Computational Phraseology

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

We shed light on aspects of the relation between the semantics and the syntactic flexibility of multiword expressions by investigating fixed adjective similes (FS), a predicative multiword expression class not studied in this respect before. We find that only a subset of the syntactic structures observed in the data are related with idiomaticity. We identify and measure two aspects of idiomaticity, one of which seems to allow for predictions about FS syntactic flexibility. Our research draws on a resource developed with the semantic and detailed syntactic annotation of web-retrieved Modern Greek material, indicating frequency of use of the individual similes.

The precise identification of light verb constructions is crucial for the successful functioning of several NLP applications. In order to facilitate the development of an algorithm that is capable of recognizing them, a manually annotated corpus of light verb constructions has been built for Hungarian. Basic annotation guidelines and statistical data on the corpus are also presented in the paper. It is also shown how applications in the fields of machine translation and information extraction can make use of such a corpus and an algorithm.

MultiWord Expressions (MWEs) repesent a key issue for numerous applications in Natural Language Processing (NLP) especially for Machine Translation (MT). In this paper, we describe a strategy for detecting translation pairs of MWEs in a French-English parallel corpus. In addition we introduce three methods aiming to integrate extracted bilingual MWES in MOSES, a phrase based Statistical Machine Translation (SMT) system. We experimentally show that these textual units can improve translation quality.

Specific-domain bilingual lexicons play an important role for domain adaptation in machine translation. The entries of these types of lexicons are mostly composed of MultiWord Expressions (MWEs). The manual construction of MWEs bilingual lexicons is costly and time-consuming. We often use word alignment approaches to automatically construct bilingual lexicons of MWEs from parallel corpora. We present in this paper a hybrid approach to extract and align MWEs from parallel corpora in a one-step process. We formalize the alignment process as an integer linear programming problem in order to find an approximated optimal solution. This process generates lists of MWEs with their translations, which are then filtered using linguistic patterns for the construction of the bilingual lexicons of MWEs. We evaluate the bilingual lexicons of MWEs produced by this approach using two methods: a manual evaluation of the alignment quality and an evaluation of the impact of this alignment on the translation quality of the phrase-based statistical machine translation system Moses. We experimentally show that the integration of the bilingual MWEs and their linguistic information into the translation model improves the performance of Moses.

In this paper, we present a hybrid approach to align single words, compound words and idiomatic expressions from bilingual parallel corpora. The objective is to develop, improve and maintain automatically translation lexicons. This approach combines linguistic and statistical information in order to improve word alignment results. The linguistic improvements taken into account refer to the use of an existing bilingual lexicon, named entities recognition, grammatical tags matching and detection of syntactic dependency relations between words. Statistical information refer to the number of occurrences of repeated words, their positions in the parallel corpus and their lengths in terms of number of characters. Single-word alignment uses an existing bilingual lexicon, named entities and cognates detection and grammatical tags matching. Compound-word alignment consists in establishing correspondences between the compound words of the source sentence and the compound words of the target sentences. A syntactic analysis is applied on the source and target sentences in order to extract dependency relations between words and to recognize compound words. Idiomatic expressions alignment starts with a monolingual term extraction for each of the source and target languages, which provides a list of sequences of repeated words and a list of potential translations. These sequences are represented with vectors which indicate their numbers of occurrences and the numbers of segments in which they appear. Then, the translation relation between the source and target expressions are evaluated with a distance metric. The single and compound word aligners have been evaluated on a subset of 1103 sentences in English and French of the JOC (Official Journal of the European Community) corpus . The obtained results showed that these aligners generate a translation lexicon with 90 % of precision for single words and 84 % of precision for compound words. We evaluated the idiomatic expressions aligner on a subset of the Canadian Parliament Hansard corpus and we obtained a precision of 81%.

