intransitive bases and (2) for the transitive bases (Spencer, 1991:252-253).1+ As noted from (2), there are three possibilities for causative constructions which can retain all the participants of the original transitive verbs. The facts about Indonesian causatives, however, suggest that a fourth type should be entered: a causative with a transitive base where the original agent (NP1) becomes an obligatory oblique in the derived causative form. Thus, it is just like (2b3) but differs in that the NP1 OBL should not be put within brackets, as shown in (3). An important guiding principle of grammar introduced by Baker is UTAH Uniformity of Theta Assignment Hypothesis) as formulated in (5). and technical (see Baker 1988 for details). Structural configuration is the heart of GB. Its sub-theory of X-bar system requires the recognition of the notion of projections. A syntactic category XP is the projection of its lexical head X. A clause is also argued to be the projection of INFL I). Thus a clause symbolised as S is equivalent with IP. In short, causative as verb incorporation (VI) is not as simple as that in (4b). The complex structure of VI is represented in (7)°. The issue of barrier is inherent here; much _ stipulation is herefore needed and Baker's discussion consequently gets more and more abstract and technical (see Baker 1988 for details). therefore needed and Baker's discussion consequently gets more and more abstract represented in (7)°. The issue of barrier is inherent here; much stipulation is (V2 to C) is shown in (8c). structure of the non-incorporated structure is shown in (12ci) and the PI in (12c ii). requires that they both must be represented as having the same D-structure. The D- structure. It moves to adjoin the matrix verb, leaving a trace (structure preservation) while at the same time because of the demands of the case theory, the applied NP Notice that the N buku is abstractly incorporated into the verb. This is shown by ‘oindexation. This is called Abstract Incorporation because the syntactic novement here occurs in LF(logical form), a level which has no direct link to PF phonetic form) in the GB model. Thus this process does not yield any norphological realisation at PF. Being incorporated, the second NP escapes from he case filter. It does not need any case again. Nevertheless, recasting the solutior is the Abstract Incorporation/Reanalysis, at least as far as Indonesian is concerned. oes not resolve the real problem: suppose we do have the Abstract Incorporation in APPL -kan, why don't we have the same case in APPL -i ? Why can it not be »xtended to the causative as well? The incorporation in (12c), can be now represented as abstract incorporation as special case of incorporation, what he calls "Incorporation without Incorporation’. Table 4. Properties of Selected Simple G Fs in Indonesian 4.3 Indonesian Phrase Structure Rule In the sentence above there are three V-bars; thus our tests must correctly predict that there are three ‘possible’ clefting constructions using -lah and lakukan. "He/she bought a book in Denpasar yesterday afternoon with his/her friends’ realised by S. A few words are needed for the PS rule in (61). First of all, the selected GF: BJ, OBJg and (X)COMP and OBL, being linked to the subcategorised argument: f V, are annotated to the nodes that are the sisters of the V® (i.e. the lexical heac ’). Second, the ADJ rule (i.e the one that expands V' to V') is put within bracket: » encode the idea that the structural configuration (and the annotated GF) i: ptional. Third, OBL may be annotated in two nodes. This is to capture the fact tha 1ere are two kinds of OBL. The OBL function which is an argument function anc hich is therefore obligatorily present (e.g OBL in the derived causative) i: nnotated to the sister node of V°. The other OBL that is optionally present (e.g the /BLag in passive or what Grimshaw (1990) calls the a-adjunct)) is structurally spresented in the same way as ADJ. It is annotated to the sister of V'. This is tc apture the idea that such an OBL in some sense patterns more with a non-argumen inction. That is, the OBLag is suppressed by the lexical rule of PASS and it: resence is, like ADJ, optional. However, we want to extend the PS rule to capture the fact that ADJ and COMP are conventions. The PS rule in (60) can be annotated with the functional schemata. In this chapter, the properties of GFs in Indonesian have been discussed. Diagnostic tests for selected functions (SUBJ, OBJg, OBJ and OBL) have been identified. The tests are associated with various characteristics of the selected GFs in Indonesian. They have to do with the structural properties (i.e their relative positions in the c-structure with respect to the predicate (V or P)), their grammatical properties (e.g related with control and reflexive), discourse notions of TOP and FOC and other language specific characteristics such as the possible substitution with dia vs ia ~3SG’. In