Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
Recent studies have suggested a theoretical distinction between active elaboration and passive storage in visuospatial working memory, but research with older adults has failed to demonstrate a differential preservation of these two abilities. The results are controversial, and the investigation of the active component has been inhibited by the absence of any appropriate experimental procedures. A new task was developed involving the mental reconstruction of pictures of objects from fragmented pieces, and this provides a useful procedure for exploring active visuospatial processing. Significant differences in terms of both correctness and response latency were obtained between young and older adults and between younger old and older old adults. Performance also varied with visual complexity, mental rotation, and processing load. It is concluded that this ecologically relevant procedure constitutes a very powerful, sensitive, and reliable tool for identifying individual differences in visuospatial working memory.
1 Jonas Kuhn was at the University of Stuttgart when the main work reported in this paper was performed. Creation of high-quality treebanks requires expert knowledge and is extremely time consuming. Hence applying an already existing grammar in treebanking is an interesting alternative. This approach has been pursued in the syntactic annotation of German newspaper text in the TIGER project. We utilized the large-scale German LFG grammar of the PARGRAM project for semi-automatic creation of TIGER treebank annotations. The symbolic LFG grammar is used for full parsing, followed by semi-automatic disambiguation and automatic transfer into the treebank format. The treebank annotation format is a ‘hybrid ’ representation structure which combines constituent analysis and functional dependencies. Both types of information are provided by the LFG analyses. Although the grammar and the treebank representations coincide in core aspects, e.g. the encoding of grammatical functions, there are mismatches in analysis details that are comparable to translation mismatches in natural language translation. This motivates the use of transfer technology from machine translation. The German LFG grammar analyzes on average 50 % of the sentences, roughly 70 % thereof are assigned a correct parse; after OT-filtering, a sentence gets 16.5 analyses on average (median: 2). We argue that despite the limits in corpus coverage the applications of the grammar in treebanking is useful especially for reasons of consistency. Finally, we sketch future extensions and applications of this approach, which include partial analyses, coverage extension, annotation of morphology, and consistency checks.
This paper presents our basic approach to creating Proposition Bank, which involves adding a layer of semantic annotation to the Penn English TreeBank. Without attempting to confirm or disconfirm any particular semantic theory, our goal is to provide consistent argument labeling that will facilitate the automatic extraction of relational data. An argument such as the window in John broke the window and in The window broke would receive the same label in both sentences. In order to ensure reliable human annotation, we provide our annotators with explicit guidelines for labeling all of the syntactic and semantic frames of each particular verb. We give several examples of these guidelines and discuss the inter-annotator agreement figures. We also discuss our current experiments on the automatic expansion of our verb guidelines based on verb class membership. Our current rate of progress and our consistency of annotation demonstrate the feasibility of the task.
One of the most difficult issues within corpora annotation on an underlying syntactic level is the restoration of nodes omitted in the surface shape of the sentence, but present on the underlying or deep syntactic level. In the present paper we concentrate on such type of nodes which are omitted due to the phenomenon usually called grammatical control with regard to their respective anaphoric relations. In particular, we extend the notion of control to nominalization and demonstrate how this relation is captured in the Prague Dependency Treebank. The theory of control is present within Chomskys framework of Government and Binding (using the terms verb of control, controller and controllee, cf. Chomsky, 1980), but also within many other formal frameworks, e.g. GPSG (Sag and Pollard, 1991) or categorial grammar (Bach, 1979). We analyse this phenomenon within the framework of the dependency grammar, theoretically based on the Functional Generative Description (FGD, cf. Sgall, Hajicova and Panevova, 1986). In FGD, on the underlying or tectogrammatical level, control is a relation of an obligatory or an optional referential dependency between a controller (antecedent) and a controllee (empty subject of the nonfinite complement (= controlled clause)). The controller is one of the participants in the valency frame of the governing verb (Actor (ACT), Addressee (ADDR), or Patient (PAT)). The controlled clause functions also as a filler of a dependency slot in the valency frame of the governing verb, being labeled as Patient or Actor. The empty subject of the controlled clause may have the function of different dependency relations to its head word (the infinitive): Actor, or, with passivization of the controlled clause, Addressee or Patient (cf. Koktova, 1992).
