Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
with a preface by George Miller WordNet, an electronic lexical database, is considered to be the most important resource available to researchers in computational linguistics, text analysis, and many related areas. Its design is inspired by current psycholinguistic and computational theories of human lexical memory. English nouns, verbs, adjectives, and adverbs are organized into synonym sets, each representing one underlying lexicalized concept. Different relations link the synonym sets.The purpose of this volume is twofold. First, it discusses the design of WordNet and the theoretical motivations behind it. Second, it provides a survey of representative applications, including word sense identification, information retrieval, selectional preferences of verbs, and lexical chains.Contributors: Reem Al-Halimi, Robert C. Berwick, J. F. M. Burg, Martin Chodorow, Christiane Fellbaum, Joachim Grabowski, Sanda Harabagiu, Marti A. Hearst, Graeme Hirst, Douglas A. Jones, Rick Kazman, Karen T. Kohl, Shari Landes, Claudia Leacock, George A. Miller, Katherine J. Miller, Dan Moldovan, Naoyuki Nomura, Uta Priss, Philip Resnik, David St-Onge, Randee Tengi, Reind P. van de Riet, Ellen Voorhees.
Treebank is a very important language resource. Because much syntactic information should be tagged, we have to pay much annotation cost to create large-scale and high quality treebanks. This paper proposes a treebank development tool to speed up the construction and reduce the potential inconsistency. 1.
The WordNet lexical database is now quite large and offers broad coverage of general lexical relations in English. As is evident in this volume, WordNet has been employed as a resource for many applications in natural language processing (NLP) and information retrieval (IR). However, many potentially useful lexical relations are currently missing from WordNet. Some of these relations, while useful for NLP and IR applications, are not necessarily appropriate for a general, domain-independent lexical database. For example, WordNet’s coverage of proper nouns is rather sparse, but proper nouns are often very important in application tasks. The standard way lexicographers find new relations is to look through huge lists of concordance lines. However, culling through long lists of concordance lines can be a rather daunting task (Church and Hanks, 1990), so a method that picks out those lines that are very likely to hold relations of interest should be an improvement over more traditional techniques. This chapter describes a method for the automatic discovery of WordNetstyle lexico-semantic relations by searching for corresponding lexico-syntactic patterns in large text collections. Large text corpora are now widely available, and can be viewed as vast resources from which to mine lexical, syntactic, and semantic information. This idea is reminiscent of what is known as “data mining” in the artificial intelligence literature (Fayyad and Uthurusamy, 1996), however, in this case the ore is raw text rather than tables of numerical data. The Lexico-Syntactic Pattern Extraction (LSPE) method is meant to be useful as an automated or semi-automated aid for lexicographers and builders of domain-dependent knowledge-bases. The LSPE technique is light-weight; it does not require a knowledge base or complex interpretation modules in order to suggest new WordNet relations.
Machine learning techniques can be used to make lexicons adaptive. The main problems in adaptation are the addition of lexical material to an existing lexical database, and the recomputation of sublanguage-dependent lexical information when porting the lexicon to a new domain or application. Inductive lexicons combine available lexical information and corpus data to alleviate these tasks. In this paper, we introduce the general methodology for the construction of inductive lexicons, and discuss empirical results on a case study using the approach: prediction of the gender of nouns in Dutch. 1. Introduction In computational lexicography, lexicons of language engineering applications should come with acceptable lexical coverage, and with the information necessary for the intended applications. They should also come equipped with methods for the automatic extension and adaptation of the lexicon with new or modified lexical entries. Computational lexicology should therefore try to solve t...
Finding simple, non-recursive, base noun phrases is an important subtask for many natural language processing applications. While previous empirical methods for base NP identification have been rather complex, this paper instead proposes a very simple algorithm that is tailored to the relative simplicity of the task. In particular, we present a corpus-based approach for finding base NPs by matching part-of-speech tag sequences. The training phase of the algorithm is based on two successful techniques: first the base NP grammar is read from a "treebank" corpus; then the grammar is improved by selecting rules with high "benefit" scores. Using this simple algorithm with a naive heuristic for matching rules, we achieve surprising accuracy in an evaluation on the Penn Treebank Wall Street Journal.
Social externalism is a thesis about the individuation-conditions of thoughts. Actually, the thesis applies only to a special category of ‘trained’ thoughts, thoughts which issue from trained thinking. It isn't that the thinker of such a thought has to have had special training about the subject-matter. It is rather that he or she needs to have acquired certain basic linguistic skills and values. For trained thoughts are thoughts whose contents are tailored to the demands of communication. Social externalism, as I understand it, says that people who are competent in a public language are equipped to have certain thoughts whose contents are fixed (in part) by the lexical semantic norms of their language.
