Multiword Expressions

Il lavoro espone i risultati di due indagini contrastive italiano-inglese e italiano-francese riguardanti le espressioni idiomatiche (EI) gestuo-cinesiche, cioè caratterizzate dal fare riferimento a gesti e ad altri comportamenti cinesici quali movimenti di parti del corpo, espressioni, posture (girarsi i pollici, storcere il naso, mettersi in ginocchio). Scopo dell'indagine era anzitutto verificare in quale misura le EI gestuo-cinesiche italiane abbiano un equivalente in inglese e in francese. Emerge che ciò si verifica per il 50% delle EI nel caso del francese e per il 61% di esse nel caso dell'inglese. Il maggior numero di equivalenze si ha nelle EI che fanno riferimento a comportamenti cinesici ritenuti universali, mentre le specificità culturali per lo più ostacolano l'esistenza di corrispettivi nelle diverse lingue. La prospettiva del confronto interlinguistico consente di far emergere appieno, e meglio comprendere, l'intreccio di elementi biologici, cognitivi e culturali del quale le EI gestuo-cinesiche sono un esempio paradigmatico.

This paper describes an algorithm for automatically extracting multiword expressions (MWEs) from a corpus. The algorithm is nodebased, i.e. extracts MWEs that contain the item specified by the user, using a fixed window-size around the node. The main idea is to detect the frequency anomalies that occur at the starting and ending points of an ngram that constitutes a MWE. This is achieved by locally comparing matrices of observed frequencies to matrices of expected frequencies, and determining, for each individual input, one or more sub-sequences that have the highest probability of being a MWE. Top-performing sub-sequences are then combined in a scoreaggregation and ranking stage, thus producing a single list of score-ranked MWE candidates, without having to indiscriminately generate all possible sub-sequences of the input strings. The knowledge-poor and computationally efficient algorithm attempts to solve certain recurring problems in MWE extraction, such as the inability to deal with MWEs of arbitrary length, the repetitive counting of nested ngrams, and excessive sensitivity to frequency. Evaluation results show that the best-performing version generates top-50 precision values between 0.71 and 0.88 on Turkish and English data, and performs better than the baseline method even at n=1000.

In this paper, we will focus on the structural, cognitive, and cross-cultural properties of proverbs. The first part examines the structural characteristics of proverbial sentences, while the second part explores the cognitive framework that underlies their structure. The data includes 300 proverbs from three different languages (English, Croatian and Italian). This kind of analysis stands at the crossroads between cognitive and structural linguistics. The cognitive aspect includes two key theories: conceptual metaphor theory and relevance theory. Our primary assumption is that the basic features of proverbs-phonetic, syntactic, and semanticarise from an adjustment of cognitive patterns to the demands of language use. The high frequency of many communicative situations, coupled with the natural desire for brevity (cognitive economy), influences the structure of proverbs. A significant portion of the proverbs in our dataset (62%) rely on metaphors. This analysis leads to an underlying motivation for the figurative meanings conveyed by proverbs. By mapping concrete experiences (source domains) onto abstract concepts (target domains), metaphors enhance our understanding of abstract ideas and universal truths. While current research sheds light on several key aspects, it also highlights important conceptual and linguistic gaps-particularly regarding the effects of associative word networks and the roles of binarism and symmetry in the memorization and retrieval of proverbs. Our study also emphasizes the structural distinctions between literal and metaphorical proverbs. To conclude, our data analysis confirms that the syntactic, semantic, and phonological markers in proverbs are interdependent. There is a high degree of correlation between these features, leading to a potentially universal pattern utilized across different languages. Furthermore, the principles of memorability and cognitive economy are confirmed to play pivotal roles in the structure of proverbs. This study aims to address theoretical challenges and deepen understanding of the universal cognitive features inherent in proverbs, encouraging further exploration of conceptual power of proverbs across diverse cultures. Danica Skara is retired professor emeritus at the University of Split, Croatia. She recieved her MA from the University of Zagreb; her PhD from the University of Zadar, Croatia. Her area of specialization includes linguistics and related disciplines: semantics, cognitive linguistics, psycholinguistics, paremiology. Apart from a variety of articles (43) on the subjects, she has published 5 books and participated in the work of 45 conferences. She was the University's Vice-Chancellor (2003. In 2000 she founded a doctoral programme in linguistics at the University of Zadar and acted as the director of the programme. She was a Head of postgraduate programme: European studies. In 2003 she was awarded a Fulbrigt Fellowship at the Cornell University (USA). She acted as a plenary speaker/coorganizer of several international conferences; she was a visiting scholar at

