arat 'spot'); anptul 'bearing
no fruit' (ptul 'fruit'); apasnorh 'graceless' (snorh 'grace').
In determinative compounds, the 2nd member preserves its own inflection.
Possessive compounds often follow the i declension, e.g. antpul (I, III), as against
ptul (I); anhawat (III) 'incredulous', as against hawatk' (I) 'faith'.
A few nominal compounds are formed by reduplication, with or without the
connective vowel. This pattern is applied to monosyllables only: mecamec 'very
great'; c'arac'ar 'very wicked'; pespes 'various' (Middle Iranian pes 'shape').
Cf. also the derivative valvalaki, 3.332 e.
3.34. Our corpus, limited though it is, allows for a statement about the
productivity of the patterns of derivation. Barring the repeated occurrences of
the same word, we find that nouns in -ut'iwn amount to 28 % of the total number
of derivatives (about 160); causative verbs to 12 %; denominative verbs in
-anam to 8 '}'0 ; adjectives in -awor to 6.25 %- Several suffixes are represented in
only one or two words, e.g. -iwn, -urd, -und (3.331 b); -anoc', -eni (3.332 c); -in,
-elen (3.332 d); -apes (3.332 e). Some are lacking : -st (3.3~1 b.); -astan, -aran
(3.332 c); -ayin (3.332 d); -abar (3.332 e), but would not fail to appear in a larger
corpus. At any rate, the contrast of such favourite suffixes as -ut'iwn, -umn,
-ai1Jor, -akan, etc. as opposed to less productive ones is significant. On the other
hand, the collective derivates mardik, mankti (3.332 c), though occurring more
frequently than the corresponding plurals (mardk', mankunk'), are isolated: no
parallels would be found in any corpus.
Among, compounds, there are 'learned words', often coined from Greek models.
Composition however, was by no means an artificial literary device: jkncrs
'fisher', barekam 'friend', mecatun 'rich', and many other usual, basic words
plainly show that it did play a [Link] in everyday spoken language .

61

[4)

SECOND PART

A DIACHRONIC APPROACH
.

4. The framework for a historical and comparative study of ClArm. can be

outlined as follows :

* Proto-Indoeuropean (PIE)
I

Ancient Near Eastern

languages ----~

I
I
I

Proto.-Armenian (PA)

I
I

[Link] \Vestern Iranian

Classical Armenian

t
Modern Armenian (including dialects)
Of PIE and PA we have no immediate [Link]. PIE phonology, morphology
and lexicon, however, can be reconstructed to a certain [Link] through the
comparative study of the extant languages, either dead, such as Avestic, Hittite,
Gothic, etc. or still developing, such as Persian, Greek and the Slavic languages.
All these appear to have issued from a [Link]. proto-language, just as the
Romance languages are divergent developments of Latin. In applying the
comparative method, however, one must bear in mind the well ascertained fact
that the unity of PIE was. a relative one, involving dialectal diversity. Furthermore, recent studies have brought to light traces of successive developments
within the PIE period. It :is now possible, and even necessary, to contrast
archaic and late PIE.
The position of ClArm. among the IE languages is that of a single idiom
apparently destitute of dialecta;l varieties, which does not belong to a gi:oup of
close cognates, such as the Indo-lranian, the Slavie, the Germanic or t he Celtic
group. Its particular affinity with Phrygian, or the Thraco-Phrygian group,
though repeatedly claimed by some comparativis~s, remains . hypothetical, for

62

Sound Change

[4-4.11)

want of sufficient data on the latter languages. Old Phrygian inscriptions, in

particular, have not yielded decisive evidence to this effect, so far.
We therefore have the right to use the term 'Proto-Armenian' in referring to
the development of a. certain PIE dialect from (approximately) 1500 B.C. to
400 A.D. In the course of so many centuries, the language doubtlessly underwent
considerable changes, so that it does not seem entirely advisable to include the
whole period under one denomination. But we have no other choice, since our
data consist, on the one hand, of the knowledge we possess of PIE phonology
[Link] morphology, and on the other hand, of ClArm. texts from the 5th century
onwards. The enormous gap will never be filled.
4.1. The peculiar position of Armenian as an IE language was not perceived
by early comparativists (Bopp, Windischmann, Fr. Miiller, etc.), who mistook
it for a marginal Iranian dialect, until H. [Link], in 1877, asserted its
autonomy on phonological evidence. The issue, it must be granted, was rather
perplexing. The ClArm lexicon is a mixture of inherited PIE (or genuine
Armenian) words, and of loanwords from various languages. [Link] aside the
latest ones, i.e. Greek and Syriac words, the borrowing of which wa.s a consequence
of christianization, we are confronted with two different layers of foreign
words : on the one hand, words of unknown or uncertain origin, presumably
borrowed from ancient Caucasian or Anatolian languages (4.11); on the other
hand, Middle Iranian loans (4.12). The latter are fairly as numerous as Norman
words are in English; more numerous, at least, than the genuine Armenian
words, from which they have to be discriminated. Much work has been done
to this purpose ever since the publication of H iibschn1ann's fundamental book
(1897), but the inventory of the Iranian loans, though considerably furthered,
still awaits completion.
4.11. At this juncture, something has to be said on the historical circumstances
that account for this unusual number of foreign words in the Armenian lexicon.
The Armenian people is supposed to have migrated from the Balkan peninsula
towards Asia Minor. According to Armenian scholars, they had settled, perhaps
as soon as 1500 B.C., in the Hayasa-Azzi land, in the North-eastern quarter of
Anatolia.. Later on, they moved East- [Link] Southwards, [Link] eventually fixed in
Urartu (Armenia). On their long way, as well as in their new home, they must
have had intercourse with various autochthonous and neighbouring nations.
Most of the early loanwords in ClArm., therefore, would have to be traced back
to the languages of those nations. The decipherment of Hurrian tablets and
Urartean inscriptions, however, did not prove very helpful in this respect. As to
the South Caucasian languages, Georgian and Lazo-Mingrelian, they do yield

(4.11:4.121)

Sound Change

63

parallels to a number of obscure Armenian words (Vogt, 1938, 1961); but, since
the earliest Georgian texts do not go farther back than ClArm. literature itself,
the study of these lexical connections does not throw much light on the remote
origin of the words at issue : so far, their only characteristic feature is a negative
one: these words lack etymology.
4.12. At the turn of the 7th century B.C., Armenia became a province of
the Median empire. From that time on, it was uninterruptedly exposed to the
powerful influence of Iranian civilization and culture. Even in the days of
Alexander the Great and of the Diadochs, hellenistic culture did not prevail;
Armenia was never totally drawn into the Greco-Roman world. Later on,
Christianity opened the way to new cultural trends (1.3); but by that time, the
Iranian stamp on the Armenian language was too strong to be obliterated by
a new one.
The massive borrowing of Iranian words seems to have been started by the
foundation of the New Iranian empire (247 B.C.-22.6 A.D.). The phonological
features of these words, indeed, point to their provenance from a NW. Iranian
dialect, Parthian, i.e. from the language of the new monarchs. Even during the
period of self-government (66-387 A.D.), Armenia was ruled by a [Link] of
Parthian origin, the Arsacid (U.p?_wqndifi) dynasty. The Parthian dialect, as can
be expected, supplied words pertaining to government, social organization, law
and religion, but also to everyday life. Several occur in our short text sample
(1.5) : anapat 'desert', patasxani 'answer', asxarh 'world', zamanak 'time', p'aik'
'glory', past(em) 'I worship', astarak 'tower', tacar 'temple', hrestak 'angel',
patuir(em) 'I command', vasn '(on) behalf'. Not only is patasxani an Iranian
word, but the phrase: patasxani tam 'I answer (give an answer)' is itself a loan
translation from Iranian, and so are many formular phrases.
4.121. The Parthian influence was not restricted to lexical items: it also
affected morphology. A striking example is t'agawor 'king' (hence t'agaworut'iwn
'kingdom', Luke 4.5) : the genuine Armenian suffix -awor (3.312) has been
substituted for the 2nd member of a Parthian compound : *tag(a)-bar 'crown
bearing'. The corresponding feminine is reflected in t'agu(r)hi 'queen' 43 The
semantic relation of t'agu(r)hi to t'agauor, and of both to t'ag 'crown', could
not fail to suggest a morphemic analysis : t'ag-uhi. 'crown (bearing) woman'.
.
Thus, a new Armenian morpheme was abstracted from one Iranian loan word,
and propagated afterwards. In Modern Armenian, derivatives in -uhi are .coined
at will, as agent nouns for female beings (e.g. paruhi 'female dancer', tnorenuhi
48

On the Old Iranian forms *taga-bara- 'king', *tiiga-br8ryii 'queen', see Benveniste, 1945, p. 74.

64

Sound Change

[4.121-4.2)

'directress', etc.). The same process of abstraction and propagation was all the
more likely to take place when several Iranian derivatives or compounds of a
given pattern could serve as models. In this 'vay, such derivational morphemes
as -ak, -astan, -aran, -agoyn, -apes {3.332 c-e; cf. 3.312) became an integral part
of ClArm. word formation {cf. Meillet, 1962, p. 182-184) 44
4.122. The inventory of Parthian words in ClArm. is of course a particular
concern for l ranologists. Recent studies in this field, however, result in a somewhat
unwelcome discovery: several items in Hiibschmann's list of genuine Armenian
words have to be barred {Benveniste 1945, 1958, 1964). The number of acknowledged Iranian loans is thus increasing at the expense of the not too large amount
of inherited PIE words.
To quote only one example : mah {IV) 'death', in view of the earlier form mark,
which occurs in Ep'rem's H ymns, was supposed to contain the PIE root
*mer/mor/mr- 'to die', together with mefanim 'I die' and mard 'man (mortal)'.
According to Hiibschmann {1897, p. 472), ma(r)h could be paralleled with Ved.
mrtyu-, Av. m~a()yu 'death' {Cf. OP. uwa--mrSiyu 'suicide'), or else with Goth.
maurpr 'murder' (< *mrtro-). Anyhow, it would be the genuine Armenian reflex
of either *mrtyu- or *mrtro-. But this explanation meets with phonological
difficulties: PIE *-ty- would have yielded an affricate {-c'-) while -d- in PA
*tnardT(o) would probably have dropped, as -j- did in *darjnam {2.343). Pedersen,
{1906, p. 364-367) resorted to a PIE prototype *mrtu- : PA should have had a
*-tu- stem, instead of the *-ti- stem in Lat. mors (< *mrti-), just as it has zgest
{IV) 'dress' as against Lat. vestis 'id.'. The dim sound of the last vowel {u) is made
rt>sponsible for the uncommon change of *-t to -h-, and for the contrast between
mark {< *mrtu-) and mard {I ) 'man' ( < *mrt6-). The argument, however, is
confuted by ard {IV) 'form, attire' < *rtu- (Ved. rtu 'season' ; cf. Gk. artuo 'I
arrange').
Thus, one solution remains: ma(r)h is a loanword from the Parthian dialect,
where Old Iranian *mr8yu- regularly developed to *marh(u) {Bolognesi, 1960,
p. 17-19). This explanation is supported by the [Link] that no reflexes of PIE *mrtyuare found outside Indo-Irania.n.
4.2. The above discussion brought us close to the problems of ETYMOLOGY,
i.e. of the historical explanation of words. If the explanation is to be found
within the lexicon of the language to which the words belong, it will be given
in terms of word formation ; and insofar as regular word forms are at issue, we
44

In telrak 'quaternion; note book', -ak he.a been added to a Greek loanword : tetr

(5.14).

< Gk. tetrM

(4.2-4.22]

Sound Change

65

do not speak of etymology, but of derivation or composition. Such forms, indeed,

are not only explainable, but also predictable, according to the rules of the
language. Besides regular word forms, however, there are in every language
more or less archaic forms, or relics of earlier morphological patterns. Here
etymology comes into play: the detection of obsolete morphemes requires an
analytical and comparative procedure, as will be shown in some examples.
4.21. An archaic pattern of composition survives in mker (l ib) 'companion",
a combination of the preposition ~nd 'with' and of a radical agent noun from
ker-, the aor. stem of utem 'I eat' (3.255 b). The original meaning thus appears
to have been 'eating with (others), messmate'. The pattern was no longer
productive in ClArm. : it may have given way to the -kic' type (3.312).
Obsolete prefixes are, e.g., amb- (Cf. Gk. amphi, OHG. umbi 'all around') in
ambolj (111) 'whole' (Cf. olj (I ) 'sound', afolf 'healthy'); t(i)- in thas (111) 'unripe'
(has 'ripeness'), tkar (li b) 'weak' (kar 'power, ability'). The full form is preserved
in tiezerk (II) 'universe' (ezr 'border, limit') : ti- is parallelable to Lat. de(deformis 'shapeless, ugly'), Welsh di- (dinerth 'weak') 45.
Passing on to derivation, besides mecanam, mecut'iwn and other regular
derivatives of mec 'large, great', we find mecarem ' I magnify, honor' : doubtless,
a denominative (3.332 a) from an obsolete adjective *mecar. The same morpheme,
indeed, is easily detected in ardar (1) 'just, upright' (ard 'form, fashion'); dalar (I)
'fresh, green' (del 'medicine', formerly 'herb'; delin 'yellow'); galar (111) 'twisted,
curved' (Cf. gelu1n 'I twist, press'; glem 'I roll'). The vowel alternation in
gelum/galar, del/dalar (4.323), as well as the occurrence of similar derivatives in
ancient Greek (miar6s 'defiled', sobar6s 'haughty', etc.), point to the PIE origin
of the pattern.
Besides -ar, ClArm. has also traces of -or: molor, rnolar 'erring, false' (molim
'I err'); glor 'round' (hence glorem 'I roll', cf. glem 'id.'); bekor 'fragment' (bekanem
' I break'); stor 'low, inferior' (~st 'under, beneath'), etc. This morpheme occurs in
the Armenian reflex of a well-known P IE adjective : nor (I) 'new', unlike its
cognates, all of which are traceable to PIE *newo- (Indo-Ir. ndva-, Gk. ne(w)os,
OCSl. now, Lat. novus) or to *newyo- (Goth. niujis, Welsh newydd), reflects a
PA derivative *newor(o).
4.22. These examples, to which more might easily be added, explain the
synchronic contrast between productive and unproductive patterns of derivation
(3.4) : the latter, as a rule, are archaisms, and so is irregular inflection at large.
45

G. Bolognesi, Sul prefisso neyativo t- in anneno. Rivista degli Studi Orieutali XXI II (Roma,
1948), p. 82-86.

66

Sound Change

(4.22)

Archaisms are of great value in diachronic studies. Although they do not enable
us to reconstruct any earlier state of a language in detail, they help us to connect
it with what is [Link] of the proto-language from which it has issued. Thus,
starting from PIE, we can trace phonological and morphological developments
down to ClArm. Let us review, e.g., PIE suffixes which survive, or have left
traces, in 5th century Armenian : derivation will thus appear in a new light.

*iyo/a > -i (V), in noun derivatives: gMci 'tool' (gMc 'work'); harsanik'
'wedding' (harsn 'bride'); kogi 'butter', lit. 'cow (fat)' (kov 'cow').
*-ikya > -ic' (lib}, in agent nouns : 3.2, Note; 3.331 c.
*-omno- > -un (I}, in verbal adjectives : 3.32, 3.331 d (Meillet, 1936, p. 48).
It must be pointed out, however, that a parallel to -ic' is to be found in the Slavic languages
only (?tieillet! 1936, p. 75). In other words, we are confronted here with a Slavic-Armenian 'isogloss'.
Besides, Meillet's explanation of -un did not meet with unrestrained approval.
NOTE.

Very few PIE action nouns in *-ti and *-tu are preserved in ClArm .: bay (Ill}
'verb' ( = Gk. phdtis 'oracle', a poetic word; Attic phdsis 'utterance'); awt' (III)
'a passing the night' (aganim 'I pass the night, stay'. As to the wf g alternation,
cf. kov : kogi, quoted above); ard (IV} 'form'; zgest (IV) 'dress' (zgenum 'I dress,
put on'). Of the latter suffix, there are no other traces; but the former remained
productive in PA, though in expanded forms : -st (-ist, -ust : 3.331 b ), and -oyt',
e.g. in erewoyt' (Ill) 'appearance', from erewim 'I appear'. Later, while -st simply
subsisted in a small number of PA derivatives, -oyt', along with -ut'iwn (Meillet,
1936, p. 80), had a better lot: it has even been revived of late in Eastern
Armenian, to coin technical terms (druyt' 'deposit, thesis' ;jevuyt' 'morpheme', etc.).

*-mon/mrJ--

> ~::n /man-

(3.142 c}, in verb derivatives (action nouns}.

PA examples are : fermn 'heath, fever' (fernum 'I warm'); skizbn 46 'beginning'
(sksanim 'I begin'); erdumn 'oath' (erdnum, 2.543, 3.254); himn 'foundation' (the
corresponding verb is lost}, and some others. In ClArm., the only productive
form is -umn (3.331 b). Of another expanded form: -awn< *-amn there are but
few traces. It must have been still productive in the Parthian period, as is proved
by pastawn 'worship', GDL. pastaman, from [Link] 'I worship' : the verb root
is Iranian (Meillet, 1962, p. 173). A third variety is -iwn < *-imn, e.g. in goc'iwn
'shouting, cry', GDL. goc'man (< *goifiman). All the inflected forms, one
perceives, belong to goc'umn as well. Yet, -iwn did not merge with -umn, and
exceptionally the clash of the sg. NAcc. with the oblique cases has been removed
by a change of inflection, as in k'rt'mnfiwn 'grumbling', GDL. k'rt'mnfean (3.141 b}.
40

With -b

<

-m- after z.

[4.23-4.31)

Sound Change

67

So far, we have simply cast a. glimpse at the PA and PIE background

of ClArm. word formation. This much, however, is enough to prove that Armenian,
crowded as it is with foreign words, is unmistakably an IE language, and a quite
particular one : the developments of *-ti or *-mon/m?J., or even of a single word
like nor 'new', lack parallels in the cognate languages 47 , and so d.o many
morphemes which we cannot trace farther back than PA, e.g. the causative
-oyc'- suffix (3.244). This statement will be confirmed in a diachronic study of
CIArm. inflection. But before we approach this last topic, we have to come back
to etymology, and take up problems we did not discuss in the preceding survey.
4.23.

The genuine Armenian words, we remarked (4.1), are only one component
of the lexicon. How then can they be singled out 1 As inherited words, they are
of course traceable to PIE prototypes; but, since the same holds true of most
Iranian words, too, an additional criterion is needed. Genuine Armenian words,
when confronted to their cognates in other IE languages, display quite particular
phonological features : e.g. hayr 'father' reflects the same prototype as Skt. pita
(pitar-), Gk. pater, Lat. pater, Goth. fadar, etc. In this connection, one must bear
in mind that outward likeness is casual, and often misleading. Doubtless, Arm.
naw 'ship' can be safely equated to Gk. na'l1s, Lat. nauis, Oirish nau 'id', but
this is the exceptional case. The genuineness criterion, therefore, lies in rules of
sound change, which have to be discovered, and then operated with due regard
to morphology and semantics. Beside hayr, e.g., we find :
4.3.

Arm. hing 'five' : Skt. pafica, Gk. pente, Goth. fimf

hur 'fire' : Hitt. pa'!J'!!-ur, Gk. pU.r, Umbr. pir, AS. fyr
hast (III) 'firm' : Goth. fasti
heru '[Link] year' : Gk. perusi, Skt. parUt,
all with Arm. h- as against p- (Germanic f-). This situation can be construed in
terms of sound change: PIE *p(V)- > Arm. h(V)-. Such a rule is only tentative,
until it has been tested and checked on more parallels; and then it may turn out
inaccurate, and have to be corrected.
Armenian, in this respect, is a rather hard case. First of all, more
n1aterials would be welcome. Hiibschmann's list of genuine Armenian words
amounts to 438 items, many of which, however, are declared uncertain. Since
then, the number of reliable Armenian etymologies has been slightly on the
increase, while Hiibschmann's inventory has had to be reduced (4.122). This
4.31.

I would not equate nor with Gk. ne(w)ar~, as Meillet does (1936, p. 50): -ar ( < -aro-), unlike
-ar ( < *-Oro-). he.a no counterpart in Greek.
47

68

Sound Change

(4.31-4.314)

state of affairs is by no means surprising : one cannot expect a very large number
of PIE words or stems to survive in such a language as ClArm., in view of the
circumstances that attended its perpetuation (4.11-12), and of the late appearance
of written evidence. Yet, the scantiness of materials is the main obstacle to the
discovery and verification of certain rules of sound change, which remain unknown
or controversial for want of decisive data.
4.311. Indeed, the particulars of sound change in Armenian appear to be
more intricate than in other IE languages. It often happens that, in seeking an
etymology, one has to contemplate more than one possible development. Initial
h- e.g., is not only the Armenian reflex of PIE *p-, as in hayr, hing, hur, hast;
it also occurs in hin (I) 'old' ( < *seno-), hol (I) 'ground' ( = Lat. solum ?). On the
other hand, it seems to have developed in PA before an initial vowel, e.g. in hot
(I ) 'odour' (Cf. od~, Lat. odor), hum (I) 'raw' ( = Skt. ama-, Gk. omos), unless,
as some comparativists are inclined to believe, It- is the reflex of a PIE laryngeal
(4.322). Conversely, there is no trace of *p- in otn 'foot' (cf. het (I) 'foot print'),
nor of *s- in al (III) 'salt', ewt'n 'seven'. To account for this situation, as well
as for such 'doublets' as hogi/ogi 'spirit', some would appeal to the hypothesis
of dialectal features (so Bolognesi, 1954, p. 150). But the ClArm. texts, unlike
Ancient Greek literature, do not afford palpable evidence for dialectal diversity.
4.312. The dropping of unstressed vowels (2.221), too, raises difficulties, in
that radical vowels are often obscured. Thus, mnam 'I remain' may derive from
*men-a- or *m en-a- (through *minam), since PIE e and e .have merged to Arm.
i before a nasal consonant (Meillet, 1936, p. 48; Mann, 1963, p. 11). But *mon-awould suit as well, for, in the same position, *o changed to u, which was to drop
later. Thus, mnam may well belong to the same sub-class of -am presents as
dolam ' I tremble', hogam ' I care', orsam ' I hunt', solam 'I creep', etc .
.

4.313. In comparative studies, besides regular sound changes, one is confronted

with isolated instances of abnormous developments. Such exceptions are all the
more embarrassing 'vhen the rules of sound change are partially disputable, as
is the case of Armenian. We should expect such old \vords as akn 'eye', unkn 'ear'
(together with the pl. forms), atamn 'tooth', leard 'liver', etc. to help on the
discovery of regular sound change. Instead, one has to try ad hoc explanations
in order to bring them into agreement with the prototypes, and in spite of many
attempts, they still stand out as phonological riddles.
4.314. Theoretically, the explanation of a \VOrd, as we defined etymology (4.2),
ought to be exhaustive, and to account for each phonological and morphological
particular. P ractically, however, this requirement is seldom fulfilled. In paralleling

[4.314-4.321)

Sound Change

69

Arm. naw 'ship' with its cognates (4.3), we considered the word stem only; for,
the Armenian inflection (Ilb or IV) does not reflect the original one. The reverse
is seen, e.g., in luc (I) 'yoke' : while Skt. yugdm, Hitt. yugan, Lat. iugum, AS.
yeoc all point to PIE *yugo-, Arm. l- is amazing. Yet, in this case, the preservation
of the o declension adds weight to the parallel, and someho'v balances the
unexpected phonological development 4 8.
Thus, besides sound change, we also have to record morphological changes,
mostly resulting in a redistribution of PIE materials : so, in our example, the
word stem *nau- on the one hand, and the *-a- and *-u- morphemes on the other.
Likewise, a regular verb form: lnum 'I fill' consists of a root (li-), a present stem
morpheme (-nu-) and a 1st sg. ending (-m). All three components are traceable
to PI E, but the combination is a new one : positing a PIE form *plenumi would
not make sense, because such a form is not reflected in any other IE language,
and consequently has to be regarded as a PA creation.
It is no wonder, therefore, that ClArm. has only scattered relics of PIE root
vowel alternation (4.323). As a rule, roots are invariable 49, as li- in lnum
(< *linum), lir (Ill) 'fulness'; moreover, they are very often identical to single
word stems : li (I) 'full' ; hence : lianam, liut'iwn.
4.32.

SOUND CHANGE

In describing sound change, or linguistic changes generally, we are free to

proceed forwards or backwards, according as we choose as our starting point
the proto-language, or the language we are concerned with. The former procedure
is the usual one and seems more adequate for didactic purposes. Consequently
we need, first of all, an outline of THE PIE SOUND PATTERN. Here, the question
arises, whether we have to go as far back as 'archaic PIE', or whether we may
limit ourselves to 'late PIE' (4). Insofar as we are dealing 'vith ClArm. only, there
are reasons to believe that the latter choice will better suit the purpose. Some
of these reasons can be easily gathered from the following outline.
4.321.
Stops:

PIE CONSONANTS
*p
*t

*b

*bh

*d

*dh

(*ph)
(*th)

This etymology has been rejected of late by N. A. Mkrtc'yan {Hit tite-Armenian Parallels, in
Lraber haaarakakan gitut'yunneri 7 [Erevan, 1970), p. 58-69), who equates Arm. luc with Hitt. luzzi
{-i st-em) 'labour due to the state, forced service'. The original meaning, he believes, was 'servitude'.
Of course, he does not care about the divergent inflections. Moreover, he mistakes a metaphorical
meaning for the primary one.
49 In the acceptation specified above (3.12).
48

70

Sound Change

*k
*k

g
*g

*gh
*gh

(4.321-4.322)

(*kh)

Fricative : *s
ClArm. is one of the few IE languages that bear evidence to the existence of
voiceless aspirates ( *ph, *th, *kh). This set of phonemes admittedly developed in
a late period, and did not spread over the whole IE area.
On the other hand, the contrast between palatal and velar stops ( *k/k, etc.) is
reflected in all IE languages, though not in the same fashion. Armenian belongs
to the so-called sati1m languages, which changed the palatals to fricatives
(*k > s) or affricates ( *g > c), while other I E languages, among 'vhich Celtic,
Germanic, Latin, Greek and Hittite, contrast simple velars to labialized velars :
PI E
Lat.

*k/k
k /kw (written : c/qu)

50

In Armenian, the original contrast is neutralized after u or u- diphthongs ( = Arm.

aio, oy : 4.47). In this position, velars merged with palatals :
PIE

*leukos-*yugo-

>
>

PA

*l,ouko- >
*y/lugo- >

as against
PIE *regos- 'darkness'

ClArm. loys (I) 'light'

luc (I) 'yoke'
ClArm. erek 'evening' (in : z-erek 'on

the evening'; c'-erek 'day-time', lit. 'up to the evening').

4.322. PIE RESONANTS AND VOWELS: *m, *n, *r, *l, together with *i and *u,
were members of a particuliar class of phonemes, each of which occurred in turns
as a consonant or a vowel, according to the phonological environment. In the
usual notation, the allophones are marked off as follows :

*/m/ . m . 1l1 : ipm (or 0 m)

*/r/ r r : rr (or 0 r)
.
.
.
.
1y
* /i/ y l
*/u/ w . u : uw 58
On the other hand, *i and *u share an important feature with the genuine vowels
*e, *o, *a, namely the contrast of long vs. short : *i/i, *u/u, *e/ e, etc. This
According to certain comparativists (Bonfante, Pisani) PA, like some other PIE dialects, should
have had two sets of poetpal&tal stops, namely simple velars (*k) and labialized velars (*k"'). The
problems involved cannot be discui!sed here. See below, 4.334, Note; 4.452, and, for a detailed
discussion: G. R. Solt.a, Palaialiaierung und Labialisiuung. IF 70 (1965), p. 276-315.
51 Instead of y, w, German philologists use j, l'
50

r4.322-4.3231

Sound Change

71

situation, however, is the late reflex of a quite different one. It is currently

admitted that the archaic PIE sound system included laryngeal consonants,
which affected the development of the adjoining vowels : *H 2e > a; *eH i > e;
'*eH2 > a; *iH > i, etc. Recent studies aim at bringing to light traces of
laryngeals in various IE languages, including Armenian 52 But, although the
' laryngeal theory' is widely accepted, comparativists, as yet, hardly agree on the
particulars of its application. So we are justified in keeping up the traditional
'late PIE' vowel pattern: *e/ e, *o/o, *a/a; besides: *i, *u and *~ ( < H). This
amounts to assuming that, in 'late PIE', original lengthened vowels ( * e, *o)
had merged with the outcome of V
laryngeal ( *eH, *oH).

