Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
Η αναζήτηση πληροφορίας στο Παγκόσμιο Ιστό λόγω της ραγδαίας αύξησης του όγκου του, αποτελεί ένα δύσκολο και χρονοβόρο εγχείρημα. Το τεράστιο πλήθος πληροφοριών που εντοπίζεται σε αυτό αποτελεί ταυτόχρονα και το σημαντικότερο εμπόδιο στην εύρεση των συγκεκριμένων στοιχείων που ενδιαφέρουν κάθε χρήστη. Οι μηχανές αναζήτησης αποτελούν το αρχικό σημείο εισόδου για εκατομμύρια χρήστες στο διαδίκτυο. Το μέγεθος του διαδικτύου αυξάνεται συνεχώς και η κάθε μηχανή αναζήτησης μπορεί να χειριστεί ένα μικρό ποσοστό των συνολικών πληροφοριών που διαθέτει. Ο συνδυασμός μερικών μηχανών αναζήτησης μπορεί να οδηγήσει στην αύξηση της κάλυψης του συνολικού ποσού της πληροφορίας στο Διαδίκτυο, και επομένως αυξάνει την πιθανότητα εύρεσης περισσότερων επιθυμητών πληροφοριών. Αυτός είναι και ο στόχος των Μηχανών Μετά-Αναζήτησης. Επιπρόσθετα, η συνωνυμία και η πολυσημία συντελούν στη δυσκολία εύρεσης της επιθυμητής πληροφορίας. Στα πλαίσια αυτής της εργασίας αναπτύχθηκε μια τεχνική βελτίωσης ερωτημάτων, η οποία χρησιμοποιεί το σημασιολογικό δίκτυο λημμάτων WordNet, ώστε η εφαρμογή να προτείνει εναλλακτικούς όρους στους χρήστες του Διαδικτύου για την διατύπωση της πληροφοριακής τους ανάγκης. Στην παρούσα διπλωματική υλοποιείται η μηχανή μετά-αναζήτησης “WordNet meta-search engine”, η οποία χρησιμοποιεί τα αποτελέσματα των τριών πιο διαδεδομένων Μηχανών Αναζήτησης, της Google, της Yahoo!Search και της Bing. Καθώς επίσης, δίνει τη δυνατότητα στο χρήστη να επιλέξει, αν θέλει, και άλλες δύο μηχανές αναζήτησης, την AltaVista και την AllTheWeb. Η μετά-μηχανή αναζήτησης “WordNet meta-search engine” είναι μία εύκολη στη χρήση εφαρμογή, η οποία ομαδοποιεί τα αποτελέσματα της βάσει• των όρων αναζήτησης και των σχέσεων των συνώνυμων, αντώνυμων, υπερώνυμων και υπώνυμων όρων που παρέχει ο «Θησαυρός» λεξιλογίου του WordNet. Ο σκοπός της παρούσας μετά-μηχανής αναζήτησης είναι να βελτιώσει την απόδοση της ανάκτησης πληροφορίας με την ενσωμάτωση τεχνικών διεύρυνσης και επέκτασης των ερωτημάτων προς όφελος του χρήστη, με τη βοήθεια του σημασιολογικού δικτύου WordNet.
espanolEn la Argentina, desde 1870 se inicio una prolifica produccion de instrumentos lexicograficos que registraban singularidades lexicas. La conciencia de tal peculiaridad condujo a confeccionar, continuando con la tradicion hispanoamericana, diccionarios complementarios y contrastivos de diferentes modalidades. Por un lado, se publicaron obras descriptivas que recogian ruralismos, indigenismos, regionalismos (tanto americanismos como provincialismos o localismos) y argentinismos. Por otro, algunas normativas que recolectaban barbarismos y censuraban su uso, tomando como parametro la norma del castellano peninsular. En este trabajo, analizamos puntualmente un dominio del discurso lexicografico: los mecanismos de citacion y ejemplificacion. Primero, expondremos las diversas clases de ejemplos y sus funcionamientos. Luego, examinaremos nuestro corpus centrandonos en dos aspectos: a) las condiciones del proceso de diccionarizacion; b) los modos de funcionamiento discursivo de los ejemplos en la lexicografia monolingue argentina. Apuntamos a mostrar que dicho dominio, tanto como el paratexto, la nomenclatura y la microestructura, permite vincular el discurso lexicografico con el imaginario nacional. EnglishSince 1870, a prolific production of lexicographical instruments registering lexical singularities began in Argentina. The awareness of this peculiarity led to the elaboration following a Hispanic American tradition of complementary and contrastive dictionaries of different modalities. On one hand, descriptive works that collected ruralisms, indigenisms, regionalisms (both Americanisms and provincialisms or localisms), and Argentinisms were published. On the other hand, some normative works that gathered barbarisms and condemned their use emerged, using as a parameter the peninsular Spanish norm. In this paper, we analyze specifically a domain of lexicographical discourse: the quotation and exemplification mechanisms. First, we expose different types of examples and their functioning. Then, we examine our corpus focusing on two aspects: a) the conditions of the dictionarization process; and b) discursive functioning of examples in Argentine monolingual lexicography. We aim to show that this domain, as well as paratext, nomenclature and microstructure, allows us to link lexicographical discourse with national imaginary.