This comparative study examines the impact of WhatsApp-based instruction versus text-based instruction on intentional media-related vocabulary learning and learners' perceptions. Specifically, it investigates the effects of WhatsApp functionalities (e.g. multimodal input, media sharing and group chat) on deliberate learning, word retention and collaboration among final-year undergraduate students. Within a mixedmethods framework, a convenient sample of 70 students from a language institute in Tunisia was assigned to experimental group (EG) and control group (CG). While vocabulary achievement was assessed through a pretest-posttest design, qualitative data was gleaned through a semi-structured interview run with 10 informants from the Mobile Assisted Vocabulary Learning (MAVL) group. The eight-session MAVL intervention included the intentional instruction of commonly used idioms in media context through the different affordances of WhatsApp. Findings from paired t-tests and independent t-tests, processed with SPSS, revealed higher lexical performance among the WhatsApp group. Drawing on reflexive thematic analysis (RTA), qualitative data displayed the participants' favorable attitudes toward the app's potential to improve word recall and cultivate a spirit of collaboration. The theoretical and pedagogical implications of these findings offer clarity for educators in their ongoing search for cutting-edge vocabulary teaching approaches. This research demonstrates that using the in-built functionalities of WhatsApp (e.g., multiple input types, chat groups) in intentional vocabulary learning significantly enhances students' word knowledge and word retention and increases their collaboration. The findings provide valuable guidance for EFL teachers, curriculum developers, and policymakers on how to effectively integrate MAVL into the EFL classroom. The study's results are important because they show a practical and engaging way to improve student learning vocabulary which is often an underexposed skill in the EFL classroom. By harnessing modern technology, educators can help more students develop not only lexical knowledge but also 21st century skills such as collaoration.

This paper is aimed at describing the main differences between spoken and written English. More specifically, attention is paid to the different examples which are classified as predicative Prepositional Phrases (PPs) in the International Corpus of English-­Great Britain (ICE-­GB) and their frequency in spoken and written texts. These units can be defined as those phrases which are introduced by a preposition and
followed by a Noun Phrase (NP) acting as its complement. Furthermore, they perform the function of Subject Complement (Cs) at clause level. Such is the case of “She first fell in love with Will when she was eighteen, and she adores him still” (ICE­‐GB:W2F­‐019#47:1). Although in terms of frequency this is not the syntactic function PPs more often perform, they are taken into account because of their complexity and
due to the lack of detailed analyses. In most cases they are described as isolated examples and this phenomenon is not considered to be a very productive one. After introducing some basic notions, these structures are analyzed focusing on their presence in both spoken and oral texts within the ICE-GB. This is a one-million-word corpus which is both morphologically tagged and syntactically parsed. Moreover, it was compiled in the nineties and consists of both spoken (60%) and written material (40%).
The ICECUP (ICE Corpus Utility Program) software retrieved 3307 examples from 3223 sentences. These instances were then filtered since some of them were later classified as “noise” (in some cases the PPs were performing other functions either at phrase or at clause level and in others the element acting as the complement of the preposition was not a NP). For these reasons the final subcorpus consists of 1332 examples.
67.49% of these instances (899) are found in oral texts whereas 32.51% of them (433) belong to written texts. All these examples have been classified into different groups and subgroups corresponding to the different text categories available in this corpus (Nelson, Wallis and Aarts, 2002: 307-8). The results are presented in charts by means of both figures and percentages and different conclusions are later drawn based on the analysis of these charts.
Thus, for example, it can be noticed that, although it was expected that the amount of structures under study would be higher in spoken than in written texts because of the structure of the corpus itself, the relative frequency (which takes into account the relationship between the number of examples and the number of words) proves so, too: 0.1410% in spoken texts as opposed to 0.1022% in written texts, with an average of 0.1255% in the whole corpus. Moreover, there are more examples in dialogues (581) than in monologues (318) and in printed texts (332) than in non-printed ones (101).
This information proves especially relevant for non-native speakers of English since it allows them to become aware of the differences between speaking and writing. According to the evidence, some units are used more often in spoken language than in written English. Therefore, when producing any kind of text, students will feel more confident for they will be able to choose the appropriate structures bearing in mind these issues.