4.323. Vowel alternation. In PIE roots and derivational morphemes, the

normal *e vowel was shifted to *o, or dropped, according to morphological rules.
Adding to this fundamental variation the less frequent lengthening of *e or *o,
we obtain the following set of regular vowel alternants :
*e/ *o/ 0/ (*e/*o)
or rather, of morpheme alternants, e.g.
*men-/*mon-/*mn/ (*men-/*mon-)
*mn' m0 n
In ClArm., the *e/o/fJ alternation is best preserved in the declension of nstems (3.141-143). Only in this case is it still instrumental on the morphological
level : the stem variation -in/-un/-an reflects *-en/-on/-r,i(0 n ); in the I ., -amb < *-tibhi (Cf. Meillet, 1936, p. 78-80). The knowledge of the PIE alternation also enables
us to bring together

berem 'I bring'/ -(a)wor (3.312)/ bard (III) 'heap'

< *blter-e< *bhor-ii*bhr-ti
The *e grade is missing in: ofogem 'I irrigate' ( < *srow-e-)/afu (I, III) 'brook'
(< *sru-ti-); c'ogay ' I went'(< *kyow-a-)/c'u(I ) 'start, departure'(< *l11u-ti-) 53 ;
the *o grade in: mefanim/mard (4.122); del/dalar; gelum/galar (-al-< ot--; 4.21).
To this last pair we may add: glem 'I roll'. The root vowel is ambiguous (4.312);
but, whether we assume *gilem or *gulem as the earlier form, it certainly reflects
a present stem with a lengthened vowel (*e or *o).
Evidence for Laryngwu, ed. by W. Winter, Austin, The University of Texas, 1960 (on Armenian :
p. 27-41); W. Cowgill, Evidence .for Laryngeals. The Hague, 1965.
53 A typical example of morphological change (4.314) : the original i inflection, preserved in bard,
bay, au;t' (4.22), has been substituted by the o inflection, the latter being felt regular in monosyllables
52

of the C + i or u type (3.132).

72

Sound Change

[4.33-4.331 J

We need not dwell upon the e(a, o)/'J alternation, which, in the light of the laryngeal theory,
is currently explained as a particular instance of the general phenomenon. Indeed, it does not seem
to have left traces in CIA rm., except in the inflection of the verb 'to give' (3.255 c) : tam 'I give' displays
the zero grade (ta < a~-) as against tu. < do- in the aorist (etu) and in tur (I) 'gift'. On edi, edak',
s~e 5.432, Note.
NOTE.

4.33. As pointed out above (4.31), the Armenian developments of PIE

phonemes are not thoroughly known. Even if we disregard a few puzzling words
(4.313), and deal with normal sound change only, as we shall purposely do, we
must own that the evidence for certain rules is rather lean and leaves room enough
for such bold etymological speculation as abounds, e.g., in Mann's historical
phonology (1963). But no amount of speculation carries weight against one
unobjectionable etymology. Furthermore, a change for which there is but little
evidence can be proved regular on the strength of parallel developments.
The PA STRESS smFT has been alluded to in a previous section (2.22). After
the word stress had settled on the penult, the vowel in the last syllable weakened
and eventually was lost, together with the following consonants except l, r, and
probably n (see below, 5.213)) : *esmi > em 'I am'; *ebheret > eber 'he brought';
*genii > kin 'woman'; *mrt6s (sg.N.) > mard 'man', etc. Very likely i dropped
earlier than the other vowels : this would account for the contrast between
-n < *-nti and -nd < *-nte, -nto, -ntii (Bolognesi, 1954, p. 127) 54 I n Latin, too,
i has been lost after a stressed syllable in animal < *animiili, dos < *dotis
'dowry', etc.
4.331. The developments of the PIE vowels and resonants (except *y and *w)
can be summed up as follows :

*e > e (later > i before n, m); *e > i

*o > o (later > u before n, m); *o > u
*a, a,- ~>a
*"i, -i > i.; *u, u- > u.
Unexpectedly, a occurs instead of e in tasn 'ten' ( = Gk. deka, Lat. decem, OHG.
zehan, etc.); calr, GDL. calu 'laughter' ( = Gk. gelos); vat' sun 'sixty', as against
vec' 'six'. Other Armenian words have a as against Gk. or Lat. o :

akn 'eye', pl. ac'k' (3.152) : Gk. 6sse (Hom.), OCSL oCi 'eyes'; Lat. oc-ulus;
Gk. poli6s 'id'
ali-, in alewor 'grey, hoary'
atea1n 'I hate'
: Lat. odium 'hatred', odi 'I hate'.
In drandk' 'doorway' ( < *dur-and ), the i inflection is certainly not the original one: Skt.
'frame', Lat. antae 'doorposts' are *a stems.
$4

ata~

Sound Change

[4.331-4.332]

73

In asr, GDL. asu 'fleece' (3.151 Note) the origin of a is ambiguous: cf. Gk. p6kos
m. and pekos nt. 'id'. These parallels seem undisputable, but the conditions of
the change of *e or *o to a are not quite clear. For a tentative rule, see
Grammont, p. 223-225, 245-248.

*m, n, l, r
*tt" (Om), ti- (On), l (Ol), f (Or) >

No change

am, an, al (al), ar.

PA -(V)mn is preserved after u ( sarzumn) and in monosyllables : himn 'foundation',

kamn 'flail'. Otherwise, m was weakened to w: pastawn, koc'iwn (4.22). Thus,
anun 'name' (with u < *ow) can be traced back to the same prototype as Gk.

6noma.
On the change of -mn to -wn, see also Meillet, 1936, p. 48, and Pedersen, 1905, p. 217-218.
In fact, the matter seems to be more complicated, and our statement fails to account for the preservation
of -mn in ayce-amn 'roe' (cf. aye (Ill} 'goat'), eleamn 'hoar frost', unless these words [Link] compounds;
amn, however, never occurs aa a. single word. The relation of atanin 'tooth' to Gk. odon, G. od6ntos
involves difficulties. Supposing t he Arm. word to be derived from *od'l)I, (cf. Gk. odont-), the PA form
should have been atan(d)-mn; -m-, being preceded by a consonant, was prevented from weakening.
NOTE.

4.332. The Armenian reflexes of PIE consonants are divergent or uniform

according to whether the developments were, or were not, affected by the adjoining
phonemes. Uniform reflexes are peculiar to voiced stops and voiceless aspirates.

>p
*d > t
*g > c
*g > k/ c after u, aw, oy (4.321).

a) PIE voiced stops: *b

Examples:

stipem 'I urge, compel' = Gk. steibo 'I tread, stamp on' 55
tur (I) 'gift, present' = Gk. d0ron, OCSl. daru 'id';
hot (I) 'smell' : cf. Gk. odme, Lat. odor 'id';
sirt (I II) 'heart'< *kerdi-. The root is *kerd- (AS heorte)/*krd- (Lat. cor, G.
cordis). The i stem with the zero grade ( *krdi-) is found in Gk. kardi-a, OCSl.
srudi-ce, and, perhaps, in Arm. sartnum, I startle' (2.343), as in OCSl. sruditi s~
'to become angry' 56.
cunr 'knee' (3.151, Note) = Gk. g6nu, Skt. janu-: cf. Lat. genu, Goth. kniu, etc.
acem 'I bring, handle' = Skt. ajami, Gk. ago, Lat. ago 'I drive';
si; For the sake of brevity, I use'=' when the words, or word stems, reflect the same prototype;
'cf.' when the parallel is limited to the root.
56 Pisani, [Link]. KZ 61 (1934), p. 189.

74

Sound Change

[4.331-4.333]

mec (lib) 'great, big' = Gk. megas; cf. Goth. mikils 'id'; Hitt. mekkis 'many,
much' ( ?) ;
gore (I ) 'work, action' : cf. Gk. (w)ergon, AS. weorc (cf. 3.32, Note).
kin, I. knaw (3.17) 'woman, wife' = OCSl. kna 'woman' ; cf. AS. cw en 'wife',
Goth. qino 'woman, wife'. The pl. kanayk' has the zero grade, like Gk. gunt, G.
gunaik6s; but the alternating inflection, a most archaic feature, is preserved in
Armenian only.
erek 'evening' = Gk. erebos 'dark place, hell', Skt. rajas- 'dimness, region of
,
clouds, atmosphere', Goth. riqiz 'darkness'
ankanim 'I fall' : cf. Goth. sigqan 'to sink'. The Arm. verb is built up on the
zero grade of the root (cf. 5.431).

> p'
> t'
*kh > x

b) PIE voiceless aspirates : *ph

*th
Examples :

p'lanim 'I collapse'; p'ul 'downfall, decrease' : cf. AS. feallan, Lith. pulti 'to
fall'; Gk. sphallo 'I trip up, deceive' ( ?). The root vo,vel is ambiguous (-ul- < *ol
after a labial consonant?);
y-alt' (I , III) 'large' : cf. Skt. prthu-, Gk. platus 'broad, large' s7 ;
xacanem 'I bite' (with c < *dy : 4.352) : cf. Skt. khadami 'I chew, devor';
sxalim 'I err' = Skt. skhaliimi 'I stumble'.
For more examples (with discussions), see Meillet, 1936, p. 34-36.
4.333. The so-called voiced aspirates changed to simple voiced stops (or
affricates) when initial or preceded by a resonant, but weakened when placed
between vowels. The latter alteration took place before the dropping of last
syllable vowels :

*bh- > b;
*(V)bh(V) > w/v
*dh- > d ;
*(V)dh(V) > ?
*gh- > j;
*(V)h(V) > z
*gh- > g/f before e, i;

(Decisive evidence is lacking)

*(V)gh(V)

>

z?/j after u, aw, oy (4.321)

Examples:

berem 'I bring' = Skt. bhariimi, Gk. phero, AS. biru 'I bear'; -(a)wor < *-bhorii
(4.323). After u, w is regularly dropped : zinuor 'armed, soldier', from zen .(IV)
'weapon'. The 3rd sg. [Link]. eber does not invalidate the rule: it has its b from
berem, beri, etc. analogically.
01

The parallel is mine. I assume that y is a prefix, as in y-a/ol 'favourable, convenient', cf. ajol 'id'.

(4.333-4.334]

75

Sound Change

orb (I) 'orphan' = Lat. orlnts 'bereft'; cf. Gk. <>rphan6s 'orphan';
y-awelurn (ar-) 'I increase': cf. Gk. ophello 'id'. On a= Gk. o, cf. 4.331 .
durk' (II) 'gate' = Gk. thitra 'door'; cf. Goth. daur (o stem}, Lat. Joris 'id'.
gind (lib, III) 'ring' = Alb. veth 'ear-ring'; cf. Goth. and AS. windan 'to wind'.
jern 'hand' : cf. Gk. khefr, kher-, Hitt. kessar 'id';
barjr, GDL. barju 'high' = Hitt. parkus; cf. Skt. brhant- 'high', Gaulish
Brigantes 'the exalted ones' ( 1), name of a. people;
lize1n (also lizariem, lizurn) 'I lick' = Gk. leikho, Ol r. ligim; cf. Lat. lingo;
<>Zni (< *<>Zini) 'hedgehog' : cf. Gk. ekhf-nos, OCSl. jezi.
argand (III) 'womb' = OCSl. grQdi 'breast'. On the metathesis (*ghr > rg),
see belo,v, 4,342 ~s;
fer (I) 'fine '[Link]' = Gk. theros 'summer'; ferm (I) 'warm, hot' = Gk.
therm6s 'id'; cf. Lat. forrnits 'hot' (with the o grade, as in Skt. gharma- 'heat');
fil, fit (l ib) 'nerve' = OCSl. zila, Lith. gysla 'vein, sinew'; cf. Lat. filum
'yarn' ( 1);
iz (Ill) 'viper' : Skt. dhi 'snake' ; Gk. ekhis 'viper' and 6phis 'snake'.
The prototype is controversial (Pedersen, 1905, p. 205; Meillet, 1936, p. 75).
4.334. The developments of the simple voiceless, except *fc, [Link] more intricate.
Unlike the voiced aspirates, these phonemes, when preceded by a. resonant (R},
did not undergo the same changes as in word initial, so that three positions, at
least, have to be considered separately. Regarding the third one (intervocalic
position), it must .be remembered that the changes are prior to the loss of last
syllable vowels (4.333).
PIE voiceless stops :

*p- > h-/0;

*t- >t'-;

*(R)p > 1;
*(R)t > d;

*(V)p(V) > w, v
*(V)t(V) > -y- (> 0)/-t'- after aw, oy

*(R)k > g;

*(V)k(V) > -k'-

fc > s
*k-

>

k';

/-s-

after

u, aw, oy (4.321).

Examples:

hun (III 1) 'ford, channel' = OCSl. pQti 'road', Lat. pons (i stem) 'bridge'; cf.
Skt. pantha(n)-/pathi- 'road, path'; Gk. p6ntos 'sea' ( ?) se;
harc'anem 'I ask, seek' = Skt. ncchiimi, Lat. posco 'id' < PIE *prfc-ske(eharc' 'he asked', aor. = Skt. anccha.t 'id', impf.). Cf. OHG forsk6n 'to seek';
Lith. piisti 'to suit'.
58 ~ieiUet'a

examples: gan. (III) 'flogging', mtg (I Ib) 'mist' (1936, p. 28) have to be discarded as
Partbian loanwords (BenYeniste, 1958, p. 60-61).
6t The semantic developments of PIE *pcmlhi- (*pcmleH- fp'l)lH-) have been elucidated by
Benveniste, Word 10 (1954), p. 256-257.

76

Sound Change

(4.334]

The root is *rrefc- ([Link] 'to ask'; Lat. rreces 'request, prayer')/*pofc(OCSI. prositi 'to ask, beg'; Lat. pocus 'suitor')/*prk- (Arm. harsn 'bride, new
married woman, daughter-in-law');
ul (I ) 'kid' = Gk. polos 'foal'; cf. AS. fola 'id', Lat. pullus 'cub, chicken'. Hence
amul (I ) 'sterile, of a female' ( < *~-polo-) eo;
ew 'and, also' = Skt. api 'also, even'; Gk. epi 'upon, besides';
arciw, GDL. arcui 'eagle' = l\faced. argipus 'id'. The by-form arcui (I) can be
equated to Av. 'q'fnifya- 'hawk', Skt. rjipya- 'dashing';
stuar (l ib) 'solid, huge': cf. Lith. stipit,s 'strong'. The Arm. -ar adjective (4.21)
reflects *stipar(o)- through *stiu1ar.
t'af 'stake for drying fruit, roosting perch' 61 = Gk. tarsos 'cheese hurdle'
< *trso-; cf. tarsia 'stake for drying figs'; OHG. darra 'stake for drying fruit'
(< *torsa); Lat. torreo 'I dry, roast'.
The widespread root *ters-/tors--/trs- 'dry, be dry or thirsty' also lies in ClArm.
t'arsamim, t'afamim 'I wither'; ant'afam 'unfading, amaranth';
mard (I) 'man' = Gk. brotos (Hom.) 'mortal, man', Skt. mrta- 'corpse';
bay (I II) 'verb' = Gk. phdtis 'oracle', phasis 'utterance' (4.22);
bere ( < *berey) 'he brings' = Skt. bharati, AS. bired. The 3rd sg. ending *-ti,
always preceded by an inflectional vowel, reduced to -y, which later dropped
after i and u (3.231);
hayr (< *hayir) 'father'= Skt. pitti (pitar-), Gk. patb, Lat. pater, AS. faeder,
Ol r. athir. Intervocalic -y- < *-t- thus appears to have remained unaltered up to
the loss of the last syllable vowels. Word internally, it dropped later, as in the
sg. I. harb < *hayarb(i);
awt' 'a passing the night' : cf. aganim 'I pass the night, stay', Gk. iauo ' I pass
the night'. The action noun *auti- is not reflected elsewhere.
sar (I , III) 'top, summit, peak' = Skt. t;iras- 'head, top'; cf. Gk. kara 'id';
hars-n 'bride' < *prfcen/on- (or *prkna- 1) : see above (harc'anem);
tasn 'ten' = Skt. dat;a, Gk. deka, Lat. decem, OHG. zehan, etc. On a< *e,
see 4.331.
k'ani 'how many 1' : cf. Lat. quantus 'how great, how much';
argel (I , I II) 'obstacle, prison' : cf. Gk. arkeo 'I ward off', Lat. arceo 'I keep off,
contain';
king 'five' (the final vowel is preserved in hnge-tasan 'fifteen') = Skt. panca,
Gk. pente (Aeolic pempe), Lith. penki;
lk'anem ([Link]. lk'i) 'I leave' : cf. Gk. 'leipo ([Link]. elipon), Lat. linquo 'id'; OHG.
lihan 'to lend':; Lith. likti 'to remain, to leave';
so Meillet, 1936, p. 48. For other examples of -m- < -mp-, aee Dum6zil, BSL 39 (1938), p. 241-242.
61 A typical example of an old, genuine Armenian word which does not occur in classical literature.
The original inflection (o ?) is unknown.

Sound Change

(4.334-4.335]

77

dustr 'daughter' < *dhukter- (Lith. [Link]-, OCSl. duster-, Goth. dauluar 'id').
The relation of *dhuktb- to *dhugh ~ter- (Skt. duhitar-, Gk. thugatb) is a controversial issue. See E. P. [Link], J AOS 90 (1970), p. 228-231.
The evidence for t. > t', *k > k' being rather scarce, t he rules were not easily discovered.
Past century philologists admitted that the regular Armenian reflexes were t and k. This view is
revived by Mann (1963, 81 and 113), al though the alleged examples, as Pedersen proved long ago
(1906, p. 372-374, 381), do not stand up to scrutiny. Exceptions, however, cannot be overlooked.
Irregular reflexes do occur in some pronouns : d- < *t in du 'thou ( = Doric Gk. tu, Lat. tu, AS. pii.,
etc.); in da, G. clor-a 'that (person)'; ayd, G. aydr ( < *ayclor) 'that'; -d (2.324). all derived from the
PIE demonstrative stem *lo. In the interrogaHve, pronouns and adverbs, * /.; weakened to h (him
D. of ii 'what ?'), and finally was lost: o 'who?'; zi, I . iw 'what!' (cf. 3.112); ur 'where!'.
Arm. l'ork' 'four' (cf. lortk'-t<uan 'fourteen') somehow reflects the same prototype as Skt. catvdraJ.i,
Doric Gk. tllorea, Lat. quattuor, with o in the penult as against e in Ionic tuaere11, Lit. ketveri, OCSI.
letverll. The difficulty lies in l'. Meillets conjecture (l 'ork' < *ktworea, 1936, p. 54, cf. Grammont,
p. 252-253) is untenable: a *ktw- cluster would have reduced to *tw- > ~:., as in k'arasun 'forty'.
Starting from the normal stem *k"'etwor- (with a labialized velar), one will readily admit a dissimilation
(*k"'etor-). But the difficulty is not removed, unless we assume, as Pisani does (1950, p. 165169) that
*k"', unlike *k, changed to l' before a front vowel: then, the development should have been: kwetqr.
> *l 'e(y)or- > l'or- (Cf. Pedersen, 1906, p. 396). Pisani's assumption, however, gives rise to some
objeotions.
Finally, leard (ill) 'liver' has been paralleled to Skt. yakrt, Lat. iuur (Hiibschmann, p. 452), and
Grammont tries to prove that intervocalic * /.; did regularly vanish (p. 237-239; cf. Winter, 1955).
But he does not succeed in explaining away the [Link] equation : elik' ' he left' = Gk. Bipe.
NOTE.

4.335. PIE *s. Leaving [Link] some clusters (on \vhich see below, 4.343), and
using the symbol N for 'nasal consonant' (norm), we can state the following rules :

> eJ(h 1)
*rs
> r(rs);
*(N)s > s
*(V)s( V) > g
*s(V)-

> f;
*s(N) > N

*sr

Examples:

al (III) 'salt' : cf. Gk. hal-s, Lat. sal, OCSl. soli 'id'
ewt'n 'seven' = Skt. sapta, Gk. hepta, Lat. septem, etc.
hin (I) 'ancient, old' = Ved. sana-, Gk. henos, Oir. sen 'id'; cf. Lat. senex, G.
senis 'old man';
of (III) 'bottom' : cf. Gk. &rrhos 'rump' ( < *orsos), AS. ears '[Link]
t'afamim, t'arsamim 'I wither' (4.334);
k'eri (V) 'maternal uncle', from k'ef-, stem II of k'oyr 'sister' 8 t. Analogous
" L'oncle matemel est done d6sign6 lit~ralement comme 'celui de la soour', d'apres sa soour
qui est la milre de EGO" (E. Benveniste, Le vocabulaire dea in&titutiona i11do-europeenne11 (Paris, 1969),
vol. I , p. 231.
eta

78

Sound Change

[ 4.335-4.341]

derivatives are Skt. svasrtya- 'sister's son, nephew', Lat. sobrinus 'cousin'
(< *swesrino-);
us (I) 'shoulder' = Skt. anp,sa-, Gk. omos ( < *omso- 1); cf. Lat. umerus;
mis (I) 'flesh, meat' = Skt. m<i?JiSa-, OCSl. m{SO 'id'; cf. Lat. membrum 'limb'
(< *memsro- );
gin (I) 'price, value' = Skt. vasna- 'id', Lat. uenum do, uendo ' I sell' ;
um, D of o(v) 'who?' < *kosme (cf. ume-k', D. of o-k' 'someone'): cf. Skt.
kasmai, Umbr. pusme, Goth. hwamma;

nu (I) 'daughter-in -law' = Gk. nu6s < *[Link]-; cf. Lat._ nurus 'id' (u stem);
Skt. snu~ii-, OCSl. snuxa, AS. snoru 'id' (ii stems) ;
k'oyr 'sister' = Skt. svasa (svasar-), Lat. soror, Oir. siur 'id', all derived from
the PIE sg.N. *swesor. The loss of intervocalic -s- and the change of *6 to u
brought about a diphthong, 'vhich further developed like an original *eu (4.47).
It thus appears that intervocalic s \Vas lost at a very early stage of PA.
The paradigm (3.17) faithfully reflects the PIE stem variation: k'ef < *swesr(sg. GDL. a11d AbL) : k'or- < *swesor- (pl.N. and AccL.). The I. k'erb must have
its e from the other oblique cases : *swesrbhi would have yielded *k'arb.

Clusters. So far, '"e have been dealing with the developments of PIE
stops before vowels. But PIE, unlike ClArm. (2.31), had a variety of releasing
clusters, mostly consisting of stop + resonant. Let us consider, firstly, those in
\vhich the resonant was r or l. Here again, voiceless stops altered more seriously
than voiced, or voiced aspirates.
4.34.

Voiceless stop+ r, l in word initial. Insofar as evidence is available,

the stop has been lost; l clusters reduce to l-; r clusters to er- :
4.341.

li (I) 'full' < *pleto- (Lat. im-pletus 'filled') or *pleyo- : Gk. plews (H om.),
pleos. On lnum 'I fill', see 4.314.
erec' (IV) 'elder, priest' < *preisku-: cf. Lat. priscus 'ancient' (o stem), pr'istinus
'former'; also, perhaps, Gk. presbus, Cretan preigus 'old man';
erek' 'three' < *treyes (Skt. traya'IJ,, Lat. tres), Acc. eris < *trins (Skt. trin, Lat.
tris, Goth. f>rins); eri- < *tri- in eream (I) 'three years old ';
lu (declension unknown) 'heard, news, fame' < *klut6- (Skt. 9ruta- 'heard',
Gk. klut6s 'famous', Olr. -cloth 'was heard') or *kluti- (Skt. 9ruti- 'hearing'). The
root is *kleu-/klou-/klu-; ClArm. has reflexes of the zero grade only: lur (1, III)
'hearing, fame'; lsem ( < *lusem), [Link]. luay 'I hear'.
Word internally, *-tr-

>

-wr- :

arawr (I) 'plough' = Gk. arotron, '\Velsh aradr; cf. Lat. ariitrum;

[4.341-4.343)

79

Sound Change

hawr, GDL. of hayr 'father' : cf. the sg.D. forms Skt. pitr~, Gk. pat,ri, Lat.
patri; hawru (lib) 'step-father' : cf. Gk. patros, Lat. patruus 'paternal uncle';
amuri (V) 'unmarried' < *~-putriyo- : cf. Skt. putrd- 'legitimate son or child' 6 2
The loss of w after u is regular : see 3.183 and 4.333, on zinuor 'armed, soldier'.
It is therefore possible to equate ur 'where 1' to Skt. kutra 'id'; the final vowel
is preserved in ure-k' 'somewhere'.
A similar development of *-kr * is presumable, in view of mawruk' (II) 'beard'
< *smokru-, cf. Lit. smakras 'chin', Alb. mjekre 'beard, chin' (Skt. ymd9ru'beard' has -yr- < *-kr-), and of t'ewr (I, Ill) 'crooked, awry', if etymologically
related to t' ek'em 'I bend, shape'.
4.342. Voiced (or voiced aspirate)+ r : Such clusters have been reversed, not
only word internally, but even in word initial. In the latter position, a vowel
(e or a) is prefixed :

birt (lib or III 1) 'rigid, rude' : cf. AS. bit(t)er 'bitter, painful' < *bhidr6-;
k'irtn 'sweat' < *swidren/on- : cf. Gk. h~ros 'id';
surb (I) 'pure, holy' = Skt. 9ubhrd- (Ved.) 'bright'. The Arm. word inay be a
loan from Iranian (Benveniste, 1964, p. 2);
darbin (lib) 'smith' : cf. Lat. Jaber, G. fabri < *dhabhro-;
merj 'near' = Gk. mekhri 'up to' < *meghri;
artawsr, pl. artasuk' 'tear(s)' < *draku-. A slightly different prototype is
reflected in Gk. ddkru; Goth. tagr, AS, tear (< *dakru- or *dakro-);
argand (Ill) 'womb' = 0081. grQdi 'breast';
elbayr 'brother'= Skt. bhrata (bhratar-), [Link], Goth. bropar, Olr. brdthair
'id'. The oblique case stem elbawr- reflects *bhratr-. The substitution of l for r
is also seen in albewr 'spring, fountain' = Gk. phrear '\veil' < *bhrewr .
Evidence for l clusters is missing.

..

4.343. Stops, mostly voiceless, combined wi~h other consonants, too, in

particular with s. Reflexes of such clusters appear in the following words :

*-pn-:
tawn (III) 'feast, holiday' < *dapni-: cf. Olcel. tafn 'victim' ( < *dapno-);
Lat. dap-s 'sacrifice, ritual meal', and probably damnum 'expense', Gk. dapdne 'id';
k'un (I) 'sleep' (v.>ith u from earlier ow)= Skt. svdpna-, Lit. sapnas, Lat. somnus,
Olr. suan 'id', all reflecting PIE *sw6pno-.
62

N. Adontz, Mela11ge8 Emile B<>isac<J., vol. I, Bruxelles, 1937, p. 12.

..

80

Sound Change

(4.343]

*pt- :
t'er (lib or III ~) 'side' : cf. Gk. pter6n 'wing'. Hence : t'eri (V) 'defective
(< lateral, marginal'; )
t'eli (V) 'elm' = Gk. ptelea 'id'.
*-pt- :
ewt'n 'seven' = Skt. sapta, etc. (4.335);
ut' 'eight' (with u from ow) = Elean Gk. opto, an altered form of PIE *okto
(Skt. a~au, Gk. okt6, AS. eaht, etc.). The alteration, in both cases, is due to the
influence of *septm .

