Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
Article choice can pose difficult problems in applications such as machine translation and automated summarization. In this paper, we investigate the use of corpus data to collect statistical generalizations about article use in English in order to be able to generate articles automatically to supplement a symbolic generator. We use data from the Penn Treebank as input to a memory-based learner (TiMBL 3.0; We discuss competitive results obtained using a variety of lexical, syntactic and semantic features that play an important role in automated article generation.
This study measures comparative lexical and syntactic closure rates in annotated Chinese newspaper corpora from the Academica Sinica Balanced Corpus and the University of Pennsylvania's Chinese Treebank. It then draws inferences as to how large such corpora need be to be representative models of subject-matter-constrained language domains within the same genre. Future large corpora should be built incrementally only by combining smaller representative sublanguage collections.
In this contribution we discuss how a fuzzy querying interface can support the generation of linguistic database summaries - a special technique of data mining. Links between our approach to linguistic summaries and the well-known technique of association rules is shown. The implementation of linguistic summaries generation using the authors’ FQUERY for Access package is presented.
We present a fuzzy logic based approach to the derivation of linguistic summaries of sets of data (databases), and show that it may be viewed as an example of a new paradigm shift from computing on numbers to computing on words that has been recently strongly advocated by Zadeh. We present an implementation of linguistic database summaries for sales data of a computer retailer that clearly shows that the new approach is viable and yields a new quality by providing human consistent results.
No language in the world is homogeneous, or ever will be. Whereas earlier forms of English were characterised by extreme variation on all levels and Middle English is in fact best described as a loose conglomerate of unstable varieties, we usually lack any more detailed insight into what functions this variation had for the individual speaker. The social correlates so well known from modern sociolinguistics, such as age, sex, education, religion, can normally not be applied to the existing texts, nor can even the geographical range of recorded forms be determined with any degree of certainty. Finally, if modern dialect or other non-standard features are contrasted with (as the term non-standard implies) an accepted standard form of a language, this method would necessarily fail with Middle English even if we knew more about it than we do and, in view of the state of surviving documents, ever will. It is safe to assume that for its speakers the linguistic heterogeneity of Middle English was ordered in some way, but it was so only for continually shifting speech communities, whose number and individual geographical spread we know very little about. The scene changed dramatically in the fifteenth century: the emergence of a new standard language began to re-institute a linguistic norm for written supraregional English. This development was a natural consequence of the acceptance of English in public domains, and was speeded up by the change-over to English as the Chancery language in 1430.
This paper presents results for a maximum-entropy-based part of speech tagger, which achieves superior performance principally by enriching the information sources used for tagging. In particular, we get improved results by incorporating these features: (i) more extensive treatment of capitalization for unknown words; (ii) features for the disambiguation of the tense forms of verbs; (iii) features for disambiguating particles from prepositions and adverbs. The best resulting accuracy for the tagger on the Penn Treebank is 96.86% overall, and 86.91% on previously unseen words.
The ambiguity related to the use of movement and localisation declension cases in Basque is a serious problem in the morphological generation phase in Machine Translation. We present the approach we have developed to solve this ambiguity. Information about the [±animate] semantic feature of the lemma we want to decline is necessary to choose the appropriate suffix. The lexical database used does not contain such information. Besides, it would be very hard to add it manually to the 28.000 substantives contained in it. For this reason, we made two experiments to obtain the [±animate] feature from other resources. First, our aim was to get automatically the required knowledge from corpora, but the results were not good. Secondly, after a minimal manual tagging and using semantic relations between words extracted from definitions of a monolingual dictionary, we tagged more than half of the words in real texts with the [±animate] feature.
In this paper, we present a method for comparing Lexicalized Tree Adjoining Grammars extracted from annotated corpora for three languages: English, Chinese and Korean. This method makes it possible to do a quantitative comparison between the syntactic structures of each language, thereby providing a way of testing the Universal Grammar Hypothesis, the foundation of modern linguistic theories.