Particle Verbs (PVs) are a very frequent and productive word class in German. They can occur in different syntactic paradigms. In verb-first and verb-second clauses which do not contain auxiliary verbs they occur syntactically separated. In other cases the PV is written together as one word, but it still may occur morphologically separated, e.g. by the infinitive marker zu. Especially the syntactically separated paradigm may be problematic for NLP tasks, since parsers may not correctly identify the syntactic dependency between the verb and the particle. This has also consequences for lemmatization. In German corpora we observed that there is a notable variance between individual PVs to occur in different paradigms. For example the verb aus|sehen occurs syntactically separated in 58 % of the cases, while the verb an|sehen only occurs separated in 20 % of the cases. Our research interest lies in finding out which factors affect the preferences of different PVs for different syntactic ...

Noun-noun compounds are complex words with two simplex nouns as constituents. In English and German, the first constituent represents the modifier of the compound, and the second constituent represents the head. A compound may have different degrees of compositionality regardings its constituents. For example, the German compound Lederhose ('leather trousers') is highly compositional, because the meaning of the compound can be obtained by combining the meanings of the constituent words. The compound Sündenbock ('sin buck', meaning scapegoat), in contrast, is rather non-compositional, because there is no obvious synchronic relation between the meaning of the compound and the meanings of the constituents. Understanding the principles of compositionality is important to understand the processing of compounds in human mind, for their semantic interpretation, and for their translation to other languages. – A common NLP approach to compute the degree of compositionality ar...

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.

This study explored how second language (L2) speakers' use of multiword sequences in speech predicted perceived fluency ratings while controlling for their utterance fluency. A total of 102 Japanese speakers of English delivered an argumentative speech, which was analyzed for bigram and trigram measures (frequency, proportion, and mutual information) and utterance fluency measures capturing three subdimensions: speed, breakdown, and repair fluency (Tavakoli & Skehan, 2005). Perceived fluency was assessed by 10 experienced L2 raters. Mixedeffects regression analyses revealed that after establishing the parsimonious model solely by UF predictors (marginal R 2 = .61), a frequency-based n-gram predictor-bigram proportion-slightly but significantly explained the remaining variance of fluency rating scores (0.8%). The results indicated that multiword sequences in speech had a small but systematic impact on perceived fluency, even controlling for the effects of utterance fluency. This finding contributes to the discussion concerning the role of multiword sequences in fluent speech production.

The PARSEME Shared Task on automatic identification of verbal multiword expressions aims at identifying such expressions in running texts. Typology of verbal multiword expressions, very detailed annotation guidelines and gold-standard data for as many languages as possible will be provided. Since the Prague Dependency Treebank includes Czech multiword expression annotation, it was natural to make an attempt to automatically convert the data into the Shared Task format. However, since the Czech treebank predates the Shared Task annotation guidelines, a prior examination was necessary to determine to which extent the conversion can be fully automatic and how much manual work remains. In this paper, we show that information contained in the Prague Dependency Treebank is sufficient to extract all of the Shared Task categories of verbal multiword expressions relevant for Czech, even if these categories are originally annotated differently; nevertheless, some manual checking and annotatio...

Although the interest of literature in word combinations has significantly increased over the last decades, the full classification of their types and comprehensive collection of their forms is far from complete and flawless. This paper presents the results of a research project devoted precisely to the collection and analysis of Italian combinatorial lexicon. The investigation has both a theoretical and an empirical import. From the former point of view, it takes into account multiword units as specific forms of lexical entry. On the basis of the most recent achievements in lexicology, the notion of 'word format' is introduced and the vast variety of the combination formats is considered and classified. Various specific case studies of combinations are discussed. From the empirical point of view, the CombiNet dictionary is presented, an on-line lexicographic tool recording a sample of Italian multiword entries implementing an ample variety of entry formats. In addition to describing the lexicographic layout in detail, the paper illustrates the advanced research tool CombiNet is equipped with, as a device for both lexicographic and lexicological investigation.