This paper describes applications of stochastic and symbolic NLP methods to treebank annotation. In particular we focus on (1) the automation of treebank annotation, (2) the comparison of conflicting annotations for the same sentence and (3) the automatic detection of inconsistencies. These techniques are currently employed for building a German treebank.
Although the influence of emotional arousal on declarative memory has been documented behaviorally, the mechanisms underlying arousal-memory interactions and their representation in the human brain remain uncertain. One route through which arousal achieves its effects on memory performance is by regulating consolidation processes. Animal research has revealed that the amygdala strengthens hippocampal-dependent memory consolidation in a limited time window following participation in an arousing task. To examine whether this integrative function of amygdalo-hippocampal structures extends to the human brain, we tested unilateral-temporallobectomy patients on an adaptation of a classic paradigm in which levels of physiological arousal at encoding modulate retention over time. Subjects rated emotionally arousing (taboo) and neutral words on an arousal scale while their skin conductance responses (SCRs) were monitored. Recall for the words was assessed immediately and after a 1-hr delay. Both temporal-lobectomy patients and control subjects generated enhanced SCRs and arousal ratings for the arousing words at the time of encoding. However, only control subjects exhibited an increase in memory for the arousing words over time. This group difference in the effect of arousal on the rate of forgetting suggests that the role of medial temporal lobe structures in memory consolidation for arousing events is conserved across species.
In many Oceanic languages in northwest Melanesia the default attribute construction ('a big house') is one whose morphosyntax looks like that of a possession construction: the attribute occupies the (possessed) head slot, the noun the (possessor) modifier slot ('a big one of a house'), that is, the opposite of the cross-linguistic norm and a rare phenomenon worldwide. I briefly describe these constructions, which are morphosyntactically varied, then examine their history, proposing that a major factor in their genesis was the presence in Proto-Oceanic of a small class of adjectival nouns whose reflexes in languages scattered across Oceania either may still behave as noun phrase heads or retain features reflecting this earlier status. The adjectival noun class had a small membership but high token frequency, and provided the template for a pattern extension that in a number of northwest Melanesian languages drew in the much larger adjectival verb class. I also address the question of why this change occurred in northwest Melanesia but not elsewhere in Oceania
Abstract The relationship between multiple stigmas and others' perceptions of the stigmas was investigated. The participants received 6 stigma targets; controllability of the onset of the stigma, race of the target, and gender of the target were manipulated. The participants' affective ratings differed on the basis of perceived controllability of the onset of the stigma. After receiving additional information about the targets, the participants assigned more blame to the Black stigma targets than to the White stigma targets.
Three experiments examined contributions of study phase awareness of word identity to subsequent word-identification priming by manipulating visual attention to words at study. In Experiment 1, word-identification priming was reduced for ignored relative to attended words, even though ignored words were identified sufficiently to produce negative priming in the study phase. Word-identification priming was also reduced after color naming relative to emotional valence rating (Experiment 2) or word reading (Experiment 3), even though an effect of emotional valence upon color naming (Experiment 2) indicated that words were identified at study. Thus, word-identification priming was reduced even when word identification occurred at study. Word-identification priming may depend on awareness of word identity at the time of study.
This experiment examined individual differences in emotional responsivity by recording the startle eyeblink reflex while 57 college students viewed affect-laden pictures and then rated their pleasantness. All participants first completed measures of affect intensity, alexithymia, and depression. Startleprobes were some-times presented at 120, 300, 800, or 4,500 ms after slide onset. By 300 ns, blinks elicited during negative slides were larger than those elicited during positive ones. Negative slides were also rated as more unpleasant. Moreover, all three personality variables moderated either the valence ratings, startle modification, or both. High-affect intensity was associated with diminished modulation of startle, but more extreme ratings. Alexithymia had no effect on the startle measure, but high-alexithymia participants did show more moderate ratings. Depressed participants exhibited accelerated (120 ms) modulation of startle. The results suggest the importance of measuring both physiological responses and subjective feelings in the study of individual differences in emotion.
Four studies bridged the areas of personality-mood and mood-cognition relations by investigating the effects of Extraversion and Neuroticism on the evaluation of affectively pleasant, unpleasant, and neutral word pairs. Specifically measured were affectivity ratings, categorization according to affect, judgments of associative strength, and response latencies. A strong, consistent cognitive bias toward affective as opposed to neutral stimuli was found across participants. Although some biases were systematically related to personality and mood, effects of individual differences were present only under specific conditions. The results are discussed in terms of a personality-mood framework and its implications for cognitive functioning. Personality traits influence the frequency and intensity of experienced
Punctuation has usually been ignored by researchers in computational linguistics over the years. Recently, it has been realized that a true understanding of written language will be impossible if punctuation marks are not taken into account. This paper contains the details of a computer-aided exercise to investigate English punctuation practice for the special case of comma (the most significant punctuation mark) in a parsed corpus. The study classifies the various "structural" uses of the comma according to the syntax-patterns in which a comma occurs. The corpus (Penn Treebank) consists of syntactically annotated sentences with no part-of-speech tag information about the individual words.
