An examination of some ways in which languages vary, and how this can be addressed within the con... more An examination of some ways in which languages vary, and how this can be addressed within the context of machine translation
Proceedings of Workshop on Teaching Machine Treanslation, 2002
We discuss a framework for teaching contrastive translation-oriented linguistics to students of m... more We discuss a framework for teaching contrastive translation-oriented linguistics to students of machine translation. The framework embodies a classification of crosslinguistic differences and a demonstration of how the neutralization of such differences in abstract representations can make transfer modules simpler. We illustrate by means of examples from a range of languages, and then discuss a proposal for linguistic typology to which our system can be fruitfully related. 1
The first edition of this textbook from a team of well-known linguists at the University of Essex... more The first edition of this textbook from a team of well-known linguists at the University of Essex was published in 1999. Chapters have been updated and new material added for this revised edition. There are many introductions to linguistics available, so what does this one have to offer as a unique selling point? The first is that the multiple authorship means that the contents are reliable and accurate, with writers contributing on their areas of expertise, rather than a single author trying to cover a lot of ground. But the arrangement of chapters is the most noticeable difference from other texts. Rather than starting with phonetics, grammar, semantics, and then going on to topics such as child language and sociolinguistics, this introduction is divided into three main parts, sounds, words and sentences, each beginning with basic structural information before going on to the connections to acquisition, processing, and so on. Thus, the part on sounds has a 20-page introduction to phonetics, before diving straight into sociolinguistic variation with omission of initial (h). Then come sound change, phonemes and phonology, child phonology and speech perception and production. The part on words starts with essential information on word classes, morphology and lexical semantics, before covering words in child language, the mental lexicon, lexical disorders and lexical variation and change. I think many students will find this method of organisation attractive, as they often prefer areas such as language acquisition and variation to the 'dull' basics of morphological and syntactic analysis. It may even help get the students interested in these descriptive basics. It is also a positive aspect that the later chapters in each part refer back to the earlier ones: so the chapter 'Children and words' takes up points made in previous chapters
Budapest, Hungary 1. Introduction 1.1. Main characteristics of the Hungarian language Hungarian b... more Budapest, Hungary 1. Introduction 1.1. Main characteristics of the Hungarian language Hungarian belongs to the Finno-Ugric language family, whose other well-known representative is Finnish. The relation is obvious at the grammatical level, but very distant for vocabulary. Typologically it is generally considered to be an agglutinative language (like Finnish or Japanese). This means that grammatical roles like object or locative are often expressed by suffixes in Hungarian sentences. Consequently, word forms are morpheme-sequences consisting of a stem and several grammatical morphemes with additional syntactic information such as person, possession, and so on. The vowel harmony (as in Turkish) determines the necessary form of the suffix (e.g. házbm 'in the house', but kertbtn 'in the garden'). The rich morphological structure of the language enables a relatively free word order, but it is influenced significantly by the topic-comment structure in question. The neutral form is SOV, e.g. A fiú levelet ir.-'The boy is writing a letter.' the boy letter-ACC writes
De ideologische dimensie is meestal niet inherent aan een bepaalde vorm van spelling, maar wordt ... more De ideologische dimensie is meestal niet inherent aan een bepaalde vorm van spelling, maar wordt eraan toegekend in een politiek krachtenspel. 9 Mijn werkplek, dat is mijn hoofd. the ideas developed at that conference find their reflection in this work. Furthermore, I am grateful to the anthropologists Ted and Reggie MacDonald, who made it possible for me to do fieldwork in Arajuno, Pastaza, and to Roselyn Howard, who paved the way for me in Canar. Their help and hospitality were of great value.
Our paper focuses on the one hand on the challenges posed by the structural variability, flexibil... more Our paper focuses on the one hand on the challenges posed by the structural variability, flexibility and ambiguity found in historical corpora and evaluates methods of dealing with them on the other. We are currently engaged in a project which aims to compile a representative corpus of German for the period 1650-1800. Looking at exemplary data from the first stage of this project (1650-1700), which consists of newspaper texts from this period, we first aim from the perspective of corpus linguistics to identify the problems associated with the morphological, syntactical and graphemic peculiarities that are characteristic of that particular stage. Specific phenomena which significantly complicate automatic tagging, lemmatisation and parsing include, for instance, "abperlende" (Admoni 1980; Demske-Neumann 1990), i.e. complex and often asyndetic syntax; non-syntactic, prosodic, virgulated punctuation (Demske et al. 2004; cf. Stolt 1990), inflectional variability (e.g. Admoni 1...
