Papers reviewed and determined not to be word norm studies. Use the flag icon to report errors or suggest re-inclusion.
18625 papers
This paper presents Bayesian learning for recurrent neural network language model (RNN-LM). Our goal is to regularize the RNN-LM by compensating for the randomness of the estimated model parameters which is characterized by a Gaussian prior. This model is not only constructed by training the synaptic weight parameters according to the maximum a posteriori criterion but also regularized by estimating the Gaussian hyper-parameter through the type 2 maximum likelihood. However, a critical issue in Bayesian RNN-LM is the heavy computation of Hessian matrix which is formed as the sum of a large amount of outer-products of high-dimensional gradient vectors. We present a rapid approximation to reduce the redundancy due to the curse of dimensionality and speed up the calculation by summing up only the salient outer-products. Experiments on 1B-Word Benchmark, Penn Treebank and World Street Journal corpora show that rapid Bayesian RNN-LM consistently improves the perplexity and word error rate in comparison with standard RNN-LM.
Grammatical Change Begins within the Word: Causal Modeling of the Co-evolution of Icelandic Morphology and Syntax Fermin Moscoso del Prado Martin (fmoscoso@linguistics.ucsb.edu) Department of Linguistics, UC – Santa Barbara Santa Barbara, CA 93106 USA Abstract I introduce a combination of information-theoretical and causal modeling to study the cascading of changes between the morphology and the syntax of a language on a diachronic scale. Through the analysis of a historical treebank of Icelandic language ranging from the XII to the XXI century, I show that it is changes in the inflectional morphology of the language that triggered changes in its syntax. This offers a novel and powerful approach to draw conclusions in historical linguistics from a macroscopic perspective. In addition, these findings have implications for the dynamical properties of the linguistic system. Keywords: Granger-causality; Historical Linguistics; Icelandic; Information Theory; Morphological paradigms; Syntactic complexity. Introduction A common observation in the field of Historical Linguistics is that grammatical changes in language tend to be cascaded (e.g., Biberauer & Roberts, 2008; Lightfoot, 2002). Changes at a given level of language (e.g., phonology, morphology, syntax, …) disturb the unstable equilibrium at which human languages reside, triggering a cascade of further linguistic changes as a result. These cascaded changes continue until the system reaches a new meta-stable state (e.g, Croft, 1995; Smith, 1996). For instance, it is widely documented that the loss of the grammatical case markers (morphology) in Late Middle English led to a more rigid word order (syntax) in Early Modern English (cf., Fisiak, 1984). Traditionally, historical linguists have tracked these cascaded changes by looking for the earliest document at which a particular grammatical innovation can be found or – conversely– at the latest time in which a later extinct construction was documented. Historical linguistics often relies on hard dichotomies on the presence or absence of individual words, affixes, or constructions. This approach is clearly useful for documenting the approximate time at which particular constructions appeared or disappeared. The approach is however limited in its power to detect more subtle forms of grammatical change. For instance, as I will argue in this study, Icelandic morphological paradigms are remarkably resilient, having survived with little change for a such a long period, that most of its current system can be directly traced –virtually unchanged– all the way up to the Old West Norse of the XII century. However, despite the striking conservativeness of the Icelandic paradigms, their patterns of usage have changed along this period. One can obtain a higher degree of sensitivity by studying the frequencies (and implicitly the probabilities) of usage of different constructions in diachronic scale. These often reveal gradual changes in the grammar of a language. For instance, Ellegard (1953) documented how the usage of the English periphrastic do construction (e.g., I do not speak vs. I speak not, or Do you speak? vs. Speak you?) arose gradually –rather than abruptly– during a period of two hundred years, from the late XV century to early XVIII century. With the wide availability of diachronic corpora in electronic form, these frequency-based methods have gained much prominence in recent years (see Hilpert & Gries, in press, for a recent survey of quantitative methods in historical linguistics). Admittedly, frequency methods offer an improved sensitivity to the gradualness of grammatical changes along time. Most often, these methods are applied to relate the evolution of the usage of a particular construction in different contexts. The natural tools to achieve this type of inferences are different types of regression analysis. These analyses study the correlations between different factors that might affect the emergence or demise of a construction. For instance, Hilpert (2013) uses such tools to analyze the evolution of (among many others) the patterns of usage of the English future constructions will vs. be going to. Using a logistic regression analysis, he identifies several factors that significantly co-occur with the uses of either construction. One must –however– keep in mind the old adagio: “correlation does not imply causation”. Even in the cases when one finds significant correlation between the frequencies of use and co-occurrence of different linguistic patterns and constructions, it still remains problematic to argue that one pattern causes another. Back to example above, the causal connection between the loss of grammatical case and the fixation of word order is in fact not so trivial on the basis of the historical data alone. Evidently, I cannot infer that I have just had dinner because it stopped raining, even if I observed both events in a sequence. To make such arguments, I would require some form of statistical evidence on how reliably do I start eating whenever it stops raining. Similarly, just the temporal sequentiality between the loss of case marking and the emergence of rigid word order does not –by itself alone– necessarily warrant causality. For instance, Kiparsky (1996) discusses how fixed word order arose also in Icelandic while the case markers were preserved.
Correspondence analysis (CA) was conceived by Benzécri (1977b; Benzécri et al., 1973, 1981) as an inductive method able to infer from their own texts of interest the linguistic norms that rule them. He wanted to offer a powerful tool to tackle all kinds of issues related to the form, meaning, and style of texts.
We propose Diverse Embedding Neural Network (DENN), a novel architecture for language models (LMs). A DENNLM projects the input word history vector onto multiple diverse low-dimensional sub-spaces instead of a single higher-dimensional sub-space as in conventional feed-forward neural network LMs. We encourage these sub-spaces to be diverse during network training through an augmented loss function. Our language modeling experiments on the Penn Treebank data set show the performance benefit of using a DENNLM.
Lexical databases of semantic frames have been shown to be useful in problems related to natural language processing.However, creation of such databases is a task that is time consuming and involves many manual steps.One of these steps is selection and grouping of sentences to identify frames.However, we advocate that if sentences were previously annotated with ontological information, this grouping could be executed automatically.In this article we present tests performed with clustering sentences containing the lexeme Travel (noun and verb).Tests showed that the use of clustering algorithms on ontologically annotated sentences is a promising step towards automating construction of semantic frames databases.
This work presents an improvement on a novel Selection Method to develop applications in the context of the Service-Oriented Computing paradigm. We have defined an Interface Compatibility procedure to assess Web Services by exploring the available information from WSDL documents. Such information involves data types from return, parameters and exceptions, and identifiers from parameters and operation names. The lexical database WordNet was originally used as a semantic basis to assess terms from identifiers. In this paper we use the DISCO database as an alternative, to evaluate the independence of the approach w.r.t. the semantic basis. We made a comparative analysis through different experiments with a data-set of 465 real-life Web Services and measured the results using metrics from the Information Retrieval field.
Bilingual Base Noun Phrase (BaseNP) extraction is one of the key tasks of Natural Language Processing (NLP). This task is more challenging for the pair of English-Vietnamese due to the lack of available Vietnamese language resources such as treebanks, part-of-speech taggers, and parsers. In this paper, we propose a combination model that uses language characteristics based on statistics and the projection method to extract BaseNP correspondences from a bilingual corpus. The language characteristics used in this model include the word segmentation, word order and word classification [1]. Our model overcomes not only the lack of resources of Vietnamese, but also improves the performance of miss-alignment, null-alignment, overlap and conflict projection of the existing methods. The proposed model can be easily applied to other language pairs. Experiment on 66,646 pairs of sentences in the English-Vietnamese bilingual corpus shows that our proposed model is very satisfactory.
This chapter explores the challenges of developing the field of Latin Computational Linguistics. Computational Linguistics aims at designing, implementing, and applying computational models for natural languages. A large part of Computational Linguistics research has been developed for English, or at least tested on this language. A crucial aspect of a fruitful exchange between the disciplines of Latin Linguistics and Computational Linguistics concerns the way Latin texts are collected, accessed, and investigated for linguistic analyses. So, any attempt into Latin Computational Linguistics is likely to start from corpora. Annotation provides each word form in a sentence with one or more labels that mark its attributes; for example, a morpho-syntactic annotation would add a 'genitive' tag to puellarum. The chapter advocates the use of corpora in Latin Linguistics by reporting on research based on Latin treebanks, to show their potential for Historical Linguistics research.Keywords: historical corpora; historical languages; historical Linguistics research; Latin Computational Linguistics; morpho-syntactic annotation
In this paper I have made an attempt to describe the surface forms, usage patterns, and pronunciation variations of some Bengali consonant grapheme clusters, which are found as great stumbling blocks in language learning, optical character recognition, text-to-speech conversion, speech synthesis, and language processing. To make this study authentic, I have used a moderately large lexical database obtained from a written corpus of modern Bengali prose texts. I have tried to study how these consonant grapheme clusters are orthographically designed by way of combining consonant graphemes and consonant graphic variants; how these clusters are utilized in formation of words; and how these clusters are actually pronounced within words in standard Bengali speech. I have also tried to discuss how the pronunciation of individual consonant graphemes is largely affected when these characters are combined in formation of clusters; and how the pronunciation of these clusters, that are controlled by various phonological rules and processes that operate in standard Bengali speech, leave notable marks on orthographic representation of words in written texts.