Part-of-speech tagging methodology has succeeded, but on problems that may lack real-world application. Redirection of the field is indicated, toward potentially more useful, but harder and more sophisticated tagging tasks: (1) using much more detailed tagsets (semantically and syntactically); (2) testing performance on treebanks reflecting the huge gamut of domains, etc., characterizing real-world applications; (3) understanding the magnitude of the unknown-word and unknown-tag problems, then overcoming them. Tagging results are presented on two versions of a new, highly variegated treebank, featuring tagsets of 2720 and 443 tags, respectively, and utilizing a dictionaryless, decision-tree tagger.
This paper discusses the relation for soaps between sensory attributes and both liking and image attributes. A clear relation emerges between sensory attribute level and liking, but no clear relation emerges between sensory attributes and image attributes. There are two possible conclusions to be drawn from these results. One conclusion is that consumers can validly assign ratings to the image attribute of a soap, but that there is no way to trace this image rating back to sensory inputs. This conclusion suggests that more research is needed to understand the meaning of image attributes. The second conclusion is that consumers cannot validly rate the image attributes of a soap, even though they can complete the questionnaire. This second conclusion implies that consumers can validly rate some attributes (e.g., sensory, liking), but not others (e.g., image), and that it may be misleading to collect and attempt to analyze image ratings for health and beauty aids products.
This paper compares different methods of generating intonation for an American English Text-to-Speech synthesis system. We look at a primarily rule-based approach and two data-driven approaches. For data-driven modeling we used two separate data sets, each representing a somewhat different prosodic style. One database was recordings of a portion of 1989 Wall Street Journal text from the Penn Treebank Project. The second database was recordings of interactive prompts used in telephone network services. Both were read by the same female speaker. Approximately two and one-half hours of speech was phonetically and prosodically segmented and labeled (first automatically, and subsequently verified manually). The prosodic labeling used ToBI [7] tones and breaks. Three different intonation models were compared: (1) a predominantly rule-based model based on ToBI labels [3]; (2) a parametric model using the Tilt approach [8]; and (3) a Vector Quantized model based on an underlying parametric re...
This pilot study examined African-American psychiatric patients' reactions to the Cultural Mistrust Inventory, a measure of blacks' mistrust of white society. Twenty-two black psychiatric patients were screened for the Culturally-Sensitive Diagnostic Interview Research Project. All patients were debriefed after the screening interview including queries about their reactions to the experience, whether they would be willing to participate in the next interview, and their reasons for participating or not. Patients' responses were recorded verbatim and were categorized in terms of their valence (positive, neutral, or negative) and affectivity (yes or no) by independent raters. Agreement between raters in terms of the valence of patients' reactions was very good, but it was poor to fair in terms of affectivity ratings. The majority of these black patients' responses were positive and nonaffective. Administration of the Cultural Mistrust Inventory to black psychiatric patients does not cause negative emotional reactions.
In “Response to Elliott and Valenza, 'And Then There Were None'”, (1996) Donald Foster has taken strenuous issue with our Shakespeare Clinic's final report, which concluded that none of the testable Shakespeare claimants, and none of the Shakespeare Apocrypha poems and plays – including Funeral Elegy by W.S. – match Shakespeare. Though he seems to accept most of our exclusions – notably excepting those of the Elegy and A Lover's Complaint – he believes that our methodology is nonetheless fatally flawed by “worthless figures ... wrong more often than right”, “rigorous cherry–picking”, “playing with a stacked deck”, and “conveniently exil[ing] ... inconvenient data.” He describes our tests as “foul vapor” and “methodological madness.”
This paper describes a method in determining syntactic structure for coordinate constructions. It is based on the information taken from semantic similarities, selectional restrictions, and some other linguistic cues. We discuss the role the information plays in resolving ambiguities that appear in coordinate constructions, describe the means of acquiring the necessary information automatically from two on-line corpora and a lexical database, and devise two algorithms for disambiguating coordinate constructions. An experiment that follows shows effectiveness of our method and its applicability to resolving ambiguities in some other syntactic structures.
We describe a system for extracting concepts from unstructured text. We do this by clustering document words and then assembling a structure which relates these words semantically. The clustering process identifies words which co--occur across a set of documents and creates groups of words which suggest a semantic context common across the document set. This context is formalized by identifying semantic relationships between the cluster words using a lexical database to build a Semantic Relationship Graph (SRG). This SRG is a directed graph which conveys a robust representation of the sub-- and super--class relationships between the correct word senses. We show how this process can be applied to a user--selected set of HTML documents; the SRGs can subsequently aid in searching the World Wide Web for documents which are similar. 1 Introduction Mining textual information presents challenges over data mining of relational or transaction databases because there are no predefined fields...