This study explores great potentials of computer corpus in English writing research and pedagogy via an in-depth comparative analysis of causality devices in Chinese EFL student writings. Our corpus-based study shows better performances of senior students (CLEC ST6 as against CLEC ST5) in the overall frequency, lexical richness, positionings and semantic prosody of the causality devices though still somehow deviant from western discourse norms.
This paper aims to explore the uncertainties of English language in general and the lexical and semantic deviations in e e cummings in particular, with reference to ten of his poems, namely, “the hours rise up putting off stars,” “in just-spring when the world,” “nobody loses all the time,” “it is so long since my heart,” “as freedom is a breakfastfood,” “love is the every only god,” “somewhere i have never travelled gladly beyond,” “sweet spring is your,” “if i have made, my lady, intricate,” and “anyone lived in a pretty how town.” These poems have been studied with the aim to examine how language can break the linguistic norms and yet be communicative and effective. The paper shows how cummings goes beyond the constraints of the linguistic forms with all their conventions and insipid attitudes and carves out a fresh language for all his poetic requirements.
Abstract Especially for non-experts, translating legal texts is a complicated and multifaceted process, demanding much linguistic and technical competence from translators. Legal systems differ from one another, and each one has a specific set of norms, especially reflected at the lexical level – that is, in terminology. Understanding two legal systems is not easy, even for experts; of course, it is even more difficult for translators, who are usually not legal experts. This article focuses on how to quickly and transparently provide translators with some (basic) technical knowledge using ontologies, which in recent years have found considerable application in synthesizing and visualizing knowledge.
Proceedings of the Ninth International Workshop \non Treebanks and Linguistic Theories. \nEditors: Markus Dickinson, Kaili Müürisep and Marco Passarotti. \nNEALT Proceedings Series, Vol. 9 (2010), 127-138. \n© 2010 The editors and contributors. \nPublished by \nNorthern European Association for Language \nTechnology (NEALT) \nhttp://omilia.uio.no/nealt. \nElectronically published at \nTartu University Library (Estonia) \nhttp://hdl.handle.net/10062/15891.
The insula has been implicated as a component of central networks subserving evaluative and affective processes. This study examined evaluative valence and arousal ratings in response to picture stimuli in patients with lesions of the insula and two contrast groups: a control-lesion group (the primary contrast group) and an amygdala-lesion group. Patients rated the positivity and negativity of picture stimuli (from very unpleasant to very pleasant) and how emotionally arousing they found the pictures to be. Compared with patients in the control-lesion group, patients with insular lesions reported reduced arousal in response to both unpleasant and pleasant stimuli, as well as marked attenuation of valence ratings. In contrast, the arousal ratings of patients with amygdala lesions were selectively attenuated for unpleasant stimuli, and these patients' positive and negative valence ratings did not differ from those of the control-lesion group. Results support the view that the insular cortex may play a broad role in integrating affective and cognitive processes, whereas the amygdala may have a more selective role in affective arousal, especially for negative stimuli.