Specific-domain bilingual lexicons play an importan t role for domain adaptation in machine translation . The entries of these types of lexicons are mostly composed of MultiWord Expressio n (MWEs). The manual construction of MWEs bilingua l lexicons is costly and time-consuming. We often use word alignment approac hes to automatically construct bilingual lexicons o f MWEs from parallel corpora. We present in this paper a hybrid approach to extra ct and align MWEs from parallel corpora in a one-st p process. We formalize the alignment process as an integer linear programming problem in order to find an approximated optimal so luti n. This process generates lists of MWEs with their translations, which are th en filtered using linguistic patterns for the const ruc ion of the bilingual lexicons of MWEs. We evaluate the bilingual lexicons of MWEs pr oduced by this approach using two methods: a manual evaluation of the alignment quality and an evaluation of the impact of this ali gnment o...

We describe in this paper a hybrid approach to build automatically bilingual lexicons of Multiword Expressions (MWEs) from parallel corpora. We more specifically investigate the impact of using a domain-specific bilingual lexicon of MWEs on domain adaptation of an Example-Based Machine Translation (EBMT) system. We conducted experiments on the English-French language pair and two kinds of texts: in-domain texts from Europarl (European Parliament proceedings) and out-of-domain texts from Emea (European Medicines Agency documents) and Ecb (European Central Bank corpus). The obtained results indicate that integrating domain-specific bilingual lexicons of MWEs improves translation quality of the EBMT system when texts to translate are related to the specific domain and induces a relatively slight deterioration of translation quality when translating generalpurpose texts.

Identifying and translating MultiWord Expressions (MWES) in a text represent a key issue for numerous applications of Natural Language Processing (NLP), especially for Machine Translation (MT). In this paper, we present a method aiming to construct a bilingual lexicon of MWES from a French-English parallel corpus. In order to assess the quality of the mined lexicon, a Statistical Machine Translation (SMT) task-based evaluation is conducted. We investigate the performance of three dynamic strategies and of one static strategy to integrate the mined bilingual MWES lexicon in a SMT system. Experimental results shows that such a lexicon improves the quality of translation. Identifier et traduire correctement les Expressions Multi-Mots (EMMS) dans un texte constituent un défi majeur pour différentes applications du Traitement Automatique des Langues Naturelles, et surtout en Traduction Automatique. Ce présent travail présente une méthode permettant de construire un lexique bilingue d'EMMS à partir d'un corpus parallèle Français-Anglais. Afin d'évaluer la qualité du lexique acquis, une évaluation axée sur la tâche de Traduction Automatique Statistique (TAS) est menée. Nous étudions les performances de trois stratégies dynamiques et d'une stratégie statique pour intégrer le lexique bilingue d'EMMS dans un système de TAS. Les expériences menées dans ce cadre montrent que ces unités améliorent la qualité de traduction.

MultiWord Expressions (MWEs) repesent a key issue for numerous applications in Natural Language Processing (NLP) especially for Machine Translation (MT). In this paper, we describe a strategy for detecting translation pairs of MWEs in a French-English parallel corpus. In addition we introduce three methods aiming to integrate extracted bilingual MWES in MOSES, a phrase based Statistical Machine Translation (SMT) system. We experimentally show that these textual units can improve translation quality.

There is a great deal of knowledge available on the Web, which represents a great opportunity for automatic, intelligent text processing and understanding, but the major problems are finding the legitimate sources of information and the fact that search engines provide page statistics not occurrences. This paper presents a new, domain independent, general-purpose idiom identification approach. Our approach combines the knowledge of the Web with the knowledge extracted from dictionaries. This method can overcome the limitations of current techniques that rely on linguistic knowledge or statistics. It can recognize idioms even when the complete sentence is not present, and without the need for domain knowledge. It is currently designed to work with text in English but can be extended to other languages.

This paper presents recent developments in processing verbal idioms within a rule-based grammar of European Portuguese. It describes the automatic construction of parsing rules directly from a lexicon-grammar matrix with about 2,500 idioms and about 100 structural, distributional, and transformational properties. Transformations (passive, pronominalization, etc.) of idioms' base sentences are now taken into account within the automatic rule generation process. An intrinsic evaluation achieves 95% recall.