This paper is based on the following set of assumptions: (1) Ways of speaking characteristic of a given speech-community constitute a manifestation of a tacit system of `cultural rules', or, as the author calls them, `cultural scripts'; (2) to understand a society's ways of speaking, we have to identify and articulate its implicit `cultural scripts'; (3) to be able to do this without ethnocentric bias we need a universal, language-independent perspective; and (4) this can be attained if the `rules' in question are stated in terms of lexical universals, that is, universal human concepts lexicalized in all languages of the world. In various earlier publications Anna Wierzbicka tried to show how cultural scripts can be set out and justified with reference, in particular, to Japanese, Chinese, Polish, (White) Anglo-American, Black American and Anglo-Australian cultural norms. In this paper she applies the `cultural script' approach to German and compares German norms with `Anglo' norms (that is, norms prevailing in English-specking societies). Wierzbicka notes that in recent decades great changes have undoubtedly occurred in German ways of speaking and, it can be presumed, in underlying cultural values. For example, the dramatic spread of the use of the `familiar' form of address ( du, as opposed to Sie), and the decline in the use of titles (e.g. Herr Müller instead of Prof. Müller) point to significant changes in interpersonal relations, in the direction of more egalitarian informality. At the same time, she shows that evidence of contemporary public signs which are discussed here suggests that some traditional German values, like the value of social discipline and of Ordnung (order) based on legitimate authority, are far from obsolete. She also shows that in studying such values we can rely on concepts more precise and more illuminating than `authoritarianism' or `authoritarian personality', often used in the past in analyses of German culture and society, and that the `cultural scripts' approach offers a rigorous and efficient tool for studying change and variation, as well as continuity, in social attitudes and cultural values. Above all, the author argues that rather than perpetuating stereotypes based on prejudice and lack of understanding `cultural scripts' help outsiders grasp the `cultural logic' underlying unfamiliar ways of speaking which may otherwise look like a strange collection of idiosyncracies—or worse.
This paper is about two aspects of subcategorisation in NLP. First, it is about the automatic extraction of subcategorisation information from corpora. More specifically, we are concerned with unsupervised learning of subcategorisation information from tagged text by means of hierarchical clustering. The second aspect of the paper is the usage of this subcategorisation information for parsing, especially for the distinction between complements and adjuncts. We show that the information learned by unsupervised clustering can be exploited by a memory-based learner, to improve upon the complement-adjunct distinction. We compare the improvement gained by the use of this unsupervised information (1%) to that of different representations of subcategorisation information extracted from the tree-bank annotation (maximum 1.5%). The unsupervised information thus achieves two thirds of the improvement that can be obtained from the hand-crafted treebank information. 1 1 Introduction Subcategoris...
This article describes how an independent commercial academic publisher initiated its electronic publishing programme. It outlines the range of electronic activities under development and some of the issues addressed during the creation of electronic resources. Case studies of two early projects are included: a multimedia teaching too, A Right to Die? The Dax Cowart Case; and an SGML textbase, the Arden Shakespeare CD-ROM. In addition, the Routledge Encyclopedia of Philosophy is discussed as an example of the second generation of electronic projects at Routledge, highlighting lessons learned from previous projects and some of the issues relating to the production of a simultaneous print and electronic resource.
