Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18265 papers
We examine the problem of choosing word order for a set of dependency trees so as to minimize total dependency length. We present an algorithm for computing the optimal layout of a single tree as well as a numerical method for optimizing a grammar of orderings over a set of dependency types. A grammar generated by minimizing dependency length in unordered trees from the Penn Treebank is found to agree surprisingly well with English word order, suggesting that dependency length minimization has influenced the evolution of English. 1
ion over Parts of Speech. Lexical units are grouped into frames irrespective of their parts of speech. This allows to easily map, e. g., two text fragments onto each other that carry essentially the same meaning, but where one is headed by a verb and the other by a noun, such as ‘A bought B’ vs. ‘(the) acquisition of B by A’. In GermaNet, this mapping requires additional knowledge in the form of derivation relations (see above). Semantic Role Labelling. By semantic role labelling (‘frame elements’), syntactic variations are abstracted over: As such syntactic variants are mapped onto the same FrameNet representations, no additional relabelling mechanism is required. Frame-to-Frame Relations. Frame-to-frame relations, as recorded in the FrameNet database, list correspondences between frame elements. ‘A sold B to C.’ can be directly mapped onto ‘C bought B from A.’: The frames COMMERCE_SELL and COMMERCE_BUY are properly related, as are the participating frame elements BUYER, SELLER and GOODS. For every subtree for which a FrameNet representation can be found (based on the lemma of the node and the argument realisation), the corresponding FrameNet labels will be added: The name of the frame is added as a supplementary label to the node corresponding to the frame evoking element; the edges are labelled with the corresponding frame elements. Thus, the subtree is annotated as representing an instance of the respective frame. An example is shown in fig. 5.11. Note that frame structures are not fully disambiguated. Syntactic differences are used for disambiguation. For example, the reflexive use of a verb may be associated with a different frame than the intransitive one. In these cases, disambiguation is done. In other cases no disambiguation is performed, for example, where the correct frame can only be identified through sortal preferences on arguments. 5.2. THE LINGUISTIC KNOWLEDGE-BASE 213 Frame information provides an additional level of normalisation: syntactically different realisations, as, e. g., occasioned by dative shift, will receive the same FrameNet representation. For ‘John gave the book to Mary.’ and ‘John gave Mary the book.’, a GIVING frame with the same frame elements is derived. In particular, ‘Mary’ is identified as the RECIPIENT in both cases. Frame Relations as Sources of Inferences. The FrameNet lexical database not only defines frames as abstract semantic predicates and frame elements as abstract semantics role labels, but also a hierarchy based upon different frameto-frame relations defined both between frames and frame elements. We translate frame-to-frame relations directly into relabelling relations with corresponding relevance values. We currently use all available FrameNet frame-to-frame relations, except for the SEE_ALSO relation, even though some of them are only rather vaguely defined (cf. 4.3). It is therefore not always possible to foresee whether or not using a frame-to-frame relation will or will not result in a valid inference relation. For example, when two words evoke the same frame, this does not mean that they stand in the classical synonymy relation: ‘Good’ and ‘bad’ both evoke the DESIRABILITY frame, even though they would be considered antonyms in terms of classical lexical relations. We have decided to exploit all frame-to-frame relations as sources of inferences. From the definition of the relations (cf. 4.3), we considered that this would in most cases produce interesting, if possibly sometimes unlikely inferences. We considered, however, that the additional step of answer checking should detect and properly mark those cases. In our experiments, we did not observe any serious problems with this approach (cf. 7.2.2.3). This may to a large extent be due to the limited overall current coverage of the frame lexicon that we use (cf. 4.3.3). We expect clearer definitions of the relations to emerge together with growing coverage; at some point, it may turn out to be advisable to remove all but some core relations from consideration. This is how the relations are currently utilised for inferences: ‘Same Frame’. This is not strictly a frame-to-frame relation: Two subtrees labelled with the same frame (and frame elements) match during direct answer matching, simply because the labels are identical. No additional inference rules are required. Inheritance. We treat inheritance like a classical hyponymy relation, that is, we use it to introduce inferences in both directions (cf. 3.5.2.3). 214 CHAPTER 5. MATCHING STRUCTURED REPRESENTATIONS
French expressions like FRUIT DE MER 'seafood', MONTER UN BATEAU [a qqn] 'to fool someone' and A PROPOS [de qqch] 'about something' are idioms. Set phrases like these are considered in Meaning-Text Theory (MTT) as lexical units and are therefore described by an autonomous lexical entry in the dictionary. However, since each phrase consists of more than one word-form, these lexical units display complex behavior which has not yet been rigorously described in the Meaning-Text lexicography. In this paper, we present a MTT treatment of idioms focusing on their presentation in the dictionary, and propose tools for constructing a lexical database of idioms, which can efficiently represent form flexibility and combinatorial particularities of French idioms.