*-kt- :
dustr 'daughter'

AS. dohtor, etc. (4.334).

*-kt- :
The development, though not evidenced by comparative data, can be inferred
from such pairs as

alac'em 'I pray'

: alawt'k' (III) 'prayer'
canac'em, aor. caneay 'I know' : canawt' (Ill ) 'notice' (in canawt's tam 'I give
notice')
amac'em 'I am ashamed'
: amawt' (I) 'shame' 6s.
The present stem morpheme is traceable to *-ak-ye- (Cf. Gk. allasso 'I change'
< *allakyo, from allos 'other, else'), so that in the corresponding action nouns
-awt' may regularly reflect *-ak-ti-.

*st (and *sd)

>

st in any position :

sterf (I) 'barren, sterile' < *steryo-: cf. Skt. start- (Ved.), Gk. ste~ra (< *stery~).
Lat. steri-lis 'id' ;
astl 'star' : cf. Gk. aster (and ster-ope 'lightning'), Skt. star-, Lat. stella
(< *stelna), AS. steMra 'star';
z-gest (IV) 'dress' < *westu- : cf. Lat. uestis (4.22);
ost (I) 'branch, twig' = Gk. 6zdos, OHG. ast 'id' < *osdo-.
*sk 84 > c' :
c'elum 'I split, tear' : cf. Lit. skelti 'to split';
hac'i (V) 'ash tree' < *askia : cf. AS. aesc 'id' (< *aski-);
harc'anem 'I ask' = Skt. prcchami, Lat. posco (4.334).
es The i inflection, preserved in alawt' k', canawt', is presumably the original one.
64 The *k/lc contrast seems to be neutralized in contact with *s.

[ 4.343-4.351)

Sound Change

81

*ks > c' :

vec' 'six' < *useks (Pisani, 1951, p. 59): cf. Av. xsvas, Gk. heks, Welsh
chwech < *sweks.
*skh > sx:
sxalim 'I err' : cf. Skt. skhalami ' I stumble' (4.332). In ClArm. the cluster is
disjoined (2.311).
The regular development of ap is uncertain, in spite of Meillet's confidence: his unique
example: 8pafnam 'I threaten', cf. Lat. aperno 'I repel, despise' (1936, p. 32, 35) is questionable on
the semantic side, and so is on the phonological side the tentative equation: p'oyt' (I) 'zeal' = Gk.
8poude 'id' (Hiibschmann, p. 501), rejected by Pedersen (1905, p. 200), but repeated ever since
(cf. Godel, 1970b, p. 148).
NOTE.

4.35. The remaining PIE sounds, *y and *w, were merely allophones of *i,
*u in antevocalic position (4.322). However, while *r, l, m, n did not perceptibly
alter, *y and *w, together with the related clusters ( *ky, *tw, *sw, etc.), underwent various changes, some of which resulted in mergers with the reflexes of
single stops.
4.351. The development of initial *y- is uncertain. Intervocalic *-y- dropped
very early (erek' 'three' < *treyes, 4.341), as did, centuries later, PA *-y- <PIE
*-t- (4.334).
According to Meillet (1936, p. 52), *y > f after a resonant :

sterf (I) < *steryo- (quoted above, 4.343);

olf (I, lib) 'whole, sound' < *olyo- : cf. Olr. uile 'whole';
anurf (I) 'dream' < *oni5ryo- : cf. Gk. 6nar, 6neiros 'id'.
This view, though criticized by Pisani (1950, p. 178-179) and Mann (1963, 147,
Note 2), seems to hold good, and might be supported by additional evidence:
such reduplicated verbs as mrmnfem 'I murmur', plpfam ( < *pulpulfem) 'I
bubble' presumably reflect *-ye- presents, as certainly do Gk. morm'liro, Lat.
bullio; and, since ver (in i ver 'up, upwards') has been traced back to *[Link] (Skt.
upari 'above, upwards', Goth. ufar 'above, over'), verf (I) 'end' may well be
derived from *uperyo- (as to the meaning, cf. Lat. summus 'upmost, last'). The
only contradictory instance 65 is ayl (I) 'other' = Gk allos, Lat. alius, Olr. aile,
Goth. aljis, all,derived from *alyo-. But the reversal of resonant + y may have
been the normal development after a.
es Which Meillet vainly tries to explain away (?llSL 10, 1920, p. 81-82; cf. 1936, p. 90). Notice
that C!Arm. shares with Greek the adverbial use : ayl = Gk. alU 'but'.

82

Sound Change

[4.352-4.353)

Clusters. There is no controversy about the reflexes of *dhy and *ky:

.
mef (I) 'middle' = Skt. m<idhya-, Gk. rnesos, Lat. medius, Goth. midjis < PIE
*medhyo-. On e instead of e, see below, 4.451.
c'ogay 'I went', c'u (I) 'departure' : cf. Skt. cyave 'I move, depart', cyuti'departure'; Gk. se'liomai, aor. essumen 'I rush, run'. Cf. 4.323;
goc'em 'I shout, call' < *wok-ye- : cf. Gk. (w)6ssa (Hom.) 'rumour', voice
(< *wok-y;,), and (w)op- 'voice' (< *wok-), Skt. vac- 'voice, speech', Lat. uax,
G. uOcis 'voice';
ac'k' (II) 'eyes' : OCSl. oCi, Lith. akl, Gk. 6sse (Hom.) are old dual forms, and
Arm. ac'- may reflect a dual NAcc. *okye (Meillet, 1936, p. 52) or *okwi (Pisani,
1950, p. 165). But I would rather appeal to a pl. neuter *okya. Cf. 5.131 (on
mawruk', kelcik').
-ic' < *-ikya- (4.22) ..
4.352.

In view of this evidence, we may safely surmise that *ty, *dy, *fey like'\vise
developed to affricates :

mucanem ' I introduce' < *moud-ye-: cf. mtanem 'I enter, mutk' (III) entering,
entrance'; xacanem 'I bite' < *khad-ye- : cf. Skt. [Link] 'I chew, devor';
luc'anem 'I kindle' < *louk-ye- (rather than *louk-ske-, as Meillet believes :
1936, p. 107); cf. [Link] (I) 'light', 4.321. ee.
4.353. As a rule, *w is reflected by g (through *gw, presumably), except in
word final after a vowel. In this latter position, it appears as -w (-v), thus merging
with the reflex of intervocalic *p or *bh :

gayl (I) 'wolf' = Olr. fail 'id' (< *wailo-);

gtanem ' I find', giwt (III) 'invention': cf. Skt. vinddmi (I find, gain'. The [Link].
3rd sg. egi,t 'he found' = Skt. dvidat 'id' ;
gelum ' I twist, press' : cf. Gk. (w)elustheis 'rolled, coiled up', (w)e'lutron 'covering';
Lat. uoluo 'I roll' ; Goth. af-walwjan 'to roll back'. From the same root : glem
'I roll' (4.323) : cf. OCSl. valiti 'to roll' ;
ofogem_' I irrig~te' < *sro~e- (4.323): cf. Skt. sravami, Gk. rhe(w)o 'I flow'
< *srewe-;
taygr 'husband's brother'= Skt. devcf (devar-), Gk. dair < *daiwb-; cf. Lit.
dieveris 'id';
naw (IIb, IV) 'ship' : cf. Skt. naJ4-, Gk. natls, Lat. nauis, Oir. nau. The word
stem is *naw(V)-, as in the Skt. G. nava'fi, = Gk. ne(w)6s;
I claim to be held rll8poneible for these last ex~mplee, and for the rules of sound change involved
( *dy > c; [y > c'). Cf. Godel, 1965,p. 24-26; 1970b, p. 147.
66

Sound Change

(4.353-4.354)

83

tiw 'day time' (3.17) = Skt. diva- 'heaven' (but L. dive dive 'day by day');
Lat. -duum < *diwo-, in bi- tn-duum 'a space of two (three) days';
arew (IV) 'sun' : cf. Skt. ravi- 'id'. The g alternant is preserved in the compound
areg-akn 'sun', lit. 'sun spring' 67 Likewise in
kogi (V) 'butter' < *gowio/a-, beside kov (IV) 'cow' (Cf. Skt. gau-, Gk. boas,
G. bo(w)6s, AS. ou);
aganim 'I pass the night', beside awt' (III) 'a passing the night' ( < *auti-, 4.334).
But analogy generally prevailed: in naw, arew, kov, the sg. NAcc. form is an
invariable stem (3.13), recurring in derivatives (nawak 'boat') as well as in the
inflected forms (GDL. nawi, nawu, etc.). A trace of the original inflection (arew,
GDL. *aregi) is found in areg 'the eighth month of the Armenian year', an archaic
G. (Benveniste : cf. fn 67).
On k'san 'twenty', see above, 2.31. PA *gisa1~ ii; the expected reflex of PIE *wik'!l'ti (Av.
visaiti, Dorian Gk. wikati, Lat. uiginti).
The rule formulated above, it must be granted, does not cover all the facts. I nitial *w is reflected
by v in vay 'woe!' (Lat. uae, Goth. wai). Such a development, however, is not likely to have taken place
elsewhere (Pedersen, 1905, p. 194-196; Pisani, 1950, p. 184-185). On the other hand, there is no denying
that *w has been lost in such words as :
NOTE.

neard (Ill) sinew, fibre' < *sntwr-ti: cf. Av. 81Uit,.ara 'sinew, band' ; Gk. neilron 'fibre, nerve',
neurd 'bowstring';
nor (1) 'new' < *new-oro- (4.21, last line);
sor (I) 'cave, cavity' < *k6w-oro- : cf. Gk. k6(w)oi 'cavities', Lat. [Link] 'hollow';
erkan (llb, Ill) 'millstone' < *grctwen- : cf. Oir. brau 'id'; Skt. grdvan- 'a stone for pressing Soma';
inn 'nine' < *en'W'IJ : cf. Gk. enaios 'ninth' < *en'W'IJlo-.
It is perhaps worth pointing out that the PIE prototypes, insofar as they can be reconstructed, were
stressed on the first syllable, and to contrast, e.g.
neard
nor

<
<

imtwrti-

\vith taygr

< *daiwer-

*neworo-

Should this hold t rue 68, the loss of *w must be assigned to a remote stage of PA, previous to the
stress shift (4.33).
Finally, in view of Gk. phrlii,r 'well' ( < *bhrewr) and al~ata, aleura (pl. nt.) 'wheat meal', from an
obselete sg. *ale(w)ar, one cannot help tracing back to the same prototypes the Arm. words albewr
'spring, fountain', alewr 'flour'. Is -wr the immediate reflex of -u:r? There is no trace of the original
stem variat ion (*r/n) in the inflected forms: GDL. alber, [Link] (also GDAbl. aliwroy 67). See below, 5.142.

4.354.

Clusters. There is evidence for the following :

Benveniste, REArm II (1965), p. 5-11.

68 Contradictory instances are sin (I ) 'empty, vain', if identical to Gk. ken6a ( < *kenwo-) 'id',
and, on the other hand : cungk' 'knees', whether a reflex of the dual form gonwi (Meillet, 1936, p. 84),
or of a plural NAcc. parallelable to Homeric Gk. go12na.
69 Anyhow, aliwroy can by no means reflect *aleuro-, which would have yielded *aloyr.
67

84

Sound Change

*tw

(4.354)

>

k':
k'e- (oblique case stem of the 2nd sg. pronoun) < *twe-, as against the N.
du < *tu (4.354, Note) : D(Acc) k'ez < *twe-ghi (5.244, last paragraph); I . k'ew
< *twe-bhi. The G. k'o is traceable to *two- (5.245).
k'af- (in k'afasun 'forty', k'af<Yrd 'quarter') < *tior-, a zero grade alternant of
*kwetw<Yr- (4.334, Note).
Afters, the aspiration is missing : oskr 'bone' < *ostwer (Meillet, 1936, p. 51).
*dw- >erk-:
erku 'two' < *dwo: Ved. d(u)vci, Gk. duo (Hom.), OCSI. duva, Lat. duo (with
shortened o); erki- < *dwi- in erkeam 'two years old' = Skt. dvi-, Gk. d(w)i-,
Lat. bi-, in compounds ;
erknc'im, aor. erkeay ' I fear', erkiwl (III) 'fear' contain the zero grade of the
root *dwei-/dwoi-/dwi-: cf. Av. dvae8ii 'menace, frightening thing', Gk. deos
(< *dweyos). etc. 10 ;
erkar (lib) 'long (of time)' : cf. Doric Gk. dar6s ('id' < *dwar6-).
This development, recorded by Meillet (1936, p. 51) and Mann (1963, 99) as the
normal one, is amazing: one would expect k- < *dw-, parallel to k'- < *tw-,
and some evidence for such a reflex has been brought forth, e.g. kul (I) 'fold'
< *dwo-plo- (Pedersen, 1906, p. 398). Indeed, Meillet's description of the
historical process, as well as his etymology of krkin (I) 'double, repeated' are
liable to criticism (Grammont, p. 252, n. 1; Pisani, 1951, p. 54). The evidence.
however, is weighty enough. Word internally, after a resonant, *dw > k: melk
(III) 'soft' < *meldwi- : cf. Skt. mrdu-, f. mrdvt-; Lat. mollis 'id'.

*sw- > *k'-:

k'irtn 'sweat' < *swidren/on- (4.342) ;
k'un (I) 'sleep' < *swopno- (4.343);
k'oyr 'sister' < *swesi>r; k'efi (V) 'maternal uncle' < *swesrio/a- (4.335).
A different (dialectal ?) reflex is found in skesur (lib) 'mother-in-law'. The
Arm. word, like Gk. hekurtt, is derived from *swe!-urti-, not from *swe1Cr1i- (Lat.
socrus, OCSI. svekru, OHG. swigar). The initial cluster is altered in Skt. yvaf?''li-,
with the reflex of *kw instead of *sw; but there is no reason to assume, as Meillet
does (1 936, p. 51), that the same alteration should occur in any other IE language.
Furthermore, /Cw seems to have developed to 8 71 :

sun, GDL. san 'dog' : cf. Skt. y(u)vtt, G. yuna?i,; Gk. kuon, kun6s; Olr. cu, con.
On the whole word set (erku, erknl'im and cognates), see Benveniate, Word 10 {1954), p. 254-255.
71 The issue is discussed by E. Lid6n, Ein Beitrag zur armeniacMn Lautguchichle. Huschardzan,
Wien, 1911, p. 381-388.
10

[ 4.354-4.42)

Sound Change

85

The Arm. NAcc. form reflects *kw&n; the oblique case stem san- is the outcome
of some unknown analogical process;
es (I) 'ass' < *ekwo-, i.e. the PIE word for 'horse' (Skt. ayva-, OP asa-, Lat.
equus, Oir. [Link], AS, eoh). The semantic change, uncommon though it is, can be
accounted for: it may well have been brought about by the promotion of a less
usual, perhaps foreign word: ji (I) 'horse', the only cognate of which is Skt.
haya- 'id'.
4.4. The intricacies of sound change demanded a detailed, piecemeal description, which now has to be supplemented by some observations. Indeed, the
phonological developments, however complicated, have not resulted in a state
of confusion, but in a new system of phonemes and phoneme combinations.
Let us resume the main lines of this evolution, as well as the characteristic
features of its outcome.
4.41. The 'late PIE' vowel pattern (4.322) has been reduced to a six vowel
system without length contrast, except that of e to e towards the end of the
PA period (1.63).
4.42. Prior to the earliest borrowings from Iranian, the PIE stops underwent
a 'consonant shift' resembling the Germanic 'Lautverschiebung'. The resemblance,
however, does not seem to involve any' historical connection. Anyhow, the results
differ perceptibly: in Armenian, the voiced aspirates remained unaltered. This
fact, suspected by Pedersen (1906, p. 336-337), has been convincingly stated by
Vogt and Benveniste (see above, 2.11). The voiceless aspirates, too, did not alter,
except for the change of *kh to x. The Armenian shift, therefore, affected the non
aspirated stops only, and first of all, the voiceless. According to Meillet (1936,
p. 29-30; cf. Pedersen, 1905, p. 205-206), the change originated in aspiration.
But such a development would have entailed a merger of the simple voiceless
with the aspirates. Now, *p and *ph, *k and *kh never merged. .We must rather
assume that the articulation of the simple voiceless began to weaken (Pisani,
1951, p. 68-69) : the alteration of *p and *k went on unimpeded, resulting in
h/w(v) and s. Similarly, *t and *k were going to change to fricatives, let us say :
p and ~ 72 ; but the weakening process was somehow thwarted, so that p (initial
or preceded by a .u diphthong) hardened again to t', and ?1 to k'. By that time,
the original voiced stops ( *b, *d, *g, *g) had been devoiced. Adding to the
These symbols, needleBB to e&y, are arbitrary. By choosing ~. we simply imply that the assumed
fricative was different from both h (the reflex of *p) and x ( < *kh). Sec also below, 5.221.
12

86

Sound Change

[ 4.42-4.44)

resulting stops the affricates, reflecting either single _phonemes or clusters, we

are confronted with a new, perfectly balanced system :

p
p'

t
t'

k
k'
g

c
c'

c
c'
j

4.43. The developments of consonant clusters, various though they were,

converged towards a simplification of syllable structure. In particular, all releasing
clusters (except *st) were eliminated, either by the loss of initial stops (4.341),
or by reversal (4.342), or by the blending of the components (*sk > c'; *ky > c',
etc.). This led to the generalization of the (C)V and (C)VC patterns at a certain
stage of PA. Later, the dropping of unstressed i and u seems to have generated
new clusters (*gisan > k'san 'twenty': 2.31). But the consonant clash was mostly
[Link], as in *lik'i > l( a)k'i 'I left'; *gut' am > g( a)t'am 'I pity', etc., or eventually suppressed, as in k'san > k'(a)san. The origin of the neutral vowel, however, need not be connected with the reduction of non final i and u; but the
problem of its earliest occurrences can hardly be approached as long as the related
synchronic problem (2.3) remains unsolved.
4.44. Vowel prosthesis. The development of a vowel before initial *r is a
feature common to Greek and Armenian. In this case, Ancient Greek displays
usually e-, while in Armenian we also find a- or o- :

erek 'evening' = Gk. erebos 'dark, place, hell' : Goth. riqiz 'darkness;
arew (IV) 'sun' ; Skt. ravi- 'id' ;
orcam (< *orucam) 'I vomit', cf. Gk. ereugomai (aor. ~rugon) 'I belch' : Lit.
riaugeti 'to belch'.
(The root is *reug-/roug- /rug- ; hence the PA verb stem : *(o)rug-a-).
Furthermore, prosthetic vowels also occur before r or f of later origin :

erewim 'I appear, seem' : cf. Gk. prepo 'I am conspicuous, resemble, beseem';
erek' 'three' < *treyes (4.341);
afu (I, Ill) 'brook' < *sruti-;
ofogem (afogem, -anem) 'I irrigate' < *srowe- (4.323).
Did the phenomenon affect nasals, too? In anieanem 'l cunie' (cf. Gk. 6nMoa 'blame,
reproach') 13, ayr 'man, hUllband' (Gk. anir, Skt. nr and ?ira-, Oaco-Umbr. ner), the initial vowel
is likely to reflect a PIE laryngeal (4.322). But this is not the [Link] in
NOTE.

13

Meillet (1936, p. 106) is at a losa to account for c in anieanem. But see above, 4.352 (*dy > c).

[4.464.451)

Sound Change

87

amil (I) 'month' < *mi1180 : of. Gk. min ( < mem-), Lat. m~, Olr. m' (G. mu) 'id'.
And there are instances of optional proetheei.s before n- in naw/11'/anawar 'thin, sparse'; nawl'i/anawt'i
'hungry' {[Link]. a1W81', anot'i). Such scarce evidence does not allow for any conclusion.

Epenthesis. In the PA period there must have been a trend to epenthesis

as the anticipation of a subsequent phoneme (i/y or u/w); but its t races have
been blurred, so that the uncontrovertible evidence for epenthetic y or w
amounts to a small number of words only. As far as we can judge, epenthesis
occurred mostly between a and a resonant. The subsequent phoneme that
brought it about is either lost (as in ayl, awr, etc.), or altered (as in awj, awcaneni).
4.45.

Regarding the former case, Pedersen's conjecture deserves consideration. He

seems to discover the trace of an obsolete paradigm in the 'doublets' : dayl, dal
(III 1) 'first milk' ( < *dhali-) :

N Acc. dayl ( < *dali)

GDL. dali ( < *dali-

+ a case ending)

Both stems later evolved to lexical variants, whereas in other instances only one
survived and was generalized. Hence, on the one hand : cayr (III) 'end, point'.
sayr (III) 'sword point', jayn (III) 'sound, voice' ; and on the other : k'ar (III)
'stone', sal (III) 'anvil', ban (Ill) 'word, reason', etc. (Pedersen, 1906, p. 406-408).
Let us simply record the most demonstrative examples.
4.451.

y epenthesis :

dayl 'first, milk' beside dal (see above);

ayl (I) 'other' < *alyo- (4.351);
p'ayleni (-im) 'I shine' < *pholye- : cf. p'olp'olim 'I shine, glitter';
ayr ' man, husband' (3.17) = Gk. anb 'id'. The PA development presumably
started with the change of *e to i: *anir, hence *aynir > ay(n)r (Meillet, 1936,
p. 55).
Epenthesis is also evidenced after e in mef (I) 'middle' < *medhyo- (4.352);
iz (Ill) 'viper', if from * ez ( < *eghi-), GDL. iii, as Pedersen surmises (1905,
p. 205): the development of e to ey > e is assignable to the influence of the
following y or i in the prototype. But if es (I) 'ass' reflects PIE *ekwo-, as we
admitted (4.354, end), the same development, in this case at least, has to be
referred to some other [Link]. Finally, epenthesis after o is hardly deniable in
view of such word pairs as sol (I, III), n-soyl (Ill , Ilb) 'ray of light'; t'olum ' I let',
t'oyl tam 'I give way, permit' for which, unfortunately, comparative data are

m1ss1ng.

88

Sound Change

[4.452-4.46)

4.452. u1 epenthesis

awr (GDL. awur) 'day' < PA. *amur < PIE *dm6r : cf. Doric Gk. amar
(< *amr), amba 'id' (Meillet, 1936, p. 55; Pisani, 1951, p. 69);
artawsr (pl. artasuk') 'tear' < *drafcu- (4.342);
awj (III) 'serpent' < *angwhi- = Lith. angis 'venomous serpent', Lat. anguis.
In this and the following instances, the phenomenon originated in the labialized
velar : the labial appendix was shifted backwards, so that *angwhi- changed to
*aunghi, and lastly, after the dropping of n, to *aughi-;
awcanem 'I anoint' < *tlgwe- : cf. Lat. unguo 'id' ( < *ongwe-);
awjik' (V) 'collar' : cf. Gk. aukhin 'neck', with epenthesis, as against Aeolic Gk.
ampMn 'id', both from *angWhen- (Pisani, 1950, p. 188-192).
This explanation, firstly proposed by G. Bonfante 74, relies on the assumption of
labialized velars in the PIE dialect from which Armenian is descended (cf. fn 50).
At any rate, it is more satisfactory than Meillet's dubitative suggestion of n
changing to w (1936, p. 37, 44).
NOTE . To

my knowledge, no systematic discussion of the problem M a whole has been undertaken

since Pedersen's attempt (1906, p. 404-411), which is not totally successful. Several instances of w
epenthesis, in particular, still await an explanation, e.g. in gewl 'village' (3.17); giwt (III) 'finding,
invention', beside gtanem 'I find' (4.353); nwwt (I ) 'near', beside matnum, matc'im 'I approach', etc.

4.46. Vowel contraction. The loss of intervocalic *y and *s in early PA entailed

the contraction of e, o, and probably a, with the following vowel :

*-e(y)e*-e(s)O*-e(s)o*-o(s)o-

::>- -e*-eu- ::>- -oy-o-

-o-

erek' 'three' (4.341)

k'oyr 'sister' (4.335)
k'ork', pl.N. of k'oyr
bok 'barefoot' : cf. Lit. ba:sas, OHG. bar

< *bhoso- (Meillet, 1936, p. 38-39).

The same process was repeated much later, after the loss of PA -y(4.334), at least between similar vowels :

<

*-t-

*p~tr-bhi

(sg.I.) ::>- PA *hayarb(i) ::>- ClArm. harb (Grammont, p. 236; Pisani,

1950, p. 180) - but :
*bh~ti-bhi

::>-

*bayiw(i) ::>-

bayiw

On the other hand, the close vowels (i, u) were not contracted :

diem 'I suck' = Skt. dhayami 'id'

14

1~ ;

G. Bonfante, Lt4 iwglosaea greco-armeniennea. ?.f6langes Roiger Pedersen, Copenhagen, 1937,

p. 15-33 (on awj, awcanem : p. 25).
75 The parallel implies divergent developments of the PIE present stem *dh;,ye-. I &88ume that in

89

Sound Change

[4.464.47]

nu (I) 'daughter-in-law' (4.335) : GDAbl. nuoy, I. nuov

li (I) 'full' (4.341)

afu (III) 'brook' (4.323)

lioy
afui

GDL.

liov
afuiw

We thus gain a diachronic justification for our statements on hiatus (2.212-213).

But we also have to consider the further developments of the derivational suffixes
-io- (< *-iyo-), -ia- ( < *-iya: 4.22). The latter evolved to -ea-, according to the
general rule (2.213), while the former appears as -wo- in the inflection of polysyllables ending in -i (3.133). These changes took [Link] the reduction of -i1v
to -u- (2.221).
In word final however, before -n, we find the reverse development: *-i(y)an
> -iwn:

siwn (GDL. sean) 'column' = Gk. ktan 'id';

jiwn (GDL. jean) 'snow' = Av. zya 'winter', Gk. khi6n 'snow',
and other nouns in -iwn (ariwn 'blood', ankiwn 'corner', etc.), except those of the
goc'iwn type (4.22). This quite peculiar development will detain us [Link], in
connection with the origin of n stems (5.144).
4.47. The PIE diphthong pattern has been distorted. It eventually gave way
to a new set of 'quasi-diphthongs' (Cf, 1.641). Leaving aside the so-called long
diphthongs (*ei, *eu, etc.), for which no evidence is to be found in ClArm., we
start from a . six diphthong system, which was firstly reduced to four by the
merging of *oi with *ei, and of *eu with *ou :

*ai
*ai

*ei.

"

* .
()?,

*ei /

*au
*au

*eu

7&

"

*ou

*ou/

Later, *ou changed to *oi; but by that time, the articulation of the second phoneme
in the surviving diphthongs had begun to tighten ( *ai > ay; *a'!-' > aw, etc.).
These changes entailed a new distribution :

ay

*ey

oy

aw

However, the balance of the pattern was restored, or even maintained, by the
issuing of the quasi-diphthongs ew, iw, *ow, as well as of more occurrences of
PA * ll changed to i before y, by progressive assimilation, while in Skt. it opened to a through the
opposite process. This enables us to account for the puzzling etymological relation of Arm. ji (I) 'horse'
to Skt. Mya 'id' by positing a prototype *ghfJyo-.
7 8 A chronologica.l datum is found in the development of -eso- to tu in the wordfor 'sister' (4.335),
unless one accepts Grammont'& far fetched explanation of k'oyr (p. 244) .

90

Sound Change

[4.47-4.6)

ay, *ey, aw, in consequence of various sound changes. At that stage of the
development, there was no gap, no 'hole in the pattern' :
ay

*e:tj

oy

aw

ew/iw 77 *ow

Jn particular, *ow actually occurred in such words as *anown 'name' ( < onomti-),
*owt' 'eight' ( < *opto), *k'own 'sleep' ( < *swopno-), etc. But its rate of
frequency in the lexicon was probably low, and it merged with u previous to the
reduction of this latter phoneme in unstressed position. So there is no difference
in ClArm. between the reflex of PA *ow and that of PIE *u or *o :

*swopno- 'sleep'
*k'own
k'un, GDAbl. k'noy

*dkura- 'door'
durk', drac'

*doro- 'gift'
tur, troy

Close to the end of the PA period, *ey in its turn reduced to a single vowel (1.63,
2.211). The last phase of the development is therefore :
PA

ClArm.

ay
ay

*ey

oy
oy

aw
aw

ew/iw *ow
ew/iw (u)

I t is not easy to place the genuine diphthong ta in the above framework; for, its development
is not connected with that of the quasi-diphthongs. However, some data sUow for a chronological
statement : in non final eyllablee, ta shrank to e prior to the reduction of unstressed ey to i (2.223,
3.234, Note).
NOTE.