In support verb constructions (SVC), as 'have poise', the support verb is explicitly assumed to be a verb, here 'have'. However, during the last 50 years, the notion of SVC has been extended to a large range of new cases. With this new scope, the linguistic form that plays the role of the support verb can also be an expression not classified as a verb, such as 'be mad with', or an adjective, e.g. 있다 issta 'there is' in Korean. Calling 'support verb' a non-verb is a terminological issue. Beyond that, this article examines the validity of extending the definition of SVCs. After surveying the linguistic forms at stake and the reactions to such extension, we focus on delimiting the historical, stricto sensu SVCs based on criteria, in a way that matches the traditional intuition. Then, for each type of linguistic form proposed as an extension, we examine how far the extension distorts this notion. We find both SVCs and extensions of SVC to be of interest in a model of syntax and the lexicon. Finally, the 'verb' part of speech is not essential to any of the categories, which suggests adopting the terms of 'support construction' and 'extension of support construction' instead.

This paper aims to analyze the structure and meanings of the Latin indefinite pronouns
that can be traced back to the Indo-European root *kw-e-/*kw-i. All of them
are morphologically derivational forms: this property is supported by typological
studies. We investigate the formal types of Latin indefinite pronouns in order to
give an exact description of them, mostly by considering the Universal Asymmetries
Markedness relationship of Haspelmath (1997). The remarks are used to give
a new proposal for a vexata quaestio of Indo-European and typological studies: the
derivational relationship between indefinite and interrogative pronouns.

This study focuses on investigating semantic errors in English to Indonesian translation using DeepL Translate, with the aim of evaluating the extent of semantic accuracy of this translation tool. This study uses a qualitative approach with a case study design, where data is collected by observing translations during a conversation between a native English speaker and a native Indonesian speaker. Each translation was analysed using a qualitative descriptive method to identify semantic errors, which were classified into three categories: inappropriate word choice, loss of implicit meaning, and ambiguity of sentence structure. The results showed that out of 50 translated sentences, there were 15 semantic errors, with inappropriate word choice being the dominant category. The conclusion of this study is that while DeepL is capable of producing relatively good translations, its limitations in understanding semantic context remain a significant bottleneck. The study recommends further development of the automatic translation algorithm and training of users to use the technology critically and judiciously.

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.

The objective of this study is to investigate how learners of Italian as a second or foreign language search for new meanings in online Italian dictionaries. Using eye-tracking technology, we carried out experiments inviting users to do exercises on ‘combinations of words’ while they consulted various dictionaries, including De Mauro – Internazionale and Garzanti Linguistica. Results should suggest how to make digital dictionaries more efficient and thus more useful for learners of Italian as a second or foreign language.

UD_Greek-GUD (GUD) is the most recent Universal Dependencies (UD) treebank for Standard Modern Greek (SMG) and the first SMG UD treebank to annotate Verbal Multiword Expressions (VMWEs). GUD contains material from fiction texts and various sites that use colloquial SMG. We describe the special annotation decisions we implemented with GUD, the pipeline we developed to facilitate the active annotation of new material, and we report on the method we designed to evaluate the performance of models trained on GUD as regards VMWE identification tasks.

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.

The Szeged Corpus is the largest manually annotated database containing the possible morphological analyses and lemmas for each word form. In this work, we present its latest version, Szeged Corpus 2.5, in which the new harmonized morphological coding system of Hungarian has been employed and, on the other hand, the majority of misspelled words have been corrected and tagged with the proper morphological code. New morphological codes are introduced for participles, causative / modal / frequentative verbs, adverbial pronouns and punctuation marks, moreover, the distinction between common and proper nouns is eliminated. We also report some statistical data on the frequency of the new morphological codes. The new version of the corpus made it possible to train magyarlanc, a data-driven POS-tagger of Hungarian on a dataset with the new harmonized codes. According to the results, magyarlanc is able to achieve a state-of-the-art accuracy score on the 2.5 version as well.