Data sparsity issues are extremely severe for parser due to the flexibility of tree structures. Many tags and productions appears a little, nevertheless, they are crucial for the parse disambiguation where it occurs. Besides, when a common tag somewhat regularly occurs in a non-canonical position, its distribution is usually distinct. In this paper, we propose a metric that measures the scarcity of any phrase with arbitrary span size. To make a better compromise between training trees with high confidence and scarcity, we try to catch some constraints in response to rare but articulating categories when training latent variable grammar. We exploits the limited data more sufficiently by capturing the depicting power of rate tree structure configuration in Expectation & Maximization procedure and Split & Merge framework. The resulting grammars are interpretable as our intension. Based on this approach, we further propose a method that exploits the limited training date from multiple perspectives, and accumulates their advantages in a product model. Despite its limited training data, out model improves parsing performance on Penn Chinese Treebank Fifth Edition, even higher than some systems with extra unlabeled data and external resources. Furthermore, this method is easy to generalized to cope with data sparsity in other natural language processing tasks.
Dependency parsing has gained more and more interest in natural language processing in recent years due to its simplicity and general applicability for diverse languages. The international conference of computational natural language learning (CoNLL) has organized shared tasks on multilingual dependency parsing successively from 2006 to 2009, which leads to extensive progress on dependency parsing in both theoretical and practical perspectives. Meanwhile, dependency parsing has been successfully applied to machine translation, question answering, text mining, etc. To date, research on dependency parsing mainly focuses on data-driven supervised approaches and results show that the supervised models can achieve reasonable performance on in-domain texts for a variety of languages when manually labeled data is provided. However, relatively less effort is devoted to parsing out-domain texts and resource-poor languages, and few successful techniques are bought up for such scenario. This tutorial will cover all these research topics of dependency parsing and is composed of four major parts. Especially, we will survey the present progress of semi-supervised dependency parsing, web data parsing, and multilingual text parsing, and show some directions for future work. In the first part, we will introduce the fundamentals and supervised approaches for dependency parsing. The fundamentals include examples of dependency trees, annotated treebanks, evaluation metrics, and comparisons with other syntactic formulations like constituent parsing. Then we will introduce a few mainstream supervised approaches, i.e., transition-based, graph-based, easy-first, constituent-based dependency parsing. These approaches study dependency parsing from different perspectives, and achieve comparable and state-of-the-art performance for a wide range of languages. Then we will move to the hybrid models that combine the advantages of the above approaches. We will also introduce recent work on efficient parsing techniques, joint lexical analysis and dependency parsing, multiple treebank exploitation, etc. In the second part, we will survey the work on semi-supervised dependency parsing techniques. Such work aims to explore unlabeled data so that the parser can achieve higher performance. This tutorial will present several successful techniques that utilize information from different levels: whole tree level, partial tree level, and lexical level. We will discuss the advantages and limitations of these existing techniques. In the third part, we will survey the work on dependency parsing techniques for domain adaptation and web data. To advance research on out-domain parsing, researchers have organized two shared tasks, i.e., the CoNLL 2007 shared task and the shared task of syntactic analysis of non-canonical languages (SANCL 2012). Both two shared tasks attracted many participants. These participants tried different techniques to adapt the parser trained on WSJ texts to out-domain texts with the help of large-scale unlabeled data. Especially, we will present a brief survey on text normalization, which is proven to be very useful for parsing web data. In the fourth part, we will introduce the recent work on exploiting multilingual texts for dependency parsing, which falls into two lines of research. The first line is to improve supervised dependency parser with multilingual texts. The intuition behind is that ambiguities in the target language may be unambiguous in the source language. The other line is multilingual transfer learning which aims to project the syntactic knowledge from the source language to the target language.
While some cross-modal associations might have psychobiological basis, other patterns of association might be acquired or cultural (cf. [5], [11], [4]). Stimuli perceived through different sensory organs via parallel brain pathways may be associated at a higher level if they both happen to have the same effect on emotional state, mood, or affective state ([14]). If the perceived input under-specifies an event, more complex cognitive processing mechanisms kick in ([7]). This process is not primarily ecological and might be mediated by emotion ([12]). \nResearch on crossmodal matching has provided evidence that many non-arbitrary and universal correspondences exist. Audio-visual correspondences may be based on amodal correspondences, for example, the loudness of sound and the luminosity of light ([14]). [13] showed that most cultures display word clusters near ‘red’, ‘green’, ‘yellow’, and ‘blue’ (in addition to ‘white’ and ‘black’), and argued that “focal colours” really are universal. \nBresin ([2]) derived 24 colours from a scheme of selecting approximately equal distances in colour parameters in HSL (Hue, Saturation, Lightness) “space”. This produced a set where the colour patches are arguably more evenly distributed, from a perceptual point of view, than those in the two studies mentioned above. Bresin found correlations between colour parameters and the affective intent in music excerpts, i.e. listeners matched colours to music excerpt played with a certain ‘feeling’. As in [1] but more general, colour brightness was associated with positive emotion and darkness with negative emotion. \nPalmer ([12]) investigated colour association to classical music excerpt where tempo and tonal mode were manipulated. The authors found that colours of high saturation and brightness, and colours more towards yellow (‘warmth’) were selected for music stimuli in fast tempo, and that conversely, de-saturated (‘grayer’), ‘darker’, and blue colours were selected for music of slow tempo in minor mode. Furthermore, they claimed strong support for emotion as a mediating mechanism for the cross-modal associations. \nA review of the research provoked the idea that colour association to sound might be context-dependent. When associating colour to music, natural soundscapes, and ‘soundscape compositions’, do people use different strategies? Which musical features influence colour association? Can emotion mediate between musical features and colour association? \nWe designed an experiment to investigate a) correlations between visual colours defined by linear parameters and music stimuli with previously validated affect; b) correlations between the colour parameters and computational acoustic and musical features; and c) the multiple regressions onto colour parameters of affective ratings (emotions), psychoacoustic descriptors, and musical features.