4.5. After describing sound change from PIE to ClArm., it seems appropriate
to try the reverse approach (4.32), and to trace the Armenian phonemes back to
their origin. This is feasible insofar as we deal with genuine Armenian words, or
with loanwords of certain provenance. For, before words of unknown origin (4.11),
diachronic phonology is at a loss: the comparative method is not applicable,
and we miss such historical and/or literary data as allow for the elucidation of
Iranian, or Syriac words. Let us consider, e.g. xnjor (I , Ill) 'apple' : the provenance of the word and the time and circumstances of its reception into the PA
language being equally uncertain, one cannot tell which phonemes are reflected
by x or j, unless one supposes these to have been the original ones.
The following table presents the Armenian stops, affricates and fricatives as
the reflexes of PIE, or, alternately, of Middle Iranian, phonemes. The inclusion
of the latter is sufficiently motivated by the importance of the Parthian element
in the ClArm. lexicon. Incidentally, such phonemes as p, c, 8, z, x, for which
scarce evidence is to be found in genuine Armenian words, occur quite frequently
The spelling in the manuecripts is somewhat inconsistent (Meillet, 1936, p. 45-46; Abrahamy&n,
p . 8, 13), so that ew &nd iw may be regarded as variants.
77

91

Sound Change

[4.5)

in Parthian loans. There is, e.g., hardly more than one unquestionable example
of p < *b (an exceedingly rare PIE stop), as against numerous instances of
p < Iranian p.
The conditions under which the same PIE consonant came to be reflected by different
Armenian phonemes (e.g. j/z, g///) have been specified above (4.332-354), and consequently need
not be resumed in the table.
R = resonant; N = nasal.
NOTl!I.

Arm. p

<PIE *b
*d
t
k
*g
c
*g, *dy
*gy1
*bh
b
*dh; *(R)t
d
*gh; *(R)k; *w
g

*gh
l
*gh; *dhy; *(R)y 1
J
*ph
p'
t'
*th; *t
*k; *tw; *sw
k'
c'
*sk ' *ks, *ky
*ky
c'
*k *Ns
s
'
z
*gh
s
*kw 1; *s (after r)
z
*gh 1
x
*kh
h
*p
*st; *kt
st

Middle Iranian p
t

b
d
g

(J after n)
j
f(t)
(k)

x
h; f (before r)
st

92

Historical Morphology

5.

[[Link]

Historical morphology

Some aspects of the PIE background of ClArm. morphology have been delineated
[Link] (4.21-22), [Link] the notion of morphological change had to be introduced in
connection with the requisites of etymology (4.314). But historical morphology is
mainly concerned with noun and verb inflection. The aim is to state to what
extent the ClArm. system of declension and conjugation is derivable from PIE,
and, as far as possible, to trace the successive developments. At this juncture
the following points have to be emphasized.
1) PIE inflection cannot be reconstructed so confidently as the sound pattern.
An overall outline of PIE declension or conjugation, without chronological and/or
dialectal qualifications would be somewhat delusive. Even the concept of 'late
PIE', which proved helpful in the description of sound change, must he handled
with caution as soon as morphological change is at issue; for, ClArm. displays
archaic PIE features, alongside with striking innovations.
2) Morphological change, we remarked, mostly results in a new distribution
of inherited materials (4.314). Consequently, foreign morphemes are not expected
to appear in ClArm. inflection, and we have a right to assume, as a methodological
principle, that most of the new system is explainable in terms of PIE morphology.
This principle, however, must not be extended to word formation : [Link]
morphemes happen to be borrowed (cf. 4.121 ). Reduplication, too, as a morphological
device (Meillet, 1913, p. 43; Jensen, 125-126, 129), may well have developed in
PA under foreign influences.
5.1.

NouN INFLECTION.

While in many modern IE languages gender is still a grammatical category,

there is no gender contrast in ClArm. (3.1), even in pronouns: na 'he, she,
it' 11; ink'n 'himself, herself, itself'. This situation is doubtless the ultimate
result of a series of shifts and changes: faint traces of the masc./fem. contrast
have been discovered (Meillet, 1962, p. 145); and the -an- plural of some n
stems seems to have been peculiar to ancient neuters in *mti (Skt. -ma, Gk.
-ma, Lat. -men) : sermank' 'seeds', anuank' 'names' (3.14la).
CIArm. haa two sets of demonstrative pronollll8 : aa, da, na, and ay1, ayd, ayn. As a rule, the
former refer to persons, the latter, to things : Usti e 1ma ay1? '\Vherefrom does lhil man have lhif
(wisdom) t', Mark 6.2; Minll'der na z-ayn xawser 'While 1u was speaking that', Mark 5.35. But the
contrast ie suppressed when aa, da, na function as anaphorice: E t'e akn k' o gayt'aklec'uc'ane z-k'ez,
ban z-na i bac' 'If thy eye causes thee to sin, pluck it out', Mark 9.46 - or when ayn is the antecedent
of the relative pronoun: ... 06' z-is onduni, ayl z-ayn or ai'ak'eac'-n z-is 'He receives not me, but him
wko sent me', ltfark 9.36.
78

[5.1-5.12]

Historical Morphology

93

As Meillet suggests (1936, p. 12; cf; 92-93), the [Link] of gender in the neigh-

bouring languages may have helped on its elimination in PA.

5.11. As pointed out above (3.183), the diversity of declension types in ClArm.
is somehow balanced by the uniformity of most case markers. This latter feature
'implies a great deal of analogical levelling. Some of the problems involved will
be discussed later.
Stem variation is faithfully preserved in irregular -r nouns (3.17), and in n
stems at large (3.142-143, 4.323). The -o-, -a-, -i-, -u- declensions, too, reflect PIE
patterns, and many words still display the original inflection (1\feillet, 1936,
p. 73-76). But, while *-o-, *-ii-, *-i~ or *-u- originally belonged to the word stem,
the ClArm. inflection vowels occur in the oblique cases only, so that they function
as [Link] of the case endings. This morphological shift perhaps brought [Link], or
at least favoured, a partial redistribution of nouns and adjectives among the
vocalic declensions 79 It also accounts for later fluctuations (3.132).
Regarding the inflection of Parthian loanwords, see 'Meillet, 1936, p. 23, whose

statement, however, is controverted by Bolognesi (1954, p. 123-126).

5.12. In the largest part of the PIE area, the *-o- and *-ii- stem clas~es tended
to become complementary. Thus, in most adjectives, the masc.-nt. *-o- stem
was matohed by a fem. *-ii- stem, as in, Gk. ne(w)os, -.on/ne(w)ii 'new'; OCSI.
no'VU, -o/nova; Lat. nouos, -om/noua. This situation may have lasted in PA as
long as the gender contrast still survived; but in ClArm. the -o- and -a- declensions
[Link] separated again: ancient *-of ii- adjectives follow the -o- declension uniquely,
e.g. ayl 'other', kin 'old', nor 'new' (4.21), ferm 'warm', hum 'raw', surb 'pure,
holy', mef 'middle', eto. 80.
Adjectives in -i belong to the mixed declension : bari 'good', GDAbl. barwog,
I. bareaw (3.133), and so do the substantivized derivates, e.g.
from gore (I) 'work'
gorci 'tool',
kogi 'butter'
kov (IV) 'cow'
kalni 'oak'
kalin (I) 'acorn'
mayri 'pine wood'
mayr (III) 'pine'
matani 'ring'
matn 'f'mger,, etc.
Since -i reflects both *-iyo- and *-iyii- (4.366), the alternation of o and a endings
in the same paradigm is doubtless a trace of the former correlation of both stem
This fact, ignored by most Armenologists, is duly acknowledged by Mann (1968, p. 4-12).
Unfortunately, his argument swarms with untenable etymologies.
80 But not mec (lib) 'great, big', which reflects the same prototypeae Gk. megaB 'id'.
79

94

Historical .Morphology

[5.12-5.13)

classes; but at the same time, any connection with gender is excluded : k'efi
'maternal uncle' and ayri 'widow' are inflected alike.
The reason is that *-a- stems, originally, were not peculiar to fem. nouns, very
few of which, incidentally, survive in ClArm. (am 'year' = Skt. sama-; skesur
'mother-in-law', 4.354). The -a- inflection of compound agent nouns (3.333) is
traceable to PIE models (Meillet, 1962, p. 172-173). Furthermore, there is clear
evidence for a gradual spreading about of the -a- declension during the PA period :
it was propagated to the plural of PIE stems ending in a resonant :

jerk' (GDAbl. jefac') 'hands'

< PIE *ghesr- (*her-)
ac'k'
(GDAbl. ac'ac') 'eyes'
*okWifokyarlasuk' (GDAbl. artasuac') 'tears'
*drakumawrok' (GDAbl. mawruac') 'beard'
*smokruLater on, -a- endings (-aw, -ac', -awk') entered the inflection of most rand l words
(Meillet, 1913, p. 51-52; 1936, p. 81; Jensen, 150-152). In view of these
developments, the discrepancy in the agent noun morpheme ClArm. -iC' < *-ikya,
as against OCSl. -iti < *-ikyo- (Meillet, 1936, p. 75) can be confidently ascribed
to a PA innovation.
There are no traces of the PIE agent noun suffix -ttr/tor/tr- in CIArm.; but -awl (3.2, Note)
may reflect an expanded form of the alternative suffix -tel-, well preserved in the Slavic languages
(Meillet, 1936, p. 32). A derivation of cnawl 'parent' from gen~-tl- ie consistent with the rules of sound
change. The original inflection, then, should have run parallel to that of hayr, Jiawr (3.17). The new
paradigm is built up on the oblique ca36 stem cnawl-, and the a inflection (pl. GDAbl. [Link]') ie all
the less surprising as it is the normal one for agent nouns at large.
NOTE.

5.13. The i and u declensions did not develop at the same rate. The lexical
range of the former increased notably, in particular through the coining of radical
action nouns : sarz 'motion' (beside sarzumn), cin 'birth' (cnanim 'I beget, am
born'), k'ayl 'step' (k'aylem ' I step'), xaws-k' 'speech' (xawsim ' I speak'), etc.;
see also above, 3.32 Note. If this nou.n type goes back to PIE, as it likely does in
view of similar forms in Indo-Iranian, Slavic and Germanic 81, it must have
become far more productive in PA than in any other IE language.
Conversely, the u declension appears in a state of decay. Although it attracted
some nouns ending in -w/v (haw 'bird', cf. Lat. auis; arew 'sun', cf. Skt. ravi-;
kov 'cow', cf. Skt. gau~, gav-, Gk. bo11s, G. bo(w)os), it certainly began to lose
ground before the ClArm. period: unlike the other vocalic declensions, it plays
no part in derivation and composition, and has left no trace in ModArm. noun
inflection.
E.g. Skt. nlci- 'light, eplendor'; OCSJ. rea 'speech', vldl 'knowledge' ; Goth. alaM (pl.N. alaMw)
'blow', mun.a 'thought, decision', eto. The Germanic words are maao., the other ones fem., and so are
the Lat. i stem action nouns (caedla 'murder, slaughter', labu 'downfall'), which may well pertain here.
111

[5.131-5.14)

Historical Morphology

95

5.131. Two sets of PIE *i and *u stems have to be distinguished: barytonic

invariable stems, and oxytonic stems with vowel alternation ( *i/ei; *u/eu/ou).
Of the later, ClArm. has only one reflex :
erek' 'three' < *trey-es : Acc. eris < *tri-ns; I. eriwk' < *tri-bhi (4.341).
Everywhere else, the stem vowel, when preserved, is invariably i or u. Thus, bay
(III) 'verb', GDL. bayi; zard (IV) 'ornament', GDL. zardu reflect the same type
of inflection as Ionic Gk. p6li-s 'city', G. p6li-os, pl.N. p6li-es; neku~s 'corpse',
G. neku-os, D. neku-i.
Archaic pl. NAcc. forms of neuter u stems are still recognizable in : mawro-k',
AccL. mawru-s 'beard' < *smokrua; artasu-k'; artasu-s 'tears', cf. Gk. dakrua.
PA *drafcu-r, pl. drafcu-a was to yield artawsr, *artasu. As this latter form was
morphologically unclear, it was normalized through the addition of the usual
endings -k', -s (2.322). The parallelism of the i and u declensions allows- for a
similar interpretation of kelcik' 'feint' in connection with kelc (III) 'sham, false'.
The substantivized pl. neuter *kelci (with -i < -ia) was treated like *artasu,
*maioru, and transferred to the a declension (GDAbl. kelceac', cf. artasuac').
As a rule, however, the pl.N. is formed by adding -k' to the [Link]., as in the
other vocalic declensions: bayk' 'verbs', zardk' 'ornaments'. This can be easily
explained as a case of analogical expansion (see below, 5.221).
5.132. The [Link]. -rending is peculiar to u stems: asr 'fleece', GDL. asu;
barjr 'high', barju, etc. (3.151, Note). Ounr 'knee' and artawsr 'tear', [Link]
indeclinable in the sg., [Link] no exceptions: their belonging to the same stem class
is proved by comparative evidence, as well as by the pl.N. artasuk' (5.131). Therefore -r reflects *-ur. Of this ending, rare traces are to be found outside Armenian 12.
It seems to have originated in neuter nouns, such as the prototypes of Arm. cunr,
artawsr, melr, and possibly asr, calr (Meillet, 1936, p. 82; cf. Pedersen, 1905,
p. 230-231); but this does not account for its occurrence in adjectives.
5.14. Passing on to the ClArm. variable stems (3.14), it m"ust be pointed out
that not all are traceable to PIE *r, *l or *n stem~. Several are words of obscure
b~ckgrou~d, e.g. ezr 'border, side', elfiwr 'horn', kocl 'log', kat'n 'milk' dasn
'agreement', etc. The r and l stem classes include ascertainable loanwords: kaysr
'emperor' (1.61); tetr 'fourfolded parchment, quaternion' (Gk. tetras); sikl 'a
weight or coin' (Gk. siklos, from Hehr. seqel); litr 'pound' (Gk. lura); arkl 'chest,
coffer' (Lat. arc(u)la); vagr 'tiger' (Iranian: cf. Skt. vyaghr<i-), mostly in consequence of the loss of the original ending, as is _the case of tetr, sikl, litr, etc. As a
So, perhapl!, *d&ur (in Lat. duuria 'group of ten men'), beside d&u. (Goth. ligu- in hooi tigiw
'twenty'. etc.). The relationship between Hitt. pankur (G. pankunal) 'family' and panku- 'whole' ia
dubious.
82

96

Historical Morphology

[5.14-5.141)

similar alteration, i.e. the dropping of last syllable vowels, affected all PA words,
(4.33), it would not be surprizing that the reduction of *-lo/a, *-no/a to -l, -n
should have entailed a. change of inflection in genuine Armenian words as well 83.
In fact, etl {GDL. etel) 'place' may reflect *sedla- {Laconian Gk. hella, Lat. sella
'seat' ), or *sedlo- {Goth. sitls 'id'). Likewise, bein {GDL. befin, I. befamb) 'burden',
though inflected as an n stem, is parallelable to Gk. pherni 'that which is brought
{by a wife), dowry', and consequently traceable to PIE *bherna. The same
morphological change has been suspected in t' ekn {GDL. t'ikan, pl.N. t'ikunk')
'shoulder' : cf. OCSl. stigno 'femur', Russ stegno 'haunch'; karsn (GDL. karsin,
pl.N. harsunk') 'new married woman, bride' {< *p[k-na ?) ; safn {GDL. safin,
I. saiamb) 'ice' : cf. Oicel. hjarn, Russ. seren 'frozen snow' &4.
All this implies that quite a number of genuine *r, l and n stems had been
preserved in PA, together with their particular inflection forms, many of which
remained productive down to the ClArm. period.
5.141. Masc. and fem. *r stems are chiefly represented by kinship nouns:
kayr 'father', mayr 'mother', elbayf' 'brother', k'oyr 'sister', ayr 'man, husband'
(3.17). They display the zero grade of the (last) radical vowel in the sg. oblique
cases :
D.
I.
pl.N.

mayr
mawr
marb
mark'

<

*matb
*matr-ei
*matr-bhi
*nu'itb-es 81

k'oyr
k'ef

<

*swes0r
*swesr-ei 85

(k'erb)

k'Of'k'

*swesor-es.

In ayr < *anir, the oblique case stem has undergone metathesis : *anr- > PA
arn-. Hence GDL. am, I. aramb {< *arti-bhi).
Unlike the above kinship nouns, taygr 'husband's brother' and dustr 'daughter'
follow the normal r declension (stem II : tayger, dster). Evidence for the expansion
of the e grade to the whole paradigm is found in Greek: dab, G. daer-os, pl.N.
dab-es = Arm. taygr, tayger, taygerk' (Cf. [Link] OCSl. tieveri, Lith. dievens).
Little is known of the inflection of PIE *l stems. In ClArm. it runs parallel
to that of the regular r nouns: astl 'star', astel, astelk', (Cf. Gk. astir, astb-os,
aster-es).
aa r atema are not concerned here; for, the alteration of atop + r clusters (4.341-342) belongs to an
earlier period.
14 The Germanic and Slavic words reflect ~rn.o .. Arm. aafn (instead of * [Link]) may have its a
from aa1'num 'I freeze' : the verb root is aari < IC0 ri- or k,0 rl- (Cf. aor. aafeay).
86 The choice of D. forms is arbitrary: mawr may just all well reflect a G. form (*rniitr-ea/w).
81 A development: mater- > *ma(y)tr > mar- is conceivable at the beat, though not verifiable.

(5.142-5.144)

Historical Morphology

97

5.142. Neuter *r stems substituted n for r in the oblique cases. This most
archaic type of inflection is not preserved in ClArm. : awr 'day', GDL. awur,
pl.N. awurk' reflects a uniform PA paradigm : *amur, *amur + case endings.
As to albewr 'spring, fountain', alewr 'flour', the oblique case stems are hardly
explainable in terms of historical phonology. I would rather assume analogical
innovation: alber-, aler- may well have been coined after ezer-, osker- (oskr 'bone'),
etc., while the original [Link]. forms remained unchanged (Cf. 4.353, Note). A
radical change of inflection has been carried out in hur (I) 'fire', originally a *r/n
neuter (Hitt. pa'/}'!Jur, G. pa'!J'!Jwenas; Goth. fan, G. funins, as against AS. ]yr).
A trace of the alternative n stem (hun-) is found in hnoc' (lib) 'oven'; but the
PA word was inflected as an invariable r stem, as is proved by the archaic I.
form hur-b. The dropping of the vocalic case endings \vould have resulted in an
ambiguous paradigm: hur, GDL. *hur, not consistent with the normal pattern
of inflection. The anomaly has been removed by a transfer to the -o- declension.
5.143. The declension of *n stems did not develop along the same lines all
over the IE area. Not only are the alternating stem forms ( *-en/on/n-) distributed
differently, but in some languages, as [Link] and lndo-Iranian, the distribution
has been partially obscured by sound change. This is not the case of ClArm.
which, in this respect, stands very close to Greek. On the other hand, Armenian,
together with Indo-lranian; supplies valuable evidence for zero grade case forms,
few of 'vhich are preserved elsewhere (Gk. kuon, G. kun-6s 'dog', (w)artn, (w)arn6s 'lamb' [ = Arm. gain]; Lat. caro, G. earn-is 'flesh').
5.144. The sg.N. forms of ClArm. n stems raise a perplexing problem, which
will have to be discussed later (5.2). With regard to the [Link]. and the pl.N.,
two main types, or noun classes, can be set in contrast :
GDL.

I.
pl.N.

gain 'lamb'
gaiin (-in < *-en- + case
ending)
gaiamb (-amb < *-rJ--bhi)
garink' (-ink' < *-en-es)

jukn
jkan (-an < *- 0 n- +case
ending)
jkamb
jkunk' (-unk' < *-on-es).

Nouns in -iwn, together with sun 'dog' ( = Gk. ku6n), tun 'house' (3.141 b)
belong to the second class. The reason for the discrepancy in siwn, pl.N. siwnk
vs. jukn, jkunk' lies in vo,vel contraction (4.46). The development can be traced
as follows : *-i6n- > *-iun- > -iwn. The latest stage must have been reached
before the loss of last syllable vowels: thus, Meillet's and Pedersen's constructions a7 can be dispensed with.
87

Meillet, 1936, p. 80 (on -ut'iwn); Pedersen, 1905, p. 216-217.

98

Historical Morphology

[5.144-5.145)

ClArm. therefore has reflexes of both *-en/n- and *-on/n- stems. The Armenian
and Greek paradigms largely concord, save that in Greek the zero grade alternant
( *-n/ti--) does not occur at all, or else, exceptionally, holds through the whole
declension, as in kuan, (w)arbi (5.143) :

G. .
pl.N.

arn
= Gk. arsen
'[Link] sheep'
'male'

arin
drsen-os
n-es
afin-k'
arse

.
siwn
=Gk. kt an
'column'
'id'
(sean
kton-os)
si11m-k'
kton-es.

The proportion of *-en- and *-on- stems is fairly the same in both languages :
the former type is no more productive.
Unlike Greek, however, ClArm. has preserved a third type of inflection,
characterized by the *-en/on- alternation within the same paradigm (3.143). This
archaic feature is also reflected in the declension of masc. n stems in Germanic :

GD.

I.
pl.N.

t'orn 'grandson'
t'ofin
t'ofamb
t'ofunk'

Cf. Goth. gum.a 'man'

.
G. gumin-s

{ D. guniin

guman-s

On the substitution of *~on- for *-en- in compounds in ClArm. (pl.N. and Acc.)
and in Greek (th:..oughout the paradigm), see Meillet's comments (1936, p. 79).
5.145. The *-on/n- stem class includes almost all the words in -mn: himn
'foundation', kamn 'flail', fermn 'fever' ; erdumn 'oath', and the numerous action
nouns in -umn (3.331 b; see also 4.22). But an exact delineation of the
morphological background meets with difficulties.
The PIE morpheme *-mon/mn (1nii) yielded both masc. and neuter nouns, the
latter mostly denoting the effect of an action. At first sight, the ClArm. words
fit quite well in the second category, and consequently may reflect ancient neuters
(Meillet, 1936, p. 79, on sarzumn 'motion'). There are, however, some grounds
for doubt. Firstly, as early as the PIE period, the masc. type overlapped, or
competed with, the neuter ; Greek has, e.g., termon 'boundary' along with term.a
'id' ; st~man 'warp' as against Lat. stamen. Secondly, the confusion of genders
was detrimental to the neuter inflection, the distinctive feature of which was the
identity of N. and Acc. in both the sg. and the pl. : the ClArm. endings of pl.N.
{-k') and Acc. {-s) stem from the PIE masc. and fem. inflection. It is therefore by
no means certain, despite the merger of *o with *o before a nasal, that -un(k'),
-un(s) in ancient neuters should reflect *-an-a(a, as Meillet believes. At any rate,
this assumption conflicts with our above remark on -ank' plurals (5.1). Finally,
there is no evidence in ClArm. for invariable n stems with a long vowel, such as

[5.1455.21)

Historical Morphology

99

are found in other I E languages, e.g. Gk. thur6n, G. thuronos, [Link]. 'hall' ; Goth.
daur6ns, fem. pl. 'gate' 88,
In conclusion, the PIE masc. type reflected in Skt. dt;mii (at;man-) 'stone';
Gk. akm6n 'anvil', st~m6n 'warp'; Lat. sermo 'talk' (G. serm6nis, with 6 from
the N.); Goth. ahma 'spirit' provides the most suitable model for the inflection
of ClArm. words in -mn.
5.146. In the course of the PA period, the n inflection gained ground. Tlie
above mentioned innovations (5.14 : bein, etc.) are the most conspicuous ones,
but they do not belong to the earliest stage of this process. The n extension in
the pl. of k'ar (III) 'stone' and of several u stems (3.151) looks rather archaic;
but, for want of comparative data, the problem of its origin remains unsolved.
At any rate, a. connection with the PIE *r/n alternation in neuter nouns, hinted
a.t by Meillet (1936, p. 82), is out of the question.
The case of jefn 'hand', otn 'foot', akn 'eye' (3.152), together with that of
calik, manuk (3.142), will be discussed below, in connection with the problems
related to the [Link]. forms.
5.2.

CASE ENDINGS.

In 01.A,m., the sg.N. and Acc., being reduced to the [Link] word stem, [Link]
indistinguishable on the morphological level 89 This was the case of PIE neuter
nouns only; but, since the prototypes of most Armenian words belong to the
so-called 'animate' gender, the merger of both cases in the .sg. may be ascribed
to the loss of the original endings. This view does not raise serious difficulties
insofar as the vocalic declensions are at issue (Meillet, 1936. p. 6.9); however,,
regarding the n stems, it proves inadequate.
But does the ClArm. situation actually result from a merger of two contrasting
case forms, or from a selection of either the N. or the Acc. form 1 In other words,
does mard 'man' reflect both the N. *mrtos and the Acc. *mrton (or *mrtom),
or only one of these forms - and in this case, which one 1 The answer depends
on how one figures out the fate of final nasals in PA.
5.21. According to Meillet (1936, p. 56), the *m/n [Link] was neutralized
in word end : in this position PA., like Greek, had only -n (k'an 'than' = Lat.
quam 'id'). In polysyllables, *-n was dropped ( *mrton > mard). After a consonant, *-ti- developed to -an, as it did word internally, but, unexpectedly, the ~asal
escaped dropping: *septti- > ewt'n. This is the weak point in Meillet's theory: if

[Link]'s opinion (1968, p. 14), [Link] 'door', a normal -cm/n- stem, need not be mentioned
in this connection. Besides, o in daurim.s may reflect a as well as o.
8 9 The use of z. as a mark of the definite direct object is a matter of syntax (3.112).
88

100

Historical Morphology

[5.21-5.212)

*-dn is lost in am 'year' ( < *saman), mec 'great, big' ( < *megan), etc., how are
we to account for the discordant development :
(PIE) *podii

>

(PA) *hotan

>

(ClArm.) otn 'foot' 1

Pedersen (1905, p. 215-216) agrees with. Meillet regarding the loss of *-Vn (and
*-Vm). His comment implies that *-Oti (and *-01?") did not alter, except for the
merging of *-1!1' and *-ti- ( *sept1!1' > ewt'n; *enWti- > inn). Thus PA should have
preserved vocalic resonants in word final.
Pisani (1951, p. 47-51) puts the case in a different way: *-m was regularly
dropped, except in monosyllables, where it changed to -n (k'an), while *-n was
not affected: gafn ' lamb' = Gk. (w)aren. In other words, the NAcc. forms of
originally masc. or fem. n stems are derivable from PIE N. forms in *-en, *-on.
Genuine reflexes of these forms, consequently, need not be sought in such words
as calik, manuk (3.142 c; Meillet, 19~6, p. 79-80) or erec' (3.151 ; Pedersen, 1905,
p. 218).
In view of so conflicting arguments, a re-examination of the problem
cannot be dispensed with.
We do well to start from one reliable datum: a PIE vocalic nasal (*-nor *-m)

is reflected by -n in the numerals ewt'n 'seven', inn 'nine', tasn 'ten', and in the
relics of the neuter *-mti- nouns, e.g. anun 'name' ( = Gk. 6noma), sermn 'seed'.
Regarding jefn 'hand', otn 'foot', akn 'eye', Meillet's explanation is decidedly
preferable to Pisani's 90 In these words, the NAcc. unmistakably reflects PI E
Acc. forms : *pod-1} (-1!1') > otn (= Gk. p6da); *gher-1} (-1!1') > jefn ( = Gk. khera).
Then inflection, limited to the sg. (3.152), developed later, in imitation of genuine
n stems. The original paradigm has left a trace in the archaic I . form jerb
(< *gher-bhi) in jerb-a-kal 'caught by the hand, prisoner'.
Meillet's view of the development of the vocalic nasal (*-ti- > 7an > -n), too,
seems right: -an is preserved in ancient derivatives (ewt'an-asun 'seventy',
tasan-ord 'tithe'), and in the -ank' plurals : in sermank' /-s, as in artasuk' /-s,
kekik' /-s (5.131), the normal endings have been appended to the old pl. neuter
NAcc. form *serman ( < *sermana).
5.211.