In this paper, we present a system that automatically extracts lexicalized tree adjoining grammars (LTAG) from treebanks. We first discuss in detail extraction algorithms and compare them to previous works. We then report the first LTAG extraction result for Vietnamese, using a recently released Vietnamese treebank. The implementation of an open source and language independent system for automatic extraction of LTAG grammars is also discussed. 1
We describe a process for converting the Penn Arabic Treebank into the CCG formalism. Previous efforts have yielded CCGbanks in English, German, and Turkish, thus opening these languages to the sophisticated computational tools developed for CCG and enabling further cross-linguistic development. Conversion from a context free grammar treebank to a CCGbank is a four stage process: head finding, argument classification, binarization, and category conversion. In the process of implementing a basic CCGbank conversion algorithm, we reveal properties of Arabic grammar that interfere with conversion, such as subject topicalization, genitive constructions, relative clauses, and optional pronominal subjects. All of these problematic phenomena can be resolved in a variety of ways- we discuss advantages and disadvantages of each in their respective sections. We detail these and describe our categorial analysis of each of these Arabic grammatical phenomena in depth, as well as technical details on their integration into the conversion algorithm. 1.
Proceedings of the Workshop on Annotation and \nExploitation of Parallel Corpora AEPC 2010. \nEditors: Lars Ahrenberg, Jörg Tiedemann and Martin Volk. \nNEALT Proceedings Series, Vol. 10 (2010), 1-13. \n© 2010 The editors and contributors. \nPublished by \nNorthern European Association for Language \nTechnology (NEALT) \nhttp://omilia.uio.no/nealt. \nElectronically published at \nTartu University Library (Estonia) \nhttp://hdl.handle.net/10062/15893.
In this paper, we present several ways to measure and evaluate the annotation and annotators, proposed and used during the building of the Czech part of the Prague Czech-English Dependency Treebank. At first, the basic principles of the treebank annotation project are introduced (division to three layers: morphological, analytical and tectogrammatical). The main part of the paper describes in detail one of the important phases of the annotation process: three ways of evaluation of the annotators- inter-annotator agreement, error rate and performance. The measuring of the inter-annotator agreement is complicated by the fact that the data contain added and deleted nodes, making the alignment between annotations non-trivial. The error rate is measured by a set of automatic checking procedures that guard the validity of some invariants in the data. The performance of the annotators is measured by a booking web application. All three measures are later compared and related to each other. 1.
This paper presents a current status of Thai resources and tools for CG development. We also proposed a Thai categorial dependency grammar (CDG), an extended version of CG which includes dependency analysis into CG notation. Beside, an idea of how to group a word that has the same functions are presented to gain a certain type of category per word. We also discuss about a difficulty of building treebank and mention a toolkit for assisting on a Thai CGs tree building and a tree format representations. In this paper, we also give a summary of applications related to Thai CGs. 1
The Prague Dependency Treebank (henceforth PDT) is a large collection of texts in Czech. It contains several layers of rich annotation, ranging from morphology to deep syntax. It is unique in its size and theoretical background, especially for a language like Czech, which can be, with regard to the number of its speakers, considered a small language. In this article, we use PDT 2.0 to demonstrate that within real NLP systems, complex annotations may cut both ways. We present several issues that might pose problems when extracting data from PDT, and complex structures in general, and hint on possible solutions.
Latent variable grammars take an observed (coarse) treebank and induce more fine-grained grammar categories, that are better suited for modeling the syntax of natural languages. Estimation can be done in a generative or a discriminative framework, and results in the best published parsing accuracies over a wide range of syntactically divergent languages and domains. In this paper we highlight the commonalities and the differences between the two learning paradigms. 1
In the architecture of a natural language processing system based on linguistic knowledge, two types of component are important: the knowledge databases and the processing modules. One of the knowledge databases is the lexical database, which is responsible for providing the lexical unities and its properties to the processing modules. The systems that process two or more languages require bilingual and/or multilingual lexical databases. These databases can be constructed by aligning distinct monolingual databases. In this paper, we present the interlingua and the strategy of aligning the two monolingual databases in REBECA, which only stores concepts from the “wheeled vehicle” domain.
This paper proposes a method of correcting annotation errors in a treebank. By using a synchronous grammar, the method transforms parse trees containing annotation errors into the ones whose errors are corrected. The synchronous grammar is automatically induced from the treebank. We report an experimental result of applying our method to the Penn Treebank. The result demonstrates that our method corrects syntactic annotation errors with high precision.
Despite the prevalence of infidelity, there is relatively little research regarding the long-term effects on the children. Combining the views of transgenerational theory with the existing literature on infidelity as a family stressor and infidelity as a trauma, this project examines the potential lasting effects of parental infidelity on the adult child. Study participants included a small sample of adults over the age of 18 who were aware of parental infidelity in their family of origin. Through a series of regression analyses using a moderator model, the researchers found that higher negative self schemata and affect ratings related to the perception of the infidelity were associated with more conservative attitudes toward sex. Suggestions for future utilization of the proposed model and the clinical implications of the findings are also discussed.