This paper describes an annotated gold standard sample corpus of Early Modern German containing o... more This paper describes an annotated gold standard sample corpus of Early Modern German containing over 50,000 tokens of text manually annotated with POS tags, lemmas, and normalised spelling variants. The corpus is the first resource of its kind for this variant of German, and represents an ideal test bed for evaluating and adapting existing NLP tools on historical data. We describe the corpus format, annotation levels, and challenges, providing an example of the requirements and needs of smaller humanities-based corpus projects.
This paper describes a software system intended to help students learn about the process of taggi... more This paper describes a software system intended to help students learn about the process of tagging, the assignment of part-of-speech labels, which is basic to computational linguistics. It does this by (a) making it transparent why words have been analysed in a particular way, and (b) making it straightforward to add items to the system’s lexicon and improve its grammatical coverage. An unforeseen and welcome side-effect is that users are led to reflect on various properties of English grammar.
The goal of this study is to evaluate an 'off-the-shelf' POS-tagger for modern German on ... more The goal of this study is to evaluate an 'off-the-shelf' POS-tagger for modern German on historical data from the Early Modern period (1650-1800). With no specialised tagger available for this particular stage of the language, our findings will be of particular interest to smaller, humanities-based projects wishing to add POS annotations to their historical data but which lack the means or resources to train a POS tagger themselves. Our study assesses the effects of spelling variation on the performance of the tagger, and investigates to what extent tagger performance can be improved by using 'normalised' input, where spelling variants in the corpus are standardised to a modern form. Our findings show that adding such a normalisation layer improves tagger performance considerably.
... Manchester Peter Whitelock, Sharp Laboratories of Europe Christoph ZShner, UMIST Further titl... more ... Manchester Peter Whitelock, Sharp Laboratories of Europe Christoph ZShner, UMIST Further titles in the series Linguistic and computational techniques in Machine Translation system design Peter Whitelock & Kieran Kilby Analogical natural language processing Daniel Jones ...
In diesem Artikel wird an Hand eines repräsentativen Korpus der Beitrag der Zeitungen in der früh... more In diesem Artikel wird an Hand eines repräsentativen Korpus der Beitrag der Zeitungen in der frühmodernen Periode (1650-1800) zur sprachlichen Standisierung des Deutschen neu bewertet. Durch einige ausgewählte Beispiele aus Phonologie, Morphologie und Lexis wird gezeigt, dass die Drucker sich allmählich der überregionalen Sprache der Korrespondenten anpassten und somit zum endgültigen Selegierungsvorgang bei der Herausbildung der deutschen Standardsprache beitrugen. This article reexamines the contribution of newspapers to the standardization of German in the early modern period (1650-1800), using a new representative corpus. It is shown on the basis of selected examples from phonology, morphology and lexis that the printers gradually adapted to the supra-regional language of the newspaper correspondents and thus assisted the final process of selection in the development of standard German. 1. Zur Standardisierung der deutschen Sprache in der frühmodernen Periode. In den letzten Jahrzehnten ist die sprachwissenschaftliche Forschung zu einem vertieften Verständnis von sprachlichen Normen und Normierungsprozessen gekommen. Auf Grund einer bemerkenswert intensiven Beschäftigung mit der sprachlichen Standardisierung in letzter Zeit verfügen wir insbesondere über breite allgemeine Kenntnisse darüber, wie sich sprachliche Normen herauskristallisieren und wie und warum Standardsprachen entstehen 1. Vor allem wird dabei auf die bahnbrechenden Einsichten des norwegisch-amerikanischen 1 Sehr nützliche Übersichten zu diesem Thema, vor allem in Bezug auf die germanischen Sprachen, bieten neuerdings Vandenbussche (2007) und Deumert u. Vandenbussche (hg.) (2003).
We discuss a framework for teaching con- trastive translation-oriented linguistics to students of... more We discuss a framework for teaching con- trastive translation-oriented linguistics to students of machine translation. The frame- work embodies a classification of cross- linguistic differences and a demonstration of how the neutralization of such differences in abstract representations can make trans- fer modules simpler. We illustrate by means of examples from a range of languages, and then discuss a proposal
In this article, we outline the contents of the linguistic specifications of the Eurotra machine ... more In this article, we outline the contents of the linguistic specifications of the Eurotra machine translation system. We start in sections 1 and 2 from some of the requirements placed by multilingual MT on the overall design of the linguistic components. We then move on to a characterization of the Eurotra interface structure (section 3), the nature of transfer (section 4), and trends towards more interlingual representations within the project (section 5). Thereafter, we concentrate on the contents of the various levels beside the interface structure (section 6) before giving a brief survey of word structure (section 7) and outlining some areas for further research (section 8).' KEYWORDS: Eurotra, linguistic specifications, translation theory 'As it happens, crane1 and crane2 are distinguished m some EC languages (e.g. Spanish grulfa vs. grua and Dutch kraanvogel vs. kraan).
