Word Categories Research Articles

Toward explainable AI (XAI) for mental health detection based on language behavior.

Advances in artificial intelligence (AI) in general and Natural Language Processing (NLP) in particular are paving the new way forward for the automated detection and prediction of mental health disorders among the population. Recent research in this area has prioritized predictive accuracy over model interpretability by relying on deep learning methods. However, prioritizing predictive accuracy over model interpretability can result in a lack of transparency in the decision-making process, which is critical in sensitive applications such as healthcare. There is thus a growing need for explainable AI (XAI) approaches to psychiatric diagnosis and prediction. The main aim of this work is to address a gap by conducting a systematic investigation of XAI approaches in the realm of automatic detection of mental disorders from language behavior leveraging textual data from social media. In pursuit of this aim, we perform extensive experiments to evaluate the balance between accuracy and interpretability across predictive mental health models. More specifically, we build BiLSTM models trained on a comprehensive set of human-interpretable features, encompassing syntactic complexity, lexical sophistication, readability, cohesion, stylistics, as well as topics and sentiment/emotions derived from lexicon-based dictionaries to capture multiple dimensions of language production. We conduct extensive feature ablation experiments to determine the most informative feature groups associated with specific mental health conditions. We juxtapose the performance of these models against a "black-box" domain-specific pretrained transformer adapted for mental health applications. To enhance the interpretability of the transformers models, we utilize a multi-task fusion learning framework infusing information from two relevant domains (emotion and personality traits). Moreover, we employ two distinct explanation techniques: the local interpretable model-agnostic explanations (LIME) method and a model-specific self-explaining method (AGRAD). These methods allow us to discern the specific categories of words that the information-infused models rely on when generating predictions. Our proposed approaches are evaluated on two public English benchmark datasets, subsuming five mental health conditions (attention-deficit/hyperactivity disorder, anxiety, bipolar disorder, depression and psychological stress).

Words count in psychotherapy: Differentiating language characteristics of cognitive behavioral therapy and focal psychodynamic therapy for anorexia nervosa.

It is generally assumed that psychodynamic therapy and cognitive behavioral therapy (CBT) differ in terms of applied techniques and processes. To date, however, little is known about whether and how such differences can actually be observed at a basic linguistic level and in what the two treatment approaches differ most strongly (i.e., how psychodynamic and CBT therapists differ in what they actually say word-by-word in therapy sessions). Building on theoretical models and previous research that used observer ratings, we formulated specific hypotheses regarding which word categories psychodynamic and CBT therapists who treat patients with an eating disorder should differ in. To investigate these hypotheses, we used verbatim transcripts from 297 therapy sessions of a randomized controlled trial in which patients with anorexia nervosa (n = 88) received either focal psychodynamic therapy (FPT) or CBT. These transcripts were then examined using computerized quantitative text analysis. In line with our hypotheses, we found that CBT therapists overall spoke more than their FPT counterparts and that they used more words related to eating. Also in line with our hypotheses, FPT therapists used more words related to social processes. Contrary to our expectations, CBT therapists did not show a stronger focus on the future but talked more about emotions than FPT therapists. The latter effect, however, appears to be driven by a stronger focus on positive emotions. These findings suggest that computerized quantitative text analysis can differentiate meaningful language characteristics of CBT and FPT on spoken-word level and that it holds potential as a tool for researchers and therapists. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

An Exploration of English Translation Methods for Long Sentences Combined with Modal Analysis from the Perspective of Translation Purpose

Abstract The complex sentence structure and complicated semantic logic of English long sentences are the difficult points that need to be overcome urgently in English translation. This paper outlines the basic direction of machine translation based on the perspective of translation purpose. Based on the grammatical categories of English words, lexical annotation is completed, and regular expressions for matching English long sentences are proposed based on the annotation results. After realizing the preprocessing of English text, the features of the text are extracted and converted into the format required by the CRF model, and the long-sentence slicing is completed through the training of CRF. The cut Encoder encoding and Decoder decoding were utilized to generate the final machine translation result. In the test of the United Nations machine translation corpus, this paper’s method improves the translation performance in the English-Chinese direction by 0.29 Bleu value and the translation performance in the Chinese-English direction by 0.13 Bleu value compared with the method of Xu and Zens. The translation performance is effectively improved by combining long sentence slice translation with modal analysis.