This paper uses an information-based approach to conduct feature types selection for language modeling in a systematic manner.We describe a quantitative analysis of the information gain and the information redundancy for various combinations of feature types inspired by both dependency structure and bigram structure through analyzing an English treebank corpus and taking word prediction as the object.The experiments yield several conclusions on the predictive value of several feature types and feature types combinations for word prediction, which are expected to provide reliable reference for feature type selection in language modeling.
This article examines the reliance of U.S. campuses on international teaching assistants (ITA) for staffing undergraduate course and the strategies that may affect ratings of their speaking competence. This increasing reliance has led to student complaints about incomprehensibility of ITA. This problem has been examined by looking through the eyes of the students, administrators and taxpayers. Therefore, the responsibility had been placed on the ITA, whose burden it was to learn the language and culture more fully. The goal of having ITA learn the language and culture better was eclipsing another important issue that needs consideration, the issue of teaching assistant's feelings of loss of control over the students' perception about themselves.
Discusses the linguistic influences on an electronic publishing infrastructure in an environment with unstable linguistic standardization from the computational point of view. Essentially, in Serbia in the last half of the century (at least) publishing is based on the following facts: two alphabetic systems are regularly in use with the possibility to mix both alphabets in the same document; the various dialects are accepted as a part of a linguistic norm; orthography is unstable ‐ presently, several linguistic attitudes that have different views of the orthographic norm are under discussion; and, in Serbia, many minority languages are in use, which makes it difficult to provide efficient contact between different communities through electronic publishing. In this context, a systematic solution that responds to this complex situation has not been developed in the frame of traditional Serbian linguistics and lexicography in a way that enables the adequate incorporation of the new publishing technologies. Owing to these constraints, the direct application of electronic publishing tools frequently causes the degradation of the linguistic message. In such an environment, the promotion of electronic publishing therefore needs specific solutions. The paper discusses the general frame based on the specifically encoded system of electronic dictionaries that makes electronic texts independent of some of the mentioned constraints. The objective of such a frame is to enable the linguistic normalization of texts at the level of their internal representation, and to establish bridges for communicating with other language societies. Some aspects of electronic text representation that ensures its correct interpretation in different graphical systems and in different dialects are described. This also allows text indexing and retrieval using the same techniques that are available for languages not burdened with these problems.
本稿では, ツリーバンクを用いて入力文と類似した文の構文木から入力文に対する構文木を類推する手法を提案する. この手法は用例に基づく解析手法の1つであるが, 統計情報や意味的類似性ではなく, 複数のツリーバンク内データの問で定義される特定の類似関係に基づいて構文解析を行う. 特にここではツリーバンク内の知識表現形式をそのまま使って構文解析を行うため, 比較的容易に他の解析手法との融合を考えることができる. またこの手法は辞書などを用いず, データ間の類似性のみに基づいて解析を行うため, 未知語などを含む入力に対しても頑健に働く. ここでは特に基本原理として働く類似関係の有効性を評価するためにPenn Treebankを用いて評価実験を行った. その結果, 単語の表層情報と品詞情報を用いることで解析可能な文の約70%が一意に正しく解析でき, また誤ったものについても比較的正解に似た構文木を出力することができた.
En esta tesis se estudian las Gramaticas Incontextuales Probabilisticas y su aplicacion en problemas de Modelizacion del Lenguaje, Dos son los grandes problemas que se va a considerar en este tipo de modelos: el aprendizaje de las funciones de probabilidad asociadas a las reglas, y su integracion como modelo de interpretacion en tareas complejas de Modelizacion del Lenguaje. En primero de los problemas que se estudia es la estimacion de las funciones de probabilidad asociadas a las reglas.Se presentan y estudian dos de los algoritmos clasicos de estimacion de las GIP, el algoritmo Inside-Outside y el algoritmo basado en las cuentas de Viterbi. Despues se proponen nuevos algoritmos de estimacion en los cuales se utiliza un subconjunto especifico de derivaciones e cada cadena. Finalmente,los algoritmos propuestos se aplican al conjunto de datos del Penn Treebank para ilustrar su comportamiento en la practica. Por utlimo se aborda el problema de la interpretacion e integracion de las Gramaticas Incontextuales Probabilisticas en problemas de Modelizacion del Lenguaje. A continuacion se hace una propuesta de integracion que combina modelos de $n$-gramas a nivel de palabras con una Gramatica Incontextual Probabilistica a nivel de categorias lexicas.