Criteria for Manual Clustering of Verb Senses Cecily Jill Duffield, Jena D. Hwang, Susan Windisch Brown, Dmitriy Dligach, Sarah E. Vieweg, Jenny Davis, Martha Palmer ({cecily.duffield, hwangd, susan.brown, dmitriy.dligach, sarah.vieweg, jennifer.davis, martha.palmer}@colorado.edu) Departments of Linguistics and Computer Science University of Colorado Boulder, C0 80039-0295, USA Key words: word sense disambiguation, annotation, inter-annotator agreement, syntactic/semantic features Introduction Word sense ambiguity poses significant obstacles to accurate and efficient information extraction and automatic translation. Successful disambiguation of polysemous words in NLP applications depends on determining an appropriate level of granularity of sense distinctions, especially for verbs. WordNet, an important and widely used lexical resource, uses fine-grained distinctions that provide subtle information about the particular usages of various lexical items (Felbaum, 1998). When used as a resource for annotation of various genres of text, this fine level of granularity has not been conducive to high rates of inter-annotator agreement (ITA) or high automatic tagging performance. Annotation of verb senses as described by coarse-grained Proposition Bank framesets may result in higher ITA scores, but the blurring of distinctions between verb senses with similar argument structures may fail to alleviate the problems posed by ambiguity. Our goal in this project is to create verb sense distinctions at a middle level of granularity that allow us to capture as much information as possible from a lexical item while still attaining high ITA scores and high system performance in automatic sense disambiguation. We have demonstrated that clear sense distinctions improve annotator productivity and accuracy, which results in a corresponding improvement in system performance. Training on this new data, Chen, Schein, Ungar and Palmer, (2006) report 86.7% accuracy for verbs using a smoothed maximum entropy model and rich linguistic features (just over 70% for fine-grained senses). This paper focuses on the methodology used to create the sense groupings, with a particular emphasis on the types of features that are most accessible to human annotators who are not linguists. Various criteria are considered when disambiguating senses and creating sense groupings for the verbs, including frequent lexical usages and collocations, syntactic features and alternations, and semantic features, similarly to the groupings for Senseval2 (Palmer, Dang & Felbaum, 2007). Our highest priority is to create clear distinctions among sense groupings that will be easily understood by the annotators and consequently result in high rates of inter- annotator agreement. We have found that the most successful approach is to cluster senses intuitively on a verb-by-verb basis, distinguishing sense groupings with features that are easily grasped by all annotators. Such features include specific domain usages, as in legal, financial, and social uses (distinguishing two senses of integrate in, “Over two-thirds of the teachers report they integrated the arts into their subjects,” and “The movie is set in 1971, when TC Williams High integrated blacks and whites.”); a specific syntactic construction, such as a required locative prepositional phrase (distinguishing the sense of open in, “The master bedroom opens to a large terrace,” from that in “The door won't open.”); and the features of nominal arguments, such as an agentive subject (separating senses of indicate in “These symptoms indicate a serious illness,” and “He indicated the right road by nodding towards it.”) More theoretical features for distinguishing groupings have proven to be less successful. Annotators not familiar with linguistics were confused by concrete/abstract distinctions and such aspectual features as continuative or stative. Therefore, they are now rarely used to label sense groupings. Such concepts, when used, are more likely to be described in prose commentary. Certain compositional features of verbs such as manner and path have also proven to be confusing for annotators, and resulted in decreased annotator agreement. Verb sense groupings that do not receive high ITA scores in initial rounds of annotation are revised, often prioritizing the use of the more successful features illustrated above with examples. Acknowledgements We gratefully acknowledge the support of the National Science Foundation Grant NSF-0415923, Word Sense Disambiguation, and the Defense Advanced Research Projects Agency GALE program, Contract No. HR0011-06- C-0022, a subcontract from the BBN-AGILE Team. References Chen, J., Schein, A., Ungar, L., & Palmer, M. (2006). An empirical study of the behavior of word sense disambiguation. Proceedings of NAACL-HLT 2006. New York City, NY. Fellbaum, C. (Ed.). (1998). WordNet: An on-line lexical database and some of its applications. MIT Press, Cambridge, MA. Palmer, M., Dang, H.T., and Fellbaum, C. (to appear, 2007). Making fine-grained and coarse-grained sense distinctions, both manually and automatically. Journal of Natural Language Engineering
This study explores the question of whether native and nonnative listeners, some familiar with the language and some not, differ in their accent ratings of native speakers (NSs) and nonnative speakers (NNSs). Although a few studies have employed native and nonnative judges to evaluate native and nonnative speech, the present study is perhaps the first to include nonnative judges who were unfamiliar with the language they were judging. This study compares accent ratings of four different groups listening to Brazilian Portuguese spoken by NSs and NNSs. The 25 speakers of Portuguese were 5 NSs and 20 NNSs who were NSs of American English. The four listening groups were (a) Brazilian Portuguese listeners, (b) American English listeners with Portuguese experience, (c) American English listeners without Portuguese experience, and (d) English as a second language listeners without Portuguese experience. Although there were some differences, the results show striking similarities across groups, suggesting that listeners' first and second languages do not strongly affect ratings of foreign accents. The study also has implications for the critical period hypothesis because it raises the possibility that there are salient universal features of nonnative speech.
We present a novel method for evaluating the output of Machine Translation (MT), based on comparing the dependency structures of the translation and reference rather than their surface string forms. Our method uses a treebank-based, widecoverage, probabilistic Lexical-Functional Grammar (LFG) parser to produce a set of structural dependencies for each translation-reference sentence pair, and then calculates the precision and recall for these dependencies. Our dependency-based evaluation, in contrast to most popular string-based evaluation metrics, will not unfairly penalize perfectly valid syntactic variations in the translation. In addition to allowing for legitimate syntactic differences, we use paraphrases in the evaluation process to account for lexical variation. In comparison with other metrics on 16,800 sentences of Chinese-English newswire text, our method reaches high correlation with human scores. An experiment with two translations of 4,000 sentences from Spanish-English Europarl shows that, in contrast to most other metrics, our method does not display a high bias towards statistical models of translation.
The task we are investigating is unsupervised learning of natural language morphology for inflectional languages. The target morphological grammar consists of a lexicon of morphological base forms and transforms. A base form represents all inflections of a lexeme, and all base forms of the same POS category share the same fine-grained morphosyntactic type. Transforms are morphophonemic rewrite rules that convert base forms to derived forms, and whose context of application is limited to a specific set of base forms. We have developed a greedy algorithm to induce such a grammar. At each iteration, suffixal transforms to convert between base and derived forms of lexemes are hypothesized. The algorithm chooses the transform that maximizes vocabulary coverage, while minimizing the number of conflicts resulting from proposing as base forms words previously found to be derived forms. After base forms and transforms have been learned, a distributional clustering step assigns the base forms to POS classes. In future work, the transforms will be converted to generalized rewrite rules by inducing phonological characteristics common to the base forms. We have tested this algorithm on a version of the Penn Treebank annotated for inflectional morphology. The algorithm achieves 71.7% recall and 92.9% precision on inflectional relations, where both a base and derived form occur in the corpus. We are currently testing the algorithm on other languages, and will present results on the Morphochallenge gold standards.