The purpose of this research is to investigate the difference on clothing image evaluation in the ratings between men and women. For this study, pilot test was conducted to 50 clothing majored university students to explore the stimulus of `cute`, `casual`, `sexy`, elegant`, intelligent`, `formal`, `romantic`, `individual`, `refined` for the 9 each image styles from the 32 spring wears in fashion magazine 『FARBE』(March, 2000). On the basis of the preliminary survey, the question items explored the 15 pairs of polar adjectives as seven-point Likert Scale. The main survey was preceded 94 female and 111 man of university students from March 13 to 24 in 2000, twice for 7-days interval. There were significant differences between the two sexes each style image ratings, it was found that the female was recorded more ordinary, stable, refined, superior, plain, like than the male for intelligent style. Meanwhile, the intelligent style was evaluated well on in years by female, but male young. The female tended that elegance style was more stable, warm and less young than the male. The cute style was evaluated more light, tender, feminine, young by the female than the male, and the female looked warm while the male cool. The formal style was more stable, unrefined, solid, unfamiliar, dislike, old by the female than the male. The casual style was revealed plain and warm by the female while splender and cool by the male, the female more active, tender, familiar than the male and individual, attractive and poor quality than the female. The sexy style was evaluated more active, good appearance, young than the female, tender than the male and the female dislike a bit while the male like. The female evaluated the refined style for more stable, refined, superior, good appearance and nature than the male. The romantic style was evaluated more like, refined, superior, good appearance nature and familiar by the male, but the female a bit unfamiliar. The individual style was revealed that the female evaluated cool and a bit dislike while the male warm and like, and the male more refined, feminine, young than female.
This study is one of the consequtive stages of a macro project designed to identify the relation between language skills and functional literacy in Turkish adults with varying educational backgrounds. Within this framework, earlier stages investigated how elementary and high school graduates in the Turkish sočiety could functionally use their language skills. The consideration is that one’s level of functional literacy observed in his linguistic practices involves how successfully he expresses himself through written and oral language, as well as how successfully he comprehends and interprets written and oral discourse in line with the dynamics of the idealized social and linguistic norms in the information society (Baynham, 1995, Barton, 1987). Although schooling is assumed to be providing individuals with the basics of such linguistic skills, the development or maintenance of these skills to meet the demands of the modern society may present individual- or groupspecific characteristics. It is thought that the quality of social interactions, either during or after school, can create a potential for the development of functional language skills. Whatever is gained at school may or may not be long-lasting once the formal education is over. Depending on the level of education and the types of social networks, individuals are assumed to be displaying varying degrees of functional literacy performance. In individuals with limited education, sometimes horizons are broadened through positive social networks, and functional literacy performance, especially in the language use, appears to be more succesful than expected. The results of the previous stages of the project presented supporting evidence for this consideration. This time focusing on a more educated group, the presented study aims to investigate the relations between the written and oral language skills and functional literacy in university graduates who have been working either at the state or private institutions. The subject group in this study consists of 240 university graduate Turkish males and females over 25 years of age. The subjects, grouped according to (a) their sexes, (b) their years of experience at work, (c) the type of institution they presently work at, and finally (d) their majors at university (independent variables), have been given a personal information questionnaire and have been tested on four language skills, i.e., reading, writing, speaking and listening through specially prepared materials. The oral performance of the subjects have been evaluated in terms of (a) standard pronunciation, (b) morphological, syntactic and semantic well-formedness of utterances, (c) use of discourse markers and idea organization, and (d) communicative competence. Their written performance has also been tested in terms of (a) use of Standard language, (b) idea organization and development, (c) syntactic and semantic well-formedness, (d) use of simple and complex sentence structures, and (e) punctuation and spelling. The data have been analyzed with the SPSS programme and the results have been evaluated in the light of linguistic and sociological theories to find out if the level of functional language skills is bound to the nature of independent variables under consideration. It was hypothesized that the level of functional literacy in university graduates would tend to be somewhat higher than that of the lower-level graduates. Results present supporting evidence; however, findings also reveal statistically significant differences among university-graduate subjects, particularly when independent variables are matched with the results of language skills tests.
This paper would like to introduce the reader into those aspects of the Arabic language which require some special treatment compared to languages Europeans are more familiar with. In spite of having fresh experience in building the Prague Arabic Dependency Treebank, the authors try to take a broader view of the problems encountered under way. The topics discussed include linguistic data retrieval, morphology and morphotactics modelling, and description of the language on the analytical level.