We propose an effective approach to automatically identify predicate heads in Chinese sentences based on statistical pre-processing and rule-based post-processing. In the preprocessing stage, the maximal noun phrases in a sentence are recognized and replaced by “NP ” labels to simplify the sentence structure. Then a CRF model is trained to recognize the predicate heads of this simplified sentence. In the post-processing stage, a rule base is built according to the grammatical features of predicate heads. It is then utilized to correct the preliminary recognition results. Experimental results show that our approach is feasible and effective, and its accuracy achieves 89.14 % on Tsinghua Chinese Treebank. 1
This paper presents a method for the automatic detection and correction of malapropism errors found in documents using the WordNet lexical database, a search engine (Google) and a paronyms dictionary. The malapropisms detection is based on the evaluation of the cohesion of the local context using the search engine, while the correction is done using the whole text cohesion evaluated in terms of lexical chains built using the linguistic ontology. The correction candidates, which are taken from the paronyms dictionary, are evaluated versus the local and the whole text cohesion in order to find the best candidate that is chosen for replacement. The testing methods of the application are presented, along with the obtained results.
DeSR is a statistical transition-based dependency parser which learns from annotated corpora which actions to perform for building parse trees while scanning a sentence. We describe the experiments performed for the ICON 2010 Tools Contest on Indian Dependency Parsing. DesR was configured to exploit specific features from the Indian treebanks. The submitted run used a stacked combination of four configurations of the DeSR parser and achieved the best unlabeled accuracy scores in all languages. The contribution to the result of various choices is analyzed.
Real world audio clips contain numerous acoustic sources. The rich acoustic information they carry cannot be fully described with single or even multiple terms about the acoustic sources alone. For instance, the label birds assigned to a birds singing clip that includes sounds of trees and a small river does not properly capture the experience it creates in the person listening to it. In this paper we introduce a novel scheme where the subjective experience of listening to sound clips containing mixture of sources is captured using affective measures. Furthermore, in contrast to the conventional approach of simple label-based methods, the affective ratings are then used to evaluate the performance of an example-based audio retrieval system. We argue that audio retrieval systems can benefit from using affective measures which are well established in experimental psychology, especially when dealing with real world audio clips. We present experimental results of our pilot study to support this motivation where the latent indexing framework has been employed for example-based retrieval on a collection of clips from the BBC sound effects library. The result of our study indicates that using the scheme of affective measures for representation and evaluation is indeed a promising direction to explore.
Discriminative parse reranking has been shown to be an effective technique to im-prove the generative parsing models. In this paper, we present a series of exper-iments on parsing the Tsinghua Chinese Treebank with hierarchically split-merge grammars and reranked with a perceptron-based discriminative model. In addition to the homogeneous annotation on TCT, we also incorporate the PCTB-based parsing result as heterogeneous annotation into the reranking feature model. The rerank-ing model achieved 1.12 % absolute im-provement on F1 over the Berkeley parser on a development set. The head labels in Task 2.1 are annotated with a sequence labeling model. The system achieved
We describe an effective constituent projection strategy, where constituent projection is performed on the basis of dependency projection. Especially, a novel measurement is proposed to evaluate the candidate projected constituents for a target language sentence, and a PCFG-style parsing procedure is then used to search for the most probable projected constituent tree. Experiments show that, the parser trained on the projected treebank can significantly boost a state-of-the-art supervised parser. When integrated into a tree-based machine translation system, the projected parser leads to translation performance comparable with using a supervised parser trained on thousands of annotated trees. 1
Studies of discourse relations have not, in the past, attempted to characterize what serves as evidence for them, beyond lists of frozen expressions, or markers, drawn from a few well-defined syntactic classes. In this paper, we describe how the lexicalized discourse relation annotations of the Penn Discourse Treebank (PDTB) led to the discovery of a wide range of additional expressions, annotated as AltLex (alternative lexicalizations) in the PDTB 2.0. Further analysis of AltLex annotation suggests that the set of markers is open-ended, and drawn from a wider variety of syntactic types than currently assumed. As a first attempt towards automatically identifying discourse relation markers, we propose the use of syntactic paraphrase methods.