In the vocalic declensions, the [Link]. forms are ambiguous. This 'is
not the case, however, in the masc. and fem. r ' and l-etems: hayr, mayr, elbayr,
5.212.

Meillet, 1936, p. 83.84; PiMoi, 1951, p. 48-49. My former misgivings (Godel, 1970b, p. 144),
I eonfess, were onfounded: afte.r all, the prototype of aL-n may have been ma11e. (Cf. Lat. ocul1U). In
my opinion, however, durn 'door' does not belong here: from a radical noun *dhwor-/dhur-, one rather
expects a full grade Acc. form (*dhwor-f}/m), and durn may well be a PA -on/n- derivative of dw(.l:' )

< *dhura-. Mann firstly echoed Meillet'e interpretation (1963, p. 37), but afterwarda changed hit
mind (eee above, fn. 88).
90

Historical Morphology

[5.212-5.213)

101

k'oyr, taygr, astl, etc. doubtlessly reflect PIE N. forms (5.141). The same may be
assumed regarding the n stems: sun= Gk. kuon; siwn =Gk. kton (5.144). So,
Pisani is right in equating gafn to Gk. (w)aren. Likewise, -mn is the regular reflex
of *-man as well as of *-mti (of the latter, through *-man), as we implicitly admitted
(5.145). The contrary evidence alleged by Pedersen and Meillet really lacks weight:
not to speak of erec', a genuine u stem, it can be objected that the background
of the -ik, -(u)k morphemes (in calik, manuk 9 1 is unknown. Monosyllables, at
least, have -n (jukn 'fish', mukn 'mouse', unkn 'ear'), and armukn 'elbow' occurs
along with armuk. Anyhow, the n extension does not go farther back than PA,
s'o that the words in question do not give any clue as to the original N. forms.

5.213. Summing up the above discussion, we come to the following formulae :

1.
2.

*-n ( *-m) > -an

*-Vn

>
>

-n
-n

If Meillet is right in assuming that only -n occurred in PA, we have to conclude

that the N. forms have prevailed in all the vocalic declensions: mard < *mrtos
(not *mrton, which would have yielded *mardn). This conclusion is not
objectionable, since Acc. forms have actually been lost in r, land n stems. However,
Meillet's assumption relies on only one parallel (k'an = Lat. quam). Besides, he
overlooks an important difference: PIE *-n /ti always belonged to the word
stem, whereas *-m/m mostly occurred as an inflectional morpheme ([Link].

ending in noun declension; 1st sg. ending in past tenses of the active conjugation).
When *-m was part of the stem, it changed to -n : jiwn 'snow', as against Lat.
hiem-s 'winter', Arm. jmefn ( < *ghim-erno- ~) 'id' 9 2 ; otherwise, it may be
supposed to have dropped together with the preceding vowel. If this is the case,
a third formula has to be added to the above :

3.

*-Vm

>

-0

and the [Link]. of invariable stems turns out to be the result of a merger :
PIE

ClArm.

sg.N. *mrtos

Acc. *mrtom

~sg. [Link]/

Both solutions of the problem equally fit the facts. The former, however, implying
Very likely, manu-k 'child' is derived from manr, GDL. manu 'small'.
92 CIArm. tun 'house' is somehow related to OCSI. domu, La.t. domua, and more closely to Homeric
Gk. d6ma. Presumably, it reflects *dam, i.e. the sg.N. of an archaic radical noun *dom-/d0 ni- (Pisani,
1951, p. 50).
91

102

Historical l\iorphology

(5.2135.222)

the prevalence of N. forms at large, is less easily brought into agreement with
the survival of Acc. forms in jefn, otn, akn (5.211).
5.22. In the plural, there are different endings for the N. and the Acc. (2.232),
as was the case in the PIE inflection of masc. and fem. words (on the fate of the
neuter declension, see above, 5.145, 5.211, end). A priori, the CIArm. endings are
likely to reflect PIE case morphemes. R egarding the N., however, we are
confronted with a controversial issue. The masc.-fem. pl.N. ended in *-Vs : *-os
( *o stems), *-as (*a stems) *-es (all other stems). The problem is, therefore,
whether *-Vs was to be dropped altogether, as it admittedly was in the sg.
( *mrtos > mard; *bh~tis > bay, etc.), or whether *-s may have developed to
-k' under the circumstances. If the latter possibility is ruled out, i.e. if *mrtos
was to yield *mard, the addition of -k' has to be somehow accounted for. Several
attempts have been made to this effect, none of 'vhich can be regarded as
convincing 9 3. In fact, -k' behaves as a genuine case ending, not as a derivational
morpheme (collective suffix); nor can it reflect an enclitic particle (~Ieillet, 1936,
p. 70). Thus, we are left with the phonological issue : is a change of *Vs to -k'
conceivable in spite of the decisive evidence for a divergent development
(*-Vs > f1) in other instances~
5.221. There are some grounds for a positive answer. Let us point out, firstly,
that all the pl.N. forms of masc. and fem. nouns ended in -s ,while in the sg.N.
-s occ~rred in *o, *i and *u stems only. On the other hand, the dropping of -Vs
in the pl. would have resulted in a confusion of pl. and sg. forms. Now, the ClArm.
system of declension, however open to syncretism (3.111), admits of no ambiguity
with regard to the number contrast (plural vs. singular). This may account for
the preservation of the pl.N. endings at the cost of a particular development.
This development, as traced by Pedersen (1905, p. 226-227) and Grammont
(p. 227-229), partially coincides with what we supposed to have been that of *k :
(pl.N.)

*-Vs
*-Vh
-(V)k'

*k
*II..
k'

The dropping of the vowel, as a result of the stress shift, entailed a levelling of
the various endings: *mrtos > mard-k'; *samas > am-k'; *rsenes > afin-k';
*kiones > siiun-k'; *kWet(w)ore.s > c'or-k' (but c'orek'-tasan 'fourteen' : Meillet,
1936, p. 79), etc. Hence, analogically : bay-k', zard-k' (5.131).
5.222. The PIE [Link]. ending, more or less altered in most IE languages,
93

For details and references, see Godel, 1970,b p. 148-149.

[5.222-5.231]

Historical Morphology

103

is well preserved in Gothic: daga-ns 'days', gasti-ns 'guests', sunu-ns 'sons',

bropruns 'brothers' (< *bhrat;-ns). Since *ti developed to an in PA, the expected
reflex of both *-Vns and *-(C)tts, after the dropping of last syllable vowels, is
uniformly -s (4.335). As in Greek, the pl.N. and Acc. display the same stem form,
even in the r, l and n declensions :
pl.N.
Acc.

*m;tos > mardk'; *swesores > k'ork'; *;senes >afink'

*m;tons > mards; *swesortts > k'ors; *rsentiS> afins, etc.

H ence, the synchronic rule: the pl.N. and Acc. are derivable from each other by
substituting -k' for -s or vice versa.
5.223. In the pl., the L. is identical to the Acc. Since -s as a pl.L. ending can
hardly be traced back to any PIE case morpheme 94, the Acc. form may be supposed
to have taken on the syntactic function of the other case (Mann, 1968, p. 64).
5.23. Oblique cases. In the *o and *a declensions, many case endings were to
be dropped, together with the stem vowel (sg.D. *-Oi, *-ai; Abl. *-Od; [Link].
*-as; pl.G. *-om, etc.). In the other types of inflection, the stem, [Link] and
standing in contrast to the [Link]., could still perform the part of an inflected
case form. Yet, this might have resulted in the decay of noun inflection, as did
similar alterations in other IE languages. In Armenian, on the contrary, the
defects have been supplemented through various devices, so that, in spite of
numerous mergers, ClArm. displays a larger set of cases than Germanic, Celtic,
or even Greek.
5.231. To begin with the plainest item: the I. markers in mardo-v, ama-w,
bayi-w, zardu 9 &, ezer-b, jkam-b are the regular reflexes of PIE *-bhi. The vowel
is preserved in iwi-k', the I . of an obsolete indefinite pronoun *i-k' (cf. i-mn
'something', and o-k', o-mn 'someone'). The same morpheme is reflected in
Mycenian -pi (= /phi/) and Homeric Greek -phi. Extended forms occur in IndoIranian, e.g. Skt. -bhy-am ([Link].), -bhy-as ([Link].), -bhi-s (pl.I.). In the
opposite part of the IE area, *-bhi is documented as a pl.D. in Gaulish (gobe.d-bi
'to the smiths') and in Oir. (feraib 'to the men', etc.)' 98 Its original range of
occurrence appears to have been wider than that of a normal case marker; but
it certainly included the I. meaning, which prevailed in PA.
A derivation of 8 from the pl.L. marker -s-u (-si) meets with insuperable difficulties. Meillet'a
statement (1936, p. 70-71) is too optimistic.
9o On the regular loss of-wafter u, see 3.183 and 4.333 (on zinuor).
96 According to K . H. Schmidt, Dativ und lnalrumental im Plural (Glotta 41, 1963, p. 1-10), the
Celtic pl.D. forms should rather be traced back to -bhis (Cf. the Indo-Ir. pl.I.).
94

104

Historical Morphology

[5.231-5.234)

The pl. endings -(o)vk', -(a)tvk', etc. (2.323), though recalling the Indo-Iranian
pl.I. (Skt. -bhis, OP. -bis), need not be traced farther back than PA : judging
from the occurrences of Gk. -phi, the PIE morpheme was used for both the sg.
and the pl., perhaps mostly for the pl., as in the Mycenian inscriptions. For the
addition of -k', the pl.N. could serve as a. model.
5.232. A PIE case ending is recognizable in the o declension : -oy admittedly
reflects *-osyo, i.e. a G. ending well preserved in Greek (Homeric -oio/-oo) and
Indo-Iranian (Skt. -asya, OP. -ahya). In Armenian, it has taken on the syntactic
function of the Ahl. and D., but not of the L., which has merged with the NAcc.
in consequence of the loss of the original ending ( *-oi). In the mixed declension,
however, there is a specific L. form: aygw-of, telw-of (3.133). The ending
obviously stems from the o inflection, although ancient *-i(y)o- stems (hogi) do
not take it. As to -f, \vhether a reflex of *-dhy(V) or of *-gh followed by a front
vowel (Pedersen, 1905, p. 223-224), it is not derivable from any PIE case marker.
On the other hand, -of also occurs as a G(DL.) ending in miof (3.181, Note) and,
unexpectedly enough, in knof, an originally *-a stem (3.17, 4.332). In view of
this situation, one will readily presume an adverbial origin for the ClArm. case
ending. Unfortunately, the background of the Greek local adverbs in -othi, quoted
by Pedersen and Meillet (1936, p. 73), is not clear enough to allow for a satisfactory
explanation of -of.
5.233. In the normal a declension 97 , the GDL. form (lezu-i), though declared
'enigmatical' by Meillet (1936, p. 72), does allow for an explanation in terms of
PA morphological innovation. The loss of the original endings (G. *-as, Dl *-ai)
has been compensated by the addition of -i to the curtailed stems. As pointed
out above (3.133), the new morpheme, abstracted from the GDL. of the i
declension (on which see 5.131), was not added to the polysyllables in -i < *-i(y)a,
for which the o declension provided a more suitable ending (aygi, GD. aygwoy,
after Jwgi, Jwgwoy).
The above explanation finds support in such L. forms as mifi, giseri (3.181).
The selection and propagation of -i as a case ending is by no means surprising,
in view of the productivity of the i declension in PA.
5.234. In the other declensions, the lost endings (G. *-es/os; D. *-ei; L. *i-)
have not been supplemented: the remaining stem, standing in contrast to the
NAcc., was [Link] to function as a full-fledged case form: so bayi, zardu, as against
bay, zard (5.131); mawr, k'ef, tayger, as against mayr, k'oyr, taygr (5.141); gafin,
The other a declension (Ila.) ha.s no PIE background. I t runs parallel to the o declension, " hich
apparently waa not felt suitable for foreign names.
97

[5.234-5.235]

Historical Morphology

105

ikan, as against gafn, juk1i (5.144), etc. It is of course impossible to state which
PIE case is represented in the curtailed forms : all three case markers were to
be dropped alike. However, an impoverishment of the PIE declension ,pattern in
early PA, before the stress shift, is not very likely, since CIArm. displays a fairly
similar one, in spite of the changes that took place in the meantime.
5.235. The Ahl. ending -e ( < -ey), at first sight, seems to be somehow related
to -oy, the Ahl. ending in the o declension. Meillet (1936, p. 73) cautiously suggests
a derivation of both from the PIE adverbial morpheme *-(e/o)tos (Skt. mukhata'IJ,
'from the mouth, in front'; Lat. funditus from the bottom', intus 'from inside',
etc.). More recently, some comparativists, paralleling -oy, -e with the Abll. ending
documented in Luwian (-ati) and Lycian (-edi, -adi), claimed to have detected an
'isogloss' including Armenian together with the Anatolian IE languages 98 These
constructions, though not objectionable from the phonological point of view,
cannot be readily accepted.
In the first place, e and o, as vowel alternants (4.323), are expected to occur
in the same paradigms. Now o (in -oy) is obviously the stem vowel; e (in -e), on
the contrary, can.n ot have ever been part of a noun stem: indeed, -e is often
added to GDL. forms (3.183). The above mentioned theories, therefore, amount
to equating a complex termination (-oy < *-o- + a case marker) with a single,
indivisible morpheme (-e). At this juncture, a remarkable peculiarity, underscored
by P edersen (1905, p. 223-224), has to be taken into account: the demonstrative
pronouns ays, ayd, ayn, when followed by a noun, display the same form in the
D., the L., and the Ahl. :
D.
L.
Ahl.

aysm lerin 'to this mountain'

y-aysm lerin 'on this mountain'
y-aysm lefne 'from this mountain'

99.

The conclusion to be drawn is that -e does not reflect a case ending, but an
originally independant particle (postposition), which was added to the L. : i
lerin 'on the mountain': (Abl.) i lefn-e 'from (on) the mountain'. As Bugge
perceived long ago 100, Arm. -e is derivable from *eti (Skt. ati 'over, beyond';
OCSI. otu 'from'; Gk. eti 'also, still'.
W. l\J. Austin, Ia Arn1enian an Anatolian Language ! Lg 18 (1942), p. 22-25; V. Pisani,
Armeniache Afiacellen, DieSprache 12 (1966), p. 227-236. On the Anatolian Abll. ending, see E. Laroche,
BSL 55 (1960). p. 163-166.
tt Did the same rule formerly apply to adjectives, too, in the same position? The Abl. form canu,
beside canui, quoted by Jensen ( 147) might have occurred in such a phrase as: i canu befni 'from
a heavy burden'. Cf. i nor handerjl 'from the new garment', Luke 5.36.
100 KZ 82 (1897), p. 75.
98

106

Historical Morphology

(5.236-5.237)

According to the above explanation, the genuine Abl. form in the u declension is found in
zardu-t, xratu-l, etc. (3.151). The lack of u in gah,.e, oovl, etc. is an effect of analogy. The model has
to be sought in the a declension : before the addition of -i (5.233), the L. of am 'year' must have been
y-am (later: y-ami). Hence the Abl. y-ami. The analogical proce88 started with the substitution of
l to *ie in the ' declension : i srti 'from the heart', as against L. i arti. As a result, the paradigms
differ in the I . only (3.132, Note).
liieillet (1936, p. 95-96) and Mann (1968, p. 64) are inclined to ascribe a djfferent origin to the CIArm.
preposition i (y), according to whether it is used with the L. (or Acc.), or with the Ab!.: the meanings
('in, into'/'from') seem mutually exclusive. But Pedersen's argument removes the djfficulty.
NOTE.

5.236. The pl. GDAbl. marker -c' is admittedly [Link] to a [Link].l

suffix *-sko-, reflexes of which [Link] found in the Slavic and Germanic languages:
Goth. mannisks 'human' (from manna 'man'); AS. peodisc 'gentilish' (peod 'people,
nation'), etc. Accordingly, the prototype of, e.g., mardoc' must be figured out as
an adjective : *mrto-sko- 'human', which [Link] to be used, firstly, as a substitute
for the G. , and afterwards for the D. and the Ahl. (Cf. the development of -O'!f,
5.232). Once acknowledged as a mere case marker, -c' spread through the whole
nominal inflection, including adjectives and pronouns.
5.237. The -ean ending. Aside from the GDL. of most nouns in -iwn, which
reflect the expected zero grade of the stem (sean < *ki(y) 0 n-: 5.144), -ean also
occurs as a GDL. ending, together with the cognate Ahl. (-ene) and I. (-eanib),
in the declension of certain verb derivatives (3.152, 3.331 b). The fairly numerous
action nouns in -st, with -t < *-ti (4.22), originally belonged to the i declension;
and so do, indeed, those which display an invariable vowel (a, e) in the [Link] syllable.
Nouns in -st, therefore, fall into two contrasting paradigms :
GDL.

I.

imast 'understanding, sense'

imasti
[Link]

korust 'loss, perdition'

korstean
korsteamb

Whatever the reason of this distribution, one has to inquire whether the second
paradigm is morphologically related to the i declension. The following parallel
supplies the [Link]\ver :
GDL.

mah 'death'
mahu-an (or mahu)

I.

mahu-amb

korust
korstean <
korsteamb

*korusti-an

In korstean, hangstean, etc. -ean is an extended form of -i. [Link] accounts for
its further expansion : words in -und, -urd were formerly inflected as o stems,
as they still [Link] in the pl. In aniis (I) 'month', -ean occurs in the L. only (y-amsean,
3.181). On tumfean (from tiw 'day-time'), see above, 3.17.

(6.24-6.241)

Historical Morphology

107

According to Pisani (1951, p. 5-0-51), -ean is nothing [Link] than the derivational suffix -ean
(3.322 d), turned to a case ending. Indeed, in Iii iwlcy nardean aznui 'a jar of ointment of pure nard'
(~lark 14.3), it is difficult to decide \vhetber nardean has to be interpreted as an indeclinable adjective',
or as a G. form (instead of nardi) 101. At any rate, the origin of the suffix is uncertain.
NOTE.

5.24. PRONOUNS. The diachronic explanation of pronominal inflection raises

more perplexing problems. Besides deviations in sound change (4.334, Note), we
are confronted with puzzling stem forms, especially in personal pronouns (1st sg.
in-; 2nd pl. je-), but also in ink'n (GDL. ink'ean) 'self', inc' 'what~ (when
unstressed : 'something'). These latter look like compounds : k'n- is perhaps
derivable from the reflexive pronoun *swe-, and -c' from the interrogativeindefinite *ki- (kWi); but no etymology can be constructed on such vague
presumptions.
With regard to these difficulties, it seems appropriate to proceed from the
simplest paradigms t o the more intricate ones.
5.241. The demonstrative pronouns, together with the related local adverbs,
have developed to a close-knit system, in which three pronominal stems recur
in combination with various particles (Meillet, 1962, p. 5-29) 102 The relation
of the stems to the speech situation is more systematic than in other IE languages:

so- 'this' (with reference to the speaking person) < *fco-, along with *fci-:
Hitt. ka-, ki- 'this here'; Gk. *ki- in semeron 'today' (< *ky-ameron); 0081. si;
Goth. hi-; Lat. ci-s, ci-tra 'on this side';
do- 'that' (with reference to the addressed person) < *to- : Indo-Ir. ta-; Gk. to-;
OSOL to- ; Goth. pa;
no- 'that' (without reference to either) < *no- : cf. Hitt. eni-, uni- 'that';
0801. onu; Lith. antis, ans.
The peculiarities of the Armenian system appear at one glance: !co- instead of
the widespread alternant *fci-; *no-, as againstvarious cognate stems; and above
all, the rather uncommon use of *to- as a 2nd person demonstrative.
So-, do-, no- combine with the follo,ving particles: *ai- (ays, ayd, ayn), which
is also found in Indo-Ir. (OP. aita, Skt. e-{ui, e-td) 103 ; -in (soyn 'this very, the
The same ambiguity is perceptible in ModArm. afavotyan (from aravot 'morning') : afavotyan
lam 'morning hour'. The GD. meaning appears in afavotyan dem 'towards the morning'.
102 This most accurate study was first published in 1898 (MSL, vol. 10). Pede111en's monograph,
Lu pronoma dhn01!8trati/a de l'ancien armhlien, Kobenhavn, 1905, is meant to improve upon Meillet's
views, which on the whole remain quite valid.
103 Pede111en (1905, p. 240) parallels ayn 'that' with Skt. anyd- 'other' ( < anyo-; cf. 4.451): PA
ayno should have served as a model for the coining of ayso, ayd<T. This explanation has been taken
over by G. Bonfante (Mtlangea H. Pedtrten, Copenhague, 1937, p. 20).
101

108

Historical Morphology

[5.241-5.242)

same', G. sorin, D. smin, etc.; doyn, noyn; astin 'in this very place', etc.); -ik,
in the extended forms of ays, ayd, ayn: G. aysor-ik, D. aysm-ik, pl.N. aysok'-ik,
etc., and in awasik, ahawasik 'here is, behold', etc. 104 Regarding the anaphoric
pronouns: sa, da, na, conflicting explanations have been put forth by Meillet
(sa < PA *so-ay : 1936, p. 88) and Pedersen (sa < PA *so in unstressed
position: Les pronoms demonstratijs ... 13), none of which is safe from
objections. Meillet's theory simply amounts to positing two different arrangements
of the same constituents : *ay-so vs. *so-ay.
As a result of these developments, the simple demonstratives (PA. *so, do, no
were reduced to the status of enclitic pronouns, and eventually, of invariable
deictic particles (2.324).
5.242. The interrogative pronouns, in all the IE languages, are built up on the
correlated stems *ko- (kwo) and *ki- (kwi-), the original distribution of which is
obscured by diverse dialectal developments. In Armenian, as in the Slavic
languages, *ko- refers to persons, *ki- to things :
ov 'who?'

G.
DL.
I.

Cf. o-k' 'anyone'

oyr
um
(orov, from or)

uruk'
umek'
(omamb, from
omn)

z-i 'what ?'

z-inc'

er

(h)im

iw

Cf. *i-k' 'anything'

(supplemented
by inc')
irik'
imik'
iwik'

The pl. of ov is: oyk' 'which people?', GDAbl. oyc'. On zi, zinc', see above, 3.112.
The derivation of indefinite pronouns from the interrogative, as evidenced in
the above paradigms, stems from PIE : ok', *ik' are parallelable with Skt. kar;-ca
'someone' (kab, 'who?'); Goth. hwaz-uh 'every, each' (hwas); Lat. quis-que 'id'
(quis). The other pair of indefinite pronouns: omn (pl. omank') 'someone', imn
'something' (indeclinable, but for the Ahl. imemni) presumably results from a
similar combination: -mn, like -k', may reflect some enclitic particle. However,
for want of comparative evidence, we are left with this presumption, unless we
endeavour, as Meillet does, to connect omn with Goth. sums, AS. sum 'some';
Gk. hamos 'somehow', oud-am6s 'no one', etc., i.e. with the reflexes of PIE *s 0 mo(1962 [1898), p. 30-32 ; 1936, p. 90). But the postulated connection is not expressed
in morphological terms, as it ought to be : indeed, positing a different prototype
( *somno- ?) would be nothing more than an ad hoc device, since the CIArm. pronoun
remains isolated anyhow los.
104
105

On the Iranian descent of awa-, aliawa-, see A. Perikhanian, REArm n.s., vol. 3 (1967), p. 21, fn. 7.
Against Meillet, it must be urged that the CIArm. paradigm is the outcome of a partial, not

Historical Morphology

[5.242-5.243)

109

The interrogative adjective or (I)' \vhich, what 1' displays the 'comparative'
suffix *-(e)ro- (Meillet, 1936, p. 89-90). Its being also used as a relative pronoun
is by no means surprising : cf. Engl. who, which, so that Pisani's divergent
explanation (Of" < *yoro-, an extended form of the PIE relative pronou.n *yo-:
1950, p. 181-182) turns out to be groundless.
5.243. PIE pronominal inflection was characterized by peculiar endings or
case markers, some of which have expanded to nouns and adjectives over more
or less extensive dialectal areas, e.g. the sg.G. marker *-syo (in -oy < *-o-syo,
5.232). All the ClArm. pronouns examined so far display the [Link]. marker -m:
sma, aysm (from earlier *suma, *aysum: 3.181), u1n, (h)im, orum. This allows for
exact parallels with pronominal D. forms in the cognate languages :

dm(a), (ay)dm <PA *dom(V) = Skt. tasmai, Goth. pamma;

um .
= Skt. kdsmai, Goth. [Link], Umbr. pusme.
Thus ClArm. -m reflects *-~ml 108 A trace of the final vowel is perhaps preserved
"in the indefinite pronouns : [Link]. ume-k\ ume-mn.
On the sg.G., see below, 5.245. The Ahl. of the demonstrative pronouns displays
an extended ending -an-e. Tentatively, we may suppose it to have come about,
firstly, in the [Link]. of sa, da, na : the addition of -e ( *-ey) to the L. would
probably have resulted in vowel contraction; -n- may have been inserted after
the declension of n stems :

L.
cf.

sma
i kolman

Ahl.

i smane
i kolmane

The new ending was propagated not only to ays, ayd, ayn (y-aysmane, instead of
*y-aysme), but, quite unexpectedly, to the plural, too (3.182).
Regular I . forms are preserved in sov-in (from soyn), iw, iwik', orov, while in
sov-aw, dov-aw, nov-aw (from sa, da, na) the I . marker is repeated after the
particle. From the ay- demonstrative, the I. is aysu (aysu-ik), aydu, etc. The substitution of -u for -ov is unexplained. The pl.I. forms (sok'awk', aysok'iwk', etc.),
obviously resulting from late PA developments, need not detain us here.

total, reshaping. Whatever the origin of mn, om.n could not fail to be attracted to the n declension :
hence the I. omamb, and the pl. omank', omarn:'. But the sg.G. (urumn) and D. (umemn) did not share
in the change of inflection.
108 The final diphthongs in Skt. ta~mai, Umbr. ~ei 'to him' eeem to result from independent
developments.

110
5.244.
N. du
G. k<o

Historical Morphology

[5.244-5.245)

Personal pronouns. The best preserved paradigm is that of the 2nd [Link]. :

< *tu (OP. tuv-am,

<

Gk. su, OCSl. ty, Lat. tu, OHG. du/du, etc.)

*two- (Skt. tva-, Gk. sos 'thy').

In the remaining cases, the stem is k<e- < *twe-, as in Gk. se (Acc.), seo (G.).
I n t he 1st [Link]., the N. es must somehow reflect *ego (Gk. eg6, Lat. ego, Goth.
ik) or, perhaps, *eghom (Skt. aham, OP. adam). Anyhow, the substitution of s
for c (or z) has to be accounted for. I t has been explained in terms of 'sandhi'
(Meillet, 1936, p. 57) : e.g. *ec tam > es tam ' I give'. The Acc.L. is (z-is, y-is},
if from *ins, is traceable to *em- +a particle (cf. Gk. eme-ge, Goth. mik, Venetic
me-go); but here again-sis not the expected phoneme. In both cases, the alteration
may be ascribed with some likeliness to the influence of the deictic particle -s
(2.324). As to the stem, it is plainly reflected in the G. form im (< *emo-, cf. Gk.
emos 'my'). Before s and j (D. inj}, m changed to n. Analogically, the oblique
case stem *eme- (Gk. G. emeo} became ine- : Abl. inen, I . inew.
Barring the 2nd pl.N. duk', obviously derived from the sg., the background of
the remaining personal pronouns is obscure : neither for the 2nd pl. stem je-, nor
for the Abl. endings (sg. -yn, pl. -ynf: in en, k' in, menf, j enf} is a diachronic
explanation available. In the D. however, the contrast of -j (inj 'to me'} and -z
(k'ez, mez, jez) points to *gh; and, indeed, there is evidence for a D. marker
*-ghi/ghei, at least in the 1st [Link]. : Skt. mahy-am ( < *meghi), Lat. mihi, Umbr.
mehe ( < *meghei}. Likewise Arm. inj < *emghi is used for the D. only, while
k'ez, mez, jez serve as AccDL. forms.
5.245.