The annotation of the Prague Dependency Treebank is realized in two sub-collections which differ in the subtlety of annotation (the large collection and the model collection). In the present paper, we focus on deletions of complementations of verbs, postverbal nouns and adjectives, from the point of view of the annotators of the model collection. We inquire into the issues of deletions of participants of verbs with respect to the coreferential relations between the restored node and its antecedent, and we introduce the new type of deletion where the deleted node has no concrete antecedent (we call these restored nodes vague, unspecified anaphoric elements). We also specify the differences between the newly introduced lemma Unsp(ecified) and the lemma Gen(eral), used for deletions of General Participants. After analysing the process of nominalization, we extend the description of the particular types of deletions also to the postverbal nouns denoting action. 1. Three-layer system of tags The annotation of the Prague Dependency Treebank (PDT in the sequel) is basically conceived of in accordance with the theoretical assumptions of the Functional Generative Description (FGD in the sequel, see Sgall, Hajicova, and Panevova, 1986; Hajicova, 1993). The present paper deals with the current phase of annotation of PDT at the so-called tectogrammatical level (which captures underlying syntactic structures of sentences). This level of annotation, in contrast to the two preceding phases of annotation, namely the morphemic and the so-called analytical levels (for a description of the annotation scheme of PDT see e.g. Hajic et al., 2001, Hajicova et al., 2001), is realized in two steps (in two subcollections) which differ in the subtlety of annotation. The first, basic step of annotation is represented by the so-called basic or large collection (LC; today it contains about 25,000 sentences). The second step of annotation, the so-called model collection (MC), should provide full, detailed information about the underlying structure of the sentence. Due to the fact that this way of annotation is more detailed, therefore more elaborate, the model collection contains only about 400 sentences today. 2. General principles of annotation at the tectogrammatical level Tectogrammatical tree structures (TGTSs) are based on dependency syntax; they have the shape of a dependency tree with the verb as the root of the tree and its daughter nodes representing nodes depending on the governor (on each layer of the tree). The two dimensions of the tree represent the syntactic structure of the sentence (the vertical dimension) and the topic-focus articulation of the sentence, based on the underlying word order (the horizontal dimension). The tagging at the tectogrammatical level can be described by the following principles: (a) a single node of a TGTS may be a representation of more than one word; only autosemantic words have a node of their own, while the correlates of functional words (auxiliaries, prepositions etc.) are attached to the autosemantic words to which they belong (auxiliary verbs and subordinating conjunctions to the verbs, prepositions to nouns, etc.); (b) in the cases of deletion in the surface shape of the sentence, nodes are introduced into the tectogrammatical tree to 'recover' a deleted word; The Prague Bulletin of Mathematical Linguistics 78, 2002 38 (c) no non-projective structures are admitted at the tectogrammatical level (projectivity is a counterpart to continuity of constituents; non-projectivity appearing in the surface shape of some sentences is supposed to be handled by movement rules between the tectogrammatical tree and the morphemic string); (d) not only the direction of the dependence on the governing node (dependence to the left, dependence to the right) is taken into account, but also sister nodes are ordered (from left to right); their order reflects the scale of communicative dynamism which is relevant for the description of topic-focus articulation.
The orthographical complexity of Chinese, Japanese and Korean (CJK) poses a special challenge to the developers of computational linguistic tools, especially in the area of intelligent information retrieval. These difficulties are exacerbated by the lack of a standardized orthography in these languages, especially the highly irregular Japanese orthography. This paper focuses on the typology of CJK orthographic variation, provides a brief analysis of the linguistic issues, and discusses why lexical databases should play a central role in the disambiguation process.
Rapid and explosive progress of Web-based technologies makes developers hard to follow new ones, every time they come out. Web developers don't have enough time to understand so many up-to-date technologies and to enhance their productivity. To reduce development time in Web environments, as well as to increase efficiency, the authors defined a new XML-based language for Web developers. This language contains developer-friendly terminologies for DTD (document type definition). In addition, their concerns are focused on reducing network overload for Web servers and browsers. Using VBScript and JavaScript, they created an API (application programming interface) set, which can be accessed by other web pages. In order to test their DTD and API performance, a sample Web site was built that gets data from 20 database tables. In using their method, an approximate 70% page count reduction was evident, compared with other current Web development methods.
framework, by the Multimedia Annotation Consortium, and by the JSPS This paper describes the data management in multilingual lexical databases. Since NLP systems are using lexical data, the amount of work to build them is huge. That is the reason why it is important to use rigorous powerful systems and to be able to get data for a minimal cost. As an open source project, Papillon is reusing existing lexical data and wants to make volunteers collaborate. To make it possible, it combines state of the art concepts in the field of linguistic and computing. 1
We develop novel algorithmic ideas for building a natural language parser grounded upon the hypothesis of incrementality, which is widely supported by experimental data as a model of human parsing. Our proposal relies on a machine learning technique for predicting the correctness of partial syntactic structures that are built during the parsing process. A recursive neural network architecture is employed for computing predictions after a training phase on examples drawn from a corpus of parsed sentences, the Penn Treebank. Our results indicate the viability of the approach and lay out the premises for a novel generation of algorithms for natural language processing which more closely model human parsing. These algorithms may prove very useful in the development of efficient parsers and have an immediate application in the construction of semiautomatic annotation tools.
Dictionaries can be used as a basis for lexicon development for NLP applications. However, it often takes a lot of pre-processing before they are usable. In the last 5 years a product-independent database of formal word features has been developed on the basis of the Van Dale dictionaries for Dutch. The database has proven to be useful in various NLP applications. This paper describes the history, some advantages and the constraints in the development of this database.