Imaging work has begun to elucidate the spatial organization of emotions; the temporal organization, however, remains unclear. Adaptive behavior relies on rapid monitoring of potentially salient cues (typically with high emotional value) in the environment. To clarify the timing and speed of emotional processing in the two human brain hemispheres, event-related potentials (ERPs) were recorded during hemifield presentation of face images. ERPs were separately computed for disliked and liked faces, as individually assessed by postrecording affective ratings. After stimulation of either hemisphere, personal affective judgements of face images significantly modulated ERP responses at early stages, 80-116 ms after right hemisphere and 104-160 ms after left hemisphere stimulation. This is the first electrophysiological evidence for valence-dependent, automatic, i.e. pre-attentive emotional processing in humans.
Treebanks, such as the Penn Treebank, provide a basis for the automatic creation of broad coverage grammars. In the simplest case, rules can simply be ‘read off’ the parse-annotations of the corpus, producing either a simple or probabilistic context-free grammar. Such grammars, however, can be very large, presenting problems for the subsequent computational costs of parsing under the grammar. In this paper, we explore ways by which a treebank grammar can be reduced in size or ‘compacted’, which involve the use of two kinds of technique: (i) thresholding of rules by their number of occurrences; and (ii) a method of rule-parsing, which has both probabilistic and non-probabilistic variants. Our results show that by a combined use of these two techniques, a probabilistic context-free grammar can be reduced in size by 62% without any loss in parsing performance, and by 71% to give a gain in recall, but some loss in precision.
In this paper we develop a formalization of semantic relations that facilitates efficient implementations of relations in lexical databases or knowledge representation systems using bases. The formalization of relations is based on a modeling of hierarchical relations in Formal Concept Analysis. Further, relations are analyzed according to Relational Concept Analysis, which allows a representation of semantic relations consisting of relational components and quantificational tags. This representation utilizes mathematical properties of semantic relations. The quantificational tags imply inheritance rules among semantic relations that can be used to check the consistency of relations and to reduce the redundancy in implementations by storing only the basis elements of semantic relations. The research presented in this paper is an example of an application of Relational Concept Analysis to lexical databases and knowledge representation systems (cf. Priss 1996) which is part of a larger framework of research on natural language analysis and formalization.
This paper presents a pronoun resolution algorithm that adheres to the constraints and rules of Centering Theory (Grosz et al., 1995) and is an alternative to Brennan et al.'s 1987 algorithm. The advantages of this new model, the Left-Right Centering Algorithm (LRC), lie in its incremental processing of utterances and in its low computational overhead. The algorithm is compared with three other pronoun resolution methods: Hobbs' syntax-based algorithm, Strube's S-list approach, and the BFP Centering algorithm. All four methods were implemented in a system and tested on an annotated subset of the Treebank corpus consisting of 2026 pronouns. The noteworthy results were that Hobbs and LRC performed the best.