Contents: B.B. Kachru, Ladislav Zgusta: Illinois Years. - Bibliography of Publications by Ladislav Zgusta. - B.B. Kachru, Introduction. - I: Contextualizing Culture: D. Bartholomew, Otomi Culture from Dictionary Illustrative Sentences. - P. Corbin, Un Film, Deux Linguistes et Quelques Dictionnaires. Un Regard Particulier sur Simple Mortel de Pierre Jolivet. - J.A. Fishman, Dictionaries as Culturally Constructed and as Culture-Constructing: Reciprocity View as Seen from Yiddish Sources. - F.J. Hausmann, Allusions Litteraires et Citations Historiques Dans le Tresor de la Langue Francaise. - L.F. Lara, Towards a Theory of the Cultural Dictionary. - W.P. Lehmann, Spindle or the Distaff. - Y. Malkiel, Principal Categories of Learned Words. - E.A. Nida, Lexical Cosmetics. - II: Lexicography in Historical Context: F. Karttunen, Roots of Sixteenth-Century Mesoamerican Lexicography. - T.B.I. Creamer, Current State of Chinese Lexicography. - D.A. Kibbee, 'New Historiography', History of French and 'Le Bon Usage' in Nicot's Dictionary (1606). - N. Dinh-Hoa, On Chi-nam Ngoc-am Giai-nghia: An Early Chinese-Vietnamese Dictionary. - G. Stein, Chaucer and Lydgate in Palsgrave's Lesclarcissement. - III: Ideology, Norms and Language Use: M.A. Ezquerra,Political Considerations on Spanish Dictionaries. - D.M.T.Cr. Farina, Marrism and Soviet Lexicography. - C. Marello, Florence like Athens and Italian like Greek: An Ideologically Biased Theme in the Forewords of Some Italian Thesauri of the 19th Century. - A. Wierzbicka, Dictionaries and Ideologies: Three Examples from Eastern Europe. - R.D. Zorc, Philippine Regionalism versus Nationalism and the Lexicographer. - IV: Pluricentricity and Ethnocentricism: J. Algeo, British and American Biases in English Dictionaries. - C.W. Kim, One Language, Two Ideologies, and Two Dictionaries: Case of Korean. - B. Lewandowska-Tomaszczyk, Worldview and Verbal Senses. - C. Poirier, De la Soumission a la Prise de Parole: Le Cheminement de la Lexicographie au Quebec. - J. Whitcut, Taking it for Granted: Some Cultural Preconceptions in English Dictionaries. - V: Dictionaries across Languages and Cultures: Y. Kachru, Lexical Exponents of Cultural Contact: Speech Act Verbs in Hindi-English Dictionaries. - R.J. Steiner, Bilingual Dictionary in Cross-Cultural Contexts. - VI: Language Dynamics vs. Prescriptivism: A.P. Cowie, Learner's Dictionary in a Changing Cultural Perspective. - R.H. Gouws, Dictionaries and the Dynamics of Language Change. - F.E. Knowles, Dictionaries for the People or for People? - VII: Language Learner as the Consumer: G.M. Dalgish, Learners' Dictionaries: Keeping the Learner in Mind. - VIII: Structuring Semantics: F.F.M. Dolezal, Dictionary as Philosophy: Reconstructing the Meaning of Our Father. - M. Ritchie Key, Meaning as Derived from Word Formation in South American Indian Languages. - J.P. Louw, How Many Meanings to a Word? - IX: Ethical Issues and Lexicologists' Biases: D.L. Gold, When Religion Intrudes into Etymology (On The Word: Dictionary that Reveals the Hebrew Source of English). - T. McArthur, Culture-Bound and Trapped by Technology: Centuries of Bias in the Making of Wordbooks. - X: Terminology across Cultures: Z. Polacek, Amharic Lexicography and the Dynamics of Sociopolitical Terminology. - G.O. Richter, Grammatical Indications in Chinese Monolingual Dictionaries. - XI: Afterword: B.B. Kachru, Directions and Challenges.
This paper presents a method for developing limited-context grammar rules in order to mark up text automatically, by attaching specific text segments to a small number of well-defined and application-determined semantic categories. The Text Analysis Tool with Object Encoding (TATOE) was used in order to support the iterative process of developing a set of rules as well as for constructing and managing the lexical resources. The work reported here is part of a real-world application scenario: the automatic semantic mark up of German news messages, as provided by a German press agency, according to the SGML-based standard News Industry Text Format (NITF) to facilitate their further exchange. The implemented export mechanism of the semantic mark up into NITF is also described in the paper.
Video cameras provide a simple, noninvasive method for monitoring a subject’s eye movements. An important concept is that of the resolution of the system, which is the smallest eye movement that can be reliably detected. While hardware systems are available that estimate direction of gaze in real time from a video image of the pupil, such systems must limit image processing to attain real-time performance and are limited to a resolution of about 10 arc minutes. Two ways to improve resolution are discussed. The first is to improve the image processing algorithms that are used to derive an estimate. Offline analysis of the data can improve resolution by at least one order of magnitude for images of the pupil. A second avenue by which to improve resolution is to increase the optical gain of the imaging setup (i.e., the amount of image motion produced by a given eye rotation). Ophthalmoscopic imaging of retinal blood vessels provides increased optical gain and improved immunity to small head movements but requires a highly sensitive camera. The large number of images involved in a typical experiment imposes great demands on the storage, handling, and processing of data. A major bottleneck had been the real-time digitization and storage of large amounts of video imagery, but recent developments in video compression hardware have made this problem tractable at a reasonable cost. Images of both the retina and the pupil can be analyzed successfully using a basic toolbox of image-processing routines (filtering, correlation, thresholding, etc.), which are, for the most part, well suited to implementation on vectorizing supercomputers.