This paper reviews the first year of the creation of a publicly available treebank for Portuguese, Floresta Sintá(c)tica, a collaboration project between the VISL and the Computational Processing of Portuguese projects. After briefly describing the main goals and the organization of the project, the creation of the annotated objects is presented in detail: preparing the text to be annotated, applying the Constraint Grammar based PALAVRAS parser, revising its output manually in a two-stage process, and carefully documenting the linguistic options. Some examples of the kind of interesting problems dealt with are presented, and the paper ends with a brief description of the tools developed, the project results so far, and a mention to a preliminary inter-annotator test and what was learned from it. 1. Introduction: Motivation and
The basic argument of this paper is that multiple language use in learner output is not always and exclusively indicative of the deficient nature (V. Cook, 1999; Firth and Wagner, 1997; Kramsch, 1997, 1998) of the language learner with respect toan idealized monolingual second language (L2) linguistic norm. Multilingual written learner texts and learners’ explications of these texts are examined in detail. These data suggest that learners conceptualize themselves as multicompetent (V. Cook, 1991, 1992, 1999) speakers who regularly, playfully and creatively decouple conventionalized L2 form-meaning pairings in order to produce and use their own locally relevant L2 signs. Within mainstream Second Language Acquisition (SLA) research practices and correctness-oriented Foreign Language Teaching (FLT) methodologies, this departure from norm approximation may be interpreted to indicate the reduced (Harder, 1980) or deficient nature of the learner. Within a Vygotskian approach to the psychology of mind and language learning (e.g., Lantolf, 2000a; Lantolf and Appel, 1994; Vygotsky, 1978, 1986), however, the learner’s playful use of multiple linguistic codes may index resourceful, creative and pleasurable displays of multicompetence (V. Cook, 1991,1992).
The PAPILLON project aims at creating a cooperative, free, permanent, web-oriented and personalizable environment for the development and the consultation of a multilingual lexical database. The initial motivation is the lack of dictionaries, both for humans and machines, between French and many Asian languages. In particular, although there are large F-J paper usage dictionaries, they are usable only by Japanese literates, as they never contain both original (kanji/kana) and romaji writing. This applies as well to Thai, Vietnamese, Lao, etc.
In the field of empirical natural language processing, researchers constantly deal with large amounts of marked-up data; whether the markup is done by the researcher or someone else, human nature dictates that it will have errors in it. This paper will more fully characterise the problem and discuss whether and when (and how) to correct the errors. The discussion is illustrated with specific examples involving function tagging in the Penn treebank.
Cet article a pour objectif d’interroger les relations existant entre le système linguistique (entendu au sens large, comme englobant l’ensemble des règles ou régularités qui sous-tendent la production et l’interprétation des énoncés attestés), et la culture, et plus précisément les normes communicatives en vigueur dans une société donnée. Après avoir envisagé un certain nombre de faits langagiers (unités lexicales, formes honorifiques, actes de langage et formules rituelles) qui portent manifestement la trace de ces normes culturelles sous-jacentes, l’auteure en conclut qu’il est dans une certaine mesure possible de reconstituer à partir de ces traces l’« ethos communicatif » de la société considérée, mais que ce travail de reconstitution ne va pas sans rencontrer un certain nombre de difficultés, qui sont passées en revue.
In this paper, we present an approach to primarily achieve the semantic interpretation and the region retrieval for an attentive region in a color image. The main components of the system include image feature extraction, indexing process, as well as linguistic inference rules construction and semantic description. Based on these features, each of attentive regions in an image can be described by a global linguistic meaning. The main procedure consists of two parts: forward and recall processes. The forward process primarily performs the linguistic meaning description of objects for an image, and the recall process reconstructs the region image which is the rough mental image of human memory retrieval. Experiments confirm that our approach is reasonable and feasible. bership function) to construct a knowledge base such as human knowledge and experiences [I, 21. A useful data information may be presented by the fuzzy number, and the data presentation of a fuzzy number can be parameterized to simplify the fizzy computation. In accordance with the advantages of indexing process and fuzzy set theory, in this paper, we present an approach to perform a human-based image interpretation. This approach combines image features and linguistic database to describe semantic meaning for image region.
This paper aims at providing a general description for DIADORIM, a lexical database for Brazilian Portuguese. DIADORIM is said to successively merge two very different previous application-oriented dictionaries, increasing their user-friendliness, the reusability of their entries and their capability of incorporating new features. Besides improving the structure of the previous databases, DIADORIM also preserves their performance and functionality, as indicated in the real use and simulation tests carried out during its evaluation.