OBJECTIVE: To evaluate the efficacy of Simulated Presence, a personalized approach to enhance well-being among nursing home residents with Alzheimer's disease and related dementia's (ADRD). DESIGN: Latin-Square, double blinded, 3-factor design with restrictive randomization of three treatments (the study intervention, a placebo audio tape of a person reading the newspaper, and usual care). The three factors were treatment, time, and facility type. SETTING: Nine nursing homes in Eastern Massachusetts and Southern New Hampshire. PARTICIPANTS: Fifty-four subjects with documented ADRD who were aged 50 years or older, medically stable, had resided in their current nursing home for at least 3 months, and who had no planned discharge. All subjects had a history of agitated or withdrawn behaviors. INTERVENTION: The purpose of Simulated Presence is to provide a personalized intervention for persons with moderate to severe cognitive impairment. Through a unique testing process, some of the best loved memories of the ADRD person's lifetime are identified and then those memories are introduced to the patient in the format of a telephone conversation using a continuous play audio tape system. The intervention may be used for extended periods of time because each repetition is viewed as a fresh, live telephone call as a result of the short-term memory deficit of the person with ADRD. MEASUREMENTS: Direct observations of outcomes included using a newly developed scale, the Scale for the Observation of Agitation in Persons with Dementia, an agitation visual analog scale, the Positive Affect Rating Scale (mood and "interest"), a withdrawal visual analog scale, and facial diagrams of mood. Reported measures included daily staff observation logs of responses to interventions offered, and weekly staff surveys using the short-form Cohen-Mansfield Agitation Inventory and the Multidimensional Observation Scale for Elderly Subjects (mood and "interest"). Severity of dementia was assessed by the Mini-Mental State Exam, the Test for Severe Impairment, the Bedford Alzheimer's Nursing Scale, and the ADL Self-Performance Scale. RESULTS: Chi-square analysis of direct observations, using facial diagrams, revealed that Simulated Presence was equivalent to usual care (P =.141) and superior to placebo for producing a happy facial expression (P =.001). A positive effect was also documented in nursing staff observation logs using Analysis of Variance techniques (ANOVA) for subjects during Simulated Presence phases compared with the placebo phases (P <.001) and usual care phases (P <.001). According to ANOVA analyses of "interest" from weekly surveys, Simulated Presence was superior to both usual care (P =.001) and placebo (P =.008). We were unable to find evidence of significant differences (P <.05) among interventions for other direct observations and weekly reports of overall agitation or mood aspects of withdrawal. Subjects accepted the intervention most of the time, except for five subjects who refused it more than 50% of the time. CONCLUSION: This study provided evidence that Simulated Presence can be effective in enhancing well-being and decreasing problem behaviors in the nursing home setting as a substitute for or complement to usual care.
Research has shown that seeing another person (i.e., a model) perform at a certain level can influence the goal choice and performance of an observer. This study extended these findings by examining the model's effects on two potential explanatory mediators: expectations of reaching different performance levels and valence at those levels. The moderating effects of observer task experience and self‐esteem were also examined. In a repeated measures design, results showed that model performance influenced observers' goals, task performance, expectancies, and valence ratings. Path analyses indicated that expectancies mediated observational effects on goal choice. Results also indicated significant Model x Trial interactions, with a diminishing effect of model influence as personal experience developed. Results are discussed in terms of mediating effects of expectancies and valences on model influences on goal choice as well as the different weights given to social and personal information.
In recent work we have presented a formal framework for linguistic annotation based on labeled acyclic digraphs. These `annotation graphs' offer a simple yet powerful method for representing complex annotation structures incorporating hierarchy and overlap. Here, we motivate and illustrate our approach using discourse-level annotations of text and speech data drawn from the CALLHOME, COCONUT, MUC-7, DAMSL and TRAINS annotation schemes. With the help of domain specialists, we have constructed a hybrid multi-level annotation for a fragment of the Boston University Radio Speech Corpus which includes the following levels: segment, word, breath, ToBI, Tilt, Treebank, coreference and named entity. We show how annotation graphs can represent hybrid multi-level structures which derive from a diverse set of file formats. We also show how the approach facilitates substantive comparison of multiple annotations of a single signal based on different theoretical models. The discussion shows how annotation graphs open the door to wide-ranging integration of tools, formats and corpora.
This paper explores the role of lexicalization and prun-ing of grammars for base noun phrase identification. We modify our original framework (Cardie &amp; Pierce 1998) to extract lexicalized treebank grammars that assign a score to each potential noun phrase based upon both the part-of-speech tag sequence and the word sequence of the phrase. We evaluate the mod-ified framework on the “simple ” and “complex ” base NP corpora of the original study. As expected, we find that lexicalization dramatically improves the perfor-mance of the unpruned treebank grammars; however, for the simple base noun phrase data set, the lexical-ized grammar performs below the corresponding unlex-icalized but pruned grammar, suggesting that lexical-ization is not critical for recognizing very simple, rel-atively unambiguous constituents. Somewhat surpris-ingly, we also find that error-driven pruning improves the performance of the probabilistic, lexicalized base noun phrase grammars by up to 1.0 % recall and 0.4% precision, and does so even using the original pruning strategy that fails to distinguish the effects of lexical-ization. This result may have implications for many probabilistic grammar-based approaches to problems in natural language processing: error-driven pruning is a remarkably robust method for improving the perfor-mance of probabilistic and non-probabilistic grammars alike.