Summary A grammar-based probabilistic parser is described, and experimental results are presented for the parser as trained and tested on a 676,000-word, highly varied treebank of unrestricted English text. Probabilistic decision trees are utilized as a means of prediction, and a grammar with about 3000 semantic-and-syntactic tags, and 1100 non-terminal node labels suppl/es detailed lingzdstic information. Further such data is supplied for prediction purposes by thousands of questions about raw words, expres~ sions, and the sentence as a whole. The rich/n.formation base used for parse prediction allows the system to parse in a domain-general, totally--open-vocabulary setting, and to output highly-detailed semantic as well as syntactic information for sentences proccessed. Finally, a statistical procedure is described for converting less-detailed into more--detailed treebank, for use in increasing parser accuracy via much larger training treeb~.nlc~.
Multi-Media does not, just by itself, guarantee accelerated learning and enhanced motivation unless there is a clear pedagogical progression and learning strategy. The authors describe and analyze the didactic dimensions to be considered when designing a multi-media tool, based on their own experience as software authors and language trainers.
The authors here show that machine learning techniques can be used for designing an archaeological typology, at an early stage when the classes are not yet well defined. The program (LEGAL, LEarning with GAlois Lattice) is a machine learning system which uses a set of examples and counter-examples in order to discriminate between classes. Results show a good compatibility between the classes such as the yare defined by the system and the archaeological hypotheses.
Dictionary markup is one of the concerns of the Text Encoding Initiative (TEI), an international project for text encoding. In this paper, we investigate ways to use and extend the TEI encoding scheme for the markup of Korean dictionary entries. Since TEI suggestions for dictionary markup are mainly for western language dictionaries, we need to cope with problems to be encountered in encoding Korean dictionary entries. We try to extend and modify the TEI encoding scheme in the way suggested by the TEI. Also, we restrict the content model so that the encoded dictionary might be viewed as a database as well as a computerized, originally printed, dictionary.
The paper describes a morphological analyser forEstonian and how using a text corpus influenced theprocess of creating it and the resulting programitself. The influence is not limited to the lexicononly, but is also noticeable in the resulting algorithm andimplementation too. When work on the analyser began,there were no computational treatment of Estonianderivatives and compounds. After some cycles ofdevelopment and testing on the corpus, we came up withan acceptable algorithm for their treatment. Both themorphological analyser and the speller based on ithave been successfully marketed.
Research based on treebanks is ongoing for many natural language applications. However, the work involved in building a large-scale treebank is laborious and time-consuming. Thus, speeding up the process of building a treebank has become an important task. This paper proposes two versions of probabilistic chunkers to aid the development of a bracketed corpus. The basic version partitions part-of-speech sequences into chunk sequences, which form a partially bracketed corpus. Applying the chunking action recursively, the recursive version generates a fully bracketed corpus. Rather than using a treebank as a training corpus, a corpus, which is tagged with part-of-speech information only, is used. The experimental results show that the probabilistic chunker has a correct rate of more than 94% in producing a partially bracketed corpus and also gives very encouraging results in generating a fully bracketed corpus. These two versions of chunkers are simple but effective and can also be applied to many natural language applications.
In a study that used Hermans's (1987, 1988) valuation procedure, 40 participants each provided a highly valued experience of four types: science, religion, interpersonal and intrapersonal conflict between science and religion. They then rated these valuations on 30 affect terms, some of rich were later organized into categories: Positive, Negative, Self, and Other. Participants also filled out questionnaires, that were used to categorize them as low or high in scientific and religious orientation. Typical valuations of participants in these four science and religion categories are presented as qualitative idiographic information. In addition, quantitative analyses of affect ratings are presented as nomothetic information. Generally, affect ratings of scienfific experiences were more Self-directed while religious experience valuations involved equally high levels of Self and Other affect. Both scientific and religious experiences were evaluated as having Positive but not Negative affect. Interpersonal and intropersonal conflict were experienced as more Self-directed than Other-oriented. While interpersonal conflicts displayed more Negative affect than Positive, intrapersonal conflict was evaluated equally on these two measures.
This paper considers the choice of the medieval Aberdeen Bestiaryas the first project in Aberdeen University Library’s digitisationprogramme, and discusses some of the unusual features of themanuscript itself.