This paper describes an annotated gold standard sample corpus of Early Modern German containing o... more This paper describes an annotated gold standard sample corpus of Early Modern German containing over 50,000 tokens of text manually annotated with POS tags, lemmas, and normalised spelling variants. The corpus is the first resource of its kind for this variant of German, and represents an ideal test bed for evaluating and adapting existing NLP tools on historical data. We describe the corpus format, annotation levels, and challenges, providing an example of the requirements and needs of smaller humanities-based corpus projects.
Proceedings of the 5th ACL-HLT Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities}
We present a general and simple method to adapt an existing NLP tool in order to enable it to dea... more We present a general and simple method to adapt an existing NLP tool in order to enable it to deal with historical varieties of languages. This approach consists basically in expanding the dictionary with the old word variants and in retraining the tagger with a small training corpus. We implement this approach for Old Spanish. The results of a thorough evaluation over the extended tool show that using this method an almost state-of-the-art performance is obtained, adequate to carry out quantitative studies in the humanities: 94.5% accuracy for the main part of speech and 92.6% for lemma. To our knowledge, this is the first time that such a strategy is adopted to annotate historical language varieties and we believe that it could be used as well to deal with other non-standard varieties of languages.
Proceedings of the 5th ACL-HLT Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities}
We present a general and simple method to adapt an existing NLP tool in order to enable it to dea... more We present a general and simple method to adapt an existing NLP tool in order to enable it to deal with historical varieties of languages. This approach consists basically in expanding the dictionary with the old word variants and in retraining the tagger with a small training corpus. We implement this approach for Old Spanish. The results of a thorough evaluation over the extended tool show that using this method an almost state-of-the-art performance is obtained, adequate to carry out quantitative studies in the humanities: 94.5% accuracy for the main part of speech and 92.6% for lemma. To our knowledge, this is the first time that such a strategy is adopted to annotate historical language varieties and we believe that it could be used as well to deal with other non-standard varieties of languages.
An examination of some ways in which languages vary, and how this can be addressed within the con... more An examination of some ways in which languages vary, and how this can be addressed within the context of machine translation
Proceedings of Workshop on Teaching Machine Treanslation, 2002
We discuss a framework for teaching contrastive translation-oriented linguistics to students of m... more We discuss a framework for teaching contrastive translation-oriented linguistics to students of machine translation. The framework embodies a classification of crosslinguistic differences and a demonstration of how the neutralization of such differences in abstract representations can make transfer modules simpler. We illustrate by means of examples from a range of languages, and then discuss a proposal for linguistic typology to which our system can be fruitfully related. 1
The first edition of this textbook from a team of well-known linguists at the University of Essex... more The first edition of this textbook from a team of well-known linguists at the University of Essex was published in 1999. Chapters have been updated and new material added for this revised edition. There are many introductions to linguistics available, so what does this one have to offer as a unique selling point? The first is that the multiple authorship means that the contents are reliable and accurate, with writers contributing on their areas of expertise, rather than a single author trying to cover a lot of ground. But the arrangement of chapters is the most noticeable difference from other texts. Rather than starting with phonetics, grammar, semantics, and then going on to topics such as child language and sociolinguistics, this introduction is divided into three main parts, sounds, words and sentences, each beginning with basic structural information before going on to the connections to acquisition, processing, and so on. Thus, the part on sounds has a 20-page introduction to phonetics, before diving straight into sociolinguistic variation with omission of initial (h). Then come sound change, phonemes and phonology, child phonology and speech perception and production. The part on words starts with essential information on word classes, morphology and lexical semantics, before covering words in child language, the mental lexicon, lexical disorders and lexical variation and change. I think many students will find this method of organisation attractive, as they often prefer areas such as language acquisition and variation to the 'dull' basics of morphological and syntactic analysis. It may even help get the students interested in these descriptive basics. It is also a positive aspect that the later chapters in each part refer back to the earlier ones: so the chapter 'Children and words' takes up points made in previous chapters
Budapest, Hungary 1. Introduction 1.1. Main characteristics of the Hungarian language Hungarian b... more Budapest, Hungary 1. Introduction 1.1. Main characteristics of the Hungarian language Hungarian belongs to the Finno-Ugric language family, whose other well-known representative is Finnish. The relation is obvious at the grammatical level, but very distant for vocabulary. Typologically it is generally considered to be an agglutinative language (like Finnish or Japanese). This means that grammatical roles like object or locative are often expressed by suffixes in Hungarian sentences. Consequently, word forms are morpheme-sequences consisting of a stem and several grammatical morphemes with additional syntactic information such as person, possession, and so on. The vowel harmony (as in Turkish) determines the necessary form of the suffix (e.g. házbm 'in the house', but kertbtn 'in the garden'). The rich morphological structure of the language enables a relatively free word order, but it is influenced significantly by the topic-comment structure in question. The neutral form is SOV, e.g. A fiú levelet ir.-'The boy is writing a letter.' the boy letter-ACC writes
De ideologische dimensie is meestal niet inherent aan een bepaalde vorm van spelling, maar wordt ... more De ideologische dimensie is meestal niet inherent aan een bepaalde vorm van spelling, maar wordt eraan toegekend in een politiek krachtenspel. 9 Mijn werkplek, dat is mijn hoofd. the ideas developed at that conference find their reflection in this work. Furthermore, I am grateful to the anthropologists Ted and Reggie MacDonald, who made it possible for me to do fieldwork in Arajuno, Pastaza, and to Roselyn Howard, who paved the way for me in Canar. Their help and hospitality were of great value.