Background. Lexical knowledge, and in particular knowledge on multi-word expressions, is at the cornerstone of language applications such as syntactic parsing or machine translation. Corpus-driven lexical acquisition is one of the major means to create such knowledge, in order to build or consolidate dictionaries and similar types of lexical resources. We describe ongoing work devoted to the corpus-based extraction of multi-word expressions – in particular, collocations – for the Romanian language. Romanian is since 2002 one of the 23 official languages of the European Union; it is the native language or around 24 million people, and is currently ranked 8 th in the list of most spoken European languages worldwide, after Spanish (405 million native speakers), English (360), Portuguese (215), German (89), French (74), Italian (59), and Polish (40) 1 . This high rank contrasts, however, with the relatively scarce development of language resources and tools compared to other languages.

In this demo, we present our free on-line multilingual linguistic services which allow to analyze sentences or to extract collocations from a corpus directly on-line, or by uploading a corpus. They are available for 8 European languages (English, French, German, Greek, Italian, Portuguese, Romanian, Spanish) and can also be accessed as web services by programs. While several open systems are available for POS-tagging and dependency parsing or terminology extraction, their integration into an application requires some computational competence. Furthermore, none of the parsers/taggers handles MWEs very satisfactorily, in particular when the two terms of the collocation are distant from each other or in reverse order. Our tools, on the other hand, are specifically designed for users with no particular computational literacy. They do not require from the user any download, installation or adaptation if used on-line, and their integration in an application, using one the scripts describe...

D5.3 English-French and English-Greek parallel corpus for the Environment and Labour Legislation domains (M22) D5.5 English-French and English-Greek bilingual dictionaries for the Environment and Labour Legislation domains D7.4 Panacea Deliverable D7.4: Evaluation Report (Third cycle) D5.4 Bilingual Dictionary Extraction Tools This document is part of technical documentation generated in the PANACEA Project, Platform for Automatic, Normalized Annotation and Cost-Effective Acquisition (Grant Agreement no. 248064).

Automatic lexical alignment is a vital step for empirical machine translation, and although good results can be obtained with existent models (e.g. Giza++), more precise alignment is still needed for successfully handling complex constructions such as multiword expressions. In this paper we propose an approach for lexical alignment combining statistical and linguistic information. We describe the development of a baseline discriminative aligner and a set of language dependent post-processing functions that allow the inclusion of shallow linguistic knowledge. The post-processing functions were designed to significantly improve word alignment mainly on verb-particle constructs both over our baseline and over Giza++.

Multiword Expressions (MWEs) are used frequently in natural languages, but understanding the diversity in MWEs is one of the open problem in the area of Natural Language Processing. In the context of Indian languages, MWEs play an important role. In this paper, we present MWEs annotation dataset created for Indian languages viz., Hindi and Marathi. We extract possible MWE candidates using two repositories: 1) the POS-tagged corpus and 2) the IndoWordNet synsets. Annotation is done for two types of MWEs: compound nouns and light verb constructions. In the process of annotation, human annotators tag valid MWEs from these candidates based on the standard guidelines provided to them. We obtained 3178 compound nouns and 2556 light verb constructions in Hindi and 1003 compound nouns and 2416 light verb constructions in Marathi using two repositories mentioned before. This created resource is made available publicly and can be used as a gold standard for Hindi and Marathi MWE systems.

This system description explains how to use several bilingual dictionaries and aligned corpora in order to create translation candidates for novel language pairs. It proposes (1) a graph-based approach which does not depend on cyclical translations and (2) a combination of this method with a collocation-based model using the multilingually aligned Europarl corpus.

On the basis of Aranea Gigaword Web corpora, a family of comparable corpora intended for use in contrastive linguistic research, multilingual lexicography, language teaching and translation studies we discuss the pros and cons of comparable corpora in contrast to monolingual and parallel corpora for the analysis of multiword entities (MWEs). We demonstrate that by using large corpora for two or more languages, consisting of unrelated texts, yet created in a comparable manner, parallel language structures and phenomena like MWEs can be identified if the appropriate tools are employed. With the Aranea corpora, the "bilingual sketch" functionality of the Sketch Engine is one such tool which provides a new approach for analyses of similarities of (or differences between) collocation profiles (word sketches) for words and their translation equivalents.