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.

for my studies, since I could not pursue empirical research without their help. I am grateful to all of my committee members, Jill Morford, Joan Bybee, Bill Croft, and Andy Wedel, for their mentorship. My research is better due to the challenges they posed, and the assistance they gave me. Special thanks are due to my chair, Jill Morford, for her probing questions, feedback, and support throughout the research process. She was a reliable technical resource, voice of reason, and source of encouragement, and could always help me get half-baked ideas to be more fully-baked. I am also especially grateful to Joan Bybee, whose pivotal work has been indispensible to my psycholinguistic research. I thank her for immersing me in usage-based linguistics, first as a student, then as a research assistant, and finally as a collaborator. In addition to my committee members, I'm ever grateful to other UNM faculty, past and present, in linguistics and adjoining fields, including

Аннотация: Данная статья посвящена анализу результатов zero-shot межъязыкового переноса автоматической лингвистической разметки в стандарте CoBaLD с русского языка на близкородственные и неродственные языки. Исследование показывает, что качество переноса зависит как от генетического родства языков, так и от количества обучающих данных для языков в используемой языковой модели. Ключевые слова: автоматическая семантическая разметка, лингвистическая разметка, межъязыковой перенос Для цитирования: Ивойлова А.М. О влиянии генетического родства языков на качество межъязыкового переноса автоматической семантической разметки // Типология морфосинтаксических параметров. 2024. Том 7, вып. 2. С. 43-59. (На английском.

Korpusgesteuerte und korpusbasierte Untersuchungen führen überzeugend vor Augen, dass die Sprache zu einem viel stärkeren Grad aus konventionalisierten Mehrworteinheiten besteht als früher angenommen wurde. Daraus ergibt sich das Postulat, ihnen einen angemessenen Stellenwert im Fremdsprachenunterricht einzuräumen. Dieser Aufgabe versucht das von der EU kofinanzierte Projekt PhraseoLab gerecht zu werden, dessen Ziel es ist, eine digitale Lernressource zum Erlernen deutscher Phraseologie im Rahmen eines plurilingualen Ansatzes zu erarbeiten. Im Projekt werden drei Gruppen fester Mehrworteinheiten didaktisch aufbereitet: Kollokationen, Routineformeln und Idiome, der Schwerpunkt des vorliegenden Beitrags liegt auf Idiomen. Da der Status unterschiedlicher Gruppen von Mehrworteinheiten in der Phraseodidaktik umstritten ist, wird zuerst die Rolle der Idiome im DaF-Unterricht diskutiert. Anschließend werden die Ergebnisse einer Frequenzuntersuchung von Idiomen in der geschriebenen (DWDS) und gesprochenen Sprache (DWDS, DGD-Korpora) dargestellt, deren Ziel darin besteht, eine empirisch fundierte Selektion erwerbsrelevanter Idiome vorzulegen.

There has been a consensus among language researchers regarding the apparent advantages of learning lexical chunks. Conventional pedagogies (e.g., memorizing, drilling, input flooding, typographic enhancement) have been utilized in diverse research to determine an effective way of raising students’ awareness of and encoding chunks; however, these practices have produced mixed results. Most of these studies have positioned learners in a passive role and have focused on increasing breadth with retention of forms as the primary goal (Boers and Lindstromberg 2009). Furthermore, the students’ meaningful use of language and feedback from the teacher was rarely considered during the process. Considering how memory works, these conditions could explain the learner’s low retention rate and inconsistencies in the literature. The present exploratory, mixed-methods, classroom-based study veers away from the traditional ways of raising awareness of lexical chunks. It investigates the effects of a usage-driven feedback approach on lexical chunk use and uptake, which emphasizes the value of meaningful production and the importance of receiving feedback in the creation process. Four college students with different proficiency levels in an English-speaking course were investigated for ten weeks. Results from several quantitative and qualitative measures revealed that the function-focused production tasks accompanied by productive feedback from the facilitator and the students led to a significant increase in lexical chunk use.

In this paper, we look at the manual construction of a lexicon of emotion terms in Old English organised as a wordnet lexicon and based on a pre-existing dataset which categorises emotion terms on the basis of cognitive criteria. This is part of a wider initiative to construct a wordnet for Old English which includes both manually and (semi-)automatically constructed parts. We will describe our workflow which foresees the use of lemmas, senses and definitions taken from and in many cases adapted from a legacy Old English lexical resource, ClarkHall’s AngloSaxon Dictionary, and organised into synsets by matching the definitions of OE terms with the synset definitions found in the Open English WordNet. We argue that this is a useful way of disseminating research in lexical semantics and philology taking as an example the semantic field of GUILT. In our poster presentation we will give a visual description of the whole field; in the present submission we only focus on the most ‘core’ nominal synsets relating to GUILT.