A Morpho-Syntactic Analysis of Gender-Fair Language in Advertisements of Cosmetic Products in Ghana

Gender-fair language challenges and dismantles prevailing societal gender preconceptions. Cosmetic advertisements have been criticized for unfair language choices that perpetuate gender stereotypes. This study is a morphosyntactic examination of gender-fair language in cosmetic advertisements. The social constructionist theory underpins the study. Thirty (30) product packages and labels for the Ghanaian market were used in this qualitative descriptive survey. Gendered terms were analyzed on the morphological level using Quirk et al.’s (1985) functional principle of word categorization. The data collected revealed that Gender-unfair language is prevalent in cosmetic advertisements. Product names, descriptions and claims often use gendered nouns, pronouns, adjectives, phrases of different types and imperative clauses. This reinforces the idea that certain beauty concerns and products are exclusively tailored to specific genders. Gender-unfair language in cosmetic advertisements results in a limited representation of beauty ideals and practices. The use of inclusive language and representations that challenge gender stereotypes and promote diversity was recommended for cosmetic advertisers. The paper provides insight into academic discourse and practical approaches in gender, communication and marketing studies. Keywords: Advertisements, Cosmetics, Gender-Fair Language, Stereotype, Morphosyntactic

Accounting for the Concreteness and Neighborhood Effects in a High Frequency Word List for Poor Readers

Some poor readers show little or no progress in literacy interventions as their susceptibility to the concreteness and neighborhood effect is not accounted for during intervention. This study aims to develop a resource for poor readers by revising the Dolch list to account for the concreteness and neighborhood (orthographic, phonological and semantic) effect. Psycholinguistic techniques were employed to recategorize 220 Dolch list words according to concreteness via function and content word categories, and include the associated orthographic, phonological and semantic neighbors of each word into a new High Frequency List with Neighbors (HFLN). One-way analysis of variance (ANOVA), Bonferroni post hoc test and Levene’s test of variance homogeneity were carried out as measures of statistical significance and variability. The HFLN contains a total of 220 words with 1057 neighbors across five function and content word categories. Both measures of statistical significance and variability show that grade categories in the Dolch list contain greater mean concreteness values with overlapping similarities and higher variability. Conversely, the HFLN effectively delineates concreteness value clusters between categories with lower variability. The HFLN aids in targeted intervention of poor readers by presenting the available orthographic, phonological and semantic neighbors according to the descending order of concreteness.

95 Delving Beyond the Test Score: Linguistic Markers of Cognitive Impairment on Paragraph Recall

Objective:Cognitive tests requiring spoken responses, such as paragraph recall, are rich in cognitive-related information that is not captured using traditional scoring methods. This study aimed to determine if linguistic features embedded in spoken responses may differentiate between individuals who are and are not cognitively impaired.Participants and Methods:Participants in the Long Life Family Study completed a neuropsychological assessment which included the WMS-R Logical Memory I paragraph recall. For a subset of participants (N=709), test responses were digitally recorded and manually transcribed. We used Linguistic Inquiry Word Count, a text analysis program, to quantify word counts, grammatical features (e.g, prepositions, verb tenses), and the use of content words related to specific semantic categories (e.g., work-related, numbers) for immediate (IR) and delayed recall (DR). We used regression models with Generalized Estimating Equations adjusted by age, sex, education, and within-family correlation to select features associated with cognitive status (normal cognition [NC] versus cognitive impairment [CI]; Bonferroni-corrected threshold p<0.001). Next, we developed a “polyfeature score” (PFS) for both immediate and delayed recall, each calculated as a weighted sum of the selected linguistic features. We then built a logistic regression model to evaluate the predictive value of each PFS for identifying cognitively impaired individuals. In secondary analyses, we used regression models as above to identify features associated with mild cognitive impairment subtype (amnestic [aMCI] versus nonamnestic [naMCI]; threshold p< .05).Results:The sample included 599 participants with NC and 110 with CI (mean age = 72.3 ± 11.0 years, 54% female). The regression identified 8 linguistic features for IR and 7 for DR that significantly predicted cognitive status. Decreased use of content words related to work (e.g., employed, school, police) and biological processes (e.g., cook, cafeteria, eat) and the use of negations (e.g., no, not, can’t) were predictive of cognitive impairment in both recall conditions. In contrast, the use of other content word categories were predictive of cognitive status in only one recall condition (IR: leisure, cognitive processes, space; DR: drives, number). The use of fewer prepositions in IR, more first-person pronouns in DR, and fewer words in the past tense in DR were each associated with cognitive impairment. Word count was not predictive of cognitive status. Both PFSs were highly associated with cognitive status (PFS_IR ß= 0.74, p< 0.001; PFS_DR ß= 0.86, p= 0.001) with high discriminative value (PFS_IR AUC= 0.93, sensitivity = 0.81, specificity= 0.91; PFS_DR AUC= 0.95, sensitivity= 0.77, specificity= 0.88). In the CI subset, linguistic features differed between those classified as aMCI (n= 24) and naMCI (n= 40). Two function word categories predicted aMCI in IR whereas decreased word count, two function word categories, and two content word categories predicted aMCI in DR (all p< .05)Conclusions:Linguistic features from paragraph recall provide high predictive value for classifying cognitive status increasing its potential as a cognitive screener in clinical settings. Additionally, each recall condition identified unique linguistic features associated with cognitive impairment which may aid differentiation of cognitive impairment subtypes and elucidate processes underlying deficits in learning and recall.