It is impossible not to communicate.(Buda Bela)Communication implies transmission of information, ideas, feelings by means of symbols (words, images, graphics etc.).(Berelson-Steiner)As each language has several variants and is in a continuous process of change, I should like to compare the correct and incorrect text, which is used mainly on radio and television, by presenting authentic texts from mass media in Romania. Once television came into being, the question was whether this would not put the radio into shade. In time, it was proved that both media sources are needed. They function not in parallel, they complete each other, but each of them has its own well established role in the audio-visual mass-media. Nevertheless, it is also true that radio journalists should revise their journalistic genre - right under the influence of television. It has been necessary for them to renew their radio phonic text.Television can transmit information without a text, only by means of the language of images. While on the radio, the audio effect creates satisfaction. The nuance, the power, intonation, articulation and sound on the radio are part of the given information, having also a special role in the transmission of information. These have been the reasons for the radio phonic text and the text of the radio announcer to have developed as a distinct branch of the media science.Any person in good health can speak, but to be able to communicate in different situations, to express the nuance, rhythm, intonation of a text, you should have a certain training, you should be a specialist. Speaking in public has precise rules: how much of the text should be assimilated by the speakerwhich is the best variant, if he assimilates the text in its entirety or if he keeps a distance from it. Can he have control over his emotions or, because of these, he breathes faster and the rhythm of the speech is more rapid; does he raise the pitch of his voice and speaks louder? (e.g. during sports transmissions). What is the role of the logical intonation and of the emotive one? How does the melody of speech influence the intonation of the word, of the sentence, of the text? These and other are the questions I will try to answer. Mass-media has a decisive role in modem society, because people of our days get information about everything going on in the world by means of it. In this way, a person acquires knowledge about the world. It is very important for the language, for the speech, by means of which information is transmitted to the receptor, to observe the correct linguistic norms, to be beautiful. Speaking is the most complex activity, which is not inborn, it is learnt from parents and in school. Speaking is a system of communication by several channels. The general content of words is changing according to intonation, rhythm, volume, articulation etc. In the development of language and speaking, mass-media has a decisive role. On the radio and television, quite often we can hear texts where we can sense that the presenter concentrates a lot on the articulation of consonants and vowels, on the tone of the voice, on intonation and punctuation. We can hear all these but we cannot hear the idea in itself. Although, radio cannot transmit by images - as it is the case with televisionit has also its own informational language: the human voice, the speech of the presenter. By the harmony between the meaning of the text and the phonetic elements, a new quality can be achieved: the auditive influence leads to the visual effect. This inner visuality has a special effect. Today, it is evident that television does not put the radio into the shade.The acoustic language (and the read text), especially the spontaneous live one is related to the personality of the speaker, to the situation and the receptor. As a rule it is emphatic, it is a complex communication.Besides articulation, the live speech contributes to communication by the common and alternative use of phonetic instruments. …
哈金喜歡把玩中英文不同的語言特色,在創作中將兩者合而為一。他以本土化的英語論述重塑中文的諺語、譬喻與句構,翻譯、挪用並重組中文的語法,創造出饒富中文色彩的英語創作。閱讀哈金的詩集《殘骸》(Wreckage),受過中文教育的讀者很容易就可以發現詩中有許多中國古典詩詞的翻譯及改寫。作品中的詩人引述古典詩詞,以抒發個人情懷,而讀者不禁會思考:哈金的詩作源頭到底有多大的成分是來自中國文學?我們能夠依據哪些特點以及哪些標準來辨識所謂的創意?我們要如何區分「創新」與「改造」?而這樣的區分是否會改變我們對於哈金詩作的解讀與評斷?在哈金的詩作中,對中國古典詩詞的引述與對美國現今情境的感觸,二者或並置、或融合、或相互對照、或互相影響,這其中透露著何種訊息?被翻譯的是什麼,是中國詩詞還是美國情感?被移植的又是什麼,是文學還是文化?當中國的詩文被翻譯成英文、移植至美國之後,意義是否會有所轉化?是否會因應哈金複雜的美國移民經驗而呈現出某種迥異於以往的新義?而哈金的三重身份-詩者、評者、譯者-是彼此互補或彼此衝突?而這種種文學、文化再現,又訴說了何種華人離散的意涵,這些將是本文討論的重點。 Ha Jin's works bespeak an impressive linguistic creativity. Reconfigurating Chinese language through a nativized discourse of English, Ha Jin has translated, appropriated, and reconstructed Chinese linguistic norms and specifics into English-language literature in remarkable fashion. Studying Ha Jin's "Wreckage", a reader with Chinese education will be ready to identity fragments of renowned Chinese classical verse. Thus the reader is invited to ponder: To what extent does Ha Jin draw his poetic inspiration from the corpus of Chinese literature? How shall we measure accredited creativity? How do we distinguish innovation from renovation, and do those distinctions change our reading of the poems? Although Ha Jin has written exclusively of the reality of Chinese politics and society (with "A Free Life" the only exception so far), this material does not obviate the possibility of reading his works as belonging to a tradition of US immigrant literature. In Ha Jin's poetry, the juxtaposition, interaction and fusion of classical Chinese verse and contemporary American sensibility can be telling. Which has been translated, the Chinese verse or the American sensibility? What has been transplanted and translated? Have the Chinese poetry texts, after being transplanted into the English verse, undergone a transformation of meaning and resurfaced with new significance corresponding to the complexity of Ha Jin's immigrant experiences in America? This paper aims to explore how Ha Jin's Chinese poetry texts show significance corresponding to the complexity of his immigrant experiences in America.
In my habilitation dissertation, meant to validate my capacity of and maturity for directingresearch activities, I present a panorama of several topics in computational linguistics, linguisticsand computer science.Over the past decade, I was notably concerned with the phenomena of compositionalityand variability of linguistic objects. I illustrate the advantages of a compositional approachto the language in the domain of emotion detection and I explain how some linguistic objects,most prominently multi-word expressions, defy the compositionality principles. I demonstratethat the complex properties of MWEs, notably variability, are partially regular and partiallyidiosyncratic. This fact places the MWEs on the frontiers between different levels of linguisticprocessing, such as lexicon and syntax.I show the highly heterogeneous nature of MWEs by citing their two existing taxonomies.After an extensive state-of-the art study of MWE description and processing, I summarizeMultiflex, a formalism and a tool for lexical high-quality morphosyntactic description of MWUs.It uses a graph-based approach in which the inflection of a MWU is expressed in function ofthe morphology of its components, and of morphosyntactic transformation patterns. Due tounification the inflection paradigms are represented compactly. Orthographic, inflectional andsyntactic variants are treated within the same framework. The proposal is multilingual: it hasbeen tested on six European languages of three different origins (Germanic, Romance and Slavic),I believe that many others can also be successfully covered. Multiflex proves interoperable. Itadapts to different morphological language models, token boundary definitions, and underlyingmodules for the morphology of single words. It has been applied to the creation and enrichmentof linguistic resources, as well as to morphosyntactic analysis and generation. It can be integratedinto other NLP applications requiring the conflation of different surface realizations of the sameconcept.Another chapter of my activity concerns named entities, most of which are particular types ofMWEs. Their rich semantic load turned them into a hot topic in the NLP community, which isdocumented in my state-of-the art survey. I present the main assumptions, processes and resultsissued from large annotation tasks at two levels (for named entities and for coreference), parts ofthe National Corpus of Polish construction. I have also contributed to the development of bothrule-based and probabilistic named entity recognition tools, and to an automated enrichment ofProlexbase, a large multilingual database of proper names, from open sources.With respect to multi-word expressions, named entities and coreference mentions, I pay aspecial attention to nested structures. This problem sheds new light on the treatment of complexlinguistic units in NLP. When these units start being modeled as trees (or, more generally, asacyclic graphs) rather than as flat sequences of tokens, long-distance dependencies, discontinu-ities, overlapping and other frequent linguistic properties become easier to represent. This callsfor more complex processing methods which control larger contexts than what usually happensin sequential processing. Thus, both named entity recognition and coreference resolution comesvery close to parsing, and named entities or mentions with their nested structures are analogous3to multi-word expressions with embedded complements.My parallel activity concerns finite-state methods for natural language and XML processing.My main contribution in this field, co-authored with 2 colleagues, is the first full-fledged methodfor tree-to-language correction, and more precisely for correcting XML documents with respectto a DTD. We have also produced interesting results in incremental finite-state algorithmics,particularly relevant to data evolution contexts such as dynamic vocabularies or user updates.Multilingualism is the leitmotif of my research. I have applied my methods to several naturallanguages, most importantly to Polish, Serbian, English and French. I have been among theinitiators of a highly multilingual European scientific network dedicated to parsing and multi-word expressions. I have used multilingual linguistic data in experimental studies. I believethat it is particularly worthwhile to design NLP solutions taking declension-rich (e.g. Slavic)languages into account, since this leads to more universal solutions, at least as far as nominalconstructions (MWUs, NEs, mentions) are concerned. For instance, when Multiflex had beendeveloped with Polish in mind it could be applied as such to French, English, Serbian and Greek.Also, a French-Serbian collaboration led to substantial modifications in morphological modelingin Prolexbase in its early development stages. This allowed for its later application to Polishwith very few adaptations of the existing model. Other researchers also stress the advantages ofNLP studies on highly inflected languages since their morphology encodes much more syntacticinformation than is the case e.g. in English.In this dissertation I am also supposed to demonstrate my ability of playing an active rolein shaping the scientific landscape, on a local, national and international scale. I describemy: (i) various scientific collaborations and supervision activities, (ii) roles in over 10 regional,national and international projects, (iii) responsibilities in collective bodies such as program andorganizing committees of conferences and workshops, PhD juries, and the National UniversityCouncil (CNU), (iv) activity as an evaluator and a reviewer of European collaborative projects.The issues addressed in this dissertation open interesting scientific perspectives, in whicha special impact is put on links among various domains and communities. These perspectivesinclude: (i) integrating fine-grained language data into the linked open data, (ii) deep parsingof multi-word expressions, (iii) modeling multi-word expression identification in a treebank as atree-to-language correction problem, and (iv) a taxonomy and an experimental benchmark fortree-to-language correction approaches.