This paper starts by discussing the reasons why linguists should be interested in parallel corpora. I outline the questions that parallel corpora enable us to ask, and relate them to traditional questions in linguistics and translation theory. The paper then suggests a method for arriving at answers to some of these questions. The proposed method builds on the notion of “ modulation” from Vinay and Darbelnet (1958) and attempts to put this notion on a sounder theoretical and empirical basis. It also includes a method of sharing data from different language pairs in a “ Contrastive Linguistic Database”.
The development of large coverage, rich unification- (constraint-) based grammar resources is very time consuming, expensive and requires lots of linguistic expertise. In this paper we report initial results on a new methodology that attempts to partially automate the development of substantial parts of large coverage, rich unification-(constraint-) based grammar resources. The method is based on a treebank resource (in our case Penn-II) and an automatic f-structure annotation algorithm that annotates treebank trees with proto-fstructure information. Based on these, we present two parsing architectures: in our pipeline architecture we firstextract a PCFG from the treebank following the method of [Charniak, 1993; Charniak, 1996], use the PCFG to parse new text, automatically annotate the resulting trees with our f-structure annotation algorithm and generate proto-f-structures. By contrast, in the integrated architecture we firstautomatically annotate the treebank trees with fstructure information and then extract an annotated PCFG (A-PCFG) from the treebank. We then use the A-PCFG to parse new text to generate proto-fstructures. Currently
While the corpus-based research relies on human annotated corpora, it is often said that a non-negligible amount of errors remain even in frequently used corpora such as Penn Treebank. Detection of errors in annotated corpora is important for corpus-based natural language processing. In this paper, we propose a method to detect errors in corpora using support vector machines (SVMs). This method is based on the idea of extracting exceptional elements that violate consistency. We propose a method of using SVMs to assign a weight to each element and to find errors in a POS tagged corpus. We apply the method to English and Japanese POS-tagged corpora and achieve high precision in detecting errors.
In this paper, we introduce a new European project named OrienTel. The aim of OrienTel is to enable the project's participants to design and develop multilingual interactive communication services for the Mediterranean and the Middle East, ranging from Morocco in the West to the Gulf states in the East, including Turkey and Cyprus. These multilingual applications will be largely speech-based and will typically be implemented on mobile and multi-modal platforms such as cellular GSM or UMTS phones, personal digital assistants (PDAs) or combinations of the two. Applications of the kind targeted are unified messaging, information retrieval, customer care, banking, WAP and service portals. To achieve this aim, the consortium will produce various surveys of the OrienTel region, compile a set of 22 linguistic databases, conduct research into ASR-related problems the OrienTel languages hold and develop demonstrator applications bearing evidence of OrienTel's multilingual orientation.
One of the major problems in the implementation ofNatural Language Processing (NLP) or MachineTranslation(MT) is a complete lexicon: the place wherethe systems information about words is stored. There aredifficulties in deciding what information should bestored in a lexicon and even greater difficulties inacquiring this information in proper form. OriNetsystem designed to incorporate multiple lexical databaseand tools under one consistent functional interface inorder to facilitate systems requiring syntactic, semanticand lexical information of Oriya language. We dividethe whole work into two independent task. One task isto write the source file that contains the basic lexicaldata and the content of those files are the lexicalsubstance of OriNet. Lexicographer did the major workof this task. In the second task was to create a set ofprograms those would accept the source files andprocessing it ultimately to display for the user. Thispaper describes an ongoing work on designing an ObjectOriented model for OriNet system. The technology ofObject Oriented programming in particular the richlibrary of classes and programming principles in whichJava offers. It also provides a convenient tool toconceptualise the process of OriNet system. Thistechnique also allows flexibility and extensibility of thesystem with more robustness.
In recent years translation has come to be investigated increasingly from critical perspectives, with various studies high-lighting the translator’s mediating involvement in the construction of particular discourses. This article describes an exploratory think-aloud protocol study which attempts to observe this mediating involvement in the translation process, using degree-level language students as subjects. The study suggests that the students may not be conscious of the fact that certain lexical choices in their translations conform to common stereotypes of Britain. The relatively low level of critical discourse awareness among the students may in some ways be related to translation practice in their educational contexts, which focuses on norm compliance and may do little in most cases to encourage reflection on the effects of translational choices on readers’ perceptions of what is communicated.