Quantitative, not qualitative, differences in word classification errors as a function of language dominance.

This paper reports the results of a Cantonese word categorization task that maximized lexical competition. Cantonese-English early bilinguals were presented with a Cantonese word, followed by four images depicting the target word and a tone, rhyme, and onset competitor. English-dominant listeners made more errors than Cantonese-dominant listeners, but the proportions of error types were equivalent across language dominance profiles. Cantonese listeners' error patterns suggest a hierarchy of phonological units with respect to what most robustly differentiates lexical items, corroborating previous research.

Measuring novice-expert sense of place for a far-away place: Implications for geoscience instruction.

Individuals usually develop a sense of place through lived experiences or travel. Here we introduce new and innovative tools to measure sense of place for remote, far-away locations, such as Greenland. We apply this methodology within place-based education to study whether we can distinguish a sense of place between those who have visited Greenland or are otherwise strongly connected to the place from those who never visited. Place-based education research indicates that an increased sense of place has a positive effect on learning outcomes. Thus, we hypothesize that vicarious experiences with a place result in a measurably stronger sense of place when compared to the sense of place of those who have not experienced the place directly. We studied two distinct groups; the first are people with a strong Greenland connection (experts, n = 93). The second are students who have never been there (novices, n = 142). Using i) emotional value attribution of words, ii) thematic analysis of phrases and iii) categorization of words, we show significant differences between novice's and expert's use of words and phrases to describe Greenland as a proxy of sense of place. Emotional value of words revealed statistically significant differences between experts and novices such as word power (dominance), feeling pleasantness (valence), and degree of arousal evoked by the word. While both groups have an overall positive impression of Greenland, 31% of novices express a neutral view with little to no awareness of Greenland (experts 4% neutral). We found differences between experts and novices along dimensions such as natural features; cultural attributes; people of Greenland; concerns, importance, or interest in and feeling connected to Greenland. Experts exhibit more complex place attributes, frequently using emotional words, while novices present a superficial picture of Greenland. Engaging with virtual environments may shift novice learners to a more expert-like sense of place, for a far-away places like Greenland, thus, we suggest virtual field trips can supplement geoscience teaching of concepts in far-away places like Greenland and beyond.

Using Cluster Analysis for Author Classification of Albanian Texts: A Study on the Effectiveness of Stop Words

Cluster analysis is a statistical approach that identifies uniform clusters within data. The closeness of data is measured quantitatively using distance functions. Specifically for text data mining, clustering serves as a method of categorization of words based on the similarity of their occurrence within texts and classifying texts by topics or author. Hierarchical clustering is a powerful technique for identifying natural groupings within datasets, which can be especially useful for unsupervised text classification. This paper aims to utilize cluster analysis to establish Albanian texts clusters by authors. Using agglomerative hierarchical clustering we classify Albanian texts by authors according to the similarity of their word frequency. The similarity of texts is evaluated using cosine and Euclidean distances. Considering two study cases, respectively with and without Albanian stop words we conclude that the best clustering by authors of the Albanian documents is achieved with 87% accuracy using Ward’s method with cosine distance in the case of study by removing stop words.