FigureCompulsive hoarding is a pattern of thoughts (obsessions) about discarding items and repetitive behaviors (compulsions) of collecting things. One in 20 people experiences compulsive hoarding, with severity ranging from mild to life-threatening. According to the American Psychiatric Association, the onset of symptoms is usually between ages 11 and 15, becoming problematic in the 20s and progressively increasing in severity with advancing age. Hoarding is twice as common among men as women, but women are more likely to seek treatment. One study demonstrated a genetic vulnerability, with at least one first-degree relative also exhibiting hoarding behavior. In this article, I'll give you an overview of hoarding disorder, including how to recognize it, available treatment options, and patient and family teaching points. Classifying hoarding behavior The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) has included a new diagnostic category for hoarding disorder. In the past, hoarding behaviors were included in the obsessive-compulsive disorder (OCD) category, but research has demonstrated that hoarding is a distinct disorder requiring distinct treatment (see OCD or hoarding disorder?). By creating a separate diagnosis, the hope is that more research and development into specific treatments for hoarding behavior will occur. The DSM-5 diagnostic criteria for hoarding disorder start with persistent difficulties discarding or parting with possessions, regardless of the value. These behaviors result in an accumulation of possessions that clutter the active living areas of the home, workplace, yard, or vehicle, preventing normal use of the space. Other diagnostic criteria include: severe distress that occurs when any attempt is made to throw away items indecision about what to keep or discard suspicions of other people touching the items obsessional thoughts of running out of an item or needing it in the future checking the trash for accidentally discarded items functional impairment, such as loss of living space, social isolation, dysfunctional interpersonal relationships, financial problems, and health hazards. In addition, these symptoms can't be attributed to any medical condition, such as brain injury, or another mental illness, such as OCD or delusions from psychosis. Why can't hoarders let go? Brain imaging shows a pattern of brain changes in people with hoarding disorder. These individuals have less activity in the cingulate gyrus—the part of the brain that communicates with the limbic, or emotional, center and the neocortex, or thinking, center. If activity is reduced in the neocortex, a person may have difficulty with attention, problem solving, and decision making. And when communication between the neocortex and limbic system is compromised, a person assigns more emotional meaning to stimuli. This means that an individual with hoarding disorder can have trouble with both the emotional and cognitive aspects of stimuli.Table: OCD or hoarding disorder?Other studies have looked at the neuropsychological testing of compulsive hoarders who didn't exhibit any criteria for OCD. These studies also found brain hypoactivity, with slow decision making and difficulty with attention and spatial tasks, contributing to difficulty staying focused, organizing, and developing strategies for categorizing and sorting that are necessary for decreasing accumulation. The result for individuals with hoarding disorder? Making decisions about items, sorting them, and discarding them becomes unpleasant and insurmountable. So what drives the behavior of those with hoarding disorder? The behaviors of people with OCD are driven by reducing anxiety. However, the behaviors of the compulsive hoarder are driven by intrusive thoughts about not having something that may be deemed valuable and not being able to discard something. Individuals with OCD are usually aware of the irrationality of their thoughts or behaviors, but individuals with hoarding disorder lack this insight and don't see their behaviors as a problem. Usually, compulsive hoarders are identified by social agencies or law enforcement when the hoarding starts to present a community danger. Consequences of hoarding Hoarding disorder causes social and economic burdens, as well as environmental hazards and health risks. There's a high rate of divorce among compulsive hoarders; family members may experience frustration at and anxiety about their loved one's hoarding behaviors. People with this disorder miss more days of work and use mental health services five times more than the general population. In addition, individuals with hoarding disorder have increased medical comorbidities, such as stroke, gastric ulcers, diabetes, and heart disease. Often, compulsive hoarders fail to seek medical care either because they're embarrassed about their behavior or they don't feel that they have a problem. Environmental hazards In one study, 42% of hoarders had blocked access to the refrigerator, stove, and bathrooms, causing filthy or unsafe living conditions. Often, feces and urine were found in containers on the bed or in the house; dead animals were also found. People with hoarding disorder have clutter throughout their houses with small, navigable trails or areas that are no longer navigable. Excessively cluttered bathrooms, kitchens, and bedrooms cause sanitation, hygiene, and food preparation problems. The weight of the items may also lead to structural damage to the house that can result in collapse. Large amounts of clutter also pose a fire hazard, especially when outlets are blocked or electrical appliances are being used with things piled on or in them. If there was a fire, the house of a compulsive hoarder would be difficult to leave in an emergency and difficult for firefighters to get in quickly. Health risks Health risks include injuries from falls or objects falling from piles and infestations of fleas, rats, mice, or bed bugs. The cluttered condition makes it difficult for extermination to occur. Dust, mildew, and high ammonia levels from waste products can increase respiratory problems. Lastly, the risk of parasites and food-related illnesses are also increased. Financial burdens Individuals with hoarding disorder often spend most of their money buying items, causing problems with meeting financial obligations such as rent, heat, and food. With the cluttered environment making it hard to cook in the kitchen, the added expense of takeout food can also limit the money available for other bills. Legal difficulties are also increased for people with hoarding disorder. They may risk eviction or their homes being condemned for public safety reasons. Children may be removed from the home and the individual with hoarding disorder may be charged with child neglect. People who hoard animals may be charged with animal cruelty, which is a felony charge in some states. The cost of legal problems further increases the financial burden. Effects on children Often, compulsive hoarders have small children living with them. The children of parents with hoarding disorder may feel isolated and experience depression. Having friends over is most likely not an option because the child feels embarrassed with the living situation or may be told not to invite friends to the home. In addition, children are exposed to environmental and health hazards if there isn't a place to perform hygiene tasks, wash their clothes, or eat safely. Lastly, they often compete with the hoarded items for attention and love from the parent. Intervention mention There's no cure for hoarding, but you can help your patient develop healthier behaviors and attain a better quality of life. Nurse can screen for hoarding disorder and make a referral to mental health services. In fact, the Hoarding Rating Scale (HRS) and Clutter Image Rating Scale (CIRS) are screening tools that can be used to identify people at risk for this disorder. The HRS uses verbal cues, whereas the CIRS uses pictures to help screen for hoarding behaviors.Figure: Because the chance of relapse is high, educate the patient's family about what to expect during therapy.Mental health professionals may use family intervention, medications, and cognitive behavioral therapy to treat hoarding disorder. Family intervention When family members are interested in an intervention for their loved one's compulsive hoarding behaviors, they first meet with a therapist to learn about the disorder and treatment options. The family members must be cohesive and agreeable to the intervention because the person with hoarding disorder can't be helped if the family fears the consequences of the intervention. Sometimes the therapist will have the family practice the intervention before it actually occurs. Next, the family members arrange to talk as a unit to their loved one about the effect of the clutter on their lives and what help is available. Each of the family members explains in a nonjudgmental way why he or she is concerned. It's a critical element for all family members to communicate clearly that the treatment is mandatory. The intervention may take place in the hoarder's home or in a therapist's office. The therapist may be present during the initial intervention or available immediately after the intervention for the first therapy session. The goal is that when the individual with hoarding disorder faces a cohesive group of people who are concerned for him or her, it will be hard to hide or minimize the problem. The intervention is just the first step. Both the hoarder and his or her family members must commit to ongoing therapy to address the issues that surround the behaviors and learn how to handle future behaviors or issues that may arise during the change process. Medications Selective serotonin reuptake inhibitors (SSRIs) are effective in treating the symptoms of OCD; however, they aren't as effective in treating people with hoarding disorder. If the individual with hoarding disorder has coexisting anxiety disorder or depression, he or she may benefit from treatment with SSRIs. Treatment of anxiety and depression may increase the success of engaging the compulsive hoarder in psychotherapy. Fluoxetine, sertraline, paroxetine, citalopram, and escitalopram are the SSRIs most commonly used. Potential adverse reactions of SSRIs include sexual dysfunction, gastrointestinal (GI) upset, mild sedation, and restlessness. For many patients, these adverse reactions diminish after 2 to 4 weeks of taking the medication. If they persist, the healthcare provider may prescribe a different type of SSRI. Patients shouldn't abruptly stop taking an SSRI; if they do, they may develop discontinuation syndrome. Symptoms of discontinuation syndrome include dizziness, headache, diarrhea, insomnia, irritability, nausea, and lowered mood. The FDA has a black box warning indicating that antidepressants may increase the risk of suicidal thinking and behavior in some people. Patients