In Modern Japanese, three nominalizers, no, koto, and zero, are used in overlapping but slightly differing syntactic and semantic environments. This paper will discuss two types of strategy in nominalizer variation in Modern Japanese, koto versus no, on the one hand, and no versus zero, on the other. The two types of variation are shown to be very different in nature. The contrast between koto and no is primarily a semantic one characterizable in terms of greater versus lesser semantic specificity. Koto nominalization is the most semantically sensitive type of nominalization; it tends to be called for where its semantic contribution is expected, no matter how it is. No nominalization, which generally lacks semantic content, is the most unmarked nominalization strategy extensively used in Modern Japanese. The contrast between no and zero, on the other hand, is primarily a syntactic one with a diachronic dimension. In Old Japanese, which lacked the nominalizer no, a syntactic contrast existed between koto and zero. Zero nominalization, a remnant of Old Japanese, is now generally a nonproductive process limited to lexicalized expressions with varying degrees of archaism. No nominalization has long competed with zero nominalization and has successfully replaced it in many syntactic environments including some of the lexicalized expressions where zero was traditionally the norm. However, in spite of the overall dominance of no nominalization, zero nominalization has continued to be used fairly productively in some limited syntactic environments and is not expected to disappear from the syntactic inventory of Modern Japanese in the near future
Automatic Text Categorization (TC) is a complex and useful task for many natural language applications, and is usually performed through the use of a set of manually classified documents, a training collection. We suggest the utilization of additional resources like lexical databases to increase the amount of information that TC systems make use of, and thus, to improve their performance. Our approach integrates WordNet information with two training approaches through the Vector Space Model. The training approaches we test are the Rocchio (relevance feedback) and the Widrow-Hoff (machine learning) algorithms. Results obtained from evaluation show that the integration of WordNet clearly outperforms training approaches, and that an integrated technique can effectively address the classification of low frequency categories.
This study was designed to compare the effects of different kinds of visual presentations and music alone on university nonmusic students' affective and cognitive responses to music. Four groups of participants were presented with excerpts from the first and fourth movements of Beethoven's Symphony No. 6 in F major (“Pastoral”). Two groups heard music excerpts only, one interpretation conducted by Stowkowski and one by Bernstein. One of the video groups viewed corresponding excerpts from the movie Fantasia while listening to the Stowkowski recording. A second group viewed and listened to a performance video of the Vienna Philharmonic filmed during the Bernstein recording. All students (N = 128) completed cognitive listening tests based on the excerpts, rated the music on Likert-type affective scales, and responded to two open-ended questions. Significant effects of presentation condition were found. Cognitive scores were higher for the performance video than the music plus animation video on both movements. Scores for the two music-only presentations were not significantly different from each other or the two video presentations. Although affective ratings were not significantly different in magnitude between the presentation groups, the animation video (Fantasia.) presentation ranked consistently higher in affect than the other presentations. Implications of these results regarding the effects of different types of visual information presented to music listeners are discussed.
A treebank is a corpus of tagged and bracketed sentences capturing the linguistic properties of a (sub)language in an empirical way. The CASSANDRA treebank is developed as a sideline to the GALEN-IN-USE project in which it serves to make the relationships between natural language phenomena and semantic representations of medical expressions explicit, and to assist in the quality assurance of the modelling centres. The end result is a multilingual linguistic knowledge repository from which lexicons and grammars of various types can be derived in an automatic way.
All natural language processing systems (such as parsers, generators, taggers) need to have access to a lexicon about the words in the language. This thesis presents a lexicon architecture for natural language processing in Turkish. Given a query form consisting of a surface form and other features acting as restrictions, the lexicon produces feature structures containing morphosyntactic, syntactic, and semantic information for all possible interpretations of the surface form satisfying those restrictions. The lexicon is based on contemporary approaches like feature-based representation, inheritance, and unification. It makes use of two information sources: a morphological processor and a lexical database containing all the open and closed-class words of Turkish. The system has been implemented in SICStus Prolog as a standalone module for use in natural language processing applications.
Introduction to the special issue on computational linguistics using large corpora, Kenneth W. Church and Robert L. Mercer generalized probabilistic LR parsing of natural language (corpora) with unification-based grammars, Ted Briscoe and John Carroll accurate methods for the statistics of surprise and coincidence, Ted Dunning a program for aligning sentences in bilingual corpora, William A. Gale and Kenneth W. Church structural ambiguity and lexical relations, Donald Hindle and Mats Rooth text-translation alignment, Martin Kay and Martin Roescheisen retrieving collocations from text - Xtract, Frank Smadja using register-diversified corpora for general language studies, Douglas Biber from grammar to lexicon - unsupervised learning of lexical syntax, Michael R. Brent the mathematics of statistical machine translation - parameter estimation, Peter F. Brown et al building a large annotated corpus of English - the Penn treebank, Mitchell P. Marcus et al lexical semantic techniques for corpus analysis, James Pustejovsky et al coping with ambiguity and unknown words through probabilistic models, Ralph Weischedel et al.