This thesis attempts to examine language contact phenomena in the speech of first-generation Japanese adults in Australia and New Zealand through the analysis of interview and other spoken data. The main objectives of the thesis are: (i) to describe and analyse the types of transference and integration phenomena identified in the corpus; (ii) to identify and analyse the types of strategies employed by first-generation Japanese speakers in Japanese-Australian/New Zealand English contact; (iii) to investigate the types of lexical transfers (i.e., loanwords) peculiar to the Australian/New Zealand environment; (iv) to investigate the factors affecting lexical transference. The basic assumption underlying the thesis is that there are principles which we may call strategies at work behind language contact phenomena and that these strategies (i.e., processing, monitoring, and social) affect contact processes such as transference and integration. These three types of strategies op~rate concurrently and generate rules for transference and integration under the influence of certain more general principles (i.e., maxims and determinants) prevailing in a given contact setting. In this thesis evidence is presented to show that interdialectal differences in the types of lexical transfers are attributable to differences in rules, strategies, maxims, and determinants operating in different bilingual communities. Various factors are involved in transference. It is observed that according to length of stay and type of stay, Japanese speakers employ different contact strategies. In the interview situation with a newcomer from Japan, migrants tend to suppress lexical transference while sojourners are likely to adopt it. The choice of contact strategies depends primarily on whether the speaker and the interlocutor share the same communicative norm. In a dynamic type of bilingual situation such as that found in the Japanese communities in Australia! New Zealand, the communicative norm is in a state of flux, and therefore the Japanese speakers in these communities employ a set of contact strategies which allow them to explore an ad hoc norm for communication with respect to lexical transference.
This paper describes an approach to treebank development which relies on the manual development of annotation tools. The overall process of tree annotation is described, and a special emphasis is put on the description of the last tool which has been built, i.e. a dependency-based robust chunk parser. The modularization of the parser and the central role of verbal subcategorization is presented. Some experimental results, carried
The Specific Affect Coding System (SPAFF; J. M. Gottman & L. J. Krokoff, 1989) has led to conclusions about which types of dyadic affect predict positive and negative outcomes in marriage, yet the lack of information about collinearity among the codes limits interpretation of SPAFF results. Psychometric properties of SPAFF were examined by assessing the interactions of 172 newlywed couples with SPAFF and with an affect rating system developed for this study. For husbands and wives, factor analysis indicated 4 distinct factors of affect, representing anger/contempt, sadness, anxiety, and humor/affection. Anger/contempt and humor/affection were associated with marital satisfaction, relationship beliefs, and appraisals of the interactions. Correlations were in the expected directions. The strengths, limitations, and implications of the data are discussed.
Treebank formats and associated software tools are proliferating rapidly, with little consideration for interoperability. We survey a wide variety of treebank structures and operations, and show how they can be mapped onto the annotation graph model, and leading to an integrated framework encompassing tree and non-tree annotations alike. This development opens up new possibilities for managing and exploiting multilayer annotations. 1.
The Penn Treebank encodes valuable information such as grammatical function, semantic roles, and identification of traces. The addition of such information was intended to facilitate the process of predicate-argument extraction. However, even with the enriched annotation this task is far from trivial and, to our knowledge, no complete set of predicate argument structures derived from the Treebank exists. Our paper describes a method for retrieving predicate-argument structures that circumvents the complexity of the tree structures in the corpus, while employing few template rules. Our system operates on a flattened, morphologically enriched version of the corpus. This flattened representation allows access to all levels of the tree simultaneously and thus enables the detection of the main sentence constituents by means of simple template rules. A small number of rules apply to identify the head words of each constituent and the latter fill in the constituent templates, to build the logical forms representative of the predicate argument structure. The system is robust in the face of incomplete syntactic coverage.
In this paper, the Arabic lexicon has been investigated in the context of relational database theory. A feature analysis of lexical entities has been carried out which shows that lexical attributes can be classified into five categories comprising nineteen attributes, including form attributes, morphological attributes, functional attributes, meaning attributes, and referential attributes. Based on this analysis, eleven database relations have been identified which form the backbone of an Arabic lexical database, including: words, roots, forms, infinitives, verbs, nouns, plurals, particles, meanings, lexical functions, and cross-references. The design ideas discussed in this paper were tested using a sample of lexical items selected from a modern printed dictionary. The results of developing an experimental lexical database indicate that the relational approach provides an efficient method for storing and retrieving Arabic lexical information. It should be mentioned, however, that several problems were encountered when the printed data was translated into a database form. Some of these problems are inherent in the Arabic lexicon itself, while others are due to the way by which lexical information is presented by paper-based dictionaries.
Lexical data resources are growing rapidely thanks to the Internet. Unfortunately, despite numerous existing standards like TEI, MARTIF, GENELEX, EAGLES/PAROLE, etc. each resource has its own format and own structure. Furthermore, the existing lexical data is generally developed for a specific purpose and can't be reused easily in other applications. In this paper, we intend to define a complete framework for developing multilingual lexical database for multipurpose. The framework is generic enough in order to accept a wide range of dictionary structures and proposes for manipulating heterogeneous dictionaries a set of common pointers into these structures. We will first present the organisation of Dictionary Markup Language (DML) framework. Then we will describe more precisely the DML language based on XML schemata. Next, we explain how to describe dictionary macro and microstructures with the DML. Lastly, we will explain our concept of common pointers defined in a Common Dictionary Markup (CDM) set.