Interactionists interested in second language acquisition postulate that learners’ competences are sensitive to the context in which they are put into play. Here we explore the language practices displayed, in a bilingual socio-educational milieu, by three dyads of English learners while carrying out oral communicative pair-work. In particular, we examine the role language choice plays in each task. A first analysis of our data indicates that the learners’ language choices seem to reveal the linguistic norms operating in the community of practice they belong to. A second analysis reveals that they exploited their linguistic repertoires according to their interpretation of the task and to their willingness to complete it in English. Thus, in the first two tasks students relied on code-switching as a mechanism to solve communication failures, whereas the third task generated the use of a mixed repertoire as a means to complete the task in the target language.
An increased interest in body image (more specifically, becoming or staying thin) is a common trend that is increasing among children and adolescents. Previous research shows that the interest to become or remain thin peaks during early adolescence, particularly among females. PURPOSE: Because research on this topic is limited in younger populations, the purpose of this study was to examine the interest in weight control for a population of preadolescent children. METHODS: Subjects included 261 (122 female and 139 male) children from third (n=92), fourth (n=80), and fifth (n=89) grades. Average age of the participants was 9.5 years. The primary investigator met one-on-one with each child. Height was measured to the nearest centimeter and mass to the nearest 1/2 kilogram. Each child was asked if they judged themselves to be overweight (fat), underweight (skinny) or in-between. Also, each child was asked if they would like to lose weight, gain weight, or stay the same. Categorical data were separated through cross tabulation and significant differences were assessed with Chi-Square analyses. RESULTS: Average body mass index (BMI) for girls was 18.2 (just below the 75th percentile for age and gender) and for boys was 18.7 (the 75th percentile for age and gender). Thirty nine percent of all participants wanted to lose weight or remain thin. For boys and girls respectively, significant percentages 23.7% and 30.3% wanted to loose weight. Moreover, 33.8% and 43.4% of boys and girls respectively wanted to remain or become thin in spite of “in-between” self image ratings and normal BMIs (X2=19.742, p=0.001 and X2=16.418, p=0.003 for boys and girls respectively). No significant differences were noted between grades 3, 4, and 5 for boys or girls. CONCLUSIONS: The results of this study suggest that preadolescent children are focusing on body image and specifically on wanting to become or remain thin. This is especially of interest as these children had normal BMIs and generally viewed themselves as having “in-between” body images. The data also suggest that while interest in body image may peak in early adolescence, it clearly begins for both boys and girls in early prepubescent years.
The percentage of pupils fluent in both Swedish and Finnish in Swedish-speaking schools in Finland has grown and now includes a third of all pupils. This article focuses on the linguistic classroom discourse in a Swedish-speaking school in a strongly Finnish-dominated area. The purpose is to increase the understanding of how linguistic norms are maintained, and what these norms imply for the participation of bilingual pupils in classroom interaction. Videotaped lessons were analyzed by means of conversation analysis focusing on the interaction between the teacher and the bilingual pupils as well as on language-related sequences. The results show that the bilingual pupils cannot be regarded as victims of a language policy governed from above, but that they actively contribute to the construction and maintenance of a monolingual norm in the classroom. When using Finnish, they at the same time point at the “other-languageness” of the code-switched words. Monolingualism as a norm means a limitation restraining the pupils with gaps in their Swedish from participating with full competence in the classroom conversation. Between pupils, occasional Finnish words are used in an unproblematic manner. Code-switching in these cases works as a means to keep one's position on the conversational floor. By violating the monolingual norm, pupils can lodge a protest against the agenda of the teacher.
In the present chapter, my aim is to use the results of the previous two chapters to argue for three anti-individualistic doctrines in the philosophy of mind and language. These doctrines express anti-individualistic theses regarding speech content, linguistic meaning, and mental content/attitude individuation. The arguments themselves all share their basic structure: appealing to a thought experiment in which we vary the public linguistic norms in play while leaving intact all individualistic facts regarding the participants in a speech exchange, it is argued that these three properties – the content of the speech, the linguistic meanings of the expressions used, and the contents of the beliefs acquired in the exchange – will vary in a way reflecting the change in public linguistic norms. The result, of course, will be that the instantiation of the determinate properties in question is not fixed by the individualistic facts regarding the participants in a speech exchange. The facts regarding speech content etc. do not supervene on the set of individualistic facts regarding the speaker and hearer, respectively. (Putting the point in terms of supervenience connects our result with the standard way of formulating anti-individualistic doctrines.)
BACKGROUND: Approved for treatment of treatment-resistant depression and for epilepsy, vagus nerve stimulation (VNS) therapy involves stimulation of the vagus nerve, affecting both mood and appetite regulating systems. VNS is associated with changes in food intake and weight loss in animals. Studies of its impact on food intake and weight with humans are limited. It is not known whether or how VNS influences emotional response to food, but vagus afferents project to regions in the insula involving satiety and taste. METHOD: Thirty-three participants were recruited for three groups: depressed patients undergoing VNS therapy, depressed patients not undergoing VNS therapy, and healthy controls. All participants viewed images of foods twice in random order. When applicable, VNS devices were turned on for one viewing and off for the other. Participants were instructed to rate immediately after the viewings how each picture made them feel on a visual analog on three dimensions (unhappy to happy, calm to aroused, and small/submissive to big/domineering). RESULTS: Controlling for time since last meal, a significant main effect was found for arousal ratings in response to sweet food images. Post-hoc analyses indicated that the VNS group demonstrated significant changes in arousal ratings between paired food image viewings compared to controls. Sixty-four percent of VNS participants demonstrated increases and 36% demonstrated decreases in arousal. Higher body mass indexes and greater levels of self-reported sweet cravings were associated with increased arousal during VNS activation. CONCLUSIONS: This study was the first to examine the effects of acute left cervical VNS on emotional ratings of food in adults with major depression. Results suggest that VNS device activation may be associated with acute alteration in arousal response to sweet foods among depressed patients. Future research is needed to replicate these findings and to assess how activation of the vagus nerve affects eating and weight.