Our paper focuses on the one hand on the challenges posed by the structural variability, flexibil... more Our paper focuses on the one hand on the challenges posed by the structural variability, flexibility and ambiguity found in historical corpora and evaluates methods of dealing with them on the other. We are currently engaged in a project which aims to compile a representative corpus of German for the period 1650-1800. Looking at exemplary data from the first stage of this project (1650-1700), which consists of newspaper texts from this period, we first aim from the perspective of corpus linguistics to identify the problems associated with the morphological, syntactical and graphemic peculiarities that are characteristic of that particular stage. Specific phenomena which significantly complicate automatic tagging, lemmatisation and parsing include, for instance, "abperlende" (Admoni 1980; Demske-Neumann 1990), i.e. complex and often asyndetic syntax; non-syntactic, prosodic, virgulated punctuation (Demske et al. 2004; cf. Stolt 1990), inflectional variability (e.g. Admoni 1...
This paper describes an annotated gold standard sample corpus of Early Modern German containing o... more This paper describes an annotated gold standard sample corpus of Early Modern German containing over 50,000 tokens of text manually annotated with POS tags, lemmas, and normalised spelling variants. The corpus is the first resource of its kind for this variant of German, and represents an ideal test bed for evaluating and adapting existing NLP tools on historical data. We describe the corpus format, annotation levels, and challenges, providing an example of the requirements and needs of smaller humanities-based corpus projects.
This paper describes a software system intended to help students learn about the process of taggi... more This paper describes a software system intended to help students learn about the process of tagging, the assignment of part-of-speech labels, which is basic to computational linguistics. It does this by (a) making it transparent why words have been analysed in a particular way, and (b) making it straightforward to add items to the system’s lexicon and improve its grammatical coverage. An unforeseen and welcome side-effect is that users are led to reflect on various properties of English grammar.
The goal of this study is to evaluate an 'off-the-shelf' POS-tagger for modern German on ... more The goal of this study is to evaluate an 'off-the-shelf' POS-tagger for modern German on historical data from the Early Modern period (1650-1800). With no specialised tagger available for this particular stage of the language, our findings will be of particular interest to smaller, humanities-based projects wishing to add POS annotations to their historical data but which lack the means or resources to train a POS tagger themselves. Our study assesses the effects of spelling variation on the performance of the tagger, and investigates to what extent tagger performance can be improved by using 'normalised' input, where spelling variants in the corpus are standardised to a modern form. Our findings show that adding such a normalisation layer improves tagger performance considerably.
... Manchester Peter Whitelock, Sharp Laboratories of Europe Christoph ZShner, UMIST Further titl... more ... Manchester Peter Whitelock, Sharp Laboratories of Europe Christoph ZShner, UMIST Further titles in the series Linguistic and computational techniques in Machine Translation system design Peter Whitelock & Kieran Kilby Analogical natural language processing Daniel Jones ...