taking SSRI medications should be closely monitored for an increase in depression or the start of suicidal ideation for the first 4 weeks of treatment.Table: Managing adverse reactions of antidepressantsSSRIs use the same drug metabolizing enzyme pathway in the liver as other medications, such as anticoagulants, cardiac drugs, or drugs used to treat diabetes. Be prepared to intervene if the patient is showing signs of a drug-drug interaction while taking an SSRI. For example, if an SSRI is prescribed for a patient taking warfarin, his or her international normalized ratio should be closely monitored to ensure adequate anticoagulation. The warfarin dosage may have to be adjusted up or down. Serotonin syndrome is a potentially life-threatening drug interaction. It can occur when two medications that increase the serotonin level are combined, potentiating serotonin neurotransmission. The increased serotonin level throws off the body's autonomic regulation. Symptoms of serotonin syndrome include hyperthermia, restlessness, tachycardia, labile BP, changes in mental status, diaphoresis, and tremors. Serotonin syndrome progresses rapidly; if the early signs aren't promptly recognized, the patient can develop seizures and respiratory failure leading to coma. If your patient develops serotonin syndrome, immediately discontinue all medications, notify the healthcare provider, and treat the symptoms. The healthcare provider may prescribe medications to block the effects of the SSRI and treat hyperthermia and seizures. Tricyclic antidepressants (TCAs) may also be used to help with coexisting anxiety or depression. TCAs have a high anticholinergic effect due to their action on histamine receptors. Sedation, dry mouth, weight gain, and constipation are some of the adverse reactions that may result in nonadherence (see Managing adverse reactions of antidepressants). Cardiotoxicity can occur with slowing of cardiac conduction due to increased PR and QRS intervals. In addition, TCAs can be fatal in overdose, which may cause safety concerns for patients with suicidal ideation. Clomipramine is effective in treating the symptoms of OCD and may be effective for treating hoarding disorder. When any psychiatric medication is prescribed, thoroughly explain its actions and adverse reactions in appropriate language to the patient and family (see Medications used to treat hoarding disorder). Awareness of cultural differences is also valuable when initiating psychotropic treatment. In your assessment history, determine any genetic or dietary/herbal influences on metabolism. If English isn't the patient's primary language, take steps to ensure that he or she understands the information you provide. Enlisting the support of translators is advantageous to aid in adherence. Cognitive behavioral therapy Cognitive behavioral therapy is effective in treating hoarding disorder and considered a first-line treatment. In this therapy, the goal is to change a person's automatic thoughts, which occur spontaneously and lead to dysfunctional thinking. According to this type of therapy, psychological pain stems from what the person thinks an event means rather than what actually happened. In hoarding disorder, cognitive behavioral therapy focuses on excessive acquisition, difficulty discarding possessions, disorganization, and clutter.Table: Medications used to treat hoarding disorderThe therapist can first help the patient identify his or her beliefs and attachments to an item. Then an exploration into feelings about getting rid of possessions occurs. Finally, a strategy for organization and decision making about removal of the clutter is taught. After treatment begins, the change is slow, with relapse twice as high due to the difficulty that the compulsive hoarder may have in understanding or accepting the disorder. Because the belongings are often viewed as an extension of the person who hoards, many patients will resist the change or procrastinate. Group treatment can be an important part of recovery for the compulsive hoarder. Shame and isolation can be decreased by getting people together who have a similar problem. By listening to each other's stories, groups can increase motivation and instill hope. The Internet may be helpful in supporting compulsive hoarders. Virtual self-help groups provide advice for dealing with the clutter or shame. Web-based cameras can allow family members who live far away to monitor a relapse into hoarding behavior. Also, virtual face-to-face therapy can be facilitated through sites such as Skype. Your eyes and ears are needed Nurses are key in the identification of patients with hoarding disorder. When a person seeks healthcare for a medical problem, you can listen to the patient and family's concerns about the living environment. Observe if the patient is collecting things during his or her hospital stay, such as medicine cups, silverware, or trays. Does the patient have trouble throwing out items while hospitalized? Part of your nursing assessment can include questions about collections that the patient may have, any difficulty he or she has discarding items from the collection, trouble organizing the collection, or hard feelings when someone else touches items from the collection. Talk to family members to get an additional view on the behaviors. The first step is to build trust with a caring attitude and nonjudgmental approach that demonstrates respect for the patient. Encourage the patient to talk about the effects of hoarding behaviors. To provide this open communication, first examine your own feelings about compulsive hoarding. Empathy with the patient involves trying to understand the attachment that he or she has to the belongings. Instead of confronting the patient about hoarding behaviors, help him or her identify feelings attached to the behaviors and the reasons that giving them up may be hard. Next, discuss the consequences of hoarding behaviors, such as eviction, loss of valuable relationships, and safety hazards. Then talk about the benefits of changing these behaviors, such as reuniting with loved ones and increased living space. Help the patient identify ways that he or she can improve home safety, such as removing clutter from hallways, stoves, and heating sources. Because discussion about hoarding behaviors can cause anxiety, teaching the patient relaxation techniques may be valuable. When social agencies or law enforcement are involved, include family members or social workers so that an action plan can be communicated to the agency. Referral to a mental health professional who has expertise in hoarding disorder can be valuable for both the patient and the family. Offer a list of practitioners or help the patient call for an appointment. Because the family will probably be involved in the intervention, include them during education about hoarding disorder and offer local resources that can help. With the family, stress the importance of not just emptying the house of all belongings as a quick solution to the problem; the hoarding behaviors will return. It's important that the compulsive hoarder be involved in removing the items so he or she can develop decision-making and sorting skills and have a sense of control over the environment. Lastly, after the patient with hoarding disorder is in treatment, family members should continue to encourage and coach healthy behaviors. It's important to acknowledge the patient's need for control and emotional attachment to the items. Also, family members must recognize that progress will be slow and the patient may relapse. Healthier and happier Although there's no cure for hoarding disorder, you can help your patient develop healthier behaviors and become more content with daily living and relationships. key points Nursing interventionsFigure Build trust with the patient and be nonjudgmental. Assess the patient's hoarding behaviors. Instruct the patient about stress reduction techniques. Teach the patient about the possible consequences or risks of hoarding behaviors. Help the patient set small, realistic goals to remove clutter. Educate the patient's family members about hoarding disorder and to expect slow change and possible relapse. on the webFigure American Psychiatric Association: http://www.psychiatry.org/hoarding-disorder Anxiety and Depression Association of America: http://www.adaa.org/understanding-anxiety/obsessive-compulsive-disorder-ocd/hoarding-basics Columbia University Medical Center: http://www.columbiapsychiatry.org/hoarding/ International OCD Foundation: http://www.ocfoundation.org/hoarding/ Mayo Clinic: http://www.mayoclinic.com/health/hoarding/DS00966 University of California, San Diego: http://psychiatry.ucsd.edu/OCD_hoarding.html
We present HamleDT - a HArmonized Multi-LanguagE Dependency Treebank. HamleDT is a compilation of existing dependency treebanks (or depen- dency conversions of other treebanks), transformed so that they all conform to the same annotation style. In the present article, we provide a thorough investigation and discussion of a number of phenomena that are comparable across languages, though their annotation in treebanks often differs. We claim that transformation procedures can be designed to automatically identify most such phenomena and convert them to a unified annotation style. This unification is beneficial both to comparative corpus linguistics and to machine learning of syntactic parsing.
ILSP Dependency Parser is a tool trained on the Greek Dependency Treebank, a resource which comprises data annotated at several linguistic levels. Training data at the level of syntax consisted of ~70 KWords annotated using a dependency-based syntactic scheme that includes 25 main relations.
poster
The Penn Discourse Treebank (PDTB) was released to the public in 2008. It remains the largest manually annotated corpus of discourse relations to date. Its focus on discourse relations that are either lexically-grounded in explicit discourse connectives or associated with sentential adjacency has not only facilitated its use in language technology and psycholinguistics but also has spawned the annotation of comparable corpora in other languages and genres. Given this situation, this paper has four aims: (1) to provide a comprehensive introduction to the PDTB for those who are unfamiliar with it; (2) to correct some wrong (or perhaps inadvertent) assumptions about the PDTB and its annotation that may have weakened previous results or the performance of decision procedures induced from the data; (3) to explain variations seen in the annotation of comparable resources in other languages and genres, which should allow developers of future comparable resources to recognize whether the variations are relevant to them; and (4) to enumerate and explain relationships between PDTB annotation and complementary annotation of other linguistic phenomena. The paper draws on work done by ourselves and others since the corpus was released.