TERMINOLOGICAL ANALYSIS IN ENGLISH AND UZBEK LINGUISTICS

The study of the history of the formation of lexical-semantic means of language in general and terminological vocabulary in particular is of great scientific, theoretical, and practical importance. In connection with the intensive development of all areas of modern knowledge, the problem of a comprehensive study of nominative terminological systems is becoming increasingly urgent.The problem of terms as a special lexical-semantic category of words, as well as the relationship of terminological vocabulary with the rest of the vocabulary of the literary language, has not yet been sufficiently studied. In the study of terms, two main approaches are outlined -normative and descriptive. The normative approach, within the framework of which the basic requirements for the term were formulated (the author of the first of which was D.S. Lotte -unambiguity, accuracy, brevity, absence of synonyms, etc.) made it possible to create state, industry and international standards for terms and definitions, collections of recommended terms and terminological dictionaries.

Have media texts become more humorous?

As a research topic, humour has drawn much attention from multiple disciplines including linguistics. Based on Engelthaler & Hills’ (2018) humour scale, this study developed a measure named Humour Index (HMI) to quantify the degree of humour of texts. This measure was applied to examine the diachronic changes in the degree of humour of American newspapers and magazines across a time span of 118 years (1900-2017) with the use of texts from Corpus of Historical American English (COHA). Besides, the study also discussed the contributions of different types of words to the degree of humour in the two genres. The results show significant uptrends in the degree of humour of both newspapers and magazines in the examined period. Moreover, derogatory and offensive words are found to be less frequently used than other categories of words in both genres. This study provides both theoretical and methodological implications for humour studies and claims or hypotheses of previous research, such as infotainment and linguistic positivity bias.

Do gender differences exist in letters of recommendation for reproductive endocrinology and infertility fellowship?

Objective(s)To determine if gender differences exist in letters of recommendation for Reproductive Endocrinology and Infertility fellowship. DesignCohort study SettingAcademic medical center SubjectsCohort of applicants to a reproductive endocrinology and infertility fellowship program at single Midwestern academic institution in a single year. Intervention(s)None Main Outcome MeasureQuantitative measures included linguistics inquiry and word count analysis, with four summary variables and twenty-five word categories, to quantify tone and themes present by gender. Performing qualitative analysis in parallel to linguistic analysis allowed for exploration of themes not conveyed in quantitative methods alone. Qualitative measures included frequency of code themes in letters of recommendation by applicant gender. Result(s)There were 272 letters from 72 applicants, 54 (76%) of which were women and 17 (24%) were men. One applicant was excluded because gender information was not specified; 269 letters were included in the quantitative linguistics and qualitative coding analysis. One hundred ten letters (41%) were written by women, and 159 (59%) by men. Letters of recommendation written for men had higher mean word count than those written for women (537 vs 474, p=0.04).Linguistics Inquiry and Word Count analysis exhibited more risk words used to describe men applicants (p=0.01). When comparing word categories by letter writer gender, women letter writers more frequently used communal, affect, and home word categories while men writers more frequently used affiliation-related words.Qualitative analysis revealed that letters for men applicants described leadership, altruism, rapport with patients, and strong endorsements more frequently, while women applicants’ letters more often mentioned doubt raisers and disclosures of personal life. Additionally, letters for women applicants more often described candidates as drama-free, not easy to “fluster,” and not complainers. Further, letters for women more frequently disclosed parenthood status. Conclusion(s)Gender differences exist among both letter writers and applicants. Qualitative analysis revealed that women applicants were more likely to be described with doubt raisers and personal life disclosures, while men were more likely to be described by their altruism and with a strong endorsement. Increasing bias awareness and implementing mitigation strategies would improve gender equity in letters of recommendation.

An Analysis of Swear Words Used by Netizens Comments in Lutfi Agizal Instagram Account

In society, people communicate and show emotion in different ways. Currently, it is often found that people express their feeling and emotion in swearing. This study aims to find out the categories of swear word used by netizens comments in Lutfi Agizal’s Instagram account. This research used a theory by Hughes(1996). This study is descriptive research. The data were taken from 21 posts in Lutfi Agizal’s account from August up to October 2020. The result shows that from 90 data collected, there are six categories out of swear word that have been found from seven categories of Hughes’s theory. They are religion (15.58%), genitalia (10.66%), excretion (5.73%), stupidity (17.21%), animal term (33.61%), and general (17.21%), while anatomical insult is not found from the data. By the finding, it can be concluded that the most dominant used of swear word is animal term.