Phraseology is a key component of the development of linguisticcommunicative competence of advanced level learners of a foreign language (Wray 2002); furthermore, it represents a major teaching challenge, being a highly stereotyped and conventional part of a language. Despite their centrality in the acquisitional path, phraseological units are considered marginally in the language classroom, above all due to the lack of dedicated resources.In this regard, we will analyse the idioms of geography containing reference to place names or nationalities, for Italian and Spanish, to then use them in a teaching proposal that aims to improve intercultural communicative competence (Byram 1997) of Italian university students, studying Spanish as a FL at A2 level of the CEFR (2002). Methodologically, we started by extracting from the Dictionary of idioms of the Italian language (Lapucci 1969) some Italian geographical phraseological units. Then we made a comparison with the Spanish language, to identify total or partial structural correspondences, or semantic equivalences, although with different geographical references. The linguistic units thus obtained for the two languages were then classified in a didactic perspective, based on some defining features (Masini 2009, Sevilla Muñoz 2008). The outcome of this analysis finally made it possible to develop a didactic proposal, focused on learners' direct use and exploration of the selected phraseological units, as well as on the analysis of their interlinguistic and intercultural affinities and differences. The idioms of geography prove particularly suitable for combining linguistic analysis, pragmatic aspects, and intercultural reflections since they are tightly linked to the morphology of the reference territories and the peoples who have inhabited them over time, thus producing culturally determined communicative contexts.

MWEs and constructions in language acquisition and in non-standard language (e.g. tweets, forums, spontaneous speech)-Evaluation of annotation and processing techniques for MWEs and constructions-Retrospective comparative analyses from the PARSEME shared tasks on automatic identification of MWEs We received 37 submissions (21 long and 16 short papers). We selected 12 long papers and 8 short ones. From those, 6 papers were presented orally and the remaining 14 as posters. The overall acceptance rate was 54%. Of the 20 presented papers, 6 concerned both wordnets and MWEs, which makes us believe that the intended synergy effect has been achieved. In addition to the oral and poster sessions, the workshop featured an invited talk, given by Aline Villavicencio. We are grateful to the paper authors for their valuable contributions, the members of the Program Committee for their thorough and timely reviews, all members of the organizing committee for the fruitful collaboration, and to all the workshop participants for their interest in this event. Our thanks also go to the ACL 2019 organizers for their support, as well as to SIGLEX and GWA for their endorsement.

As a linguistic phenomenon, collocations have been the subject of numerous researches both in the fields of theoretical and descriptive linguistics, and, more recently, in automatic Natural Language Processing. In the area of Machine there is still improvements to be done, as major translation engines do not handle collocations in the appropriate way and end up producing literal unsatisfactory translations. Having as a starting point our previous work on machine translation error analysis (Costa et al., 2015), in this article we present a corpus annotated with collocation errors and their classification. To our believe, to have a clear understanding of the difficulties that the collocations represent to the Machine Translations engines, it is necessary a detailed linguistic analysis of their errors.

In this paper, we investigate the impact of context for the paraphrase ranking task, comparing and quantifying results for multi-word expressions and single words. We focus on systematic integration of existing paraphrase resources to produce paraphrase candidates and later ask human annotators to judge paraphrasability in context. We first conduct a paraphrase-scoring annotation task with and without context for targets that are i) single-and multi-word expressions ii) verbs and nouns. We quantify how differently annotators score paraphrases when context information is provided. Furthermore, we report on experiments with automatic paraphrase ranking. If we regard the problem as a binary classification task, we obtain an F1-score of 81.56% and 79.87% for multi-word expressions and single words resp. using kNN classifier. Approaching the problem as a learning-to-rank task, we attain MAP scores up to 87.14% and 91.58% for multiword expressions and single words resp. using LambdaMART, thus yielding highquality contextualized paraphrased selection. Further, we provide the first dataset with paraphrase judgments for multi-word targets in context.

Although MWE are relatively morphologically and syntactically fixed expressions, several types of flexibility can be observed in MWE, verbal MWE in particular. Identifying the degree of morphological and syntactic flexibility of MWE is very important for many Lexicographic and NLP tasks. Adding MWE variants/tokens to a dictionary resource requires characterizing the flexibility among other morphosyntactic features. Carrying out the task manually faces several challenges since it is a very laborious task time and effort wise, as well as it will suffer from coverage limitation. The problem is exacerbated in rich morphological languages where the average word in Arabic could have 12 possible inflection forms. Accordingly, in this paper we introduce a semi-automatic Arabic multiwords expressions resource (SAMER). We propose an automated method that identifies the morphological and syntactic flexibility of Arabic Verbal Multiword Expressions (AVMWE). All observed morphological variants a...