Reinforcing value of a behavior refers to the motivation to engage in the behavior. A reinforcing behavior will support more work to obtain the behavior. Individual differences in the reinforcing value of physical activity predict the usual physical activity in children. Another factor that may influence physical activity is liking of physical activity. Liking or hedonics refers to an affective rating associated with the behavior, and people are more likely to engage in physical activities that they like than ones that they do not like. Liking correlates with physical activity in youth. Although the independence of reinforcing value and liking of physical activity has not yet been tested, the motivation to gain access to a behavior and liking for that behavior are likely different constructs. PURPOSE: To determine whether liking and relative reinforcing value (RRV) of physical activity independently predict time youth spend in moderate-to-vigorous physical activity (MVPA). METHODS: Boys (n = 21) and girls (n = 15) age 8 to 12 years were measured for height, weight, aerobic fitness, liking and RRV of physical activity, and minutes in MVPA using accelerometers. RESULTS: Using multiple regression to control for individual differences in age, sex, BMI percentile, aerobic fitness, and time the accelerometer was worn, liking (P < 0.05) and RRV (P < 0.01) of physical activity independently predicted time in MVPA. When using median splits of the RRV and liking data to form subject groups, the group of children with both a high liking and RRV of physical activity participated in greater (P < 0.05) minutes per week of MVPA (1340 + 72 min) than groups with high RRV-low liking (1040 + 95 min), low RRV-high liking (978 + 89 min), or low RRV-low liking (1007 + 70 min) of physical activity. CONCLUSIONS: RRV and liking of MVPA are separate constructs as they independently predict MVPA of children. Those children who find physical activity the most reinforcing and also have a high liking of physical activity engage in 33% more MVPA than children who either find physical activity highly reinforcing or have a high liking of physical activity. Interventions that concurrently increase the reinforcing value and liking of physical activity may be the most effective for increasing youth participation in free-living MVPA. Supported by NIH Grant RO1 HD42766.
This paper investigates how the use of machine learning techniques can significantly predict the three major dimensions of learner-s emotions (pleasure, arousal and dominance) from brainwaves. This study has adopted an experimentation in which participants were exposed to a set of pictures from the International Affective Picture System (IAPS) while their electrical brain activity was recorded with an electroencephalogram (EEG). The pictures were already rated in a previous study via the affective rating system Self-Assessment Manikin (SAM) to assess the three dimensions of pleasure, arousal, and dominance. For each picture, we took the mean of these values for all subjects used in this previous study and associated them to the recorded brainwaves of the participants in our study. Correlation and regression analyses confirmed the hypothesis that brainwave measures could significantly predict emotional dimensions. This can be very useful in the case of impassive, taciturn or disabled learners. Standard classification techniques were used to assess the reliability of the automatic detection of learners- three major dimensions from the brainwaves. We discuss the results and the pertinence of such a method to assess learner-s emotions and integrate it into a brainwavesensing Intelligent Tutoring System.
Emotions are an important factor in human-computer interaction. One of the challenges in building emotionally intelligent systems is the automatic recognition of affective states. We are developing and evaluating a method for measuring user affect that incorporates psychological, behavioral, and physiological measures. During affective stimulation, breathing parameters, skin conductance level (SCL) and corrugator EMG activity correlate with self-reported levels of valence and arousal. Valence, at the level of subjective experience, summarizes how well one is doing, whereas arousal refers to a sense of energy. A stimulus activates appetitive or defensive motivation (the valence dimension) with some degree of energy mobilization (the arousal dimension). In the laboratory, moods are induced using different procedures. Only few studies have investigated the critical question of how long induced moods actually last. Further, knowledge concerning the persistence of physiological responding, when the stimuli are withdrawn, remains spare. The goals of this study were first, to assess the somato-physiological activity during affective stimulation (film clip viewing) and its relation to valence and arousal, and second, to determine if the response patterns persist, dissipate or otherwise change when the stimuli are withdrawn and the subjects perform a computer task. Seventy-six participants viewed a neutral film clip (an educational program) and completed immediately afterward the task (control condition). Then, they viewed either a positive high-arousal clip (sport scenes), a positive low-arousal clip (takes from landscapes), a negative high-arousal clip (scene depicting captives forced to play Russian roulette) or a negative low-arousal clip (a documentary about a slum in Belgium) and completed the task a second time (experimental condition). The task required participants to shop on an e-commerce website for office supplies. After each clip and each task, the participants rated their current mood. We tested valence and arousal effects during the last 90 s of the films, the first and last 90 s of the task of the experimental condition. Viewing of the selected film clips resulted in increasing defensive and appetitive activation in the expected ways both subjectively and physiologically. Corrugator EMG activity was higher at the end of the negative clips than the positive clips, and minute ventilation and SCL were higher for the arousing clips than for the less arousing clips. After the approximately 9-minute task, people who viewed the negative clips still reported more negative valence than those who viewed the positive clips. On the contrary, there were no differences in the arousal ratings. The valence effect in the mood state was paralleled by valence effects in the somato-physiological measures during the task. Increased facial frowning at the end of the negative clips was maintained during the task indicating persistence of defensive activation. SCL was lower for the negative film groups, especially at the end of the task, suggesting that sympathetic activation was lowered in subjects in the negative mood as compared with subjects in the positive mood. The findings of this study have several implications. First, they enrich our knowledge concerning the relationships between subjective feelings and their physiological correlates. Second, they inform us about the effectiveness of film clips as a mood induction instrument. Third, they suggest that induced changes in arousal are quickly overridden by the degree of activation "imposed" by the cognitive task, whereas induced changes in valence are more resistant and thus likely to impact the execution of the task. Finally, they show which physiological measures may be useful in tracking mood states during human-computer interaction (i.e., corrugator EMG activity and SCL). Feedback from these parameters could be used by computers to recognize mood states in the user.