Research has shown that seeing another person (i.e., a model) perform at a certain level can influence the goal choice and performance of an observer. This study extended these findings by examining the model's effects on two potential explanatory mediators: expectations of reaching different performance levels and valence at those levels. The moderating effects of observer task experience and self‐esteem were also examined. In a repeated measures design, results showed that model performance influenced observers' goals, task performance, expectancies, and valence ratings. Path analyses indicated that expectancies mediated observational effects on goal choice. Results also indicated significant Model x Trial interactions, with a diminishing effect of model influence as personal experience developed. Results are discussed in terms of mediating effects of expectancies and valences on model influences on goal choice as well as the different weights given to social and personal information.
The act of healing essentially includes a spiritual or religious component, and language is often a prime medium for enacting it. Personal narratives are sites for the negotiation and construction of cultural and linguistic norms; healing stories recontextualize bodily struggles as social and spiritual conflicts. This paper examines a personal narrative of spiritual healing told in the language of Rastafari, the Jamaican religious movement. A discourse analysis of the narrative focuses on elements of Rasta Talk in order to discover how Rastafarian beliefs underlie and shape the telling, which is itself an act of faith and a profession of commitment. The healing itself, however, draws primarily on a variety of nonRasta spiritual and occult traditions of Jamaican folk culture; their relation to Rastafari, and the reasons for employing Rasta religious rhetoric in the narrative, are also explored. Rasta Talk, a register of Jamaican Creole (JC) undergoing functional expansion, is characteristically (though by no means exclusively) used by Jamaicans who follow the Rastafari religion. Rastafari is a syncretic Afro-Christian faith which invokes and reinterprets Old Testament Biblical imagery in the service of particular religious, cultural and political themes. This narrative of supernatural illness and cure applies a historical critique of colonialism and racism to the healthcare system, allows the teller to reposition himself discursively to alleviate suffering and stigma, and claims the moral high ground for AfroJamaican ethnomedical practices and traditional values through the enactment of Rastafarian principles. Rastafari is briefly introduced first in relation to other Jamaican faith traditions, and Rasta Talk is described. The narrative is outlined chronologically. Subsequent analysis links linguistic features to key elements of Rasta beliefs, which – together with elements of Jamaican folk medicine and culture – provide the necessary context to understand the healing narrative as an act of identity.
This study examined the effects of performance information from the ratee's coworker both on rater's perceptions of the information and ratings of the ratee's performance. As part of a managerial role-play exercise, subjects were required to rate an employee's performance. Audiotaped interactions of the ratee and customers were manipulated to reflect either good or poor performance. In addition, the ratee's coworker provided performance information which varied on (a) who was the initiator of the report and (b) the favorability of the information. Results revealed that while performance information from the ratee's coworker did not significantly affect ratings of performance when it was consistent with the rater's direct observations, ratings were affected when the information from a coworker was inconsistent with the rater's direct observations even though the information from a coworker was perceived as less accurate and of less use. Copyright © 1999 John Wiley & Sons, Ltd.
We present a new approach to partial parsing of natural language texts that relies on machine learning methods. The approach combines corpus-based grammar induction with a very simple pattern-matching algorithm and an optional constituent verification step. The grammar induction algorithm acquires a set of rules for each level of linguistic analysis using a new technique for errordriven pruning of treebank grammars. The constituent verification step employs standard inductive learning techniques as an additional precision-enhancing device. We evaluate the approach on four partial parsing data sets and find that performance is very good (over 93% precision and recall) for applications that require or prefer fairly simple constituent bracketing. As the complexity of the partial parsing task increases, however, our approach lags the performance of competing approaches. We explain these differences in terms of the knowledge sources employed by each method and describe a number of features...