Reviewed by: The lexicon-encyclopedia interface ed. by Bert Peeters Adam Głaz The lexicon-encyclopedia interface. Ed. by Bert Peeters. (Current research in the semantics/pragmatics interface 5.) Amsterdam: Elsevier, 2000. Pp. viii, 499. $105.00. A three-decade-old debate on the distinction between lexical and encyclopedic knowledge is reviewed by Bert Peeters (1–52) in an introductory article to the volume. The fourteen papers that follow (of which eleven are novel contributions and three are reprints with newly added afterwords) present to the reader a number of views on the issue, grouped into five parts. In Part 1, entitled ‘Assessments’, Anne Reboul (55–95) discusses James Pustejovsky’s model of the generative lexicon, concluding that it is more valuable for artificial than for natural cognition. Then, Carlos Inchaurralde (97–114) claims that connectionist models corroborate the idea of an integrated ‘lexicopedia’. In a similar vein, John R. Taylor (115–41) considers Ronald W. Langacker’s network model, which does not seek a distinction between the lexicon and encyclopedia, as more promising than Manfred Bierwisch’s two-level model. Part 2, ‘Understanding understanding’, begins with Pierre Larivée’s (145–67) analysis of utterance interpretation which leads the author to conclude that experience-based knowledge and linguistic meaning constitute distinct but interacting levels of representation. Next is Keith Allan’s (169–217) study of quantity implicatures, which are said to reside in the lexicon, encyclopedia being perhaps the site of nonlexical implicatures. Finally, William Croft (219–56) argues for an encyclopedic approach to semantics in interpreting metaphors and metonymies. In Part 3, ‘Words, words, words’, Richard Hudson and Jasper Holmes (259–90), adopting the framework of word grammar, offer a unified view of language as a knowledge network in which the lexicon and encyclopedia cannot be clearly distinguished. In contrast, Eva Born-Rauchenecker (291–316), in her study of Russian transitive action verbs, attempts to formulate a principled distinction between the two, necessary in lexicographic endeavors. This distinction is upheld, albeit in a different way, by M. Lynne Murphy (317–48), who investigates what speakers know about words as language items. Next, Heidi Harley and Rolf Noyer (349–74) show that distributed morphology has an advantage over government and binding models in that it replaces the lexicon with encyclopedia, allowing for a fuller analysis of nominalizations. The framework of distributed morphology is also adopted by Rob Pensalfini (393–431) in his inquiry into a number of phenomena found in Jingulu, a language of Northern Australia. His contribution is preceded by Joseph Hilferty’s (377–92) suggestion that an interactionist view of the relationship between grammar, the lexicon, and encyclopedia is more adequate than modular conceptions of language. These two papers constitute Part 4, entitled ‘Grammar’. Part 5, ‘Further afield’, starts with Susanne Feigenbaum’s (435–61) examination of the strategies adopted by students of a reading course, in whose performance world and linguistic knowledge interact. In the last contribution to the volume, Victor Raskin, Salvatore Attardo, and Donalee H. Attardo (463–86) discuss the nature of lexical knowledge in a lexical database called SMEARR. A part from its other assets (author, subject, and language indices; rich bibliographical references; careful editorship; and elegant typesetting), the volume is an extremely valuable publication due to the wide range of issues addressed and (often conflicting) solutions proposed. To the interested reader, it is an up-to-date presentation of the state of the art in the lexicon-encyclopedia interface. Many of the points it raises may also become thought-provoking stimuli for further research. Adam Głaz Maria Curie-Skłodowska University, Lublin Copyright © 2002 Linguistic Society of America
Among the resources developed in SI-TAL (Integrated Systems for the Automatic Treatment of Language), ItalWordNet (IWN) was built as reference semantic database, enlarging the Italian WordNet developed in the framework of the European project EuroWordNet (EWN). The Italian lexical database was increased, by introducing and codifying, besides the new grammatical categories of the adjectives and adverbs, a subset of proper names. In the IWN context, the subset of proper names represents a quantitatively limited portion, about 3600 synsets, but it may become a qualitatively important extension. The ever growing amount of non-structured information, stored in natural language, requires the availability of computational instruments able to manage this kind of information where proper names show a remarkable incidence in any types of texts. The work here presented falls in this context, taking into account the proper names, and is focussed on: i) encoding in the IWN database; ii) more typical uses in either proper or metaphorical and metonymic ways such as textual corpora evidence; iii) possibility of a well reasoned and structured enlarging of this data on the basis of the recent experience carried out in IWN. 1. Building the set of proper names in the IWN database IWN was first developed within the EWN1 project (Vossen, 1999) and then extended in the framework of an