Iraj Mirza’s poetry occupies a special place in Persian literature as compared to the works of other poets of his time, due to his almost unrivalled use of language and rhetorics. Deviating from syntactic, semantic and pragmatic norms, he creates a new atmosphere with the simple language he uses, which draws his poetry close to the language of nature. The present paper examines Iraj Mirza’s poetry in terms of language function and his expert play with language. He deviates from the accepted linguistic norms of syntax, semantics and pragmatics, with an artistic courage, creating a new atmosphere in literary language: It is worth mentioning that he does so with such a simple language that one can claim, without unnecessary exaggeration that his poems are closer to the language of the nature than those of his contemporary poets. This article studies some of the language functions of Iraj's poems, revealing a small part of his skill in playing with the language.
Based on the language of 17th century Bosnian Franciscan literature and enriched with features of the Neo-Štokavian folklore koine, the language of the 18th century writers represents a consistent system. Although it was not subject to willful codification, the language of the 18th century writers has codification elements. This is primarily implied by functional distribution, pronounced independence from common speech, especially on the syntactic level, compulsory use for all users and specific prescriptiveness in grammar handbooks of the time.
Multiobjective evolutionary algorithms (MOEA) are an effective tool for solving search and optimization problems containing several incommensurable and possibly conflicting objectives. Unfortunately, many MOEAs face difficulties in solving problems when the number of objectives increases. In this paper, we investigate the efficacy of spatially structured MOEAs for scalable multiobjective problems. The algorithm is an extension of the standard cellular evolutionary algorithm, where the population is mapped to nodes of alternative complex networks. A selection regime based on a non-dominance rating and a crowding mechanism guides the evolutionary trajectory and an ε-dominance external archive is used to maintain a spread of solutions across the Pareto-optimal front. An important outcome of this work is the classification of the network models based on their impact on convergence speed and solution quality as the number of objectives increases for a given problem.
How far can we get with unsupervised parsing if we make our training corpus several orders of magnitude larger than has hitherto be attempted? We present a new algorithm for unsupervised parsing using an all-subtrees model, termed U-DOP*, which parses directly with packed forests of all binary trees. We train both on Penn’s WSJ data and on the (much larger) NANC corpus, showing that U-DOP * outperforms a treebank-PCFG on the standard WSJ test set. While U-DOP * performs worse than state-of-the-art supervised parsers on handannotated sentences, we show that the model outperforms supervised parsers when evaluated as a language model in syntax-based machine translation on Europarl. We argue that supervised parsers miss the fluidity between constituents and non-constituents and that in the field of syntax-based language modeling the end of supervised parsing has come in sight. 1
Although using ontologies to assist information retrieval and text document processing has recently attracted more and more attention, existing ontologybased approaches have not shown advantages over the traditional keywords-based Latent Semantic Indexing (LSI) method. This paper proposes an algorithm to extract a concept forest (CF) from a document with the assistance of a natural language ontology, the WordNet lexical database. Using concept forests to represent the semantics of text documents, the semantic similarities of these documents are then measured as the commonalities of their concept forests. Performance studies of text document clustering based on different document similarity measurement methods show that the CF-based similarity measurement is an effective alternative to the existing keywords-based methods. In particular, this CFbased approach has obvious advantages over the existing keywords-based methods, including LSI, in processing short text documents or in P2P or live news environments where it is impractical to collect the entire document corpus for analysis.
Words associated with perceptually salient, highly imageable concepts are learned earlier in life, more accurately recalled, and more rapidly named than abstract words (R. W. Brown, 1976; Walker & Hulme, 1999). Theories accounting for this concreteness effect have focused exclusively on semantic properties of word referents. A novel possibility is that word structure may also contribute to the effect. We report a corpus-based analysis of the phonological and morphological structures of a large set of nouns with imageability ratings (N = 2,023). High- and low-imageability nouns differed by length, etymology, prosody, affixation, phonological neighborhood density, and rates of consonant clustering. On average, nouns denoting abstract concepts were longer, more derivationally complex, and emerged in English from a different distribution of languages than did concrete nouns. We address implications for interactivity of word form and meaning as pertain to theories of word concreteness, lexical acquisition, and word processing.
This paper presents a uniform approach to data extraction from syntactically annotated corpora encoded in XML. XQuery, which incorporates XPath, has been designed as a query language for XML. The combination of XPath and XQuery offers flexibility and expressive power, while corpus specific functions can be added to reduce the complexity of individual extraction tasks. We illustrate our approach using examples from dependency treebanks for Dutch.
Fat talk, the verbal dissatisfaction that women express about their bodies, was studied in a female dyad whereby participants interacted with a female confederate who either self-derogated, self-accepted, or self-aggrandized. A 2 (participant body esteem: high vs. low) ×3 (confederate style of body image presentation) design was used. Results revealed that participants’ public disclosure of their body image varied according to confederate's style. Consistent with a reciprocity effect, participants disclosed the lowest public body image ratings in the self-derogate condition, with moderate ratings in the self-accept condition, and highest ratings in the self-aggrandize condition. Moreover, participants with low compared to high body esteem stated lower public body image. Participants’ judgments of the confederates’ likeability did not vary as a function of the confederate's body presentational style. Findings support the recursive nature of the social psychology of body image such that personal body image dissatisfaction is partially influenced by fat talk social norms.