This paper investigates whether high-quality annotations for tasks involving semantic disambiguation can be obtained without a major investment in time or expense. We examine the use of untrained human volunteers from Amazon’s Mechanical Turk in disambiguating prepositional phrase (PP) attachment over sentences drawn from the Wall Street Journal corpus. Our goal is to compare the performance of these crowdsourced judgments to the annotations supplied by trained linguists for the Penn Treebank project in order to indicate the viability of this approach for annotation projects that involve contextual disambiguation. The results of our experiments show that invoking majority agreement between multiple human workers can yield PP attachments with fairly high precision, confirming that this crowdsourcing approach to syntactic annotation holds promise for the generation of training corpora in new domains and genres.
There often exist multiple corpora for the same natural language processing (NLP) tasks. However, such corpora are generally used independently due to distinctions in annotation standards. For the purpose of full use of readily available human annotations, it is significant to simultaneously utilize multiple corpora of different annotation standards. In this paper, we focus on the challenge of constituent syntactic parsing with treebanks of different annotations and propose a collaborative decoding (or co-decoding) approach to improve parsing accuracy by leveraging bracket structure consensus between multiple parsing decoders trained on individual treebanks. Experimental results show the effectiveness of the proposed approach, which outperforms stateof-the-art baselines, especially on long sentences. 1
Two of the main corpora available for training discourse relation classifiers are the RST Discourse Treebank (RST-DT) and the Penn Discourse Treebank (PDTB), which are both based on the Wall Street Journal corpus. Most recent work using discourse relation classifiers have employed fully-supervised methods on these corpora. However, certain discourse relations have little labeled data, causing low classification performance for their associated classes. In this paper, we attempt to tackle this problem by employing a semi-supervised method for discourse relation classification. The proposed method is based on the analysis of feature cooccurrences in unlabeled data. This information is then used as a basis to extend the feature vectors during training. The proposed method is evaluated on both RST-DT and PDTB, where it significantly outperformed baseline classifiers. We believe that the proposed method is a first step towards improving classification performance, particularly for discourse relations lacking annotated data.
For centuries, scholars have explored the deep links among human languages. In this paper, we present a class of probabilistic models that use these links as a form of naturally occurring supervision. These models allow us to substantially improve performance for core text processing tasks, such as morphological segmentation, part-of-speech tagging, and syntactic parsing. Besides these traditional NLP tasks, we also present a multilingual model for the computational decipherment of lost languages. 1. Overview Electronic text is currently being produced at a vast and unprecedented scale across the languages of the world. Natural Language Processing (NLP) holds out the promise of automatically analyzing this growing body of text. However, over the last several decades, NLP research efforts have focused on the English language, often neglecting the thousands of other languages of the world (Bender, 2009). Most of these languages are currently beyond the reach of NLP technology due to several factors. One of these is simply the lack of the kinds of hand-annotated linguistic resources that have helped propel the performance of English language systems. For complex tasks of linguistic analysis, hand-annotated corpora can be prohibitively time-consuming and expensive to produce. For example, the most widely used annotated corpus in the English language, the Penn Treebank (Marcus et al., 1994), took years for a team of professional linguists to produce. It is unrealistic to expect such resources to ever exist for the majority of the world’s languages.