The main objective of the Rhapsodie project (ANR Rhapsodie 07 Corp-030-01) was to define rich, explicit, and reproducible schemes for the annotation of prosody and syntax in different genres (± spontaneous, ± planned, face-to-face interviews vs. broadcast, etc.), in order to study the prosody/syntax/discourse interface in spoken French, and their roles in the segmentation of speech into discourse units (Lacheret, Kahane, & Pietrandrea forthcoming). We here describe the deliverable, a syntactic and prosodic treebank of spoken French, composed of 57 short samples of spoken French (5 minutes long on average, amounting to 3 hours of speech and 33000 words), orthographically and phonetically transcribed. The transcriptions and the annotations are all aligned on the speech signal: phonemes, syllables, words, speakers, overlaps. This resource is freely available at www.projet-rhapsodie.fr. The sound samples (wav/mp3), the acoustic analysis (original F0 curve manually corrected and automatic stylized F0, pitch format), the orthographic transcriptions (txt), the microsyntactic annotations (tabular format), the macrosyntactic annotations (txt, tabular format), the prosodic annotations (xml, textgrid, tabular format), and the metadata (xml and html) can be freely downloaded under the terms of the Creative Commons licence Attribution - Noncommercial - Share Alike 3.0 France. The metadata are encoded in the IMDI-CMFI format and can be parsed on line.
The Norwegian Dependency Treebank is a new syntactic treebank for Norwegian Bokmål and Nynorsk with manual syntactic and morphological annotation, developed at the National Library of Norway in collaboration with the University of Oslo. It is the first publically available treebank for Norwegian. This paper presents the core principles behind the syntactic annotation and how these principles were employed in certain specific cases. We then present the selection of texts and distribution between genres, as well as the annotation process and an evaluation of the inter-annotator agreement. Finally, we present the first results of data-driven dependency parsing of Norwegian, contrasting four state-of-the-art dependency parsers trained on the treebank. The consistency and the parsability of this treebank is shown to be comparable to other large treebank initiatives.\n\nProceedings of the LREC 2014, Ninth International Conference on Language Resources and Evaluation, Reykjavik, Iceland. http://www.lrec-conf.org/proceedings/lrec2014/index.html.
Automatic prediction of emotions requires reliably annotated data which can be achieved using scoring or pairwise ranking. But can we predict an emotional score using a ranking-based annotation approach? In this paper, we propose to answer this question by describing a regression analysis to map crowdsourced rankings into affective scores in the induced valence-arousal emotional space. This process takes advantages of the Gaussian Processes for regression that can take into account the variance of the ratings and thus the subjectivity of emotions. Regression models successfully learn to fit input data and provide valid predictions. Two distinct experiments were realized using a small subset of the publicly available LIRIS-ACCEDE affective video database for which crowdsourced ranks, as well as affective ratings, are available for arousal and valence. It allows to enrich LIRIS-ACCEDE by providing absolute video ratings for the whole database in addition to video rankings that are already available.
In this paper, we report our preliminary efforts in building an English-Turkish parallel treebank corpus for statistical machine translation. In the corpus, we manually generated parallel trees for about 5,000 sentences from Penn Treebank. English sentences in our set have a maximum of 15 tokens, including punctuation. We constrained the translated trees to the reordering of the children and the replacement of the leaf nodes with appropriate glosses. We also report the tools that we built and used in our tree translation task.
Face judgments of dominance play an important role in human social interaction. Perceived facial dominance is thought to indicate physical formidability, as well as resource acquisition and holding potential. Dominance cues in the face affect perceptions of attractiveness, emotional state, and physical strength. Most experimental paradigms test perceptions of facial dominance in individual faces, or they use manipulated versions of the same face in a forced-choice task but in the absence of other faces. Here, we extend this work by assessing whether dominance ratings are absolute or are judged relative to other faces. We presented participants with faces to be rated for dominance (target faces), while also presenting a second face (non-target faces) that was not to be rated. We found that both the masculinity and sex of the non-target face affected dominance ratings of the target face. Masculinized non-target faces decreased the perceived dominance of a target face relative to a feminized non-target face, and displaying a male non-target face decreased perceived dominance of a target face more so than a female non-target face. Perceived dominance of male target faces was affected more by masculinization of male non-target faces than female non-target faces. These results indicate that dominance perceptions can be altered by surrounding faces, demonstrating that facial dominance is judged at least partly relative to other faces.
The article focuses on the changes in affective (emotional) ratings under the influence of the choice of one of the interpretations of ambiguous images and subsequent recognition task performance. Earlier studies showed that recognition decision affects subsequent ratings of the stimulus: the more information is accumulated about the stimulus, the more positive will be its ratings at recognition, and the more negative at non-recognition (Chetverikov, 2014). We hypothesized that a choice of a single interpretation of a stimulus becomes a source of information for subsequent decision concerning recognition or non-recognition of the unambiguous interpretation of that stimulus. Thus, this decision will affect subsequent ratings of stimuli the same way as in the case of initially unambiguous stimuli. The experimental results confirmed our hypothesis. Ratings of unambiguous stimuli corresponding to selected and non-selected interpretation of ambiguous stimuli varied depending on the recognition decision in the same way as did ratings of previously presented and new unambiguous stimuli. When a stimulus is «old» and is recognized, it is liked more, than a recognized «new» stimulus; when it is not recognized, the effect is opposite. Thus, the more information about the stimulus has been accumulated, the higher is the influence of a decision concerning stimulus recognition on subsequent ratings. Similar results were found for confidence ratings. These were higher in the case of recognition than in the case of non-recognition, but the difference between the two situations was more pronounced for «old» stimuli than for «new» ones.
BACKGROUND: Cystic fibrosis (CF) is a progressive, life-shortening disease currently treated symptomatically. Among the consequences of daily treatments and long-term progression of the disease is significant discomfort and pain. This study measured pain systematically in adolescents with CF using a pain diary, and evaluated its associations with treatment adherence, psychological symptoms, and health-related quality of life (HRQOL). METHODS: This study was part of a larger, multi-center study. Ninety-five adolescents with a mean age of 15.8 were enrolled. A total of 413 online pain diaries were completed for 6 days following a routine clinic visit. Diaries measured pain intensity, location, duration, affective ratings, and coping responses. Adherence to pulmonary medications was measured using prescription refill data; other measures included ratings of depression, anxiety and HRQOL. RESULTS: Pain was reported by 74.5% of participants, generally in the mild range, averaging 2.1 on a 10-point scale. Daily pain ratings were highly variable both within and between participants. Pain was significantly associated with worse adherence, more psychological distress, and worse HRQOL. CONCLUSIONS: Results indicated that pain is a common problem for adolescents with CF and negatively affects their disease management, psychological symptoms and health outcomes. Routine assessment of pain and systematic studies of interventions to treat pain are recommended. Pediatr Pulmonol. 2015; 50:244-251. © 2014 Wiley Periodicals, Inc.
Objectives: This study seeks to provide normative data on the production frequency and semantic typicality and familiarity of noun exemplars by semantic category for elderly individuals. Methods: A total of 198 individuals participated in the study. In Experiment 1, participants were categorized into two groups: professionals involved in dementia care and primary care-givers of patients with dementia. They ranked 17 semantic categories based on the priority of the stimuli required for naming treatment. Forty-five normal elderly individuals (NEIs) participated in Experiment 2, in which they were administered a generative naming task. Seventy-eight NEIs participated in Experiment 3, where they rated the semantic typicality and familiarity of each noun exemplar obtained from Experiment 2. Results: Experiment 1 revealed that 'food (dish)' was ranked as a top priority followed by 'clothes', 'fruits', and 'body parts'. In Experiment 2, 'body parts' generated the highest mean of production frequency among 17 semantic categories. Results from Experiment 3 suggested that the semantic typicality was positively and strongly correlated with the familiarity ratings. Conclusion: These results are clinically important and meaningful given that they provide preliminary guidelines to select stimuli for the naming treatment of individuals with neurogenic language disorders.
AIMS: To explore the enhancing effect of alcohol consumption on attractiveness ratings, in that few studies on the Beer Goggles effect control the stimuli attractiveness level and researchers have seldom considered extending the effect to stimuli other than faces. METHODS: Male and female participants (n = 103) were randomly assigned to alcohol consumption or placebo groups. Both groups were asked to assess the attractiveness of two types of pictures (faces and landscapes) with three levels of attractiveness for each stimulus category (high, moderate and low). RESULTS: We found significant interactions between beverage type and attractiveness level. Attractiveness ratings for moderate- and low-attractiveness faces were significantly higher in the alcohol compared with placebo condition, while there was no significant difference for high-attractiveness stimuli between these two conditions. As for landscapes, only low-attractiveness stimuli were rated significantly higher in the alcohol condition. CONCLUSION: Whether or not alcohol consumption leads to an increase in attractiveness ratings depends on the initial attractiveness of the stimulus materials. Alcohol consumption tends to affect ratings for stimuli with relatively low attractiveness. Furthermore, this effect is not limited to faces; it extends to other types of stimuli like landscapes.