Morphology and phonology can in many cases be used to figure out which words correspond to which in Scandinavian. For instance, it is rather easy to figure out which Norwegian personal pronoun corresponds to which in Danish, and even Icelandic or Faroese. However, when it comes to prepositions and modal verbs we cannot rely on morphology or phonology alone. For example, Norwegian and Danish måtte do not always have the same meaning, and similarly, Icelandic vilja is not used as the future modal as Norwegian and Danish ville is. Instead of relying on morphology or phonology, we can use parallel corpora. Unfortunately, there are not many parallel corpora that include all of the Scandinavian languages, and those that exist are maybe not large enough to give reliable results. Nevertheless, to get a picture of what it could look like, the Danish, Faroese, Icelandic, Norwegian, Swedish, and English parts of a small treebank, The Sophie Treebank, were used to find out which modal verbs correspond to which in the various languages.
This paper reports on a hybrid architecture for computational anaphora resolution (CAR) of German that combines a rule-based pre-filtering component with a memory-based resolution module (using the Tilburg Memory Based Learner – TiMBL). The data source is provided by the TüBa-D/Z treebank of German newspaper text (Telljohann et al. 04) that is annotated with anaphoric relations. The CAR experiments performed on these treebank data corroborate the importance of modelling aspects of discourse structure for robust, data-driven anaphora resolution. The best result with an F-measure of 0.734 achieved by these experiments outperforms the results reported by (Schiehlen 04), the only other study of German CAR that is based on newspaper treebank data. 1
In Example Based Machine Translation research, many researchers apply Structure Based EBMT approach to annotate sentence structure in tree format. During training process, corpus become large and large and human tagging in Treebank creation becomes unrealistic. Therefore, it needs a tool to simplify and unify tagging process in order to enhance tagging performance and ensure sentence tree correctness. In this paper, we propose an automatic tool to create Treebank in TCT annotation schema.
The linguistic quality of a parallel treebank depends crucially on the parallelism between the source and target language annotations. We propose a linguistic notion of translation units and a quantitative measure of parallelism for parallel dependency treebanks, and demonstrate how the proposed translation units and parallelism measure can be used to compute transfer rules, spot annotation errors, and compare different annotation schemes with respect to each other. The proposal is evaluated on the 100,000 word Copenhagen Danish-English Dependency Treebank.
This paper describes an effective approach to adapting an HPSG parser trained on the Penn Treebank to a biomedical domain. In this approach, we train probabilities of lexical entry assignments to words in a target domain and then incorporate them into the original parser. Experimental results show that this method can obtain higher parsing accuracy than previous work on domain adaptation for parsing the same data. Moreover, the results show that the combination of the proposed method and the existing method achieves parsing accuracy that is as high as that of an HPSG parser retrained from scratch, but with much lower training cost. We also evaluated our method in the Brown corpus to show the portability of our approach in another domain.
We present results that show that incorporating lexical and structural semantic information is effective for word sense disambiguation. We evaluated the method by using precise information from a large treebank and an ontology automatically created from dictionary sentences. Exploiting rich semantic and structural information improves precision 2–3%. The most gains are seen with verbs, with an improvement of 5.7% over a model using only bag of words and n-gram features.
Reviewed by: Antología conmemorativa: Nueva Revista de Filología Hispánica. Cincuenta tomos ed. by Alejandro Rivas and Yliana Rodríguez Natalya I. Stolova Antología conmemorativa: Nueva Revista de Filología Hispánica. Cincuenta tomos. Vol. 2. Ed. by Alejandro Rivas and Yliana Rodríguez. (Publicaciones de la Nueva Revista de Filología Hispánica 9.) Mexico City: El Colegio de México, 2003. Pp. viii, 651. ISBN 9681211154. The present work is the second of the two anniversary collections published by Nueva Revista de Filología Hispánica (NRFH) to celebrate the appearance of its fiftieth volume. Volume 2 contains thirty-two selected articles that have been published in the journal over the years. These include eighteen papers on literary topics and fourteen papers concerned with linguistics. I limit my description to the linguistic articles, giving in parentheses the original publication date. A series of papers in the collection adopt a diachronic or philological perspective. Eugenio Coseriu (1961) challenges the Arabic origin of several Spanish and Rumanian expressions, arguing that these originated within the Romance language family. Rafael Lapesa (1961) traces the development of the Latin demonstratives into the Spanish and French articles. Margherita Morreale (1963–64) offers a philological commentary on the Evangelio de San Mateo según el manuscrito escurialense I-j-6: Texto, gramática y vocabulario published by Thomas Montgomery in 1962. Germán de Granda (1978) examines the history behind the verbal diphthongized voseo forms. Yakov Malkiel (1988) describes the demise of Old Spanish nozir, nuzir ‘harm’ during the Late Middle Ages. The different varieties of Spanish constitute the focus of a cluster of five papers. María Josefa Canellada de Zamora and Alonso Zamora Vicente (1960), as well as Juan M. Lope Blanch (1963–64), treat the reduction and the loss of unstressed vowels in Mexican Spanish. Tracy D. Terrell (1978) focuses on the aspiration and the elision of the implosive and final /s/ in the Spanish of Puerto Rico. Manuel Alvar (1988) challenges the notion of el dialecto andaluz ‘the Andalusian dialect’. Guillermo L. Guitarte (1992) and María Beatriz Fontanella de Weinberg (1995) take up the phenomenon of rehilamiento in the nineteenth-century Spanish of Buenos Aires. Two articles employ Spanish data to consider issues related to linguistic terminology and linguistic theory. Bernard Pottier (1961) explores the notion of auxiliary verb. José Pedro Rona (1973) addresses the question of linguistic norm in the context of the different local, regional, national, and pan-Latin American features. Finally, Antonio Quilis (1982) provides a description of the grammar of Tagalog Arte y reglas de la lengua tagala (1610) written by missionary Francisco de San José Blancas and places this work within the context of the missionary linguistics of the Philippines. This volume is a valuable source for Hispanists, and for linguists interested in key works on the Spanish language published from the 1960s to the 1990s. Natalya I. Stolova Colgate University Copyright © 2007 Linguistic Society of America