Genitive and possessive adjectives.

In no pronominal paradigm is the sg.G. identical to the DL. (3.181). But while
the latter case forms are easily traceable to P I E models, the G. marker -r (sor-a,
aysr, oyr, er, etc.} seems to have developed in the course of the PA period, so that
the comparative method has to be supplemented by internal reconstruction.
To start with the personal pronouns: im, k<o, mer, jer occur as genuine G.
forms,

JO) when governed by a preposition (practically md 'instead of, for') :

tac'es noc'a and im ew and k'o 'thou shalst pay to them for me and for thyself',
Math. 17 ,26 (27).
or by a quasi preposition (or prepositional phrase) :
aiaJi im 'before me', Luke 4.7; vasn k<o 'for thee', Luke 4.10;
ert' y-ets im 'go back away from me', Luke 4.8;
ekayk' z-kni im 'come and follow me', Math. 4.17;

Historical Morphology

Lo.245)

Ill

2). in verb phrases (on which see above, 3.26) :

na. ew ayl ews oc'xark' en im 'and I have also other sheep', John 10.16;
oc' ... lueal e mer 'we have not heard', Act. 19.2.
In noun phrases, the same words behave as possessive adjectives, in that they
agree with the inflected noun: ordi im 'my son', G. ordwoy imoy, D. ordwoy imum,
etc. The syntactic connexions, therefore, depend upon intersecting paradigms, e.g. :

es 'I '
AccL.
(G.)

D.

is

im - G. imoy, D. imum ... pl.N. imk' 'mine', etc.

inj, etc.

The morphological background, ho,vever, is quite plain : im, k'o reflect adjectives :

im (I) < *emo- (Gk. em6s 'my')

k'o, k'oy- (I) < *two- (Skt. tva-; Gk. s6s 'thy'; cf. Lat. tuus),
and so do mer, jer. As in the Germanic languages, the comparative suffix (5.242,
last al.) has been worked up to derive possessive adjectives from the 1st and 2nd
pl. pronouns: cf. Goth. unsar 'our', izwar 'your'; AS. [Link], ecwer, etc. We thus
hold a. clue as to the origin of the G. marker -r.
The interrogative pronouns thro'v some light on the PA developments : the
sg.G. oyr, as Meillet rightly assumes (1936, ,p. 87), is somehow related to Skt.
kasya, Lat. quoiu(s), i.e. to the reflexes of *kosyo; and so is er to Av. eahya,
Homeric Gk. teo, OCSl. ceso, Goth. hwis, i.e. to those of *kesyo (*keso) 101 Not
only do these parallels fully account for the stem forms oy- (< *kosyo) and *ey(< *kesyo), but at the same time they allow for a chronological statement: it
appears that -r has been added to those forms after the dropping of final vowels.
At that stage, the PA adjectival suffix *-ro had changed to a bare case marker.
We also have to conclude that the G. forms of the derived indefinite pronouns
(ur-uk', ur-umn, ir-ik') have been created still later.
The place of the demonstrative pronouns in this development remains uncertain :
a vowel may have been lost after r in *sor, dor, nor (ClArm. sor-a, sor-in, ayM,
etc.), but this cannot be verified. Another perplexing item is the 3rd p. reflexive
adjective iwr (I) 'his, her'. Meillet's etymology ( *sewe/o-ro- : 1936, p. 92), implying
_the loss of the penultimate, is untenable, and Pisani's surmise of a 'crossing' of
G. and D. forms (1950, p. 186) raises suspicion. A PA adjective *ew < *sewo(Gk. he(w)6s, Lat. suus) may have come to be interpreted as a pronominal G.
io7 The problems tha.t [Link] from the occurrence of *&O beside -ayo, [Link] of a stem .form ,u.
( k"'t) in the ki- ( kwi.) [Link] need not be [Link] here.

112

Historical Morphology

[5.3-5.32)

Hence the addition of the -r marker, as in oy-r, *ey-r. In ClArm., iwr is again
a possessive adjective following, expectedly, the o declension lo8.
5.3.

VERB INFLECTION.

Our conception of the PIE verb system mainly relies upon Indo-Iranian and
Greek data. In these languages, all verb forms are derived from three fundamental tenses : present, aorist and perfect, expressing different 'aspects' (Cf. 3.222).
The same pattern, variously reshaped, underlies the conjugation in most
cognate languages, including Armenian. Only the Anatolian IE languages reflect
a different, presumably more archaic, situation.
Armenian is closely related to Greek and lndo-Iranian in that, [Link] these, it
has reflexes of the PIE imperfect, a past tense derived from the present stem
(Hom. Gk. leipo 'I leave', ~pon 'I left'), and of the e- prefix (augment), which
was added to the past tenses in the indicative (Gk. impf. eleipon; [Link]. elipon, etc.).
In the oldest texts, its use is optional : Hom. Gk. pMron/epheron; Vedic bharam/
abharam 'I bore, brought'. This may account for the fact that, in ClArm., the
augment is not prefixed to polysyllabic aorist forms : beri 'I brought', 1st pl.
berak', etc. as against 3rd sg. eber (= Gk. ephere, Skt. abliarat). Its original
function, however, was soon obscured 109 : in some anomalous verbs, indeed, the
augment has expanded to the aor. subjunctive and to the participle (3.255 c, and
Note). Neither in lndo-Iranian nor in Greek did a similar expansion take place.
5.31. The Armenian conjugation, as we saw (3.22), hinges on the contrast of
present and aorist stems. The PIE perfect is lost. But it may have survived in
early PA as a marginal member of the paradigm : such presents as gitem 'I know',
goy (3.223) are traceable to PIE perfect stems: *woid- (Skt. veda, Gk. (w)oUa,
Goth. wait 'I know'), *wos- (AS. waes 'I was') 110 Similarly, in OCSl the
inherited perfect vUe ' I know' (= Lat. uidi) has been converted to a present :
vemi.
5.32. There is substantial evidence for the mutual independance of the PIE
present and aorist. Some roots yielded present stems only, e.g. *es- 'to be', *ei'to go', *ag- 'to lead, bring', *blier- 'to bear'. In other instances, the comparative
data point the to priority of the aorist: an archaic root (or, alternately, -s-) aorist
is reflected in Skt. &lam, I gave', Gk. edo-ka, 1st pl. edomen, OCSI. daxu
[Link] the [Link] form *ew, the objection is tha.t ewr instead of iwr does not ooour in the
manuscripts. See however, fn. 77.
109 Yet, in some modern [Link] of the Van area., the augment still occurs as a mark of the aor.
3rd eg., not only in inherited forms, but also in analogical innovations : e.t'al, from t'alem 'I plunder';
ek'ic', from l;'c'em 'I throw").
110 The o vowel in c'ogay 'I went' is amazing. ls this aorist derivable from a perfect stem *kyow t
108

Historical 'Morphology

(5.32-5.321)

113

(< *do-s-}, Arm. etu. The divergent present stems in Skt. dadami, Gk. didomi,
OCSl. dami, Arm. tam belong to a more advanced state of dialectal diversity.
Indeed there has been a general tendency to 'conjugation', i.e. to the coupling
of present and aorist (or perfect) stems, either by deriving a new stem from the
same root, as in the above example, or by associating stems from different roots.
Both processes can be easily illustrated. The PIE present *bhero is reflected in
Arm. [Link], Gk. phero, Germanic *bera, etc. In the Germanic languages, the
paradigm has been completed by a preterite (Goth. bar, 1st pl. berum; AS. baer,
baeron), the model of which was supplied by such inherited perfects as was,
nam, sat, etc. In Greek and Latin, the past tenses are not built up on the root
*bher- : Gk. aor. enenkon, Lat. perf. tuli. Such 'suppletive' forms are not missing
in ClArm. : 'ampem, arbi; unim, kalay, etc. (3.255).
But in berem, beri we witness a third process : the aorist is traceable to the PIE
imperfect, and so are other Armenian root aorists. Neither in Greek nor in IndoIranian is there any example of such a tense shift, which implies a thorough
recasting of the former verb pattern. Besides, it appears that the CIArm. imperfect
is not directly related to the PIE imperfect, since the latter, where preserved,
holds the post of the aorist in the new paradigm :
Gk.

phero

(impf.) (e)pheron

(aor.) enenkon

I
Arm.

berem

(impf.) berei

([Link].) beri, 3rd sg. eber

In a. previous section of this Introduction, the CIArm. verbs have been classified
without regard to the origin of the present and aorist stems (3.25-3.255). Now,
taking into account the comparative and historical evidence, and trying to check
the inherited stem forms, we can outline a somewhat different classification.
5.321. Inherited PIE presents
a) The present is coupled with a root aorist, which reflects the PIE imperfect,
as in berem, beri and a few similar verbs (3.252). But a number of verbs in -anem
also belong to this class : in hare'anem 'I ask', aor. harc'i, one cannot fail to recognize
the Arm. reflex of a PIE present stem *n(k)-ske/o- (Cf. Skt. n-cchami, Lat.
posci5), and the aorist 3rd sg. eharc' exactly answers to the Skt. imperfect aprcchat
(Cf. OP. aprsam 'I asked'), just as does eber to Skt. abharat. Consequently, the
-an- extension in the present must be ascribed to a morphological innovation :
the earlier paradigm was doubtless : *harc'em, harc'i. I n fact the ancient present
sometimes occurs along with the extended one : lizem (= Gk. leikhi5), lizanem 'I
lick'; ofogem (afogem}, afoganem 'I irrigate'; nerkem, nerkanem 'I dye', etc.

114

Historical Morphology

(5.321-5.323)

b) The present is paired with a -c'- aorist, e.g. p'aylem ' I shine', aor. p'aylec'i.
Insofar as comparative evidence is available, the simple -em/im and -am presents
stem from PIE, while the corresponding aorists have mostly developed during
the PA period. A great number of these verbs are denominative. Now, PIE
denominative verbs had no aorists; the -c'- aorists, therefore, must be derived
from extended imperfects (see below, 5.433).
Some verbs have, exceptionally, an -ac'- aorist opposite to an -em present (3.252). These
do not belong here: on gilem 'I know', aor. gitac'i, see above, 5.31.
As to the verb 'to be' (3.223), its background is partially obscure. Although em reflects the athematie
preeent umi, ti can by no means be traced to the imperfect; and Meillet'e attempt to derive it from
the PIE perfect is not convincing (1936, p. 127). The paradigm, however, can be brought into
agreement with that of berem (5.332).
NOTB.

5.322.

Inherited PIE aorists

The corresponding presents mostly display the -(a)n- extension :

dnem ( < *dinem) ' I lay, put' : aor. edi (Cf. Skt. ddham, Gk. ethe-ka; OCSI. dkU
< *dhe-s-)

afnem 'I make, do'

arari (= Gk. &-aron 'I fitted, constructed') 111

lsem ( < *lusem) ' I hear'

luay (Cf. Vedic d9rot 'he heard', imper. yrudhi;

Gk. 3rd pl. ekluon, imper. klathi, klute)

.
gtanem 'I find'
gti, 3rd sg. egit (= Skt. avidam 'I found'; Gk.
eUon, (w)idon 'I saw')

lk'anem 'I leave'

lk'i, 3rd sg. elik' (= Gk. e'lipon)
On tam, etu, see above, 5.32. Several other verbs, certainly or presumably, belong
to this class, e.g. klanem 'I swallow', [Link]. kli, ekul; linim 'I become', aor. ele, etc.
R egarding the 'suppletive' forms, it can be assumed that both the aorist and the
present are equally archaic; the comparative evidence, however, is deficient :
ekn 'he came' doubtless equates with Vedic dgan 'he went' ( < *egemt); but
neither for eki 'I came' nor for ganl- 'I come' are reliable parallels to be found.
From the above examples, it appears that one, at least, of the regular verb [Link] (3.253 f)
has originated in the merging of two formerly different paradigms : harc'anem, harc'i, on the one hand,
and gtanem, gti; lk'anem, lk'i on the other do not have the same background.
NOTE.

5.323. Not all CIArm. verbs can be assigned to such or such class. In many
instances, both the present and the aorist are somehow traceable to PI E stem
patterns; but the 'conjugation', as defined above (5.32) seems, and sometimes
proves, to result from PA developments and arrangements. In this respect, the i
111

Thia verb is not documented elsewhere. \Ve have to do here with a Greek-Armenian isogloee.

115

Historical Morphology

(5.323-5.331)

stem aorists (2.213, 2.243) r aise a particularly perplexing problem, which will be
discussed below (5.352).
5.33. THE p A VERB SYSTEM : A TENTATIVE REOONSTRUOTION. The CIArm.
imperfect is not directly traceable to any PIE past tense. As to the pres. subjunctive, it looks like a combination of the pres. stem with the subjunctive of
the verb 'to be' : *bere- + ic'em > *1Jereyc'em > beric'em; orsa- + ic'em > orsayc'em, etc. (Meillet, 1936, p. 121, 126) 1 12 Regarding the -im and -um presents,
the developments can be supposed to have been:

*hayi + ic'em > *hayiyc'em > hayic'im

*t'olu- + ic'em > *t'oluyc'em > t'oluc'um
This amounts to assuming that *-iy- > -i-, *-uy- > -u- (Cf. 3.231 c, Note}, and
that in the endings e was eventually substituted bu i, 1i, in agreement with the
pres. indicative.
Consequently, in reconstructing the earliest PA verb system, we need not
take into account these secondary formations. The paradigm thus reduces to the
present and aorist indicative, the imperative, and the aorist subjunctive 113
According to our classification, we shall consider separately what we may term
'one stem verbs', e.g. berem (5.321), and 'two stem verbs', e.g. dnem (5.322). These
denominations, of course, do not suit t he CIArm. conjugation as described above
(3.22) : they simply refer to the original difference.
5.331.

One stem verbs

Indicative
Present
Aorist

berem
*hare' em
(harc'aneni)

beri
harc'i

I mperative

Subjunctive

ber
hare'

.
.
,
L--'
benc , ut:f'C es .. .
harc'ic'

These paradigms concord with :

Gk.
Skt.

phero
prcchami

epheron
ciprccham

phbe

prccha

But, while Greek, Vedic and Old Iranian have preserved the PIE subjunctive,
Arm. beric' is obviously the result of innovation. The -ic'- morpheme does not
Such quasi compound verb form.a, it must be granted, are not in agreement with what we know
of PIE morphology. But J'lleillet's alternative explanation (berie'em < bher-oi-ake) does not improve
the caee. See E. Evangelisti'a criticism : L'impufello armeno e l'uo pt'deritok dell'oUaliw indoeuropeo
(Arona, 1955), p. 11 and fn. 5.
113 The infinitive and participle will be dealt with later (5.44-45).
112

116

Historical Morphology

[5.331-5.332)

go back to any known mood marker : at the best. -i- might be related to the
optative suffix *-ye/i (Meillet, 1936, p. 122; Pedersen, 1905, p. 207); yet the
morphological environment, as Meillet admits, does not favour this conjecture.
On the other hand, evidence for a pres. stem extension *-ske/o- is found all
over the IE area., e.g. in harc'(an)em and its [Link].n and Latin cognates
(5.321). A variant *-iske- is documented in Greek (keurisko 'I find', aor. heU-ron)
and Latin (reminiscor 'I remember'). Supposing it to have developed in Armenian,
too, we might parallel berem, beric' (< *bherisko) with such verb pairs as :
Gk.
Lat.

steromai 'I lack, lose' : sterisko 'I deprive'

*paco 'I contract'
: pacisco(r) 'I contract, stipulate'
(3rd sg. pacit, pl. pacunt, in archaic texts).

The formal resemblance, it must be granted, has no semantic counterpart, and

might be regarded as casual. Yet, since the original meaning or function of *-ske/ocannot be assessed precisely enough, it is not a priori impossible that, in PA,
derived *-iske- presents should have
taken over the function of the former sub,
junctive, and been worked up for a modal contrast 114
5.332. Let us now return to the verb 'to be'. Its paradigm never included an
aorist (5.32), and, in its ClArm. state, it is not unlike that of berem, as reconstructed
above:
Indicative
Present
Imperfect

em
ei

Imperative

Subjunctive

c
c
er ek'
ic em, w es ...
'
It will even prove quite similar if we say '[Link]' instead of 'aorist' or 'imperfect',
and if we assume that in the subjunctive the 1st sg. (*ec') and pl. (*ic'uk') have
been substituted by ic'em, ic'emk' in analogy of the pres. indicative.
Most of the past tense endings [Link] hard to explain in terms of PIE verb inflection.
We only know that beri reflects the PIE imperfect. As to ei, its closest cognate
is the Greek optative eien, 1st pl. eimen (< *esi-men) l1s. Now, in both tenses,
the 1st sg., 1st and 3rd pl. endings are the same :

e-i

: ber-i

e-ir

er (<

*eyr)

ber-er
eber (<:: *ebheret)

Godel, 1965, p. 31.

115 See Evangelisti's study (quoted above, fn. 112), especially p. 30-38. On the relationship between
optative and past, generally: E. Benveniste, Pretb-it et optalif en [Link]~en. BSL 47 (1950), p. 11-20.
A modern parallel is the English idiom he' would do = he used to do.
114

117

Historical Morphology

( 5.332-5.333)

e-ak'
e-ik'

ber-ak'
ber-ek'
ber-in

Only in the third column is the pres. stem *bhere/o- recognizable. The common
endings -i, -ak', -in, whatever their origin, point to convergent developments,
and the conclusion to be dra\vn is that, at a remote stage of PA, only one [Link]
tense stood opposite to the present : the PIE 'aspect' contrast had been
obliterated.
5.333.

Two steni verbs

Cf. Gk.

Indicative
Present
Aorist
dnem
edi
lk'anem
lk'i, elik'
(lelpo)
elipon

Imperative

Subjunctive

dir, dik'
lik'
lipe

edic', die' es ...

lk'ic'

In ClArm., these paradigms are on a par with that of berem, save that edi, lk'i,
etc., unlike beri, are genuine aorists. The subjunctive is built up on the aorist
stem, as is shown by the following parallel :

ara, ararek'

(arnem)

ararisko

Gk.

baron

araric

*arare

Both paradigms bear evidence to an *-iske/o- present, derived from the reduplicated [Link]. stem *arare/o-. In Greek, this present is used in the indicative,
whereas in Armenian -ic'- has become a subjunctive marker: araric' stands in
contrast to the pres. indicative afnem.
Genuine aorists, \vith few exceptions (3.255 c), are inflected like beri : the
thematic aorist stem *like/o-, e.g., is well preserved in lk'er, elik' ( < *eliket),
lk' ek'; the 1st sg., 1st and 3rd pl. have taken on the usual preterite endings. Our
above statement on the PA unique past tense is thus borne out.
NOTE.

Two verbs deviate from the above pattern :

ert'am ' I go'

lam 'I give'

c'ogay

etu

ert', ert'ayk'
tur, tuk'

-''
_,,. c
.,,.
ayec , .,.
ic u ...
tac', tac'u ...

Regarding the former, although no satisfactory etymology has been brought forth so far, it is worth
pointing out that the PIE verb *eim.i , for which it has been substituted, had no aorist. We may therefore
presun1e t hat ert'ani originally belonged to the one stem verb class : barring the aor. indicative, the
paradigm runs parallel to berem, ber, beric'. The question then arises whether c'ogay is the substitute

118

Historical Morphology

[5.333-5.34)

of an earlier past tense built up on the present stem, or whether that [Link] tense survives 88 the CIArm.
imperfect. I n this latter case, ert'ayi would have to be traced as far back as ei. For want of other clues
a.a to the origin and development of the ClArm. imperfect at large, the case must be left undecided.
The subjunctive tac' ( < *da-ako : Meillet, 1936, p. 133) displays the zero grade of the root vowel,
as [Link] the full grade in etu, tur, tuk'. This anomaly can be explained in terms of preservation of
an archaic feature i1s, or, alternately, by supposing tac' to be derived from the present. Such a. derivation,
unexpected though it is in a. two stem verb, would [Link] for the difference between tac' [Link] eL-ic',
edic', elec' : the [Link] tac', being originally [Link] to the [Link] indicative, could not receive
the augment.
At any [Link] a greater difficulty lies in t he pres. indicative, for which no [Link] is to be found except
in [Link] : do, let pl. damua, like tani, [Link] reflects a.n [Link] present without vowel [Link].
This quite unusual inflection is all the more surprising a.a Armenian and [Link] do not belong to the
same dialectal area..

5.34-. THEMATIC AND ATHEMATIC INFLECTION. PIE present as well as aorist

stems fall into two main classes, according to whether they do, or do not, contain
the so-called 'thematic vowel', i.e. the *-e/o- extension (Gk. pher-o-men 'we bear'
as against i-men 'we go'). In ClArm., except for some irregular aorists, the contrast
is no longer perceptible. The inflection of the -em presents, in particular, proves
to be a mixture of original thematic forms, as well as athematic, coined in analogy
of the verb 'to be', which in turn has been partially altered under the influence
of the thematic type.
The following table is a tentative illustration of this mutual overlapping :
Athematic inflection

Thematic inflection

em < *es-mi
es < *es-si
e+-- - - - - -- - - emk' < *es-mes ----+
ek'
en < *s-enti - - - - --+

berem
beres
ber e < *bher-e-ti
beremk'
berek' < *bher-e-te (+ -k') 111
beren

The thematic inflection is better preserved in the imperative and the subjunctive:
1st sg. beric' < *bherisk-o; 1st pl. berc'uk' < *bherisk-o-mes 118. Reflexes of
thematic inflection also occur in the aorist, whether derived from the PIE imperfect
(berer, eber, berek' : 5.332) or from the thematic aorist (lk'er, elik<, lk'ek': 5.333).
As I assumed in a previous study (1965, p. 36).
111 If -k' in the 1st pl. reflects *8, a.a in the pl.N. (5.221), its occurrence in the 2nd pl. must be
ascribed to analogy.
118 The development can be roughly traced as: -om,ea > -ow(e)~ > -uk'. This implies, at a certain
stage, a change of intervocalic (or postvocalic 7) m to w, 88 in anun (4.331), awur (4.452). In emk'
< *eamea, the change wa.s not to take place.
118

[6.345.35)

119

Historical Morphology

On the whole, the changes resulted in a levelling of verb inflection. The variety
of present and aorist stems has no bearing on the personal endings (3.23). A
similar statement, we remember, had to be made regarding noun inflection (5.11).
5.341. Many terminal morphemes remain unexplained : so -r in eir, er, berer;
-w in the mediopassive aor. 3rd sg. (beraw, elew); the 1st pl. past tense ending
-ak' (eak', berak'); the 2nd pl. -fik' in the aor. subjunctive, together with its
cognates (3.236); and the imperative endings -(i)r, -(a)ruk'. Instead of discussing
the various hypotheses that have been made as to their origins, I would rather
underscore two facts which may account for the unavoidable uncertainty of
such attempts.
Firstly, in consequence of the stress shift, many verb forms were to lose their
endings, and would have been confused, had not the original terminations been
extended or supplemented (Meillet, 1936, p. 123-124, on the thematic past tenses).
Furthermore, the PIE contrast of active endings vs. middle has been obliterated.
The aor. subj. 1st pl. ending -uk', common to the active and the mediopassive
inflection (berc'uk' 'we shall bring, or be brought'), is traceable to a PIE active
termination {see above), and so is the 1st sg. zero ending {< *-o) in berayc' as
well as in beric'. On the other hand, the imperfect 3rd sg. berer 'he brought, or
'va.s brought' is supposed to reflect a PIE middle verb form: *bher-e-tor {Meillet,
1936, p. 127). This explanation, though consistent with the rules of sound change,
cannot be recorded without discussion : not only is the survival of only one middle
ending a rather strange case, but, above all, any conjecture about berer ought
to be [Link] with respect to er 'he \Vas' ; for, all the imperfect forms seem somehow
related to, or reshaped after, the past tense of the verb 'to be' (Cf. 5.332).
5.35.

Thus, the next topic to be approached is the expression of the VOICE

CONTRAST. In the ancient IE languages, as a rule, the personal endings serve as
voice markers :
Gk.

phb-o 'I bear'

: phb-o-mai 'I win; I am borne, carried'
ephere 'he bore' 119 : epher-e-to 'he won; he was borne, carried'

This situation apparently continues in the ClArm. aorist (3.232). But if we look
at the mediopassive terminations from the diachronic point of view, our analysis
proves inadequate: in reconstructing the earlier inflection pattern, we have to
separate the personal endings from the -a- morpheme with which they have
blended (3.234, Note). It then appears that, in P A, voice was chiefly expressed
In phbo, the ending is no longer separable from the thematic vowel; in
dropped.
119

~phere,

*I ha.s been

120

Hist-0rical Morphology

(5.35-5.352)

by the alternating morphemes e/i in the present, f!J/a in the aorist: the voice
markers had been shifted from the last constituents of the verb forms to the
penultimate :

ber-e-m 'I bear'

eber 'he bore'

ber-i-m 'I am borne'

: ber-a-iv 'he was borne'

Contrasting personal endings are peculiar to the aor. indicative 3rd sg. (-0/w),
and to the imperative. Besides, a mediopassive ending -iwr sometimes occurs in
the impf. 3rd sg: koc'iivr 'he was called' (instead of the common form koc' er) 120.
The mediopassive inflection, as we know, does not cover the whole conjugation.
Tracing its development amounts to inquiring about the origin of the -i- and -amorphen1es.
5.351. The -im presents are ultimately traceable to a class of intransitive verbs
denoting physical or mental states. The comparative evidence, however, is not
homogeneous. There are, in the Baltic and Slavic languages, semi-thematic -ipresents : Lith. sedi 'he is seated', turi 'he has' ; OCSl. seditu 'he is seated', buditu
'he is awake' (these latter with -i- < *-i-); on the other hand, -e- presents are
found in Italic and Germanic: Lat. sedeo 'I am seated', OHG. habem 'I have'.
The ambiguity of Arm. i does not allow for a decision (Meillet, 1936, p. 107-108).
Yet, although such verb pairs as Lat. pendo 'l suspend', pendeo ' I am hanging'
can explain the original relation of berem to berim, one will rather relate Armenian
to Baltic and Slavic, and parallel beri to Lith. sedi, OCSl. seditu.
Only during the PA period did -i- evolve to a mediopassive present morpheme,
and expand from the indicative to the present subjunctive (beric'im, -is .. .) and
eventually to the aor. subjunctive (berc'is, berc'i, berc'in : 3.234, Note).
5.352. In this connection, we must also consider the mediopassive i stem
aorists (2.213, 2.343). As the related presents display various extensions 121 :
hangc'im 'I rest' (hangeay), martnc'im 'I fight' (marteay), zart'num 'I awake'
(zart'eay), yafnem ' I arise' (yareay), etc., these verbs very likely belong to the two
stem class (5.333). Consequently, hangi-, marti-, etc. may well be regarded as
genuine aorist stems.
This is not Meillet's opinion. He parallel<> y-areay with the Lat. present oriOT
'I arise', and accordingly derives the -eay aorists from PIE imperfects (1936,
Against 11feillet's construction (1936, p. 127), I am inclined to regard -iwr as an extended form
of *-iw, i.e. of an ending which might be related to t he [Link]. 3rd sg. ending -w (bera-o, ele-w), and to
the isolated 3rd sg. ic'iw '"ould that' : cf. the pres. subj. of the verb 'to be', 3rd sg. ic'e.
121 On -num and -c'im presents, see above, 3.254. In C!Arm. these extensions [Link] mosUy exchangeable
(Abrahamyan, p. 105-106), and we have no clue as to their original distribution.
120

[5.3525.353]

Historical Morphology

121

p. 115). The parallel, limited to Armenian and Latin, may turn out delusive.
I ndeed the most archaic representative of the PIE root *or- ( *H 3 er) is a root
aorist, preserved in Vedic (t'trta 'he arose') and in Homeric Greek ( 6rto, part.
6rmenos). The presents do not concord, and orior hardly goes farther back than
Arm. y-afnem (Cf. 5.32). On the other hand, there is no comparative evidence
for ancient *i stem aorists; but *-e- stems do occur in Greek, as aorist stems of
intransitive or mediopassive verbs : rhue-nai 'to flow', khare-nai 'to rejoice',
mige-nai 'to be mixed up, or have intercourse, with', and similarly in OCSL :
sede 'he sat', bUde 'he awoke' 122 ?11eillet quotes Gk. mane-nai 'to go mad' in
connection with the -im presents (1936, p. 108). But the Greek aorists seem more
closely related to the Armenian i stem aorists, save that in the latter the *-amorpheme (5.353) has been added to the *-e- stem. Now in Lithuanian, too, the
past tense of the i present verbs displays the same extension : -ejo (in turejo 'he
had', sedejo 'he sat') is a perfect match to Arm. -ea- (in hangeay, yareay). The
Armenian development need not be historically connected with the Lithuanian :
-eay aorists never correspond to simple -im presents. The coincidence, however,
gives weight to the above argument.
Some of the -eo.y aoriata have to be explained otherwise: e.g. 8<lrteo.y, if aartnum 'I startle'
is actually derived from the noun stem [jdi- (4.332). In erkeo.y ' I feared' (4.354), -e-, whether from
-ei- or from *-i-, is certainly the root vowel.
NOTE.