recent GB work, there is a tendency to rely as much as possible on X- bar principles and state the constituency in terms of licensing conditions such as Heads Initial/Final Parameter mentioned in (28). In this view, PS rules are 4 X-bar is, in fact, a PS schema. In GB it specifies the tree- structures in the D-structure. The structural representation is crucial in the other sub-theories (such as theta theory, bounding theory) to constrain move-a. It is assumed that it is universal (for example all languages have V and the maximal projection is of the same bar level (VP ), Similarly all languages have I = and IP etc.). They differ only in the choice of position of the heads with respect to their sisters. In other frameworks, such as LFG, where semantic information is not’ structurally represented in tree-structures the X-bar schema is associated with the PS rule of surface syntax. Consequently, it is not Note the prefixes” are not the morphological heads. The heads are the stems, which are in fact lexical items. The stem in the last two examples above (per-besar) has its own internal structure as well. This shows the recursive® nature of W-structure as predicted by our rule here. The W-structure rule in (38) can now be modified as (42) and a representative example of the word tree is displayed in (43). be represented by using the template in (38) because the head is, even though its own internal structure as well. This shows the recursive® nature of W-structure The internal structure of the adjective in (39b) and the verbs in (40) canno ke-depan-kan is the result of an incorporation process. That the PP is moved up and incorporated to the matrix is motivated by the morphological subcategorization frame of the causative. Because it is a suffix, in the S-Structure it must satisfy its subcategorization frame. It must be affixed to a word or base. It is this requirement that makes the PP move and incorporate to its lexical head (i.e the matrix -kan). The movement must respect various constraints, as assumed in GB, such as ECP. Note that the CFS now carries the categorial information (i.e V]-JN ) from the lexical entry of the suffix -ing. That is, its morphological frame says that it must be attached to a verbal stem. This information can be viewed as a redundancy check on the [+N] feature in the CFS and the terminal node of the word tree (N-!). The PRED value slot of the CFS is still empty (unspecified) because -ing has no PRED value in its lexical entry. It is therefore unable to supply any lexical content to the slot. Let us take a derived -ing nominal in English: the killing of the people. 'S This is a language specific property. English and Indonesian apparently share this property. Indonesian differs from English however in that it allows a noun to act as an argument taking predicate. e.g. guru ~* teacher' in Ia guru “he is a teacher’ (see Section 3.3). Even in this case a restriction is observed. It only allows the SUBJ function, never OBJ. In languages with case affixes (e.g Warlpiri, Greenlandic Eskimo, etc) where case suffixes can act as argument taking predicates the f-structures associated with nominals may allow SUBJ,OBJ and other GFs that the corresponding f-structures of verbs in English permit. feature and predicate percolation involved is shown in the word tree below. By the Head Percolation Convention the CFS that now contains the PRED value of the verb tabrak passes up to the dominating node. By the Back up Convention the classification features [-r,-o] are transferred up. Being the highest role, the semantic role <ag> can be assigned the classification features. Hence it will be the SUBJ.” Later in the GF-mapping process, the monotonicity constraint dictates that the features must be preserved. Since the prefix di- operates on the same argument, an attempt to attach it after meN- is already affixed results in an ill-formed verb. Therefore the sentence with the verb form * di-men-(t)abrak is not acceptable. functional and morphological head. It propagates the CFS. The process of the ber- and pakai. The prefix is semantically transitive. When it emerges from the affixation with the APPL -kan, yielding bertopikan (cf. 105e and 105f). Let u: compounds in that the immediate dominating node is not an XP. 122) The Indonesian Word Structure Rule, Version 8 (Final) (157) mendatangkan The single argument of datang (156) has no other argument to compete with. By the SUBJ principle it is assigned [-r,-0]. The <th> is then mapped onto SUBJ. However, when datang “COME' undergoes causativisation (157), the <th> argument is now fused with the causee (i.e <pt> ). It is outranked by the causer (i.e <ag> ). Note that the derived form has got the actor oriented prefix meN- which assigns the same specified feature classifications as the SUBJ principle does. This leaves the causee with no other choice but [-r, +0], making it map onto OBJ. Thus, the verb datang and mendatangkan emerge from the lexicon with the GF STRs of <SUBJ> and <SUBJ,OBJ> respectively. (158) mendarat As already mentioned in (5.6.7 ex. 131 & 132) darat can never be a verb. The verbal prefix meN- introduces a PRED ~DO' with an agent argument and an uP signifying the unspecified thing done. The base PRED (DARAT’) is a constant coindexed with the <go> argument of the PRED ~TO<[th][go]> that is introduced by the RI (Rule of Inference). Being a constant it must not be linked to any GF. The agent is also understood to be the theme. That is, the two semantic roles actually refer to a single participant. By the SUBJ principle it is mapped onto SUBJ naturally. 