Youngster violence/violence on youngsters. Crossed views on the perception of verbal violence among immigrant school populations Among the different forms of violence, from and addressed to the youth, those exerted within the school framework are many and various. The form that has more particularly caught our attention, is the so-called normative violence connected to the language practices among immigrant school populations.Treating the issue of violence in relation to the norm, whether it be of a linguistic kind or another, is legitimate as far as the former constitutes and destroys the latter. Indeed, the school presents itself as « the place favouring a struggle to impose linguistic norms ». The way of speaking of school actors – a.o. learners and teachers – is an indicator of the relationship that the latter have with social rules in general and school norms in particular. Beyond the description of obscenities, language coarseness and vulgarity of the former, and of the French language « in the manner of Charles-Henri » of the others, these forms of verbal violence in the school framework are worth thinking about.Moreover, the social imagery related to immigration is so virulent that the equation « practice of languages different from &#34;correct&#34; French (the so-called &#34;bon usage&#34;), immigration and delinquency » may seem astonishing to some while it is admitted by others. The practice of the language would explain acts of delinquency: to act on the language would be preventive. « Young immigrants had better behave themselves », or even « they’d better talk correctly »! The eradication of violence seems to be at the cost of this simple solution.It is precisely beyond this naive optimism that we would like to go in this article, centred on the notion of youngster verbal violence, in the way it is perceived and lived in the school framework. To do so, the contribution of sociolinguistics can help in order to delimit and redefine language practices and to supply didactics of French with useful marks for efficient school norm teaching.
Current parameters of accurate unlexicalized parsers based on Probabilistic Context-Free Grammars (PCFGs) form a two-dimensional grid in which rewrite events are conditioned on both horizontal (head-outward) and vertical (parental) histories. In Semitic languages, where arguments may move around rather freely and phrase-structures are often shallow, there are additional morphological factors that govern the generation process. Here we propose that agreement features percolated up the parse-tree form a third dimension of parametrization that is orthogonal to the previous two. This dimension differs from mere "state-splits" as it applies to a whole set of categories rather than to individual ones and encodes linguistically motivated co-occurrences between them. This paper presents extensive experiments with extensions of unlexicalized PCFGs for parsing Modern Hebrew in which tuning the parameters in three dimensions gradually leads to improved performance. Our best result introduces a new, stronger, lower bound on the performance of treebank grammars for parsing Modern Hebrew, and is on a par with current results for parsing Modern Standard Arabic obtained by a fully lexicalized parser trained on a much larger treebank.
espanolEn este estudio hemos querido trazar un primer acercamiento al estudio de la lengua poetica del autor italiano Tommaso Stigliani y lo hemos hecho analizando su mayor obra, Il Mondo Nuovo (1628), poema epico sobre el descubrimiento de America. Hemos analizado los principales fenomenos de la ortografia, la fonetica, la morfologia, la sintaxis y el lexico que aparecen en el texto, teniendo en cuenta la norma linguistica del siglo XVII y las opiniones y teorias de los principales linguistas de la epoca. Con este estudio, por tanto, ofrecemos un acercamiento a la concepcion de lengua poetica que Tommaso Stigliani, a traves de su poema y de sus escritos teoricos, puso de manifiesto a lo largo de su carrera literaria. EnglishIn this study, we have attempted to make inroads into the study of the poetic language of the Italian author, Tommaso Stigliani, analysing his principal work, Il Mondo Nuovo (1628), an epic poem on the discovery of America. We have examined the main phenomena of orthography, phonetics, morphology, syntax and lexicon that appear in the text, taking into account the linguistic norms of the 17th century, as well as the opinions and theories of the major linguistic experts of that period. Our study, therefore, gives an insight into Tommaso Stigliani?s conception of poetic language, which, through his poem and his theoretical writings, he maintained for the duration of his literary career
Abstract. In this paper we explore an unsupervised approach to classify video content by analyzing the corresponding subtitles. The proposed method is based on the WordNet lexical database and the WordNet domains and applies natural language processing techniques on video subtitles. The method is divided into several steps. The first step includes subtitle text preprocessing. During the next steps, a keyword extraction method and a word sense disambiguation technique are applied. Subsequently, the WordNet domains that correspond to the correct word senses are identified. The final step assigns category labels to the video content based on the extracted domains. Experimental results with documentary videos show that the proposed method is quite effective in discovering the correct category for each video.
This study compared two technical sign movie formats, Quicktime (2D) and threedimentional animations (3D) and examined recall, comprehension, and rehearsal time of signs. There was no significant difference found between the formats for comprehension and recall. However, for rehearsal the high school students had significantly higher sign production scores after viewing 2D signs as comparaed to 3D signs. In addition, there was a significant difference found relating to gender. Girls scored significantly higher on all three measures, regardless of the movie format. In addition, students with more than two years experience with ASL scored significantly higher than those with less than two years on both recall and comprehension using both movie formats. Implications of these findings will be discussed in this paper.
Summary form only given. Many natural language processing (NLP) applications could benefit from a richer model of text meaning than the bag-of-words and n-gram models that currently predominate. Despite theoretical interest since the 1960s, however, no large-scale model exists; in fact, it is not even clear what such a model should minimally include. However, the introduction of large-scale public resources such as the Penn TreeBank and WordNet have generated a great deal of progress in the NLP community, and so it seems increasingly important to create some kind of meaning-oriented model and build a corresponding corpus that is large enough to support adequate machine learning. This talk argues for the necessity of (even shallow) semantics-based NLP, describes the contents and operation of the OntoNotes project, and in so doing introduces and explains the general issues facing annotation projects. Our hope is that other people not only try to use the OntoNotes corpus in their own work, but also create their own annotations on the same material, so that more layers of shallow semantics can be included into OntoNotes.