This paper’ presents a novel methodology of resolving prepositional phrase attachment ambiguities. The approach consists of three phases. First, we rely on a publicly available database to classify a large corpus of prepositional attachments extracted from the Treebank parses. As a by-product, the arguments of every prepositional relation are semantically disambiguated. In the second phase, the thematic interpretation of the prepositional relations provides additional knowledge. The third phase is concerned with learning attachment decisions from word class knowledge and relation type features. The learning technique builds upon some of the most popular current statistical techniques. We have tested this methodology on (1) Wall Street Journal articles, (2) textual definitions of concepts from a dictionary and (3) an ad-hoc corpus of Web documents, used for conceptual indexing and information extraction.
This study tested anticipated affect as a potential strategy for reducing risky single-occasion drinking (RSOD). The hypothesis was that asking respondents to focus on their anticipated affect following RSOD would lead to higher ratings of negative affect than those obtained when asking respondents to focus on their feelings towards RSOD. In turn, these negative affect ratings were hypothesized as leading to safer behavioural estimates and reductions in RSOD. The study is based on a self-report questionnaire administered at two time points. At Time 1, measures of past drinking and demographic information were collected, along with affect ratings of drinking within safer single-occasion limits and affect ratings of RSOD (within-subjects condition). Time perspective was manipulated whereby the experimental group was asked to focus on affective reactions after RSOD and the control group to focus on affective reactions towards RSOD (between-subjects condition). Two weeks later, drinking behaviour was measured. The findings showed that the time perspective manipulation resulted in significantly higher negative affect ratings in the feeling after condition than in the feeling towards condition. Further, females reported lower negative affect than males. No other main or interaction effects were found. The time perspective manipulation, however, failed to produce safer behavioural estimates and RSOD reduction at follow-up. No significant differences were found between ratings of negative affect when drinking within safe limits as compared with ratings of affect when drinking above such limits. Despite greater negative affect 'after' rather than 'toward' the target behaviour, anticipated affect following RSOD did not yield safer behavioural estimates and subsequent drinking reduction at follow-up. These findings are interpreted in the context of risk perception associated with RSOD. The implications of this study for design of interventions aimed at reducing RSOD are discussed. In particular, ways of intensifying negative affect for RSOD are considered.
This paper describes a compound unit (CU) recognizer as a pattern‐based approach and its hybridization with rule‐based translation. A compound unit is a combined concept including collocations, idioms, and compound nouns. CU recognition reduces part of speech ambiguities by combining several words into a unit and consequently lessening the parsing load. It also provides pretranslated natural equivalents. Our focus in this paper is to obtain flexibility and efficiency from pattern‐based machine translation, and high‐quality translation by hybridization. A modified trie, our search index structure using “method” strategy is used to manage heterogeneous property of the constituents. Syntactic verification is integrated to obtain precise CU recognition by means of pruning wrongly recognized units that are caused by improper variable hypotheses. The experimental result with verification shows that the precision of CU recognition is increased to 99.69% with 31 CFG rules on the cyclic trie structure for 1,268 Wall Street Journal articles of the Penn Treebank. Another experiment with CU recognition also shows that it raises the understandability of translation for Web documents.
This paper proposed the chunk parsing scheme, a shallow syntactic knowledge representation system with the descriptive ability between part of speech serial and parse tree representation, and discussed some basic concepts and automatic identification algorithms for its two main parts: word boundary stems and constituent groups. Based on this scheme, this paper also proposed a new treebank annotation strategy, i.e. from chunk bank to treebank, and carried out several syntactic parsing and knowledge acquisition experiments, such as 1) the automatic identification of Chinese maximal noun phrases, 2) the automatic acquisition of Chinese probabilistic context free grammar knowledge. All these work shows its usefulness and efficiency for natural language processing research and development.
Electronic texts are claimed to exhibit features distinct from their more tangible cousins. The Snapshot project aims to observe and capture language usage in an electronic medium by creating an open corpus of World Wide Web documents. These documents are re-encoded using the TEI guidelines to create a flexible, persistent and portable data repository. This report gives an overview of the decisions made with respect to the re-encoding of HTML documents, and with the structuring the overall corpus.