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.

We present here the enhancement of the Romanian wordnet with a new type of information, very useful in language processing, namely types of verbal multi-word expressions. All verb literals made of two or more words are attached a label specific to the type of verbal multi-word expression they correspond to. These labels were created in the PARSEME Cost Action and were used in the version 1.1 of the shared task they organized. The results of this annotation are compared to those obtained in the annotation of a Romanian news corpus with the same labels. Given the alignment of the Romanian wordnet to the Princeton WordNet, this type of annotation can be further used for drawing comparisons between equivalent verbal literals in various languages, provided that such information is annotated in the wordnets of the respective languages and their wordnets are aligned to Princeton WordNet, and thus to the Romanian wordnet.

Рецензенти проф. д-р Йовка Тишева доц. д-р Атанас Атанасов СЪДЪРЖАНИЕ Използвани знаци и съкращения / 9 Въведение / 11 Глава I Българската лингвистика за предложните изрази / 15 84 Приема се, че е възможно съществуването и на хибридни типове (РГ II 1980: 16).

We present MWE-Finder, an application that enables a user to search for multiword expressions (MWEs) in large Dutch text corpora. Components of many MWEs in Dutch can occur in multiple forms, need not be adjacent, and can occur in multiple orders (such MWEs are called flexible). Searching for such flexible MWEs is difficult and cannot be done reliably with most search applications. What is needed is a search engine that takes into account the grammatical configuration of the MWE. MWE-Finder is therefore embedded in GrETEL, a treebank search application for Dutch. A user can enter an example of a MWE in a specific canonical form, after which the system searches for sentences in which the MWE occurs, using queries generated automatically from the canonical form. We will describe in detail how the queries for this MWE are derived from the canonical form. The MWE can also be selected from a list of approximately 10k canonical forms for Dutch MWEs that MWE-Finder offers. We will show that MWE-Finder also offers facilities to find examples with unexpected modifiers or determiners on components of the MWE, and that it will yield statistics on the arguments, modifiers and determiners that occur with the MWE and its components.

This paper introduces and demonstrates MWE-Finder, an application to search for flexible multiword expressions (MWEs) in Dutch text corpora, starting from an example. If the example is in canonical form, the application automatically generates three queries to search for sentences that contain an occurrence of the MWE and thus enables efficient analysis of its properties. The application offers canonical forms for more than 11k MWEs. Searching is done in treebanks, so the grammatical structure of the sentences is taken into account.

This paper proposes a canonical form for Multiword Expressions (MWEs), in particular for the Dutch language. The canonical form can be enriched with all kinds of annotations that can be used to describe the properties of the MWE and its components. It also introduces the DUCAME (DUtch CAnonical Multiword Expressions) lexical resource with more than 11k MWEs in canonical form. DUCAME is used in MWE-Finder to automatically generate queries for searching for flexible MWEs in large text corpora.

In this paper we showcase and evaluate MWE-Finder, a system that allows users to search for occurrences of an MWE in a large Dutch text corpus. To this end, we conduct three small case studies, and discuss the results in detail. We make use of the MWEs 0geen *+haan zal naar iets kraaien 'no one will say anything about something', iemand zal 0dat *+varken wassen 'someone will deal with that problem' and iemand zal iemand het hemd van het lijf vragen 'someone will want to know all the ins and outs of something from someone', which are all in canonical form following Odijk (2023) and Odijk and Kroon (2024). The results show that MWE-Finder is very accurate in retrieving the target MWEs, reaching an accuracy of 93.7%, and an F 1-score of 95.2%. The case studies additionally lay bare points of improvement of MWE-Finder, specifically concerning the enrichment of syntactic parses by making the object relation explicit in certain constructions. 'I thought Saab wanted to declare itself bankrupt yesterday.'