Key to fast adaptation of language technologies for any language hinges on the availability of fundamental tools and resources such as monolingual/parallel corpora, annotated corpora, part-of-speech (POS) taggers, parsers and so on. The languages which lack those fundamental resources are often referred as under-resourced\nlanguages.\n\nIn this thesis, we address the problem of cross-lingual dependency parsing of under-resourced languages. We apply three methodologies to induce dependency structures: (i) projecting dependencies from a resource-rich language to under-resourced languages via parallel corpus word alignment links (ii) parsing under-\nresourced languages using parsers whose models are trained on treebanks of other\nlanguages, and do not look at actual word forms, but only on POS categories. Here\nwe address the problem of incompatibilities in annotation styles between source side parsers and target side evaluation treebanks by harmonizing annotations to a common standard; and finally (iii) we add a new under-resourced scenario in which we use machine translated parallel corpora instead of human translated corpora for\nprojecting dependencies to under-resourced languages.\n\nWe apply the aforementioned methodologies to five Indian languages (ILs): Hindi, Urdu, Telugu, Bengali and Tamil (in the order of high to low availability of treebank data). To make the evaluation possible for Tamil, we develop a depen\ndency treebank resource for Tamil from scratch and we use the created data in\nevaluation and as a source in parsing other ILs. Finally, we list out strategies that\ncan be used to obtain dependency structures for target languages under different\nresource-poor scenarios.
We report of the procedures of developing a large representative corpus of 50,000 sentences taken from clinical notes. Previous reports of annotated corpus of clinical notes have been small and they do not represent the whole domain of clinical notes. The sentences included in this corpus have been selected from a very large raw corpus of ten thousand documents. These ten thousand documents are sampled from an internal repository of more than 700,000 documents taken from multiple health care providers. Each of the documents is de-identified to remove any PHI data. Using the Penn Treebank tagging guidelines with a bit of modifications, we annotate this corpus manually with an average inter-annotator agreement of more than 98%. The goal is to create a parts of speech annotated corpus in the clinical domain that is comparable to the Penn Treebank and also represents the totality of the contemporary text as used in the clinical domain. We also report the output of the TnT tagger trained on the initial 21,000 annotated sentences reaching a preliminary accuracy of above 96%.]
The authors use a self-built Chinese Discourse Treebank(80% relations are implicit) to recognize implicit relations. In this corpus, discourse relations are divided into three layers, the first layer has four types: causality, coordination, transition and explanation. Based on this corpus, maximum entropy classifier is employed to identify four types relations with context, lexical and dependency parse features. Experimental results show that total accuracy is 62.15% and the identification effect of coordination is the best, F1 reaches 75.26%.
Research in Sentiment Analysis has shown rapid progress since late 90s. It is an important research area as analyzing user’s feedback is useful for business analysis, product comparison, counter intelligence, and poll prediction. Despite the rapid surge of Sentiment Analysis research, many unresolved research questions remain. One of the biggest concerns is the Semantic Gap, which involves translating machine understandable form to human understandable form. Though research has been carried out for machine to understand human language, it is still not capable to address the problem mentioned as human languages are diverse and complex. WordNet, for example, attempt to address this issue by incorporating large lexical database for English, with various functionalities to manipulate this database. Recently, WordNet provides multilingual support, which is very helpful to address the diverse human languages. In this paper, we propose a novel multilingual common ontology tool to analyze user’s feedback and opinion. Unlike other existing state of the art tools, our tool is capable of handling multi languages regardless of the webpage layout. Experimental results show that our tool is highly efficient in analyzing opinion from social networking sites.
Facial expressions are one of the most important ways of non-verbal communication for humans. To date, most research in this field has focused solely on emotional aspects, largely neglecting the communicative and conversational aspects of expressions. Furthermore, although there is evidence for some degree of cross-cultural universality among emotional expressions, much less is known about how facial expressions in general are perceived across cultures. Here, we investigate the structure of the complex space of both emotional and conversational expressions in a cross-cultural context. The two experiments reported here used matching video sequences of 27 expressions from both the KU (Korean) facial expression database and the MPI (German) facial expression database (each expression was shown by 6 actors, totaling 162 videos from each database). In the first experiment, four groups (each n=20) of native German and Korean participants were asked to group the sequences of the German or Korean databases into clusters based on similarity. This grouping data yielded four different confusion matrices. In the second experiment, another four groups of participants (each n=20) from both cultures were asked to rate each video according to 13 emotional and conversational attributes. This rating data yielded an averaged 13-dimensional vector for each sequence. For each of the four grouping/rating data-pairs, we then used kernel canonical correlation analysis (KCCA) to determine a two-dimensional embedding of expressions that best explained both grouping and rating data. Although other attributes contributed as well, the two dimensions recovered by KCCA showed maximal correlation with valence and arousal ratings – this was true regardless of participants' cultural backgrounds or of the database that was used. Our results show that evaluative dimensions for both German and Korean cultural contexts are highly similar, confirming that cultural universals exist even in this complex space of emotional and conversational facial expressions. Meeting abstract presented at VSS 2014
My study assessed the relationship between the colour of depicted clothing and ratings of emotional intensity in drawings of emotional scenarios. Participants (N = 42) viewed a set of drawings in one of seven colours, labeled the Actor and Cause, and rated the intensity of the emotions depicted on 11 emotional scales. Participants also listed colours they associated with specific emotions. Colour of clothing in drawings was not found to affect ratings of emotional intensity, but in drawings of sadness, embarrassment, and empathy the Actor obtained higher ratings than did the Cause. Colour-emotion associations obtained separately were, however, generally consistent with previous findings. COLOUR AND EMOTION 3 Colour and Emotional Intensity Emotion recognition is an evolutionary advantage that can inform behavioural decisions (Elfenbein, Foo, White, Tan & Aik, 2007) and help individuals to effectively navigate social situations (Van Kleef, 2010). It is a process that is largely automatic and unconscious, thus requiring limited cognitive resources (Tracy & Robins, 2008). However, researchers in the areas of social, cognitive and perceptual psychology are still attempting to determine the factors contributing to the automaticity of emotion perception.