In this paper, we propose an error correction method using text corpora. In this method, recognition errors are corrected using phonetically similar examples in the text corpora. The reliability of the correction hypotheses are judged according to their semantic consistency and their phonetic similarity to the original input. We previously proposed an error correction method that uses a treebank [1]. However, the previous method was not flexible in its use of examples, because structural mismatches occurred between the input and examples due to recognition errors. In our new proposal, examples are treated as morpheme sequences. This enables us to use examples partially when there are no useful full-sentence-examples. We built our proposed method into a speech translation system and compared the translation quality for simple translation and translation with error correction. The rate of acceptable translation increased about 10% with our proposed method compared to simple translation.
We tested new analytic procedures for combining an observer's image-ratings of lesion-likelihood with localization reports that are incomplete (unavailable on images rated as 'normal') and/or imprecise (possibly scored as 'correct' by chance), and for fitting a constrained ROC formulation to the rating data alone. Eight radiologist readers in a previous study had rated the likelihood of nodular lesions on each of 250 chest-film cases (39 with subtle nodules, 36 with 'typical' nodules and 175 normal cases) that were presented in two display modes (original films or on video workstation). Ratings in the four positive categories (2 to 5) were accompanied by reports that grossly localized the suspected nodules into one of 7 film- regions (upper, middle or lower portions of left or right lung field, or retrocardiac), but there was no localization for the cases rated as 'normal' (category 1). In each of 29 sets of data, we estimated the area below the ROC curve (A<SUB>z</SUB>) and its standard error using three different fits: (1) the usual ROC formulation, (2) the constrained ROC formulation and (3) the new procedure that included incomplete and imprecise localization data (I&I). Estimates of A<SUB>z</SUB> from the usual and constrained ROC fits were quite similar unless the standard ROC exhibited an upward 'hook,' but standard errors of A<SUB>z</SUB> were always the same or smaller for the constrained ROC fit. The I&I fit that included localization data often estimated A<SUB>z</SUB> to be either larger or smaller than the usual or constrained ROC fits that considered only the rating data, but its A<SUB>z</SUB> had substantially smaller standard errors in 28 of the 29 sets of observer data.
The digitization of library documents and archives increasingly extends to audiovisual (AV) document repositories. As a consequence, new computer-aided techniques are being devised, providing opportunities for new uses of AV documents. As scholars work mainly by reading, annotating, reusing, and producing documents they are directly concerned by these changes. The first part of this article describes AV document use in the humanities, as well as the current and future influence computers might have on evolving practices. After establishing that “full-indexing” (indexing of the content for random access to any segment of an AV document) is a necessary condition if scholars are to develop new practices in using AV material, we will focus on the specific problems raised by AV indexing as opposed to text indexing, followed by a discussion of related AV indexing projects as well as standardization issues. The third part will propose a representation model for the description of AV material (AI-Strata) and an exchange format of AV annotations (AEDI), based on a free segmentation approach. An example of annotation is also provided. The last part is devoted to a discussion regarding potential long-term influences of digital AV indexing techniques on scholarly uses of AV documents.
The use of the Inside-Outside (IO) algorithm for the estimation of the probability distributions of Stochastic ContextFree Grammars is characterized by the use of all the derivations in the learning process. However, its application in real tasks for Language Modeling is restricted due to the that it needs to converge. Alternatively, several estimations algorithms which consider a certain subset of derivations in the estimation process have been proposed elsewhere. This set of derivations can be chosen according to structural criteria, or by selecting the k-best derivations. These alternatives are studied in this paper, and they are tested on the corpus of the Wall Street Journal processed in the Penn Treebank project.
Thesauri have been widely used in bibliographic databases for 30 years. Recently, CD-ROMs of a variety of dictionaries with their GUIs are spreaded to current users. On the other hand, End users dose not use thesauri as for their bulky printed matter. The king of software browsing graphically and managing thesauri does not appear in PC environment.. The browsing tool for lexical database with hierarchical structure has been developed using Java.. This paper describes the functions, the components, and examples of its usage. The problems of the browser and the functions to be extended are discussed.