163) meng-obat-kan ~ have someone treated medically by someone else’ The causer of the most inclusive ARG STR (i.e the agent of the current (or matrix) CAUSE predicate) is the most prominent role. This is the purer agent. Being the highest in the thematic hierarchy, it wins out and is therefore mapped onto SUBJ (by virtue of the SUBJ principle and the actor oriented prefix meN-). The lower causer cannot be classified as [-r] because this classification must be assigned to the lowest role, namely the <loc>. The <loc> argument must get it matrix) CAUSE predicate) is the most prominent role. This is the purer agent. assigned to the lowest role, namely the <loc>. The <loc> argument must get it
![Note that the CFS now carries the categorial information (i.e V]-JN ) from the lexical entry of the suffix -ing. That is, its morphological frame says that it must be attached to a verbal stem. This information can be viewed as a redundancy check on the [+N] feature in the CFS and the terminal node of the word tree (N-!). The PRED value slot of the CFS is still empty (unspecified) because -ing has no PRED value in its lexical entry. It is therefore unable to supply any lexical content to the slot. Let us take a derived -ing nominal in English: the killing of the people.](https://figures.academia-assets.com/46406300/figure_013.jpg)
![feature and predicate percolation involved is shown in the word tree below. By the Head Percolation Convention the CFS that now contains the PRED value of the verb tabrak passes up to the dominating node. By the Back up Convention the classification features [-r,-o] are transferred up. Being the highest role, the semantic role <ag> can be assigned the classification features. Hence it will be the SUBJ.” Later in the GF-mapping process, the monotonicity constraint dictates that the features must be preserved. Since the prefix di- operates on the same argument, an attempt to attach it after meN- is already affixed results in an ill-formed verb. Therefore the sentence with the verb form * di-men-(t)abrak is not acceptable. functional and morphological head. It propagates the CFS. The process of the](https://figures.academia-assets.com/46406300/figure_015.jpg)
![The single argument of datang (156) has no other argument to compete with. By the SUBJ principle it is assigned [-r,-0]. The <th> is then mapped onto SUBJ. However, when datang “COME' undergoes causativisation (157), the <th> argument is now fused with the causee (i.e <pt> ). It is outranked by the causer (i.e <ag> ). Note that the derived form has got the actor oriented prefix meN- which assigns the same specified feature classifications as the SUBJ principle does. This leaves the causee with no other choice but [-r, +0], making it map onto OBJ. Thus, the verb datang and mendatangkan emerge from the lexicon with the GF STRs of <SUBJ> and <SUBJ,OBJ> respectively.](https://figures.academia-assets.com/46406300/figure_018.jpg)
![(158) mendarat As already mentioned in (5.6.7 ex. 131 & 132) darat can never be a verb. The verbal prefix meN- introduces a PRED ~DO' with an agent argument and an uP signifying the unspecified thing done. The base PRED (DARAT’) is a constant coindexed with the <go> argument of the PRED ~TO<[th][go]> that is introduced by the RI (Rule of Inference). Being a constant it must not be linked to any GF. The agent is also understood to be the theme. That is, the two semantic roles actually refer to a single participant. By the SUBJ principle it is mapped onto SUBJ naturally.](https://figures.academia-assets.com/46406300/figure_019.jpg)
![163) meng-obat-kan ~ have someone treated medically by someone else’ The causer of the most inclusive ARG STR (i.e the agent of the current (or matrix) CAUSE predicate) is the most prominent role. This is the purer agent. Being the highest in the thematic hierarchy, it wins out and is therefore mapped onto SUBJ (by virtue of the SUBJ principle and the actor oriented prefix meN-). The lower causer cannot be classified as [-r] because this classification must be assigned to the lowest role, namely the <loc>. The <loc> argument must get it matrix) CAUSE predicate) is the most prominent role. This is the purer agent. assigned to the lowest role, namely the <loc>. The <loc> argument must get it](https://figures.academia-assets.com/46406300/table_008.jpg)