5.353. The aorist morpheme -a- is not ambiguous. Evidence for *-a- in past
tenses is found in the Baltic and Slavic languages, on the one hand (Lith. likau
'I left'; OCSL biraxu 'I gathered'), in the Italic languages on the other (Lat.
eram 'I was', legebam 'I was reading', legerarn 'I had read'; Oscan fufan.s 'they
were') 12s. In Armenian, -a-, once specialized as an aorist marker (beray 'I was
brought'), expanded from the indicative to the other moods : imper. pl. ber-a-ruk'
(but sg. ber-ir). In the subjunctive, -a- occurs in the 1st sg. only: ber-a-yc', 2nd
berc'is, etc. ; but the subjunctive of the i stem aorists still bears evidence to the
earlier, regular inflection: p'axeayc', p'axic'es (< *p'axeayc'es: 3.234, Note).
Originally, the -a- aorist \Vas not bound to the mediopassive conjugation,
and traces of its former independence are not missing. Leaving aside the past
tense 1st pl. ending -ak' (eak', berak'), which raises particular problems, we only
need to mention such paradigms as :
The -t morpheme demanded the zero grade of the root vowel, as is proved by the Greek aorista
and, to some extent, by the OCSI. [Link] and infinitives (bildlU 'to be awake', -lfplli 'to adhere') as
well as the Latin presents uideo 'I see', clueo ' I am reputed', maneo 'I remain'.
128 On the modal use of *ii- (optative > subjunctive). evidenced in Old Irish and the Italic
languages, see Benveniste's study (quoted fn. 115).
lH

122

utem 'l eat'

lsem 'I hear'
melancem 'I sin'
yafnem 'l arise'

Historical Morphology

keray (3rd sg. eker)

luay
melay
yareay

[5.353-5.4)

imper. ker, kerayk'

lur, luaruk
ari, arik'

or, conversely, with an -im present and no -a- aorist :

linim 'I become'

ele

ler, leruk',

or again, the optional use of active or mediopassive forms in the aorist of some
a stem verbs (3.251 c).
The correlation of -im presents (hayim, hecanim) with -a- aorists (hayec'ay,
hecay) settled in the course of the PA period. The one stem verb nstim 'I sit' is a
good illustration. In view of Skt. ni-~tdami, Gk. hizzo, Lat. -sUo ( < *sisde/o-},
we should expect Arm. *nstem, aor. *nsti, and the imperative is actually nist,
But, unlike berem and the other transitive verbs (acem 'I bring', hanem 'I draw').
the intransitive present *nstem passed over to the -im type, and the corresponding
past, accordingly, was attracted to the mediopassive inflection: nstay. In some
archaic verbs, an -eay aorist is still associated with an active present. In view of
lanac'em 'I know' ( < *canac'em}, aor. caneay, there can be no doubt that p'ax' cim,
hangc'im and the other -cim presents have come [Link] through morphological
change: the earlier present forms can be safely reconstructed as *p'ax(i)c'em,
*hang(i)c'em.
5.4. After elucidating, as far as possible, the development of the inflection.
let us review the present and aorist stems in the light of the comparative evidence,
Regarding the present, it must be remarked that the morphemic analysis we
applied in a previous section (3.22) proves inadequate as soon as we take into
account the PIE background : the inflection vowel was originally part of the
stem. As a rule, -e- stems (bere-m, dne-m) are traceable to PIE thematic presents.
The occurrence of -e- instead of -o- where Greek has the latter (Gk. phbo, pheromen,
West Gk. pheronti) has already been accounted for (5.34). In -i- stems (beri-m,
hayi-m, etc.}, the vowel, though ambiguous, is likely to reflect *-i- (5.351); the
original inflection, accordingly, may have been [Link] or semi-thematic.
However, the ClArm. -em and -im presents need not be treated separately : not
only are most of the latter derived, as berim from berem (3.241), or secondary,
as nstim, p'axc'im, but the e/i contrast does not hold through the whole
paradigm:
Pres. ind. gre-m
hayi-m

impf. gre-i
haye-i

[Link]. gre-cI
inf. gre-1
haye-c'ay
haye-1

[5.4-5.411)

Historical Morphology

123

Both -em and -im presents will therefore be brought together as direct or indirect
reflexes of PIE thematic presents.
Athematic inflection is presumable for u stems at large and for one part of the
a stems, the remainder being traceable to derived thematic presents (see below,
5.41 end).
5.41.

ARMENIAN REFLEXES OF PIE [Link] PRESENTS:

*-e/o stems : acem, berem; nstim (5.353). With a secondary -an- extension :
lizanem 'I lick', beside lizem (= Gk. leikho).
*-ye/o- stems: p'aylem, -im (4.451); mrmnfem 'I groan, murmur' ( < *murmurye/o- : cf. Gk. [Link]). With a secondary -an- extension : xacanem, luc'anem
(4.352); anicanem 'I curse' ( < onei,d-ye/o- : 4.44, Note).
In denominative and derived presents, -e- < *-eye/o- (Meillet, 1936, p. 105) :
gorcem 'I work, do' < *woreye-; ofogem, afog(an)em 'I irrigate' < *sroweye(*sreu-/srou-/sru- 'to flo,v'). As to utem ' I eat', substituted to the old athematic
present *edmi (Skt. <idmi; Lat. edo, 3rd sg. [Link]; Hitt. 3rd pl. adanzi), it seems
traceable to an [Link] present *odeye/o-: some evidence for *o in
similar instances is to be found in Gk. (w}Otheo ' I thrust, push' (Cf. Skt. dvadh'it
'he slew'), poleomai 'I go to and fro', cf. amphi-polos 'moving around, attendant'
(and Skt. carami 'I move'). It is therefore appropriate to trace glem 'I roll' through *gulem - back to *woleye/o- (Cf. OCSL valiti < *woli-). Examples of
denominative verbs have been given in a previous section (3.332 a) and need
not be resumed.
Simple -am presents, as a rule, contain *-aye/o- stems, either derived or
denominative (Meillet, 1936, p. 110). On mnam 'I remain', see also above, 4.312:
such presents as orsam, dolam, etc. seem parallelable to Gk. pot<iomai 'I fly'
(beside petomai); Lat. domare, OHG. zam0n 'to tame'.
We have to place here the C!Arm. reduplicated verbs: t'at'awem 'I dip', k'rk'reni 'I search,
stir', Mleletn, oroleni 'I overflow' (cf. helum 'I pour'), p'olp'olim 'I glitter', iawiap'em 'I touch, feel', etc.
These are not directly [Link] to PIE models, nor do they belong to the -ye/o stem claaa. Meillet'a
parallel, therefore, is not to the point: the Greek verbs he quotes (1936, p. 113) rather correspond to
Arm. mrmnfem (see above), trtn/em 'I moan, complain', pip/am (for earlier pulpul/etn ?) 'I bubble'.
NOTE.

5.411. *-(O)ne/o- stems: afnem, dnem, lk'anem (5.322); y-afnem 'I [Link]', unim
' I take hold, have', tanim ' I carry' 124, usanim 'I learn' (Cf. OCSL 'lyYknQ ' I get
used, learn'), etc. This extension has been largely worked up to provide presents
for ancient [Link], not only in Armenian, but also in other IE languages, especially
The relation of the aorist tar-ay to the present tanim is a morphological riddle. Etymological
data are missing.
124

124

Historical ?iforphology

[5.411-5.421)

in Slavic and Greek: 0081. vyknQ, [Link]. vyku; stanQ 'I stand up', [Link]. staxu; Gk.
dakno 'I bite', aor. edakon; hamartano ' I miss the mark, fail', aor. himarton, etc.
At a far later stage, -an- was added to inherited present stems, too, as in lizanem,
anicanem, harc'anem (5.321). As the -anem/im presents are all coupled with root
aorists (3.252 f-g), it is often impossible to tell whether they belong to the earlier
or to the later formation.
5.412. *-ske/o- stems: harc'anem (5.321). Besides, -c'- occurs in the compound
morpheme -oy-c'- of the causative verbs (3.244). For want of comparative evidence,
however, neither the original function of -oy- ( < *-eu/ou-) nor the relation of
-oyc'- to -oys- (korusanem 'I make to perish, lose') and -oyz- (p'luzanem) 'I pull
down') can be elucidated 1 2 5.
The Armenian subjunctive, we admitted ,is derived from *-iske/o- present
stems (5.331). Once the derivative suffix had evolved to a modal morpheme,
nothing prevented its being added to non modal -c'- stems; hence harc'ic',
usuc'ic', grec'ic', etc.
5.413. The -c'em (-c'im) presents doubtless belong to the *-ye/o- stem class.
Their only cognates, as Meillet perceived, [Link] the Greek presents in -sso, Attic
-tto (Meillet, 1936, p. 109). But one need not assume, as he does, a. blending of
*-ye/o- with *-ske/o- : here as elsewhere, Arm. c', Gk. SS (U) [Link] the expected
reflexes of PIE *ky (4.343). Whatever the ultimate background of this type, its
expansion in connection with the -eay aorist belongs to the gloomy history of PA.
5.42.

REFLEXES OF ATHEMATIC PRESENTS

Of simple root presents there are but fe,v relics in ClArm. : em 'I am' (5.34),
the defective verb bam 'I say' ( = Doric Gk. phami), and probably the monosyllables kam 'I stand' (Cf. Doric Gk. eban 'I stepped', [Link].), gam 'I come', lam
'I cry'. On tam 'I give', see 5.333, Note; on gelum : 5.423.
*-nafn~-

stems: banam 'I open' ([Link]. bac'i), t'anam 'I \vet' (t'ac'i);
bafnam, dafnam (2.343), stanam 'I get, acquire', mofanam ' I forget', etc.
Arm. -na- reflects either *-na- or *-n~- 126 The alternation is preserved in
Indo-I ranian (Skt. -na/ni-) and in Greek. e.g, punemi, 1st pl. pitnamen ' I spread
out'. It originated in the growing together of t he 'infixed' morpheme *-ne/n5.421.

See fn. 33. -oyB- also occurs in heluBem 'I nail, fix' (aor. helU8i) and %1'axu8em 'I cheer, exhort
(aor. ec'i), from %1'ax = urax 'cheerful'.
126 The latter alternant ( *n~-) has prevailed in Germanic : Goth. gawaknan 'to wake up'.
125

Historical Morphology

[5.421-5.423]

125

and a. follo"ing laryngeal : *pot-ne-H2- > Gk. pitnii-; *pot-n-H2- > pitna- 1 27 In
late PIE *-nafn~- had become a. bare present stem extension , susceptible of
various dialectal developments. The above quoted Arm. verbs have no direct
cognates. But the inchoative type in -anam (3.332 a), still productive in ModArm.,
reminds one of the analogous Germanic formation preserved in Gothic : cf. Arm.
li-anal 'to become full' and Goth. fullnan 'id'.
5.422. -nu- stems : afnum 'I take' ( = Hom. Gk. arnumai 'I ,vin'), z-genum
' I dress, put on', cf. Gk. -(w)liennumi 'I clothe', both from *u1es-nu-; erdnurn 'I
swear', lnum 'I fill', mkenum 'I cause to fall, t hrow down' (Cf. ankanim ' I fall') 128,
sartnum ' I startle', etc.
On helum 'I pour' (and t'olum 'I let'), see l'i1eillet, 1936, p. 48 (l < ln) and 112. If he is right,
l should have been propagated from the present to the aorist (heli, t'oli). But the preaent stems may
reflect *pelu-, t'olu- as well.
NOTE.

The original alternation *-neu/nu-, preserved in Indo-Iranian (Skt. apn6mi,

1st pl. apnuma~ 'I reach, obtain'), has given way to -nu/nu- in Greek. So unusual
a. change is not likely to have come [Link] elsewhere, so that \Ve can confidently
assume that in PA the full grade [Link] *-neu- had been lost.
This formation has been fairly productive: most of the -nu1n presents lack
cognates in the other IE languages (Cf. 4.314, on lnum). I n ClArm. not only do
the -nu- and -c'i- extensions behave as free variants (3.254; fn. 121), but other
variations also occur, e.g. zbalim, zbalnum 'I am busy'; hecanim, hecnum 'I ride'.
5.423. -u- stems : gelum ' I twist, press'. Armenian, as it seems, has preserved
an archaic present *welu-mi, which in Latin has passed over to the thematic
inflection : uoluo ' I roll' ( < *u elu-o). It is difficult to decide whether -u- is part
of the root. At any rate, it does not recur in the aorist (get-i). As, on the other
hand, there is no evidence for a. PIE -u- extension in present stems, we cannot
guess how it happened t hat -um presents were regularly paired with root aorists
(3.253 e). I n one instance, at least, the priority of the aorist cannot be doubted :
[Link] \vith the irregular present harkanem 'I strike' (aor. hari : 3.255 a}, there is
a doublet harum. All that we can state is that the paradigm, once established,
became productive: -u-, like -ane-, has been substituted t o -e- in ancient thematic
presents, e.g. lizu1n, lizanem (5.321); ger<ntm, gercane1n, gercem 'I shave'. Some of
the -um presents stem from the P arthia.n period: dizum (dizanem) 'I pile up';
zenum 'I immolate' 1 2 9 are built up on I ranian words or roots.
1

The position of t he infixed morpheme clearly appears in such Skt. presents as bhi-na-d-mi 'I
cleave, split', lat pl. bhi-n-d-nui[I., aa against bhit-ti 'splitting'. Cf. Lat. /ind-0 'I cleave', pf. /[Link].
128 Godel, 1965, p. 26.
129 G. Bologneai, RicLing 5 (1962), p. 132-135; E. Benveniste, 1964, p. 33-34.
127

126
5.43.

Historical Morphology

REFLEXES OF

PIE

[ 5.43-5.432)

AORISTS

In describing the OlArm. conjugation, we had to distinguish, on the one hand,

root aorists, and on the other, -c'- aorists (3.22). But, when considered from the
historical point of view, the former class breaks up: some of its members reflect
PIE imperfects (5.321); the other ones are derived from genuine aorists, either
thematic or [Link], and only these will concern us here. Besides, the phrase
'root aorist' will henceforth be used with reference to PIE only, not to OlArm.
5.431.

Th&matic aor-ists

As the distinctive feature of the PIE thematic aorists was the zero grade of
the root vowel (Gk. elipon, as against the impf. eleipon), we have a criterion for
bringing together with lk'i, gti (5.322) a number of other aorists for which exact
parallels are wanting: rnti, 3rd sg. emut 'I entered' ( < *mude/o- : cf. mucanem,
4.352); kli, ekul 'I swalloved' ( < *gule/o-, perhaps from *golejo- : Skt. girami
'I swallow'); arbi 'I drank' ( < *s1bhe/o- : cf. Lat. sorbeo 'I suck'); hari 'I stroke'
(< *p0 re/o- : cf. 0 081. perQ ' I beat'), and, with [Link] inflection : ankay
'I fell' ( < *[Link]/o-). The present ankanim, being derived from the aorist, does
not concord with the genuine present *senge/o-, preserved in the Germanic
languages (Goth. sigqan 'to sink').
Evidence for reduplicated thematic aorists is found in Greek and Indo-Iranian.
OIArm. has only arari 'I [Link], did' (5.322, 5.333).
5.432.

Athematic root (and -s-) aorists

In spite of sound change and of the general levelling of verb inflection (5.34,
end), OlArm. still [Link] unmistakable relics of this archaic formation. Such
reflexes, as can be expected, appear in anomalous verbs (3.255 c). To begin with
the plainest parallels, let us review the 3rd sg. forms :

ekn 'he came'

et 'he gave'
ed 'he laid, put'

Vedic dgan ( < *egemt)

= Skt. ddat (< *edot); cf. Gk. edoke; 0081. da
= Skt. ddhiit (< *edhet); cf. Gk. etheke; 0081. de.

There is also a Vedic 3rd sg. dhiis (without augment}, and the same -s instead
of -t recurs in Hitt. dais 'he set, placed', Phrygian e,daes 'id'. It is of course
impossible to tell \vhich has been lost in Armenian.
The preservation of the root vowel in the other persons implies the loss of a
following syllable : in the prototypes, the vowel certainly belonged to the
penultimate; but of the original [Link] syllable not the slightest trace is left. Therefore, neither Skt. ddam, nor Gk. edoka are suitable parallels to Arm. etu. On the

[5. 432-5.433)

Historical Morphology

127

other hand, 0081. daxu, Alb. dase 'I gave' bear evidence to -s- aorist forms, which
enables us to trace etu back to *edosom 130 This explanation applies to edi
(< *edhesom : OCSl. dexu), not however to eki, whose relation to ekn remains a
myst&y.
Other reflexes of root (or -s-) aorists can be suspected in luay 'I heard' (5.322),
erkeay 'I feared' (< *dwei/dwi-: 5.352, end), and erduay 'I swore', although the
etymology is uncertain. The dropping of intervocalic -s- does not allow for any
statement regarding the stems: lu- (in luay), e.g., may be derived from *klu-sas well as from *klu-.
On presumable reflexes of [Link] *-e- aorists, see 5.352.
With regard to inflection, the anomalous aorists differ perceptibly from the regular [Link]
At first sight, it seen1s that the full grade root vowel prevailed in PA as it does in Sanskrit (am,
1st pl. <ima). However, there is [Link] a trace of vo,vel [Link] in edi, lst pl. edak' : cf. Gk.
ethe-ka, ethe-~n (the root is *dM{dh'il ). If this surmise proved right, it might throw some light on the
origin of the let pl. ending -ak' : according to [Link], edi is the only source of the lst eg. -i (in ber-i,
grec'-i, etc) 131.
NOTE.

5.433. -c'- aorists. In the first place, it must be remarked that -c'- does not
occur after consonants : lc'i 'I filled' (< *lic'i), xc'i 'I stopped up' (< *xic'i) are
no exceptions. Cf. also ba-c'i, from banam 'I open', as against barj-i, from bafnam
'I lift, raise'. In -Jnt'erc'ay 'I read', either i ot u may have been dropped; but
for want of a reliable etymology, the case cannot be settled. The stems, as a rule,
end in -ec'-, -ac'- (3.251-252; besides, in the -num verb class: ankec'i 'I threw',
yec'ay 'I leaned', zgec'ay 'I dressed').
In Ionic and Homeric Greek, the *-ske/o- extension frequently occurs in derived
(iterative) past tenses. It is added to imperfect or aorist stems, e.g. impf. ephaske,
from ephe, (e)phato 'he said'; phbeske, from (e)phere 'he bore, brought'; aor.
stdske, from (e)ste 'he stood'. The peculiar semantic shade of these preterites does
not preclude a parallel with the Armenian -c'- aorists: these, too, prove to be
extended aorists or imperfects. Except for the root vowel grade, e.g., elic' 'he
filled' is formed like Gk. stdske, and traceable to *eple-ske. In this instance,
*-ske/o- has been added to the old root aorist *ple-(s)- : Vedic dpras, Hom. Gk.
pleto; cf. also the Latin perfect -plevi, 2nd sg. plesti. Likewise, zgec'ay 'I dressed'
must somehow be derived from a genuine [Link]. stem. On the other hand, as we
already pointed out (5.321 b), the -c'- aorists, when joined to simple presents,
reflect extended imperfects: mnac'i 'I remained' < *mona(ye)-ske/o- 132
1so G. Bonfante, The Armenian Aorist. JAOS 62 (1942), p. 102-105. This is a strong argument for
Pisani's thesis on the divergent developments of . Vm and * Vn (Cf. 5.213, end).
131 Godel, 1970b, p. 149. ?tly previous statement (1965, p. 33) was perhaps too skeptical.
132 The question whether the contraction of *-aye to -a- should be dated farther back than the
earliest instances of extended imperfects must be left unanswered.

128

Historical Morphology

[5.433-5.441]

However, such a formula does not seem adequate for grec'i, hayec'ay, etc. The
difficulty lies in the 3rd sg. greac' (3.233, end) and in the imper. sg. grea, hayeac'
(3.235 b). I believe it can be removed: the regular alternation ea/e may well
have been resorted to in order to avoid uncommon endings, firstly in the
imperative : grea instead of *gre (final -e never occurs except in the conjunction
et'e, t'e 'that'). This conjecture is supported by one isolated, and therefore presumably archaic, paradigm : 'ilnkec'i 'I threw', 'ilnkec'er, 'ilnkec' (later : 'Jnkeac'). In
the 3rd sg., e is substituted by e in the last syllable, as in alues 'fox', GDL. aluesu
and a few other words (l\feillet, 1913, p. 19; Jensen, 41) 133, Only in the imper.
sg. does ea turn up : 'ilnkea. Thus, 'ilnkec'i is traceable to *songe(ye)-ske/o-, and
likewise gorcec'i 'I vvorked, achieved' to *wore(ye) -ske/o-, p'aylec'i 'I shone' to
*pholye-ske/o-, etc. Generally speaking, the regular -ec'- aorists can be equated
with the Ionic imperfects in -eske/o-.
There ren1ain the -ac'- aorists that stand opposite to -em presents (3.252), and
those which correspond to -(a)nam presents. The latter, like k'i, zgec'ay, certainly
reflect genuine aorists, not imperfects; but in both cases the PA developments
cannot be reconstructed with certainty.
5.44. Both the INFINITIVE and the PARTICIPLE belong to the 0 declension (3.2) :
in bereal as well as in berel, -l < *-lo-. Evidence for primary adjectives in *-lois found in several IE languages: Gk. deil6s 'cowardly', tuphl6s 'blind'; Lat.
pendulus 'h~nging', etc. In OCSl. this morpheme supplies the active past participle,
mostly used in compound tenses (bilu jesmi 'I have struck'). Verbal adjectives in
-l < *-lo- also occur in Tokharian (Eastern dialect; instead of -l, Western
Tokharian has -lye, -lle < *-lyo-). Thus, we have a frame of reference for the
Armenian participle in -eal 134 As, on the other hand, adjectives do not evolve
to infinitives, the above evidence does not account for berel. Although there are
only faint traces of PIE action nouns in *-lo-, such a formation has to be
postulated in order to explain the Armenian infinitive: it may have been
productive in some limited dialectal area.
5.441. The development of infinitive forms as an integral part of the verb
paradigm cannot be traced back to the PIE period. Armenian in this respect is
on a par >vith all the cognate languages. The nominal origin of the infinitive is
still ascertainable, in spite of its manifold syntactic uses las. The original action
This did not seem important enough to be mentioned in our survey of normal vowel alternation
(2.22-2.223).
134 Meillet's interpretation of the syntagm 1wra bereal e 'he has brought' ie wrong: see above, 3.26.
135 Besides Meillet, 1913, 123-125, see also H. Vogt, 1937.
138

[5.441-5.442)

Historical l\1orphology

129

nouns were presumably root derivatives: so, perhaps, tal < *da-lo- '(a) giving'.
The derivation of the infinitive from the present stem (3.231 e) is the result of a
not uncommon development : in the Germanic languages, too, the infinitive
belongs to the present group.
It is no wonder, therefore, that the verbal adjectives are explainable in terms
of noun derivation: sireli, zarmanali, etc. (3.2) are formed in exactly the same
way as akani 'clear-sighted' (akn 'eye'), ali 'salty' (al 'salt'), hawasti 'certain'
(hawast 'certainty'). Likevise, banali 'key' (from banal 'to open') is parallelable
to gorci, goki (4.22): many derived adjectives in -i < *-iyo/ii- have been
substantivized.
The other type of verbal adjective is used as predicate : bereloc' em 'I am to
bring, or be brought' (Cf. 3.21, 3.243). The compound morpheme -loo' obviously
contains the infinitive suffix *-lo. As to -c', I would not hesitate to declare it
identical to the case marker -c' in mardo-c', srti-c',jkan-c', etc. (5.236). The common
opinion regarding the origin of this latter thus proves right: in bereloc', galoc',
etc., -*sko- still occurs as a derivational suffix, as it regularly does in Germanic
and Slavic iss.
5.442. The background of the participle is not perfectly clear: -eal is doubtless
a complex morpheme, but neither in Slavic nor in Tokharian do we find any clue
as to the development of the diphthong. As a rule, though not constantly, -eal
is attached to the aorist stem (3.221). Presumably, this rule originated in the
relationship between the participle and the root aorist : beri, bereal; ankay,
ankeal, etc. ; hence analogically: asac'i, asac'eal; morac'ay, morac'eal. The latest
stage of this process is reflected in koc'ec'eal, hayec'eal, as against greal, bazmeal.
On the other hand, some participles seem archaic : in edeal 'put, having put';
leal 'being, having been' (3.255 c), -e- is apparently the root vowel: *edi-al,
*ley-al, and this analysis is the only suitable one for the isolated participle k'alc'eal
'hungry' < *k'alc'i-al (on k'alc'num, aor. k'alc'eay, see 2.213, 2.243). Such scarce
data do not allow for a decisive statement. Yet 've can tentatively assume that the
PA participle ended in -al, and that the CIArm. ending -eal has been abstracted
from such forms as the above, 'vhich may have been more numerous in earlier
times, e.g. before the substitution of p'axuc'eal for *p'axial (3.221, Note).

iss It would. be worth while to search for other traces of its original function: e.g. a1nroc' (Ilb}
'st ronghold', from amur (1) 'solid, strong' is likely to reflect a substantivized adjective in *o-8kc/a-.

130
6.