Introduction: The last decade has witnessed a growing interest in human social cognition. Social neuroscience has differentiated at least two functions and the respective neural networks that support successful interaction: a network that allows for affect sharing (empathizing) and underlies the understanding of others’ affective states and a network that enables the inference of thoughts, beliefs, and goals and underlies understanding others’ mental states (Theory of Mind (ToM)). Empathy paradigms have mainly focused on witnessing pain or suffering and have revealed increased activation in the anterior insula (AI) and the anterior midcingulate cortex (aMCC) [1]. ToM tasks typically involve inferring others’ thoughts or intentions and induce increased activation in the temporoparietal junction (TPJ), the temporal poles and the dorsomedial prefrontal cortex (dmPFC) [2]. Although these different routes of social cognition show differential time courses during development [3] and are differentially impaired in psychopathology [4], these routes usually interact in healthy brains. So far, fMRI paradigms are lacking, which can reliably dissociate mentalizing and empathizing routes within a single person. The aim of the present study was to develop and validate an fMRI task in which demands on the affective and the cognitive social cognition route are manipulated independently and that therefore allows the investigation of their interaction as well as differential impairment. Methods: We developed the EmpaToM, a 30 minute paradigm presenting participants with naturalistic video stimuli (~15 seconds) in which people recount auto-biographical episodes that are either emotionally negative (e.g. loss of a loved one) or neutral (control condition; e.g., commuting to work). Each video is followed by ratings of affect and of empathic concern. Subsequently, specific questions about the content of the video are probing either ToM (questions about the mental states of people in the video) or no ToM (control condition; factual reasoning). The task, hence, follows a 2 (negative emotional load versus no emotional load) x 2 (ToM requirements versus factual reasoning) factorial design. In a first validation study, 19 participants were tested in the EmpaToM and classical tasks measuring empathy (the Social Video Task (SoVT) [5]) and ToM [2] (3T Siemens Verio Scanner; TR = 2000; TE =; 37 slices (2 mm)). Results: Viewing negative emotional compared to neutral videos increased activation in a distinct network including the bilateral AI and aMCC. This empathy network overlapped with activity elicited by the SoVT [5] and coordinates derived from a meta-analysis on empathy for pain paradigms [1]. Affect ratings and emotional concern ratings after the emotionally negative videos were correlated with the IRI subscale ‘empathic concern’ and with SoVT concern ratings. Furthermore, activity in the aMCC and right AI was parametrically modulated by the subjective rating of negative emotion experienced after each video. Contrasting ToM- and factual reasoning during the questions activated the well-described network including bilateral TPJ, temporal poles and mPFC. Again, these clusters largely overlapped with a classical false-belief ToM-task [2]. A subsequent study with 200 participants replicated these patterns. Conclusions: The present results confirm that the EmpaTom is able to reliably identify and separate the networks underlying our abilities to empathize (affective route) and to mentalize (cognitive route) within individuals and one task. This makes the EmpaToM a suitable task for the efficient differentiation of empathy and ToM related networks and for investigating their interaction. The EmpaToM could be of tremendous use for both the comprehensive characterization of different types of deficits in psychopathology as well as for the measurement of differential training effects and interventions both on the level of behavior and brain functioning. References [1] Lamm, C., Decety, J., & Singer, T. (2011). ‘Meta-analytic evidence for common and distinct neural networks associated with directly experienced pain and empathy for pain’, NeuroImage, vol. 54, pp. 2492-2502. [2] Dodell-Feder, D., Koster-Hale, J., Bedny, M., Saxe, R. (2011). ‘fMRI item analysis in a theory of mind task’, Neuroimage, vol. 55, pp. 705-12. [3] Singer, T. (2006). ‘The neuronal basis and ontogeny of empathy and mind reading: Review of literature and implications for future research’, Neuroscience & Biobehavioral Reviews, vol. 30, pp. 855-863. [4] Bird, G., Silani, G., Brindley, R., White, S., Frith, U., & Singer, T. (2010). ‘Empathic brain responses in insula are modulated by levels of alexithymia but not autism’, vol., 133, pp. 1515-1525. [5] Klimecki, O.M., Leiberg, S., Lamm, C., & Singer, T. 2013). ‚Functional Neural Plasticity and Associated Changes in Positive Affect After Compassion Training’. Cerebral Cortex, vol. 23, pp. 1552-1561.
Machine assistance is vital to managing the cost of corpus annotation projects. Identifying effective forms of machine assistance through principled evaluation is particularly important and challenging in under-resourced domains and highly heterogeneous corpora, as the quality of machine assistance varies. We perform a fine-grained evaluation of two machine-assistance techniques in the context of an under-resourced corpus annotation project. This evaluation requires a carefully controlled user study crafted to test a number of specific hypotheses. We show that human annotators performing morphological analysis of text in a Semitic language perform their task significantly more accurately and quickly when even mediocre pre-annotations are provided. When pre-annotations are at least 70 % accurate, annotator speed and accuracy show statistically significant relative improvements of 25–35 and 5–7 %, respectively. However, controlled user studies are too costly to be suitable for under-resourced corpus annotation projects. Thus, we also present an alternative analysis methodology that models the data as a combination of latent variables in a Bayesian framework. We show that modeling the effects of interesting confounding factors can generate useful insights. In particular, correction propagation appears to be most effective for our task when implemented with minimal user involvement. More importantly, by explicitly accounting for confounding variables, this approach has the potential to yield fine-grained evaluations using data collected in a natural environment outside of costly controlled user studies.
This article analyses the structure of Yoruba numerals and their derivation. Data are collected from the compilation of Yoruba numerals and observation of its use coupled with the researcher's intuitive knowledge of the language. The work dwells on the existing literature on numerals too. The author adopts a descriptive method in analysing the data. The work looks at the roles of affixes in realising odd numbers, multiples of 20, centenary, bicentenary, and so on in their order of increase. It is discovered that the direction of counting in Yoruba is largely progressive. Besides, the language adopts base 5, decimal (base 10) and vigesimal (base 20) systems of counting. It is equally discovered that the choice of either of the two variations is largely dependent on the articulatory parameter of the first vowel (V1) of the root word. It is noted that the Yoruba numeral system offers a suitable linguistic database for both the theoretical and empirical domains of linguistic study especially documentary linguistics. The current study has general pedagogic implications for the teaching and learning of Yoruba numerals.
In the context of processing Bengali words through a computer, there may arise several issues that are directly linked with surface structure of words. These issues may create problems in manual and computer-based counting of number of words in a corpus. They can also create problems in morphological processing of words. These issues come up because there is hardly any consistency in orthographic representation of words in written Bengali texts. The high irregularities in writing of inflected words, proper names, adjectival forms, adverbial forms, compound words, reduplicated words, onomatopoeic words, hyphenated words, etc. present a daunting task before an investigator in normalizing the surface forms of words for generating a lexical database as well as developing a word processing system for the works of language technology.
The French Lexical Network (fr-LN) is a global model of the French lexicon presently under construction. The fr-LN accounts for lexical knowledge as a lexical network structured by paradigmatic and syntagmatic relations holding between lexical units. This paper describes how morphological knowledge is presently being introduced into the fr-LN through the implementation and lexicographic exploitation of a dynamic morphological model. Section 1 presents theoretical and practical justifications for the approach which we believe allows for a cognitively sound description of morphological data within semantically-oriented lexical databases. Section 2 gives an overview of the structure of the dynamic morphological model, which is constructed through two complementary processes: a Morphological Process--section 3--and a Lexicographic Process--section 4.
The morphological constituents of English compounds (e.g., “butter” and “fly” for “butterfly”) and two-character Chinese compounds may differ in meaning from the whole word. Subjective differences and ambiguity of transparency make judgments difficult, and a computational alternative based on a general model might be a way to average across subjective differences. In the present study, we propose two approaches based on latent semantic analysis (Landauer & Dumais in Psychological Review 104:211–240, 1997): Model 1 compares the semantic similarity between a compound word and each of its constituents, and Model 2 derives the dominant meaning of a constituent from a clustering analysis of morphological family members (e.g., “butterfingers” or “buttermilk” for “butter”). The proposed models successfully predicted participants’ transparency ratings, and we recommend that experimenters use Model 1 for English compounds and Model 2 for Chinese compounds, on the basis of differences in raters’ morphological processing in the different writing systems. The dominance of lexical meaning, semantic transparency, and the average similarity between all pairs within a morphological family are provided, and practical applications for future studies are discussed.
Collocations in the sense of idiosyncratic binary lexical co-occurrences are one of the biggest challenges for any language learner. Even advanced learners make collocation mistakes in that they literally translate collocation elements from their native tongue, create new words as collocation elements, choose a wrong subcategorization for one of the elements, etc. Therefore, automatic collocation error detection and correction is increasingly in demand. However, while state-of-the-art models predict, with a reasonable accuracy, whether a given co-occurrence is a valid collocation or not, only few of them manage to suggest appropriate corrections with an acceptable hit rate. Most often, a ranked list of correction options is offered from which the learner has then to choose. This is clearly unsatisfactory. Our proposal focuses on this critical part of the problem in the context of the acquisition of Spanish as second language. For collocation error detection, we use a frequency-based technique. To improve on collocation error correction, we discuss three different metrics with respect to their capability to select the most appropriate correction of miscollocations found in our learner corpus.
Crowdsourcing services—particularly Amazon Mechanical Turk—have made it easy for behavioral scientists to recruit research participants. However, researchers have overlooked crucial differences between crowdsourcing and traditional recruitment methods that provide unique opportunities and challenges. We show that crowdsourced workers are likely to participate across multiple related experiments and that researchers are overzealous in the exclusion of research participants. We describe how both of these problems can be avoided using advanced interface features that also allow prescreening and longitudinal data collection. Using these techniques can minimize the effects of previously ignored drawbacks and expand the scope of crowdsourcing as a tool for psychological research.
The Timed Antagonistic Response Alethiometer (TARA) is a true–false statement classification task that diagnoses lying on the basis of slower average response speeds. Previous research (Gregg in Applied Cognitive Psychology, 21, 621–647, 2007) showed that a computer-based TARA was about 80 % accurate when its statements conveyed demographic facts or religious views. Here, we tested the TARA’s diagnostic potential when its statements conveyed attitudes—here, toward both branded and generic consumer products—across different versions of the TARA (Exps. 1a, 1b, and 1c), as well as across consecutive administrations (Exp. 2). The results generalized well across versions, and maximal accuracy rates exceeding 80 % were obtained, although accuracy declined somewhat upon readministration. Overall, the TARA shows promise as a comparatively cheap, convenient, and diagnostic index of lying about attitudes.