Corpus-based Translation Studies have promoted research on the features of translated language, by focusing on the process and product of translation, from a descriptive perspective. Some of these features have been proposed by Toury (1995/2012) under the term of laws of translation, namely the law of growing standardisation and the law of interference. The law of standardisation appears to be particularly at play in diatopy, and more specifically in the case of transnational languages (e.g. English, Spanish, French, German). In fact, some studies have revealed the tendency to standardise the diatopic varieties of Spanish in translated language (Corpas Pastor, 2015a, 2015b, 2017, 2018). This paper focuses on verb + noun (object) collocations of Spanish translations of The Portrait of Dorian Gray by Oscar Wilde. Two different varieties have been chosen (Peninsular and Colombian Spanish). Our main aim is to establish whether the Colombian Spanish translation actually matches the variety spoken in Colombia or it is closer to general or standard Spanish. For this purpose, the techniques used to translate this type of collocations in both Spanish translations will be analysed. Furthermore, the diatopic distribution of these collocations will be studied by means of large corpora.

This paper describes the first phase of a project whose ultimate goal is the implementation of a practical tool to support the work of language learners and translators by automatically identifying multiword expressions (MWEs) and retrieving their translations for any pair of languages. The task of translating multiword expressions is viewed as a two-stage process. The first stage is the extraction of MWEs in each of the languages; the second stage is a matching procedure for the extracted MWEs in each language which proposes the translation equivalents. This project pursues the development of a knowledge-poor approach for any pair of languages which does not depend on translation resources such as dictionaries, translation memories or parallel corpora which can be time consuming to develop or difficult to acquire, being expensive or proprietary. In line with this philosophy, the methodology developed does not rely on any dictionaries or parallel corpora, nor does it use any (biling...

Due to the limited availability of parallel data in many languages, we propose a methodology that benefits from comparable corpora to find translation equivalents for collocations (as a specific type of difficult-to-translate multi-word expressions). Finding translations is known to be more difficult for collocations than for words. We propose a method based on bilingual context extraction and build a word (distributional) representation model drawing on these bilingual contexts (bilingual English-Spanish contexts in our case). We show that the bilingual context construction is effective for the task of translation equivalent learning and that our method outperforms a simplified distributional similarity baseline in finding translation equivalents.

In recent years, language models (LMs) have become almost synonymous with NLP. Pre-trained to "read" a large text corpus, such models are useful as both a representation layer as well as a source of world knowledge. But how well do they represent MWEs? This talk will discuss various problems in representing MWEs, and the extent to which LMs address them: • Do LMs capture the implicit relationship between constituents in compositional MWEs (from baby oil through parsley cake to cheeseburger stabbing)? • Do LMs recognize when words are used nonliterally in non-compositional MWEs (e.g. do they know whether there are fleas in the flea market)? • Do LMs know idioms, and can they infer the meaning of new idioms from the context as humans often do? Vered Shwartz is a postdoctoral researcher at the Allen Institute for AI (AI2) and the University of Washington. She will join the Department of Computer Science at the University of British Columbia as an Assistant Professor in fall 2021. Her research interests include computational semantics and pragmatics, multiword expressions, and commonsense reasoning.

Tourism discourse as a domain-specific discourse is characterized by a set of linguistic, pragmatic, and function features that make it different from other discourses and the general language. One of its essential elements is the usage of appealing, innovative, exotic-sounding words in order to attract potential tourists by "persuading, luring, wooing and seducing" [6]. In this context, formulaic language plays a key role. To date, research into chunks of language used in tourism have mostly focused on collocations [1, 8, 23], with a few works on longer sequences [11, 12, 13]. Bearing this in mind, this paper aims to contribute to the analysis of 4-word bundles in this domain, more specifically, in the segment of adventure tourism. To do so, a corpus-driven analysis was undertaken. As for our methodology, a specialized corpus containing English promotional texts was compiled. After that, the software Sketch Engine was used to extract a list of potential 4-word bundles. Next, manual verification was performed to ensure the validity of the units. Finally, the resulting list was classified according to their structural framework and their function in the text. The findings show that, in terms of the structure, the most typical sequences were verbal bundles; on the other hand, in terms of the function, a significant amount of the units was mainly used to address readers directly.