We examined male and female abservers' reactions to the use of touch by a nurse towards a patient in a hospital situation. The results suggest that an overrall pattern for observers to react more favorably to the nurse touch compared to no touch in interacting with patients may be judged in part by the attitudes of males and females about the use of touch. The generally favorable reaction to the use of touch by a nurse is consistent with Lewis et al's (1995) study. However, in contrast to their study, there were no differences in the nurse's supportiveness, competence, affect ratings and nurse's confidence in the low touch and high touch conditions.
This article discusses the principal problems of delimiting the scope of adverbs and their classification in terms of lexicology, lexicography and natural language processing, with a view to making the exhaustive lexical database for adverbs as well as to developing the description models which lead to the construction of the lexicon. Our discussion is not limited to the adverbs in traditional sense, but extended to the mono-/poly lexical words which can be considered as adverbials morphologically, syntactically and lexicologically. For instance, N + postposition(distributionally very limited), limited and/or productive inflectional forms of verbs and adjectives, and fixed forms of phrases and sentences which we consider as compound adverbs. To support our explanation, the morphological and syntactic properties of adverbs are discussed with typological and language universal point of view. Semantic classification is not included in this article because we think it can not be made rigorously without delimiting and analyzing possible various forms of adverbs(and adverbials) and their syntactic functions.
The traditions of grammatical terminology are quite different in the German and the French educational system. While the French ministry of education issues official regulations which are binding on a national level (they last did so in 1997), the federal system in Germany prevents the standardisation of the grammatical nomenclature for all the Länder of the FRG. Moreover, the historico-cultural background is different in both countries: the French public has been used to an active and centralist language policy for several centuries; in Germany governmental interference with the linguistic norm is often met with resistance, and the regulation of many details is in fact left to the important publishing houses. For this very reason it was left largely to the school-book publishing houses to decide how to put the recommendations of the German Secretaries of cultural affairs issued in 1982 into practice. The present article is based on a number of selected examples and provides a critical analysis of the usage of grammatical terminology in German and French school-books, and it pleads for a re-orientation – particularly in Germany: the inconsistencies and ad hoc solutions which can be observed quite frequently can only be overcome if the terminology is based on a well reflected linguistic theory. This is the only way to achieve an effect of synergy between grammar lessons in different school languages and ultimately a standardisation of the grammatical terminology on a European level.
Consciousness of the arbitrary nature of language, its essential conventionality (in the sense not of conformity, but of operating in accordance with tacitly agreed codes which have no naturally inherent laws) has become a major feature of twentieth-century poetry. It is almost, one is tempted to say, the defining feature of the modern, were it not that post-modern poetry is still more marked by it than classical modernism of the Pound-Eliot-Williams upheaval. By freeing poetry from the demands of consecutive syntax and claiming as its own the modern cinematic technique of juxtaposing images for primarily emotional/dramatic effect, modernism opened the way to further experiments in the breaking down of accepted assumptions governing verbal expression. Once the standard of 'correct', transparent English, whether spoken or, still more significantly, printed, was breached, it became possible to question the conventions of presentation which educated writers and readers had come to take as inviolable rules, and which are still treated as such in the language of scientific, journalistic and critical discourse. It became possible for transgression, or non-observance, of the rules to function on a positively sophisticated, rather than vulgarly negative, level — though this, of course, also presupposes general familiarity and conformity with them, since the abnormal effects of dispensing with them, or operating them in unfamiliar ways, depends on the existence of a strong feeling for them as the linguistic norm.
The Internet presents a potentially revolutionary tool in the dissemination of scientific information, offering many advantages to authors and audiences. However, this resource has been underutilized in psychological research because of several factors: unfamiliarity with required technology, lack of peer review, absence of an efficient centralized accessibility resource, concerns about copyright issues, and financial considerations. The present article describes the advantages of on-line presentation of research, as well as discusses various concerns about on-line publishing and the developing solutions to deal with those concerns.
Lexicon definition is one of the main bottlenecks in the development of new applications in the field of Information Extraction from text. Generic resources (e.g., lexical databases) are promising for reducing the cost of specific lexica definition, but they introduce lexical ambiguity. This paper proposes a methodology for building application-specific lexica by using WordNet. Lexical ambiguity is kept under control by marking synsets in WordNet with field labels taken from the Dewey Decimal Classification.