Discontinuities occur especially frequently in languages with a relatively free word order, such as German. Generally, due to the longdistance dependencies they induce, they lie beyond the expressivity of Probabilistic CFG, i.e., they cannot be directly reconstructed by a PCFG parser. In this paper, we use a parser for Probabilistic Linear Context-Free Rewriting Systems (PLCFRS), a formalism with high expressivity, to directly parse the German NeGra and TIGER treebanks. In both treebanks, discontinuities are annotated with crossing branches. Based on an evaluation using different metrics, we show that an output quality can be achieved which is comparable to the output quality of PCFG-based systems. In most constituency treebanks, sentence annotation is restricted to having the shape of trees without crossing branches, and the non-local dependencies induced by the discontinuities are modeled by an additional mechanism. In the Penn Treebank (PTB) (Marcus et al., 1994), e.g., this mechanism is a combination of special labels and empty nodes, establishing implicit additional edges. In the German TüBa-D/Z (Telljohann et al., 2006), additional edges are established by a combination of topological field annotation and special edge labels. As an example, Fig. 1 shows a tree from TüBa-D/Z with the annotation of (1). Note here the edge label ON-MOD on the relative clause which indicates that the subject of the sentence (alle Attribute) is modified. 1
Habituation is a fundamental form of learning manifested by a decrement of neuronal responses to repeated sensory stimulation. In addition, habituation is also known to occur on the behavioral level, manifested by reduced emotional reactions to repeatedly presented affective stimuli. It is, however, not clear which brain areas show a decline in activity during repeated sensory stimulation on the same time scale as reduced valence and arousal experience and whether these areas can be delineated from other brain areas with habituation effects on faster or slower time scales. These questions were addressed using functional magnetic resonance imaging acquired during repeated stimulation with piano melodies. The magnitude of functional responses in the laterobasal amygdala and in related cortical areas and that of valence and arousal ratings, given after each music presentation, declined in parallel over the experiment. In contrast to this long-term habituation (43 min), short-term decreases occurring within seconds were found in the primary auditory cortex. Sustained responses that remained throughout the whole investigated time period were detected in the ventrolateral prefrontal cortex extending to the dorsal part of the anterior insular cortex. These findings identify an amygdalocortical network that forms the potential basis of affective habituation in humans.
Respondents can vary strongly in the way they use rating scales. Specifically, respondents can exhibit a variety of response styles, which threatens the validity of the responses. The purpose of this article is to investigate how response style and content of the items affect rating scale responses. The authors develop a novel model that accounts for different types of response styles, content of items, and background characteristics of respondents. By imposing a bilinear parameter structure on a multinomial logit model, the authors graphically distinguish the effects on the response behavior of the characteristics of a respondent and the content of an item. The authors combine this approach with finite mixture modeling, yielding two segmentations of the respondents: one for response style and one for item content. They apply this latent-class bilinear multinomial logit model to the well-known List of Values in a cross-national context. The results show large differences in the opinions and the response styles of respondents and reveal previously unknown response styles. Some response styles appear to be valid communication styles, whereas other response styles often concur with inconsistent opinions of the items and seem to be response bias.
We investigate parsing accuracy on the Korean Treebank 2.0 with a number of different grammars. Comparisons among these grammars and to their English counterparts suggest different aspects of Korean that contribute to parsing difficulty. Our results indicate that the coarseness of the Treebank’s nonterminal set is a even greater problem than in the English Treebank. We also find that Korean’s relatively free word order does not impact parsing results as much as one might expect, but in fact the prevalence of zero pronouns accounts for a large portion of the difference between Korean and English parsing scores. 1
Internet dating is now ranked third as the way people meet, behind meeting at work or school, and through a friend or family member. This study researches the use of social and linguistic norms in online dating advertisements. Previous research has posed that social groups create unique identities and group members will selectively present themselves in ways consistent with these identities. Using Craigslist to assess the similarities and differences between genders and sexualities in online personal postings, an online quiz-like survey was created. This research reports on people's abilities to predict the sexual orientation and gender of the writer based on linguistic cues.
Parser disambiguation with precision grammars generally takes place via statistical ranking of the parse yield of the grammar using a supervised parse selection model. In the standard process, the parse selection model is trained over a hand-disambiguated treebank, meaning that without a significant investment of effort to produce the treebank, parse selection is not possible. Furthermore, as treebanking is generally streamlined with parse selection models, creating the initial treebank without a model requires more resources than subsequent treebanks. In this work, we show that, by taking advantage of the constrained nature of these HPSG grammars, we can learn a discriminative parse selection model from raw text in a purely unsupervised fashion. This allows us to bootstrap the treebanking process and provide better parsers faster, and with less resources. 1
Proceedings of the Ninth International Workshop \non Treebanks and Linguistic Theories. \nEditors: Markus Dickinson, Kaili Müürisep and Marco Passarotti. \nNEALT Proceedings Series, Vol. 9 (2010), 55-66. \n© 2010 The editors and contributors. \nPublished by \nNorthern European Association for Language \nTechnology (NEALT) \nhttp://omilia.uio.no/nealt. \nElectronically published at \nTartu University Library (Estonia) \nhttp://hdl.handle.net/10062/15891.