Although MWE are relatively morphologically and syntactically fixed expressions, several types of flexibility can be observed in MWE, verbal MWE in particular. Identifying the degree of morphological and syntactic flexibility of MWE is very important for many Lexicographic and NLP tasks. Adding MWE variants/tokens to a dictionary resource requires characterizing the flexibility among other morphosyntactic features. Carrying out the task manually faces several challenges since it is a very laborious task time and effort wise, as well as it will suffer from coverage limitation. The problem is exacerbated in rich morphological languages where the average word in Arabic could have 12 possible inflection forms. Accordingly, in this paper we introduce a semi-automatic Arabic multiwords expressions resource (SAMER). We propose an automated method that identifies the morphological and syntactic flexibility of Arabic Verbal Multiword Expressions (AVMWE). All observed morphological variants a...

Automated Compliance Checking (ACC) systems aim to semantically parse building regulations to a set of rules. However, semantic parsing is known to be hard and requires large amounts of training data. The complexity of creating such training data has led to research that focuses on small sub-tasks, such as shallow parsing or the extraction of a limited subset of rules. This study introduces a shallow parsing task for which training data is relatively cheap to create, with the aim of learning a lexicon for ACC. We annotate a small domain-specific dataset of 200 sentences, SPAR.txt 1 , and train a sequence tagger that achieves 79,93 F1-score on the test set. We then show through manual evaluation that the model identifies most (89,84%) defined terms in a set of building regulation documents, and that both contiguous and discontiguous Multi-Word Expressions (MWE) are discovered with reasonable accuracy (70,3%).

This paper analyses the support (or light) verb constructions (SVC) in a publicly available, manually annotated corpus of multiword expressions (MWE) in Brazilian Portuguese. The paper highlights several issues in the linguistic definitions therein adopted for these types of MWE, and reports the results from applying STRING, a rule-based parsing system, originally developed for European Portuguese, to this corpus from Brazilian Portuguese. The goal is twofold: to improve the linguistic definition of SVC in the annotation task, as well as to gauge the major difficulties found when transposing linguistic resources between these two varieties of the same language.

The paper reports on a study aimed at defining the limits between fixed expressions and Support Verb Constructions. To this end, a set of formal criteria that are applicable for the efficient classification of verbal MWEs were checked consistently against data in Greek and French. Ultimately, the delineation of the two types of constructions and their intermediate class is crucial not only for linguistic and lexicographic purposes, but also for Natural Language Processing tasks.

The study explores the implementation of corpora in teaching of the four courses of Croatian as a foreign language. Croatian, as a less-resourced and less commonly taught language, shares the same issues as other less resourced languages: studies about using corpus-are still missing. Student satisfaction and class engagement as well as teachers' course delivery methods (indirect and data-driven approach) were analysed in four groups of participants: two groups of beginner level of the foreign language learning (A1) and two groups of intermediate level (B1, B2) The experiment was conceived in three phases. In phase one, four teachers of Croatian as a Foreign Language (CFL) held an experimental lesson in their course groups. In phase two, the students were asked to complete a survey to examine degree of satisfaction with corpus-based approach in teaching CFL. In phase three, in-depth interviews were conducted with all teachers to examine their perceptions about the role of corpora in teaching, as well as their expectations regarding corpus-assisted language teaching. Based on the analysis, it was concluded that degree of students' satisfaction with experimental classes prepared using corpus was mostly positive. Experiment revealed that approach to language teaching depends largely on perception of teachers about integration and use of corpora in teaching. Corpora can be used directly in class of absolute beginners.