In diesem Artikel wird an Hand eines repräsentativen Korpus der Beitrag der Zeitungen in der früh... more In diesem Artikel wird an Hand eines repräsentativen Korpus der Beitrag der Zeitungen in der frühmodernen Periode (1650-1800) zur sprachlichen Standisierung des Deutschen neu bewertet. Durch einige ausgewählte Beispiele aus Phonologie, Morphologie und Lexis wird gezeigt, dass die Drucker sich allmählich der überregionalen Sprache der Korrespondenten anpassten und somit zum endgültigen Selegierungsvorgang bei der Herausbildung der deutschen Standardsprache beitrugen. This article reexamines the contribution of newspapers to the standardization of German in the early modern period (1650-1800), using a new representative corpus. It is shown on the basis of selected examples from phonology, morphology and lexis that the printers gradually adapted to the supra-regional language of the newspaper correspondents and thus assisted the final process of selection in the development of standard German. 1. Zur Standardisierung der deutschen Sprache in der frühmodernen Periode. In den letzten Jahrzehnten ist die sprachwissenschaftliche Forschung zu einem vertieften Verständnis von sprachlichen Normen und Normierungsprozessen gekommen. Auf Grund einer bemerkenswert intensiven Beschäftigung mit der sprachlichen Standardisierung in letzter Zeit verfügen wir insbesondere über breite allgemeine Kenntnisse darüber, wie sich sprachliche Normen herauskristallisieren und wie und warum Standardsprachen entstehen 1. Vor allem wird dabei auf die bahnbrechenden Einsichten des norwegisch-amerikanischen 1 Sehr nützliche Übersichten zu diesem Thema, vor allem in Bezug auf die germanischen Sprachen, bieten neuerdings Vandenbussche (2007) und Deumert u. Vandenbussche (hg.) (2003).
We discuss a framework for teaching con- trastive translation-oriented linguistics to students of... more We discuss a framework for teaching con- trastive translation-oriented linguistics to students of machine translation. The frame- work embodies a classification of cross- linguistic differences and a demonstration of how the neutralization of such differences in abstract representations can make trans- fer modules simpler. We illustrate by means of examples from a range of languages, and then discuss a proposal
In this article, we outline the contents of the linguistic specifications of the Eurotra machine ... more In this article, we outline the contents of the linguistic specifications of the Eurotra machine translation system. We start in sections 1 and 2 from some of the requirements placed by multilingual MT on the overall design of the linguistic components. We then move on to a characterization of the Eurotra interface structure (section 3), the nature of transfer (section 4), and trends towards more interlingual representations within the project (section 5). Thereafter, we concentrate on the contents of the various levels beside the interface structure (section 6) before giving a brief survey of word structure (section 7) and outlining some areas for further research (section 8).' KEYWORDS: Eurotra, linguistic specifications, translation theory 'As it happens, crane1 and crane2 are distinguished m some EC languages (e.g. Spanish grulfa vs. grua and Dutch kraanvogel vs. kraan).
This paper describes an annotated gold standard sample corpus of Early Modern German containing o... more This paper describes an annotated gold standard sample corpus of Early Modern German containing over 50,000 tokens of text manually annotated with POS tags, lemmas, and normalised spelling variants. The corpus is the first resource of its kind for this variant of German, and represents an ideal test bed for evaluating and adapting existing NLP tools on historical data. We describe the corpus format, annotation levels, and challenges, providing an example of the requirements and needs of smaller humanities-based corpus projects.
Proceedings of the 5th ACL-HLT Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities}
We present a general and simple method to adapt an existing NLP tool in order to enable it to dea... more We present a general and simple method to adapt an existing NLP tool in order to enable it to deal with historical varieties of languages. This approach consists basically in expanding the dictionary with the old word variants and in retraining the tagger with a small training corpus. We implement this approach for Old Spanish. The results of a thorough evaluation over the extended tool show that using this method an almost state-of-the-art performance is obtained, adequate to carry out quantitative studies in the humanities: 94.5% accuracy for the main part of speech and 92.6% for lemma. To our knowledge, this is the first time that such a strategy is adopted to annotate historical language varieties and we believe that it could be used as well to deal with other non-standard varieties of languages.
Proceedings of the 5th ACL-HLT Workshop on Language Technology for Cultural Heritage, Social Sciences, and Humanities}
We present a general and simple method to adapt an existing NLP tool in order to enable it to dea... more We present a general and simple method to adapt an existing NLP tool in order to enable it to deal with historical varieties of languages. This approach consists basically in expanding the dictionary with the old word variants and in retraining the tagger with a small training corpus. We implement this approach for Old Spanish. The results of a thorough evaluation over the extended tool show that using this method an almost state-of-the-art performance is obtained, adequate to carry out quantitative studies in the humanities: 94.5% accuracy for the main part of speech and 92.6% for lemma. To our knowledge, this is the first time that such a strategy is adopted to annotate historical language varieties and we believe that it could be used as well to deal with other non-standard varieties of languages.
Uploads
Papers by Paul Bennett