Over the last decade, the prominence of statistical NLP applications that use syntactic rather than only word-based shallow clues increased very significantly. This prominence triggered the creation of large scale treebanks, i.e., corpora annotated with syntactic structures. However, a look at the annotation schemata used across these treebanks raises some issues. Thus, it is often unclear how the set of syntactic relation labels has been obtained and how it can be organized so as to allow for different levels of granularity in the annotation. Furthermore, it appears questionable that despite the linguistic insight that syntax is very much language-specific, multilingual treebanks often draw upon the same schemata, with little consideration of the syntactic idiosyncrasies of the languages involved. Our objective is to detail the procedure for establishing an annotation schema for surfacesyntactic annotation of Spanish verbal relations and present a restricted set of easy-to-use criteria which facilitate the decision process of the annotators, but which can also accommodate for the elaboration of a more or a less fine-grained tagset. The procedure has been tested on a Spanish 3,500 sentence corpus, a fragment of the AnCora newspaper corpus. 1 "Surface-syntactic" is used here in the sense of the Meaning-Text Theory (Mel'čuk, 1988).

Genuine lexical writing assistants that attempt to detect lexical errors such as miscollocations are traditionally less common in Computer Assisted Language Learning than spell and grammar checkers. However, there is empirical evidence of the importance of capturing and correcting miscollocations in the writings of language learners, and therefore an increasing number of proposals deals with the detection of errors in collocations and the delivery of lists of correction suggestions. However, very few of these proposals take into account the varying ease with which learners can master different collocation and miscollocation types, or the fact that some collocation type errors might be more common than others, given that a writing assistant should be capable of handling at least the most common types of miscollocation. Furthermore, existing proposals explore collocation error-specific strategies, implicitly assuming that with one universal strategy all types of miscollocations can be detected and corrected. Our preliminary study, conducted on Spanish and French material, highlights one type of collocation in which learners err the most: support verb constructions (SVCs). To account for this, we explore a SVC-specific collocation error detection and correction strategy.

Genuine lexical writing assistants that attempt to detect lexical errors such as miscollocations are traditionally less common in Computer Assisted Language Learning than spell and grammar checkers. However, there is empirical evidence of the importance of capturing and correcting miscollocations in the writings of language learners, and therefore an increasing number of proposals deals with the detection of errors in collocations and the delivery of lists of correction suggestions. However, very few of these proposals take into account the varying ease with which learners can master different collocation and miscollocation types, or the fact that some collocation type errors might be more common than others, given that a writing assistant should be capable of handling at least the most common types of miscollocation. Furthermore, existing proposals explore collocation error-specific strategies, implicitly assuming that with one universal strategy all types of miscollocations can be detected and corrected. Our preliminary study, conducted on Spanish and French material, highlights one type of collocation in which learners err the most: support verb constructions (SVCs). To account for this, we explore a SVC-specific collocation error detection and correction strategy.

We report on UD_Greek-GUD (henceforth GUD), the most recent Universal Dependencies (UD) treebank of Standard Modern Greek (SMG). GUD adheres to UD.v2 (de Marneffe et al., 2021) and is the first SMG UD treebank to annotate Verbal Multiword Expressions (VMWE). We describe the language material and the special annotation features of GUD. We also report on ongoing work on training models on GUD and establishing an evaluation method for their predictions about VMWE.

Medieval French is known to be relatively hard to parse, with several possible sources of confusion for automatic parsers, among which its flexible word order and its graphical and syntactic variation, both synchronically and diachronically. In this work, we study in particular the influence of word order, by comparing the performances of two state-of-the-art syntactic parsers trained and evaluated on two treebanks: the Syntactic Reference Corpus of Medieval French (SRCMF), a treebank of Old French (9th—13th century) and the Google Stanford Dependency treebank of contemporary French.

We introduce a new method to tag Multiword Expressions (MWEs) using a linguistically interpretable language-independent deep learning architecture. We specifically target discontinuity, an under-explored aspect that poses a significant challenge to computational treatment of MWEs. Two neural architectures are explored: Graph Convolutional Network (GCN) and multi-head self-attention. GCN leverages dependency parse information, and self-attention attends to long-range relations. We finally propose a combined model that integrates complementary information from both, through a gating mechanism. The experiments on a standard multilingual dataset for verbal MWEs show that our model outperforms the baselines not only in the case of discontinuous MWEs but also in overall F-score. 1 * *The first two authors contributed equally.