This technical report is an appendix to Eisner (1996): it gives superior experimental results that were reported only in the talk version of that paper. Eisner (1996) trained three probability models on a small set of about 4,000 conjunction-free, dependency-grammar parses derived from the Wall Street Journal section of the Penn Treebank, and then evaluated the models on a held-out test set, using a novel O(n^3) parsing algorithm. The present paper describes some details of the experiments and repeats them with a larger training set of 25,000 sentences. As reported at the talk, the more extensive training yields greatly improved performance. Nearly half the sentences are parsed with no misattachments; two-thirds are parsed with at most one misattachment. Of the models described in the original written paper, the best score is still obtained with the generative (top-down) "model C." However, slightly better models are also explored, in particular, two variants on the comprehension (bottom-up) "model B." The better of these has an attachment accuracy of 90%, and (unlike model C) tags words more accurately than the comparable trigram tagger. Differences are statistically significant. If tags are roughly known in advance, search error is all but eliminated and the new model attains an attachment accuracy of 93%. We find that the parser of Collins (1996), when combined with a highly-trained tagger, also achieves 93% when trained and tested on the same sentences. Similarities and differences are discussed.
Psychophysiological changes during long-distance driving may be associated with driving fatigue and morbidity. Measures of stress and arousal, including heart rate, blood pressure, catecholamines, cortisol, state anxiety, and self-ratings of stress and arousal were collected from 10 long-distance bus drivers during 12-hour driving shifts and at matched times on nondriving rest days. Cardiovascular and catecholamine data were elevated across the entire work day, compared with rest days. Self-reported stress and state anxiety were elevated only at the preshift measure, and these elevations were interpreted as the result of anticipatory anxiety and additional work demands at the beginning of the shift. Decelerating activation from the 9th to the 12th hours of driving were reflected in slower heart rate and lower subjective arousal ratings. Suggested explanations for these findings are that drivers experience a release of tension when they anticipate the end of the shift and therefore deactivation is a signal or precursor to the onset of fatigue in physiological adjustment mechanisms.
Abstract Previous studies examining the association between social comparison processes and body image dissatisfaction have yielded inconsistent findings. This study examined whether such discrepancies are due to either the use of identical comparison targets for all subjects or variability in body mass. Specifically, 216 subjects were randomly assigned to one of three experimental conditions: self-generated upward comparison group, self-generated downward comparison group, or control group. Dependent variables were measures of body image. Results indicated that increasing body mass and trait comparison tendencies were associated with increased body dissatisfaction. However, the experimental manipulation did not affect body image ratings. Results suggest that social comparison processes may operate similarly over a range of body mass index (BMI) values.
After presenting a novel O(n^3) parsing algorithm for dependency grammar, we develop three contrasting ways to stochasticize it. We propose (a) a lexical affinity model where words struggle to modify each other, (b) a sense tagging model where words fluctuate randomly in their selectional preferences, and (c) a generative model where the speaker fleshes out each word's syntactic and conceptual structure without regard to the implications for the hearer. We also give preliminary empirical results from evaluating the three models' parsing performance on annotated Wall Street Journal training text (derived from the Penn Treebank). In these results, the generative (i.e., top-down) model performs significantly better than the others, and does about equally well at assigning part-of-speech tags.
The increasing availability of corpora annotated for linguistic structure prompts the question: if we have the same texts, annotated for phrase structure under two different schemes, to what extent do the annotations agree on structuring within the text? We suggest the term tree alignment to indicate the situation where two markup schemes choose to bracket off the same text elements. We propose a general method for determining agreement between two analyses. We then describe an efficient implementation, which is also modular in that the core of the implementation can be reused regardless of the format of markup used in the corpora. The output of the implementation on the Susanne and Penn treebank corpora is discussed.
Mexican Spanish is often described as a conservative variety whose distinguishing features go back only to the 19th century. This dissertation proposes that the origins of Mexican Spanish, at least with regard to verbal paradigm reduction, may be traced to the first century of the colonial period. My analysis is based on the collection of documents compiled in Documentos linguisticos de la Nueva Espana: Altiplano Central by Concepcion Company. The documents in this book are non-literary sources written during the colonial period. The following subjects were studied in this dissertation: language policy and Castilianization during the colonial period in Mexico, forms of address and the elimination of the second person plural, the -ra form and its function as an imperfect subjunctive and, finally, a historical analysis of the temporal values of the present perfect. It was found that the Mexican Spanish verbal paradigm, since the first century of the colonial period, has been characterized by the reduction of plural forms and a tendency to prefer the ending -ra form for past subjunctive, and a temporal value of an open past for the present perfect. However, there were two specific periods of time when this tendency was broken, at the beginning of the 17th and of the 19th century. At the same time, these two periods were characterized by a reassessment of the European values and imposition of its linguistic norms. This study confirms that, during the colonial period Mexican Spanish was caught between two strong opposing tendencies. One of these tendencies is the process of linguistic simplification, common in colonial territories; and the other is the process of monocentric standardization, by means of which the linguistic norm of Spain exerted pressure in the colonies.