In this paper we explore two strategies to incorporate local morphosyntactic features in Hindi dependency parsing. These features are obtained using a shallow parser. We first explore which information provided by the shallow parser is most beneficial and show that local morphosyntactic features in the form of chunk type, head/non-head information, chunk boundary information, distance to the end of the chunk and suffix concatenation are very crucial in Hindi dependency parsing. We then investigate the best way to incorporate this information during dependency parsing. Further, we compare the results of various experiments based on various criterions and do some error analysis. All the experiments were done with two data-driven parsers, MaltParser and MSTParser, on a part of multi-layered and multi-representational Hindi Treebank which is under development. This paper is also the first attempt at complete sentence level parsing for Hindi.
Chinese parsing has received more and more attention, and in this paper, we use toolkit to perform parsing on the data of Tsinghua Chinese Treebank (TCT) used in CIPS, and we use Conditional Random Fields (CRFs) to train specific model for the head recognition. At last, we compare different results on different POS results. 1
In this paper we demonstrate how Prolog can be used to query linguistically annotated corpora, combining the ease of dedicated declarative query languages and the flexibility of general-purpose languages. On the basis of a Prolog representation of the German Tuba-D/Z Treebank, we show how one can tally arbitrary features of (groups) of nodes, define queries that combine information from different layers of annotation and cross sentence boundaries, query ‘virtual annotation’ by transforming annotation on-thefly, and perform data driven error analysis. Almost all code needed for these case studies is contained in the paper.
Factorization is the operation of transforming a production in a Linear Context-Free Rewriting System (LCFRS) into two simpler productions by factoring out a subset of the nonterminals on the production’s righthand side. Factorization lowers the rank of a production but may increase its fan-out. We show how to apply factorization in order to minimize the parsing complexity of the resulting grammar, and study the relationship between rank, fanout, and parsing complexity. We show that it is always possible to obtain optimum parsing complexity with rank two. However, among transformed grammars of rank two, minimum parsing complexity is not always possible with minimum fan-out. Applying our factorization algorithm to LCFRS rules extracted from dependency treebanks allows us to find the most efficient parsing strategy for the syntactic phenomena found in non-projective trees. 1
We present the first effort towards producing an Arabic Discourse Treebank, a news corpus where all discourse connectives are identified and annotated with the discourse relations they convey as well as with the two arguments they relate. We discuss our collection of Arabic discourse connectives as well as principles for identifying and annotating them in context, taking into account properties specific to Arabic. In particular, we deal with the fact that Arabic has a rich morphology: we therefore include clitics as connectives as well as a wide range of nominalizations as potential arguments. We present a dedicated discourse annotation tool for Arabic and a large-scale annotation study. We show that both the human iden-tification of discourse connectives and the determination of the discourse relations they convey is reliable. Our current annotated corpus encompasses a final 5651 annotated discourse connectives in 537 news texts. In future, we will release the annotated corpus to other researchers and use it for training and testing automated methods for discourse connective and relation recognition. 1.
The Quranic Arabic Dependency Treebank (QADT) is part of the Quranic Arabic Corpus (http://corpus.quran.com), an online linguistic resource organized by the University of Leeds, and developed through online collaborative annotation. The website has become a popular study resource for Arabic and the Quran, and is now used by over 1,500 researchers and students daily. This paper presents the treebank, explains the choice of syntactic representation (), and highlights key parts of the annotation guidelines. The text being analyzed is the Quran, the central religious book of Islam, written in classical Quranic Arabic (c. 600 CE). To date, all 77,430 words of the Quran have a manually verified morphological analysis, and syntactic analysis is in progress. 11,000 words of Quranic Arabic have been syntactically annotated as part of a gold standard treebank (). Annotation guidelines are especially important to promote consistency for a corpus which is being developed through online collaboration, since often many people will participate from different backgrounds and with different levels of linguistic expertise. The treebank is available online for collaborative correction to improve accuracy, with suggestions reviewed by expert Arabic linguists, and compared against existing published books of Quranic Syntax. 1.
Automated conversion has allowed the development of wide-coverage corpora for a variety of grammar formalisms without the expense of manual annotation. Analysing new languages also tests formalisms, exposing their strengths and weaknesses. We present Chinese CCGbank, a 760,000 word corpus annotated with Combinatory Categorial Grammar (CCG) derivations, induced automatically from the Penn Chinese Treebank (PCTB). We design parsimonious CCG analyses for a range of Chinese syntactic constructions, and transform the PCTB trees to produce them. Our process yields a corpus of 27,759 derivations, covering 98.1 % of the PCTB. 1