Ambiguity resolution in the parsing of natural language requires a vast repository of knowledge to guide disambiguation. An effective approach to this problem is to use machine learning algorithms to acquire the needed knowledge and to extract generalizations about disambiguation decisions. Such parsing methods require a corpus-based approach with a collection of correct parses compiled by human experts. Current statistical parsing models suffer from sparse data problems, and experiments have indicated that more labeled data will improve performance. In this dissertation, we explore methods that attempt to combine human supervision with machine learning algorithms to try and extend accuracy beyond what is possible with the use of limited amounts of labeled data. In each case we do this by exposing a machine learning algorithm to unlabeled data in addition to the existing labeled data. Most recent research in parsing has shown the advantage of having a lexicalized model, where the word relationships mediate knowledge about disambiguation decisions. We use Lexicalized Tree Adjoining Grammars (TAGs) as the basis of our machine learning algorithm since they arise naturally from the lexicalization of Context Free Grammars (CFGs). We show in this dissertation that probability measures applied to TAGs retain the simplicity of probabilistic CFGs along with its elegant formal properties and that while PCFGs need additional independence assumptions to be useful in statistical parsing, no such changes need to be made to probabilistic TAGs. The main results presented in this dissertation are: (1) We extend the Co-Training algorithm (Yarowsky 1995; Blum and Mitchell 1998), a machine learning technique for combining labeled and unlabeled data previously used with classifiers with 2/3 labels to the more complex problem of statistical parsing. Using empirical results based on parsing the Wall Street Journal corpus we show that training a statistical parser on the combined labeled and unlabeled data strongly outperforms training only on the labeled data. (2) We present a machine learning algorithm that can be used to discover previously unknown subcategorization frames. The algorithm can then be used to label dependents of a verb in a treebank as either arguments or adjuncts. We use this algorithm to augment the Czech Dependency Treebank with argument/adjunct information. (3) We extend a supervised classifier for automatically identifying verb alternation classes for a set of verbs so that it can be used on minimally annotated data. Previous work (Merlo and Stevenson 2001) provided a classifier for this task that used automatically parsed text. With the use of learning of subcategorization frames we construct the same type of classifier which now requires text annotated with part-of-speech tags and phrasal chunks. In each of these results we use some existing linguistic resource that has been annotated by humans and add some further significant linguistic annotation by applying statistical machine learning algorithms.
We present a flexible approach for extracting hierarchical classifications from linguistic data. To this end, the framework of observational logic is introduced, which extends the logic that underlies standard Formal Concept Analysis by allowing disjunctive rules and exclusions. We give a rigorous mathematical characterization of how the chosen rule type affects the structure of the induced hierarchy. The framework is applied to the induction of hierarchical classifications from linguistic databases. The pros and cons of several types of hierarchies are discussed in detail with respect to criteria such as compactness of representation, suitability for inference tasks, and intelligibility for the human user.
This work describes how a statistical parser which was originally designed to parse English was adapted to parse Chinese with little human effort. The only modifications were a head table for Chinese and a set of automatically derived word features for dealing with unknown Chinese words. When trained on the Chinese Treebank, the resulting parser achieves state of art performance on parsing tokenized Chinese sentences.
This paper describes a national cooperational project STO, which has the aim of developing a large-scale Danish lexical database for computational use. We discuss some organisational aspects of the project and present the current project structure and main activities. Further, we discuss in more detail some of the linguistic issues that have required thorough consideration before a large-scale encoding could be initiated, encompassing topics such as the morphological encoding of compounds and proper names, as well as the syntactic encoding of there-constructions, phrasal verbs and reflexive verbs.
The aim of this work is the construction of a syntactically annotated treebank for Basque. In this paper we present first, the basis of the annotation. After examining several options we chose the scheme presented in (Carrol et al., 1998). It follows the EAGLES standards and it is based on the idea of adding to each sentence in the corpus a series of grammatical relations specifying the dependencies between modifiers and their nucleus. After the formalism has been presented, we will describe the problems we have found and the decisions we have taken to solve them. Next we present an example showing the application of the scheme to an initial corpus. Finally, we present the main conclusions about the applicability to Basque and future work.
We summarize here the results of a series of evaluations of the annotators’ assignments of tectogrammatical (i.e. underlying syntactic) tree structures and of the values of the edges as well as the values of the attribute representing the topic-focus articulation of the sentences, within the large-scale project of the Prague Dependency Treebank.
This paper describes the CTB Coreference Annotation Guidelines for annotating pronominal anaphoric expressions in the Penn Chinese Treebank. The goals of the annotation are: to provide training data for learning-based pronoun resolution tools, and to provide a "gold" standard to be used in the evaluation of pronoun resolution algorithms. The choices that were made concerning the coindexing of pronominal anaphors and their antecedents are discussed, as are some questions that arose in trying to categorize those pronominal expressions that did not refer to specific nominal entities in the text.