Currently, the high volume of international information exchange involves a wide range of localities. As each locality comes with its own distinctive dialect, the need for an effective means of language translation is becoming more and more apparent. Among the concerns of information professionals is the capacity of an interested party to access web information offered in an unfamiliar language. Classified under the wide field of artificial intelligence, machine translation (MT) is an approach related to natural language processing. The machine translation technique involves the use of software for the conversion of documents or verbalized information from one natural language into another. Of late, a substantial number of procedures have been proposed for the fashioning of an efficient MT system. While these procedures were observed to be capable in certain areas, they were found wanting in others. The objectives of this endeavour are to (a) conduct a thorough investigation on mach...

Clichés, as trite expressions, are predominantly multiword expressions, but not all MWEs are clichés. We conduct a preliminary examination of the problem of determining how clichéd a text is, taken as a whole, by comparing it to a reference text with respect to the proportion of more-frequent n-grams, as measured in an external corpus. We find that more-frequent n-grams are over-represented in clichéd text. We apply this finding to the "Eumaeus" episode of James Joyce's novel Ulysses, which literary scholars believe to be written in a deliberately clichéd style.

This paper describes the adaptation and extension of an existing morphological system and its integration into an intranet service of a large international bank. The system includes a tool for the analysis and extraction of simple and complex terms. As a side-effect the procedure for the definition of new terms has been consolidated. The intranet service analyzes HTML pages on the fly, compares the results with the vocabulary of an inhouse terminological database (CLS-TDB) and generates hyperlinks in case matches have been found. Currently, the service handles terms in both German and English. The implementation of the service for Italian, French and Spanish is under way.

Having a quality annotated corpus is essential especially for applied research. Despite the recent focus of Web science community on researching about cyberbullying, the community dose not still have standard benchmarks. In this paper, we publish first, a quality annotated corpus and second, an offensive words lexicon capturing different types type of harassment as (i) sexual harassment, (ii) racial harassment, (iii) appearance-related harassment, (iv) intellectual harassment, and (v) political harassment.We crawled data from Twitter using our offensive lexicon. Then relied on the human judge to annotate the collected tweets w.r.t. the contextual types because using offensive words is not sufficient to reliably detect harassment. Our corpus consists of 25,000 annotated tweets in five contextual types. We are pleased to share this novel annotated corpus and the lexicon with the research community. The instruction to acquire the corpus has been published on the Git repository.

This paper describes the combinatorial properties of agere in association with nouns that designate a scenic activity. I therefore examine combinations such as fabulam, tragoediam, comoediam, partes, gestum, personam agere. The aim of the work is to show that these combinations are light verb constructions, providing linguistic
evidences. For this purpose, examples were collected through the examination of the Thesaurus Linguae Latinae and some of the tests in the literature were used (Langer, 2004) for the identification of such constructions (passivization, pronominalization, relativization and expansion of the predicative name). Moreover, the degree of syntagmatic cohesion between the elements of the construction and the semantic features of the combinations are taken into account. The choice of the tests responds to the needs of a research based on a closed corpus without the possibility of using the judgment of the native speaker.

This paper addresses Persian Complex Predicates (CPs) from an Applied/Pedagogical Construction Grammar (PCxG) stance. PCxG is an approach to foreign language pedagogy that emphasises the importance of constructions (form-meaning pairings), which are patterns of words and grammatical structures that have meaning beyond the sum of their individual parts. According to Goldberg (2006: 3), it is 'an attempt to describe language in a way that is both descriptively accurate and pedagogically useful'. Persian CPs are multi-word predicates comprised of twenty so-called light verbs and a nonverbal element (noun, adjective, adverb, preposition, verbal particle, complex noun, noun plus adverb) forming a single conceptual unit (e.g. pakhsh kardan, lit. scatteredADJ do, 'to spread'; and charkh zadan, lit. wheelN hit, 'to stroll'). Persian CPs present a compelling challenge to linguistics due to their lexical and phrasal properties. For example, they can undergo derivational processes, but they are also syntactically separable by the negation prefix, future auxiliary, or the direct object clitics. In this study, I argue that for teaching Persian CPs to English speakers a PCxG approach can be construed as a multidisciplinary effort aiming to elicit those aspects of Construction Grammar (CxG) that can be tied in more explicitly with Applied Linguistics, teacher education, and foreign language pedagogy.

Pakzadian, M. (2024). Persian (Farsi) Complex Predicates from an Applied Cognitive Construction Grammar Stance. Cahiers du Centre de Linguistique et des Sciences du Langage, n 68, 171-198.