<h3>Introduction</h3> Natural language applications like machine translation, question answering, and summarization currently are forced to depend on impoverished text models like bags of words or n-grams, while the decisions that they are making ought to be based on the meanings of those words in context. That lack of semantics causes problems throughout the applications. Misinterpreting the meaning of an ambiguous word results in failing to extract data, incorrect alignments for translation, and ambiguous language models. Incorrect coreference resolution results in missed information (because a connection is not made) or incorrectly conflated information (due to false connections). Some richer semantic representation is badly needed. The OntoNotes project is a collaborative effort between BBN Technologies, the University of Colorado, the University of Pennsylvania, and the University of Southern California's Information Sciences Institute to produce such a resource. It aims to annotate a large corpus comprising various genres of text (news, conversational telephone speech, weblogs, use net, broadcast, talk shows) in three languages (English, Chinese, and Arabic) with structural information (syntax and predicate argument structure) and shallow semantics (word sense linked to an ontology and coreference). OntoNotes builds on two time-tested resources, following the Penn Treebank for syntax and the Penn PropBank for predicate-argument structure. Its semantic representation will include word sense disambiguation for nouns and verbs, with each word sense connected to an ontology, and coreference. The current goals call for annotation of over a million words each of English and Chinese, and half a million words of Arabic over five years. The authors wish to make this resource available to the natural language research community so that decoders for these phenomena can be trained to generate the same structure in new documents. Lessons learned over the years have shown that the quality of annotation is crucial if it is going to be used for training machine learning algorithms. Taking this cue, we ensure that each layer of annotation in OntoNotes will have at least 90% inter- annotator agreement. Our pilot studies have shown that predicate structure, word sense, ontology linking, and coreference can all be annotated rapidly and with better than 90% consistency. <h3>Samples</h3> The following screen captures provide examples of the data contained in this corpus. <ul> <li> <a href="./desc/addenda/LDC2007T21_eng_tbk.jpg" rel="nofollow">English tree</a>. </li> <li> <a href="./desc/addenda/LDC2007T21_sense_pred.jpg" rel="nofollow">English sense predicate structure</a>. </li> <li> <a href="./desc/addenda/LDC2007T21_chi_comp.jpg" rel="nofollow">Chinese tree and sense predicate structure</a>. </li> </ul><h3>Sponsorship</h3> This work was suppported in part by the Defense Research Advanced Projects Agency, GALE Program Grant No. HR0011-06-C-0022. The content of this publication does not necessarily reflect the position or policy of the Government, and no official endorsement should be inferred. </br> Portions © 1989 Dow Jones & Company, Inc., © 1996-2001 Sinorama Magazine, © 1994-1998 Xinhua News Agency, © 1995, 2005, 2006, 2007 Trustees of the University of Pennsylvania
One may need to build a statistical parser for a new language, using only a very small labeled treebank together with raw text. We argue that bootstrapping a parser is most promising when the model uses a rich set of redundant features, as in recent models for scoring dependency parses (McDonald et al., 2005). Drawing on Abney’s (2004) analysis of the Yarowsky algorithm, we perform bootstrapping by entropy regularization: we maximize a linear combination of conditional likelihood on labeled data and confidence (negative Rényi entropy) on unlabeled data. In initial experiments, this surpassed EM for training a simple feature-poor generative model, and also improved the performance of a feature-rich, conditionally estimated model where EM could not easily have been applied. For our models and training sets, more peaked measures of confidence, measured by Rényi entropy, outperformed smoother ones. We discuss how our feature set could be extended with cross-lingual or cross-domain features, to incorporate knowledge from parallel or comparable corpora during bootstrapping. 1
We aim to improve the performance of a syntactic parser that uses a part-of-speech (POS) tagger as a preprocessor. Pipelined parsers consisting of POS taggers and syntactic parsers have several advantages, such as the capability of domain adaptation. However the performance of such systems on raw texts tends to be disappointing as they are affected by the errors of automatic POS tagging. We attempt to compensate for the decrease in accuracy caused by automatic taggers by allowing the taggers to output multiple answers when the tags cannot be determined reliably enough. We empirically verify the effectiveness of the method using an HPSG parser trained on the Penn Treebank. Our results show that ambiguous POS tagging improves parsing if outputs of taggers are weighted by probability values, and the results support previous studies with similar intentions. We also examine the effectiveness of our method for adapting the parser to the GENIA corpus and show that the use of ambiguous POS taggers can help development of portable parsers while keeping accuracy high. 1
Due to the data sparseness problem, the lexical information from a treebank for a lexicalized parser could be insufficient. This paper proposes an approach to learn head-modifier pairs from a raw corpus, and to integrate them into a lexicalized dependency parser to parse a Chinese Treebank. Experimental re-sults show that this approach not only enlarged the coverage of bi-lexical de-pendency, but also improved the accuracy of dependency parsing significantly.
This paper describes practical issues in the framework-independent evaluation of deep and shallow parsers. We focus on the use of two dependencybased syntactic representation formats in parser evaluation, namely, Carroll et al. (1998)’s Grammatical Relations and de Marneffe et al. (2006)’s Stanford Dependency scheme. Our approach is to convert the output of parsers into these two formats, and measure the accuracy of the resulting converted output. Through the evaluation of an HPSG parser and Penn Treebank phrase structure parsers, we found that mapping between different representation schemes is a non-trivial task that results in lossy conversions that may obscure important differences between different parsing approaches. We discuss sources of disagreements in the representation of syntactic structures in the two dependency-based formats, indicating possible directions for improved framework-independent parser evaluation.