In this paper we analyze the problem of recognition of temporal expressions and their translations. Starting from named entities describing time (e.g. of the TIMEX type) and using FST cascades, temporal expressions were extracted from a text in French, followed by an analysis of (human) translations of these expressions in several languages, by means of an aligned corpus. The goal of this paper is to develop a formal description of such expressions, which enables the extraction of equivalent temporal expressions from aligned texts

Fixedness in language has been extensively studied in areas like multi-word units, idiomatic expressions, collocations and verb-particle constructions. These have often been treated as relatively fixed non-compositional sequences, which allow for little variation. In our paper we will focus on co-occurrence phenomena between elements in syntactic relations. Specifically, we focus on subject-verb and verb-object relations in active and passive constructions. Looking for fixedness in these syntactic relations where compositionality is expected to hold to a large degree may strike the reader as a strange undertaking. Our main interest lies in establishing how far an open choice principle holds for these relations and to what degree we can find fixedness in these syntactic relations. The identification of syntactic relations requires syntactically annotated corpora. Most standard corpora of sufficient size are either not annotated at all, or annotated at the non-hierarchical level of part-of-speech tags only. They typically contain no hierarchical information about the syntactic organisation of sentences. Parsing approaches to fixedness are still quite rare. Exceptions are Lin (1998) and Seretan and Wehrli (2006). Robust broad-coverage syntactic parsers, for example Schneider (2007) or Andersen (2008), have now become available, offering new perspectives for this research. This paper describes the syntactic annotation of over 160 million running words with the help of Pro3Gres, a dependency parser. See Schneider (2007) for a more detailed description. We document the extraction of a database with verb centres and their dependents. We then explore the possibilities and limitations of this dependency database for the study of fixedness in syntactic relations.

This synopsis presents the application of computational linguistic tools and approaches which were developed by the author for Descriptive Linguistics, Text Mining, and Psycholinguistics. It also describes how the computational linguistic tools, which are originally based on linguistic insights and assumptions, lead to new and detailed linguistic insights if applied to different research areas, and can in turn again improve the computational tools. The computational tools are based on models of language, predicting part-of-speech tags or syntactic attachment. These models, which were originally designed for the practical purpose of solving a computational linguistics task, can increasingly be used as models of human language processing. A large-scale syntactic parser is the core linguistic tool that I am going to use. I further also employ its preprocessing tools, part-of-speech taggers and chunkers, and approaches learning from the data, so-called data- driven approaches. The use o...

Fixedness in language has been extensively studied in areas like multi-word units, idiomatic expressions, collocations and verb-particle constructions. These have often been treated as relatively fixed non-compositional sequences, which allow for little variation. In our paper we will focus on co-occurrence phenomena between elements in syntactic relations. Specifically, we focus on subject-verb and verb-object relations in active and passive constructions. Looking for fixedness in these syntactic relations where compositionality is expected to hold to a large degree may strike the reader as a strange undertaking. Our main interest lies in establishing how far an open choice principle holds for these relations and to what degree we can find fixedness in these syntactic relations. The identification of syntactic relations requires syntactically annotated corpora. Most standard corpora of sufficient size are either not annotated at all, or annotated at the non-hierarchical level of part-of-speech tags only. They typically contain no hierarchical information about the syntactic organisation of sentences. Parsing approaches to fixedness are still quite rare. Exceptions are Lin (1998) and Seretan and Wehrli (2006). Robust broad-coverage syntactic parsers, for example Schneider (2007) or Andersen (2008), have now become available, offering new perspectives for this research. This paper describes the syntactic annotation of over 160 million running words with the help of Pro3Gres, a dependency parser. See Schneider (2007) for a more detailed description. We document the extraction of a database with verb centres and their dependents. We then explore the possibilities and limitations of this dependency database for the study of fixedness in syntactic relations.

In this paper we present a rule-based method for multi-word term extraction that relies on extensive lexical resources in the form of electronic dictionaries and finite-state transducers for modelling various syntactic structures of multi-word terms. The same technology is used for lemmatization of extracted multi-word terms, which is unavoidable for highly inflected languages in order to pass extracted data to evaluators and subsequently to terminological e-dictionaries and databases. The approach is illustrated on a corpus of Serbian texts from the mining domain containing more than 600,000 simple word forms. Extracted and lemmatized multi-word terms are filtered in order to reject falsely offered lemmas and then ranked by introducing measures that combine linguistic and statistical information (C-Value, T-Score, LLR, and Keyness). Mean average precision for retrieval of MWU forms ranges from 0.789 to 0.804, while mean average precision of lemma production ranges from 0.956 to 0.9...