Conclusory Remarks

(6-6.2)

Conclosory remarks

The IE character of Armenian, suspected since the very beginning of comparative philology, was definitely ascertained in the 2nd edition of Franz Bopp's
Comparative Grammar (1857-1861). A further, decisive step was taken by,
H . Hiibschmann in his 1877 article (KZ 23, p. 5-49), in which he demonstrated
that Armenian is not an offshoot of the Iranian branch, as it had been thought
to be up to then, but the descendant of a peculiar IE dialect.
6.1 . From that time on, comparative studies on Armenian were carried on
with more rigor and consistency, and the main lines of its evolution were brought
to light. Yet there is still a great deal of work to do, but not only in the domain
of sound change 2 a1, which apparently is more attractive or more accessible to
comparativists at large. Indeed, the most recent contributions deal with etymology
or with phonological problems, while little interest is bestowed on the development
of Armenian morphology.
Current research, be it said in passing, might sometimes profit by certain
suggestions of earlier philologists, which have passed unnoticed or been totally
forgotten. In a previous section, e.g., I cautiously asserted that decisive evidence
for intervocalic *dh is not to be found in Armenian {4.333). I might, however,
have mentioned that Bugge long ago brought forth some interesting parallels,
especially with Greek, to the effect that PIE *dh between vowels should have
merged with *gh : awaz (I) 'sand' {Gk. dmathos, OHG. sant); azn 'nation' (Gk.
ethnos); suzanem 'I plunge, cover' {Gk. keatho 'I cover, hide') 138 R egarding the
identity of the pl.L. and Acc., I advocated Mann's view against Meillet's {5.223),
but I might have referred to the former's presumable source as well 1ae. This
remark, however, is not meant to justify in any way Mann's prejudiced attitude
toward the successive generations of Armenologists: while he exalts Hiibschmann's
predecessors, he underestimates the contributions of such scholars as Pedersen
and Meillet, not to speak of Pisani's and Bolognesi's studies, which he does not
seem to be fully acquainted 'vith.
6.2. Hiibschmann's discovery had another consequence: the placing of
Armenian among the IE languages became a problem that had to be solved
by measuring, as precisely as possible, the degrees of affinity between Armenian
See in this respect the Notes to 4.334, 4.353, 4.452.
138 S. Bugge, Etnumch und .Arn1eniJI. Christiania 1890, p. 79. On a- in am as against Gk. e,
eee 4.331; on the transfer of the word to then declension: 5.14.
189 H. Osthoff, Zur armeniaclun Laut- und TJ' ort/orachung. Sprachwissentachaftliche Abhandlungcn
II. 4, Budapest, 1901, p. 121, fn.
187

(6.2-6.3)

Conclusory Remarks

131

and its various cognates. In this respect, some weight is attributed to the testimony
of Herodotus, who mentions the Armenians as Phrygian emigrants or colonists
(Hdt. VII 73). This assertion, though confirmed by a later one from Eudoxus,
must not be overrated. Anyhow, the comparative method, insofar as it is applicable
in this instance, has not yet succeeded in either proving or disproving the presumed
affinity of the languages in question 140.
On the other hand, attempts have been made to discover traces of dialectal
kinship between Armenian and the ancient IE languages that have come to light
during the 1st quarter of this century. Both Tokharian and Hittite --- or rather,
the Anatolian branch of IE - are marginal languages. Each displays quite
peculiar features, some of which look archaic, while others are likely to reflect
separate developments. Armenian is certainly a comparable case. But what can
be expected in such a situation is rather the survival of archaisms in the eccentric
or isolated languages than the appearance of common innovations. This opinion,
however, is not shared by those who, like Pisani, reject the conception of PIE
as a more or less unified language, and account for dialectal affinity in terms of
successive networks of isoglosses. Yet, however one figures out the IE background, no significant isogloss including Armenian and Hittite (or Anatolian)
only has been traced so far. Affinity between Armenian and Tokharian has been
claimed with better grounds : a quite curious feature, common to both languages,
is the use of the morpheme *-ske/o- in causative verbs. The parallel, however,
does not throw any light on the development of the Armenian causative suffix
-oyc'- (5.412).
6.3. Evidence for dialectal grouping is to be gained from different data. There
are phonological isoglosses, two of which have been alluded to in connection with
the PIE consonant pattern (4.321). Another holds between the languages that
preserve the distinction of e, o and a, and those in which all three vowels have
merged to a (Indo-Iranian), or in which the contrast of a and o has been effaced,
as in Baltic, Slavic and Germanic. Armenian in this respect stands closer to
Greek, Italic and Celtic, except for the sporadic occurrence of a instead of e or o
(4.331). Whether this is enough to assign to Armenian an intermediate place is
a question that does not seem to have ever been raised.
Some comparativists deliberately give the primacy to lexical isoglosses .141 It
cannot be denied that this procedure leads to valuable results. It is significant
that quite a number of words are reflected in Greek and Armenian only, e.g.
See 0. Haas, Zur VorguchichU der armeniachen. Sprache. HA 76 (1961), p. 663-612, and
W. Dressler'e criticism : Armeni8ch und Phyrgiach, HA 78 (1964), p. 486-498.
141 So W. Porzig, Die Gliederung du indogermani8chen SprachgebieJa. C. Winter, Heidelberg, 1954;
G. R. Solt&, Die Stellung du Arn1eniachen im Krme der indogermaniBchen Sprac/aen, Wien, 1960.
140

132

Historical Morphology

[6.3)

*mega- (Gk. megas, Arm. mec); *amr/dm0r- (Dor. Gk. amar, Arm. aior); *kiyon,
(Gk. kton, Arm. siwn); the reduplicated aorist *arare/o- (5.322, cf. 5.333), etc.
The weight of these lexical concordances is increased by as numerous morphological
ones 142 Indeed, the preservation or the loss of a word is often a matter of chance.
Our knowledge of the lexicon of an ancient language depends upon written evidence.
Now, many words that never occur in the literature may have been usual in the
spoken language, or have survived in unwritten dialects. Neither in the Avesta
nor in the OP. inscriptions does the ancient word for 'night' appear; but it has
turned up of late through a systematic investigation on modern Iranian patois.
A similar case is Arm. t'af (4.334). On the other hand, words are easily borrowed,
and the borrowing may entail the loss of an old genuine term. There are reasons
to believe that Armenian in this respect is no exceptional case.
I am therefore inclined to lay more weight on morphological isoglosses. The
most significant ones unite Armenian with Greek and Indo-Iranian. Only these
languages bear evidence to the so-called PIE imperfect. The augment is also
documented in Phrygian aorist forms, and the prohibitive negation *me (Arm.
mi) in Albanian and Tokharian, too. Other isoglosses include Western languages :
see 5.231 ( *-bhi-), 5.232 ( *-osyo). This latter G. ending is also found in Italy, in
old Faliscan inscriptions. These case endings can be regarded as archaic; but the
development of a past tense (imperfect) from present stems is more likely to be
an early common innovation of Eastern dialects 143 A thorough discussion of the
problem would exceed the scope of this Introduction.

lleillet's inventory (1936, p. 142-143) is not quite up to date: some items would have to be
corrected or discarded, while others might be added.
143 E. Evangelisti, L'imperfetto arnieiw ... (referred to. above, fn. 112), p. 29. Strangely enough,
the Italian scholar ignores the Armenian reflexes of the imperfect.
142

INDEX

Ablative 3.182-183; 3.142 d (n stems); 5.235;

6.243
Accusative 3.112; pl.: 2.322; (3.183); 5.222
Action nouns 3.331 b
Radical action nouns 3.32 Note
Adjectives 3.1; 3.151 (adj. in -r)
Derived adjectives 3.32; 3.332 d
Pol!8688ive adjectives 5.245
Verbal adjectives 3.2; 5.441
Adverbs 3.312 ; 3.332 e
Affricates 2.1; 4.42
Agent nouns 3.2 Note; 3.331 c; 3.333 (compound agent nouns)
Alternation
Consonant alternation 2.231-232; 3.234
Vowel alternation 2.22-223 ; 3.12; 3.233 b
(in root aorists); 4.323 (PIE)
Anatolian languages 5.235; 5.3 ; 6.2
Aorist 3.222; 3.232 (indic.); 3.234 (subj.);
5.322 (PIE)
Aorist stems 3.22; 3.25-255; 5.43-433 (PIE)
i stem aorists 2.213; 3.22 Note; 3.234;
5.352
Aspect 3.222; 5.3 (PIE)
Athematic inflexion 5.34; 5.4
Athematic presents 5.42-423
Athematic aorista 5.432
Augme.n t 2.213; 2.31; 3.233 a; 3.265 Note; 5.3

Case system 3.11

Case endings 3.18-183; 5.2-5.237; 5.243-244
Causative verbs 3.244; 5.412
Clusters See : con8onanta
Cohortative 3.236
Common endings 3.24-241 (in the pres. group);
3.232, 3.242 a (in the aor. group); 5.332 (impf.
and active aor.)
Common verbs 3.24; 3.242 b
Compound tenses 3.243; 3.26
Compou nd words 3.31-312; 3.333-334
Conjugation 3.2-3.222; 3.243; 5.32

Consonants l.6-1.62; 2.1; 2.341 (double consonants)

Consonant clusters 2.32-323 (in word end);
2.34-345 (word internally); 4.34-343 (PIE);
4. 43
Consonant shift 4.42
See also : alternation
Contraction 4.46
Dative 3.181-182; 5.243-244 (pronouns)
Declension 3.12; 3.13-132 (invariable stems);
3. 14-143 (variable 11telllll); 3.133, 3.15-152
(mixed decl.)
Declension markers 3.16
Deictic particles 2.324; 2.331; 2.42; 5.241
Denominative verbs 3.332 a
Diphthongs l.631 (aw); l.641; 2.213 (ea); 4.47
Epentheaie 4.45-452
Fricatives 2.1; 4.321 (PIE a); 4.42
(Future) 3.21
(Gender) 3.1; 5.1; 5.145
Genitive 3.181-182; 3.26; 5.232; 5.243 (-oy);
5.245 (pronouns)
Georgian 4.11
Hiatus 2.21-212; 4.46
Hurrian 4.11
Imperative 3.221; 3.235; 5.433 (grea, hayeae' )
Imperfect 2.211; 3.2; 3.221; 3.231 b; 5.33;
5.332 (ei); 5.3, 5.32 (PIE)
Inchoative verbs 3.332 a; 5.421
Indicative 3.21
See also : preatnt, imptr/ect, aorist
Infinitive 3.2; 3.221; 3.231 e; 5.441
Instrumental 2.323; 3.183; 5.231; 5.243 (pronouns)
lsogloss 4.22 Note; 5.235; 5.322 fn 111 ; 6.2-3
Laryngeals 4.322; 4.44 Note ; 5.421
Loanwords l.61; 2.32; 4.1-4.122; 4.333 fn 58;
4.5; 5.14

134
Locative 3.111; 3.133 {in -o/: cf. 5.232};
3.181; 5.235 Note; pl. : 5.223
Mediopassi\'e 3.241 (pres.}; 3.242 (aor.}; 5.351
(-im presents}; 5.353 (-ay aorists}
Middle 5.341 (PIE)
?tionoeyllables 2.221 {VC type}; 3.132 (C + i,
'U type)
Morphological change 4.314; 5
Nominative pl. : 2.322; 5.131; 5.22-221; 5.222
Parthian 4.12-122
Participle 3.2; 3.221 ; 3.~22 fn 28 (used ae a
main verb}; 5.442
Paet 5.332
Perfect 5.3-5.31 (PIE)
Phrygian 4; 6.2-3
Polysyllables in -i 2.344; 3.133; 4.22; 5.12;
5.233
Prefixes 3.222; 3.31; 4.21 (obsolete prefixes}
Negative prefixes 4.334
See also : augment
Present 3.231 a (indic.}; 3.231 c {subj.); 5.321
(PIE}

Present stems 3.22; 3.23; 5.41-423 (PIE)

Prohibitive 3.2; 3..221; 3.231 d
Pronouns
Demonstrative pronouns 2.324; 3.11 (na,
nok'a); 3.181-182; 4.334 Note (da, ayd);
5.1 fn 78; 5.241
Indefinite pronouns 5.242

Index
Interrogative pronouns 3.112 {zi, zinc!');
3.181 (011}; 4.334 Note; 5.242; 5.245
Personal 3.11 (u, du); 3.182 (mek', duk');
4.334 Note (du}; 5.24; 5.244-245
Relative pronouns 3.181; 5.242
Reduplication 3.333 (in noun stems}; 5.41
Note {verb stems); 5 (2.)
Resonants 2.1; 4.322 {PIE}
Stem ' ' ariation 3.12; 5.11
Stops 2.1; 4.321 (PIE); 4.42
Stress shift 2.22; 4.33; 5.241
Subjunctive 3.21; 5.33 (pres.}; 5.331-333 (aor.}
See also : pruenl, aori8t
Suppletive verb stems 5.32; 5.322
Syllable 2.2; 2.3; 4.43
Syncretism 3.111
Thematic inflection 5.34; 5.4
Thematic presents 5.41-413
Thematic aorists 5.431
Tokharian 5.44; 5.442; 6.2-3
Transliteration l.4 ; l.42; l.642-643
Verb classes 3.251-255; 5.321-323
Voice 3.24-243; 5.35
Vowels l.63-632; 2.1; 4.322 {PlE)
Inflection vowel 3.22; 3.231 ; 3.241 ; 5.4
Neutral vowel 2.3 ;4.43
Prosthetic vowel 4.44
Thematic vowel 5.34
See also : alternation, contraction.

'

A LIST OF CLASSICAL ARMENIAN WORDS AND MORPHEMES

(The alphabetic order is Latin, not Armenian)

-a (in mediopaesive aoriste) 3.234 Note; 5.35;

5.353
-abar 3.312
-ac 3.2 Note; 3.32
aeem 3.252; 4.332
aganim (3.253 g) ; 4.22; 4.353
-agoyn 3.321; 4.121
-ak 4.121
-akan 3.32; 4.121
-akic' 3.312
akn, pl. a~'k' 3.152; 4.331; 5.211 ; pl. : 4.352;
5.12
alewor 4.331
alewr 4.141 a; 4.353 Note; 5.142
al 1.62; 4.311; 4.335
alal\'em, alawt'k' 4.343
albewr 4.342; 4.353 Note; 5.142
am 5.12
amall'em, amawt' 4.343
amb- 4.21 (ambolJ)
amis 3.181 (L.); 4.44 Note
amroc' 5.441 fn 136
amuri 4.341
anicanem (3.253 f); 3.32 Note (ankk'); 4.44
Note; 5.41
ankanim (3.253 g); 4.332; 5.431 (aor.)
anun 3.141 a; 4.331; 5.211
anurJ 4.351
(a)pes 3.312; 4.121
-ar 4.21
-aran 4.1 21
-arar 3.312
arawr 4.341
areiw 4.334
ard 4.122; 4.21; 4.22
arew 4.353; 4.44; 5.13
argand 4.333 ; 4.342
argel 4.334
artawsr, pl. artasuk' 3.151 Note; 4.342; 4.452;
5.132; pl.: 5.12; 5.131
am 5.144

arnem, aor. arari 3.255 b; 2.231 ; 5.322; 5.333;

5.411
arnum 3.254; 5.422
arogem See ofogem
aru 4.323; 4.44; 4.46
asr 3.151 Note; 4.331; 5.132
-astan 3.312; 4.121
astl 3.141 a; 4.343; 5.141
Astuao 1.42
atamn 4.331 Note
ateam, aor. atec'i (3.251 c); 2.223; 4.351
awaz 6.1
awcanem (3.253 f) ; 4.452
awj 4.452
awjik' 4.452
-awl 3.2 Note; 5.12 Note
-awor 3.312; 4.323
awrhnem 2.345
awt' 4.22; 4.334; 4.353
ayl 3.181 (dee!.); 4.351 ; 4.451
ayr 3.17 (dee!.); 4.44 Note; 4.451; 5.141
azn 3.142 d; 6.1
bagin 1.61
banali 5.441
bard 4.323
barjr 3.151 ; 4.333; 5.132
bamam, aor. barji 3.255 a; 2.343; 5.421
bay 4.22; 4.334 ; 5.131
berem 3.252; 4.323; 4.333; 5.32; 5.331
bern 3.142 d (dee!.); 5.14
birt 4.342
bok 4.46
ealik 3.142 c (dee!.); 5.212
calr 3.151 Note; 5.132
cnawl 5.12 Note
cnelay 1.61
cunr, pl. cungk' 3.151 Note; 4.332; 5.132: pl. :
4.353 Note, fn 68

136

A List of Clasaic:al Armenian Words and l\forphemes

e' (prepos.) 2.311

-e' ([Link].) 3.183; 5.236. Cf. 5.441
e' . (aor. extension) 3.22; 5.321 b; 5.433
e'elum (3.253 e); 4.343
Ilana(\'em, aor. caneay 3.255 a; 4.343 (canawt');
5.353
ilrag 1.61
llmarit 1.61; 3.1
<l'oga.y 4.323; 4.353; 5.31 fn 110; eee ert'am
<l'ork' 4.334 Note
il'u 4.323; 4.352
il'ua.r 2.31
dalar 1.62; 4.21 ; 4.323
dar 1.61
darbin 4.342
darnam, aor. darjay 3.255 a; 2.231 ; 2.343 ;
5.421
day!, dal 4.45
de! 4.21 ; 4.323
detin 1.62; 4.21
diem 4.46
dnem, aor. edi 3.255 e; 5.322; 5.333; 5.411
(pres.); 5.432 (aor.); 5.44.2 (part.)
drandk' 4.33 fn 54
durk' 3.152; 4.333
dutn 3.142 c (dee!.); 5.145 fn 88; 5.211 fn 90
dustr 3.141a(dee!.);4.334; 4.343; 5.141

See augmt:nl
-ea.I 3.2; 5.442
-ea.n (GDL.) 3.152; 3.17 (lu~n/ean) ; 5.237
-ea.y (aor.) 2.213; 2.343; 5.232
elbayr 3.17; 4.342
eln 1.62
em 'I am' 3.223 (conjug.); 4.33; 5.321 Note;
5.332; 5.42
erdnum, aor. erdua.y 3.254; 2.343 ;. 5.422
(pres.); 5.432 ([Link].)
erdumn 3.142 c (dee!.); 4.22; 5.145
erek 4.321; 4.332; 4.44
erek' 4.341; 4.44; 4.46; 5.131
erewim 4.44
eree' 3.151 (dee!.); 4.341; 5.212
erkan 4.353 Note
erkar 4.354
e-

erkn<l' im (3.253 h); 4.354; 5.352 (aor.); 5.432

(aor.)
erku 4.354
eri'a~, aor. 6'ogay 3.255 b; 3.221; 5.333 Note
et! 5.14
ew 4.334
ewt'n 4.31 1 ; 4.335; 4.343; 5.21-211

-e (Ahl.) 3.183; 5.235

ea 4.354; 4.451
empem, aor. [Link] 3.255 a.; 5.32; 5.431 (aor.)
enkenum, aor. enkee'i 3.254; 5.422; 5.433
(aor.)
enker 4.21
ent'[Link] 3.252 d; 3.32 Note (m'ac' k' )
ent' emum 3.254; 5.433
galar 4.21; 4.323
ga.m, [Link]. eki 3.255 e; 5.322; 5.42 (pres.); 5.432
(aor.)
gafn 5.143-144; 5.21
gay! 4.353
gel um (3.253 e); 4.21 ; 4.323; 4.353; 5.423
gewl 3.17; 4.452 Note
gin 4.335
gind 4.333
[Link] 3.252; 5.31; 5.321 Note
giwt 4.353; 4.452 Note
glem 4.21; 4.323; 4.353; 5.41
goil'em 4.352
goil 'iwn 4.22; 4.331
gore 3.32 Note; 4.332
gorcem 3.32 Note; 5.41 (pres.) ; 5.433 (aor.)
gorci 4.22
goy 3.223 ; 5.31
gtanem (3.253 f); 4.353; 5.322
hac' i 4.343
[Link] 3.151 Note
ha.ng6'im 3.253 h; 5.353
[Link]'anem (3.253 f); 3.32 Note (hare'); 4.334;
4.343; 5.321 a; 5.331; 5.412
harkanem, aor. hari 3.255 a; 5.423; 5.431
[Link] (3.143); 4.334; 5.14
[Link] 4.S
haw 5.13
[Link] 3.17 (dee!.); 4.3; 4.334; 4.341 (hater)

A Ll8t ofClaeaical Armenian Words and Morphemee

helum (3.253 e); 5.422
heru 4.311
himn (3.142 c); 4.22; 4.331 ; 5.145
bin 4.311 ; 4.335
bing 4.3; 4.334
hiwl 3.131
hnoc' 5.142
hol 4.31 1
hot 4.311 ; 4.332
hum 4.31 1
bun 4.334
hur 4.3; 5.142
i/y (prepoe.) 1.64; 3.112; 5.235 Note
-i (GDL.) 5.233
-i See polysyllables in -i
-ic' /yc' (subj.) 3.231 c (Cf. 5.33); 3.234; 5.331;
5.333
-ill' 3.132; 3.2 Note; 4.22; 4.352; 5.12
inn 4.353 Note; 5.21-211
-iwn 3.141 b; 4.46; 5.144
-iwn < -imn 4.22
iwr (po88. adj.) 5.245
-iwr 5.35
il 2.221 ; 4.333; 4.451
jern, pl. jerk' 3.152 (decl.); 4.333; 5.211; pl. .:
5.12
ji 3.132; 4.354; 4.46 fn 75
jiwn 4.46; 5.212
jmem 5.212
jukn 3.142 c (dee!.); 5.144; 5.212
]er, ]erm 4.333
)ermn 4.22; 5.145
Jil, ]ii 4.333
kam (3.251 c); 2.341 (aor. subj. 1.:ac:'ie', kae'
c:'ea); 5.42
kayer 1.61; 5.14
keam, aor. kec'i (3.251 c); 3.213; 3.233 b{aor.);
2.22 fn 10 (pres. subj.)
kelc, kclcik' 5.131
kin 3.17 (decl.) ; 4.33; 4.332 ; 5.232.(kno/)
klanem (3.253 f); 5.332; 5.431 {aor.)
kogi 4.22; 4.353
.
kov 4.22; 4.353; 5.13

137

-k' (pl.N.) 2.322; (3.183); ll.22-221

k'alc'num 3.254; 2.343; 5.442 (k'ak'eal)
k'an 5.21 ; 5.212
k 'ani 4.334
k'ar 3.151 (decl.); 5.146
k'ar- 4.354 (k'ar~un, 1.:'arord)
k'eri 4.335; 4.354
k'irtn (3.142 c); 4.342; 4.354
k'oyr 3.17 (deol.); 4.335; 4.354; 4.46; 5.141
k'san 1.61; 2.31; 4.353 Note; 4.43
k'un 4.343; 4.354; 4.47
Jeard 4.334 Note
learn 2.231; 3.142 d (decl.)
Ii 3.132; 4.314; 4.341 ; 4.46
-Ii (verbal adj.) 3.2; 5.441
linim, aor. ell! 3.255 o; 3.223; 5.322 ; 5.353;
5.442 ([Link].)
lizem (3.252); 4.333; 5.321; 5.41 (lizantm);
5.4.23 (lizum)
lk'[Link] 3.253 C; 4.334; 5.322; 5.333; 5.411
lnum 3.254; 2.341 (aor. eubj. lc'ic', lc'c'u);
4.314; 5.422 (pres.); 5.433 (aor.)
-Joe' (verbal adj.) 3.2; 3.21 ; 5.441
loys 4.321
lsem, aor. luay 3.255 b; 4.341 ; 5.322; 5.353;
5.432 ([Link].)
Ju 4.341
Jue 4.314; 4.321
luc'[Link] (3.253 f); 4.352; 5.41
mah (marh) 4.122; 5.237
[Link] 3. 142 (decl.); 5.212
[Link] 4.122; 4.323; 4.33 ; 4.334
martn6' im 3.253 h; 5.352
mawruk' 4.341; 5.12; 5.131
[Link] 3.17; 5.141
mec 4.332 ; 5.12 fn 80
mecarem 4.21
mek'enay

1.61

melk 4.354
merj 4. 342
meranim (3.253 g); 4.122; 4.451
meta.k's 1.61
me] 4.352; 4.451
mi (neg. 3.2
mie 4.335

138

A List of Claaaical Armenian Words and Morphemee

mnam (3.251 o); 4.312 ; 6.41 (pree.); 6.433

(aor.)
mrmnJem 4.351 ; 6.41
mucanem (3.253 f); 4.352

aain (3.142 d); 6.14

aermn (3.141 a); 6.1; 5.211
sin 4.353 Note, fn 68
sirt 4. 332
eiwn 4.46; 5.144
skesur 4.364; 6.12
skizbn 3.142 c; 4.22
sor 4.353 Note
spainam 3.25 Note; 4.343 Note
-st (-ist, Wit) 3.152; 4.22; 5.237
stanam (3.251 d); 2.31 (aor. subj. atac'ayc',

naw 4.3 ; 4.314 ; 4.353

neard 4.363 Note
nor 3.181 (decl.); 4.21; 4.23 fn 47
natim 3.252; 5.353
nu 4.335; 4.43
OJ 3.133 (aygi, ttli ); 3.17 (kin); 3.181 Note
(mi); 5.232
Ol 3.2 Note
olJ 4.21 ; 4.351
omn (pron.) 5.242
or (p,ron.) 3.181 (decl.) ;5.242
Or 4.21
orb 4.333
oroam 4.44
orsam 4.312; 6.41
or 4.336
orogem '4.323; 4.353; 4.44; 6.321 a; 5.41
oekr 4.364
otn, pl. otk' 3.162 (decl.); 4.311; 5.21-211
ov, o (pron.) 3.181 (decl.); 6.242
oy (GDAbl.) 6.232; 6.236
-oyo' (caua. suffix) 3.244; 4.23; 5.412
-oyt' 4.22
[Link] 2.344; 4.333

afa4c'ia)

sterJ 4.343 ; 4.351

stipem 4.332
stuar 4.334
surb 4.342
suzanem (3.252 f); 6.1
exalim 4.332 b; 4.343

M>I 4.451 (nloyl)

iiun 3.141 b (dee!.); 4.364; 5.144
tam, aor. etu 3.255 c; 4.323 Note; 6.32; 5.333
Note (aor. subj.); 5.432 (aor.)
[Link], aor. taray 3.255 b; 6.411
tari 3.133
t&l!n 4.331; 4.334; 5.211
tawn 4.343
taygr 4.353; 5.141
teli 3.133
~r, tikin 2.213 Note; 3.17
t(i)- 4.21
titloe l.62
tiw 3.17 (GD. tumfean); 4.353
tun 3.141 b (decl.) ; 5.144; 5.212 fn 92
..tur 4.323 Note; 4.332
......

palat l.61
paltawn 4.22; 4.331
plpJam (3.251 c); 4.351; 5.41 Note
p'atk' 1.61
p'aeian l.61
p'axll'im 3.263 h; 3.221 Note (part.); 5.352;
6.363
p'aylem 3.32 Note(p'ayl); 4.461; 6.3~1b ;6.41
(pres.); 5.433 (aor.)
p'lanim (3.253 g) ; 4.332 b
p' oyt' 4.343 Note
p' liat' l.61
([Link].) 2.322; (3.183); 6.222-223 .
ear 4.334
eartnum (3.264); 2.343; 4.332; 6.352 (aor.);
6.422

..

or ":
1

,;

, .'
.'"

til~ga.'Wor 4.121
t.;;.t ~. 4.334
t'aramim, t'ariiamim 4.334; 4.336
t'ek'ein 4.341
t'eli 4.343
t 'em l.61
t'er 4.343
t'ewr 4.341
t'olum (3.263 e); 4.461 (t'oyl); 6.422 Note
t'ofn 3.143 (decl.); 6.144
t'iiuar 2.31

A List of Claesioal Armenian Words and Morphemes

-uhi (-urhi) 4.121
ul 2.22 l ; 4.334
um 4.335; 5.243
-um (DL.) 3.181; 3.183:; 5.243
-umn 3.142 c; 4.22; 5.145
-un 3.32; 4.22
unim, aor. kalay 3.255 a; 5.32
unkn, pl. akanjk' 3.152; 4.131; 5.212
ur 4.334 Note; 4.341
us 4.335
utem, aor. keray 3.255 b; 5.353; 5.41 (pres.)
ut' 4.343
-ut'iwn 3.141 b; 3.32; 4.22
vat'eun 4.331
vay 4.353 Note
vec' 4.331; 4.343

139

verJ 4.351
xacanem (3.253 f); 4.832 b; 4.352; 5.41
:mjor 4.5
yalt' 4.332 b
yamem, aor. yareay 3.255 b; 5.352; 5.353;
5.411
yawelum (3.253 e); 4.333
z. (prepoe.)

2.311; 3.112

zard 5.131
zenum (3.253 e); 5.428
zgenum, aor. zgec'ay (3.254); 5.422 (pres.);
5.433 (aor.)
zgeet 4.122; 4.22; 4.343
zi, zin6' (pron). 3.111; 3.181; 5.242.