This article describes an automatic system for writing specialized texts in Spanish. The arText prototype is a free online text editor that includes different types of linguistic information. It is designed for a variety of end users and domains, including specialists and university students working in the fields of medicine and tourism, and laypersons writing to the public administration. ArText provides guidance on how to structure a text, prompts users to include all necessary contents in each section, and detects lexical and discourse problems in the text.

For NLP applications that require some sort of semantic interpretation it would be helpful to know what expressions exhibit an idiomatic meaning and what expressions exhibit a literal meaning. We investigate whether automatic word-alignment in existing parallel corpora facilitates the classification of candidate expressions along a continuum ranging from literal and transparent expressions to idiomatic and opaque expressions. Our method relies on two criteria: (i) meaning predictability that is measured as semantic entropy and (ii), the overlap between the meaning of an expression and the meaning of its component words. We approximate the mentioned overlap as the proportion of default alignments. We obtain a significant improvement over the baseline with both measures.

The paper reports on a study aimed at defining the limits between fixed expressions and Support Verb Constructions. To this end, a set of formal criteria that are applicable for the efficient classification of verbal MWEs were checked consistently against data in Greek and French. Ultimately, the delineation of the two types of constructions and their intermediate class is crucial not only for linguistic and lexicographic purposes, but also for Natural Language Processing tasks.

This work presents an initial investigation on how to distinguish collocations from free combinations. The assumption is that, while free combinations can be literally translated, the overall meaning of collocations is different from the sum of the translation of its parts. Based on that, we verify whether a machine translation system can help us perform such distinction. Results show that it improves the precision compared with standard methods of collocation identification through statistical association measures.

This paper discusses the treatment of fixed word expressions developed for our ITS-2 French-English translation system. This treatment makes a clear distinction between compounds-i.e. multiword expressions of X°-level in which the chunks are adjacent-and idiomatic phrases-i.e. multiword expressions of phrasal categories, where the chunks are not necessarily adjacent. In our system, compounds are handled during the lexical analysis, while idioms are treated in the syntax, where they are treated as "specialized lexemes". Once recognized, an idiom can be transfered according to the specifications of the bilingual dictionary. We will show several cases of transfer to corresponding idioms in the target language, or to simple lexemes. The complete system, including several hundreds of compounds and idioms can be consulted on the Internet (ht tp ://latl.unige.ch/itsweb.html).

Identifying of compound nouns is important for a wide spectrum of applications in the field of natural language processing such as machine translation and information retrieval. Extraction of compound nouns requires deep or shallow syntactic preprocessing tools and large corpora. This paper investigates several methods for extracting Noun compounds from Malay text corpora. First, we present the empirical results of sixteen statistical association measures of Malay <N+N> compound nouns extraction. Second, we introduce the possibility of integrating multiple association measures. Third, this work also provides a standard dataset intended to provide a common platform for evaluating research on the identification compound Nouns in Malay language. The standard data set contains 7,235 unique N-N candidates, 2,970 of them are N-N compound nouns collocations. The extraction algorithms are evaluated against this reference data set. The experimental results  demonstrate that a group of ...

We present and make available a large automatically-acquired all-words list of English multiword expressions scored for compositionality. Intrinsic evaluation against manually-produced gold standards demonstrates that our compositionality estimates are sound, and extrinsic evaluation via incorporation of our list into a machine translation system to better handle idiomatic expressions results in a statistically significant improvement to the system’s BLEU scores. As the method used to produce the list is language-independent, we also make available lists in seven other European languages.

In this paper, the authors address the significance and complexity of tokenization, the beginning step of NLP. Notions of word and token are discussed and defined from the viewpoints of lexicography and pragmatic implementation, respectively. Automatic segmentation of Chinese words is presented as an illustration of tokenization. Practical approaches to identification of compound tokens in English, such as idioms, phrasal verbs and fixed expressions, are developed.