Semantic changes in the diachronic perspective from the viewpoint of ethnic experience categorization in the English language

The article considers semantic changes in English words as certain movements in accordance with figurative schemes within the framework of the conceptual structure of the word and lexical categories. We believe this aspect to be relevant as the traditional approach to semantic changes of words in the diachronic perspective does not reveal the mechanism of transition from one meaning to another. In terms of cognitive linguistics, it is possible to more profoundly describe the phenomenon of recategorization in the English word semantics in the diachronic perspective and understand the essence of these changes. The scientific novelty of this study is the evaluation of semantic changes in the vocabulary of the English language from the standpoint of categorization, which is considered within the framework of diachronic changes influenced by a variety of external, extralinguistic factors, including the level of the development of civilization, culture and thinking of the English-speaking nation. We have provided rationale for the hypothesis that every shift of the meaning towards “improvement” or “deterioration” in the diachronic perspective will depend on the direction of the movement (to the core or periphery) of the word conceptual base: deterioration of the meaning is often associated with its movement towards the periphery, and improvement - with its movement to the core. The expansion and narrowing of the meanings of words is associated, respectively, with the movement from its part to its whole and from its whole to its part in the conceptual structure of words. In each specific case, the language selects the optimal way to change the value. This is the theoretical significance of our research. The possibility of using research materials in courses in the history of the English language, linguistic and cultural aspects of the English language, lexicology and stylistics of the English language determines the practical significance of the study.

Reaction time as an outcome measure during online fear conditioning: Effects of number of trials, age, and levels of processing

Recent studies have shown that fear conditioning experiments can be successfully conducted online. However, there is limited evidence that measures other than subjective ratings of threat expectancy can be collected, which means that online research may not be able to adequately replace laboratory experiments. In the current study, we conducted an online fear conditioning experiment consisting of habituation, acquisition, extinction and 48 h delayed extinction recall using ratings of threat expectancy and conditional stimulus pleasantness, and probe reaction time as outcome measures. The conditional stimuli were categories of words and a levels of processing manipulation explored whether words that were processed at a deeper level during extinction evoked smaller differential threat responses during extinction recall. Although the levels of processing manipulation did not produce a significant outcome, we found that extinction recall was successfully operationalised in our study. Reaction time indicated differential responding during both acquisition and extinction recall, and age of participants was correlated in one of two experiments with differential threat expectancy and reaction time, such that older participants showed better safety learning. The outcomes of this experiment provide multiple novel tools for researchers to explore fear conditioning, especially in an online environment.

Analysis on Deeply Derogatory Terms in the Chinese LanguageTaking Man Yi as an Example

One of the many types of concurrent meanings of words is the positive and negative concomitant meaning of words, which represents the positive and negative attitudes of the linguistic community about the accompanying rational meanings of words. There are numerous sorts of concomitant meanings of words. The notion of extremely derogatory terms proposed by Tirrell gave rise to a brand new category of words as well as an innovative approach to the study of slurs. The paper does a literature analysis to investigate deeply derogatory terms in Chinese, and then it uses that information to examine the Chinese word "Man Yi". It solves the problem of the theory's inability to adapt to China's circumstances.

An Analysis Of Derivational And Inflectional Morphemes In The Lyrics Of Calum Scott’s Songs In The Album “Bridges”

Morphology studies the intricate forms of words and the influence of changes in word forms on word categories and meanings. Morphemes, as the smallest grammatical units with meaning, are the primary focus of morphology. Morphemes cannot be further divided into smaller linguistic units. This research aims to investigate derivational and inflectional morphemes in the lyrics of Calum Scott’s songs in the album “Bridges”. The qualitative method is employed in this study, which reveals the presence of derivational and inflectional morphemes in the lyrics. The analysis demonstrates the existence of various types of derivational and inflectional affixes. Inflectional morphemes, particularly affixes, are found to be more dominant in the lyrics of Calum Scott’s songs in the album “Bridges”. The findings of this research identify 50 derivational and 99 inflectional morphemes. In derivations, 21 types of affixes are discovered, including prefixes such as re-, pre-, a-, sur- and suffixes such as -ent, -le, -ion, -ful, - less, -ing, -ly, -ain, -ed, -ness, -ment, -y, -ry, -er, -ty, -st, and -ical. Furthermore, there are 8 types of inflectional morphemes, namely plural, past tense, past perfect, comparative, possessive, third-person singular, gerund, and progressive.

A Study of Gender Difference of Using Taboo Words in The Euphoria Series

Taboo word is a forbidden word to use in any situation because it can cause discomfort to others, however taboo word can easily be spotted in everyday conversation. With that phenomenon, the analysis is conducted to investigate the use of taboo words and identify the difference between women and men character with the theory from Jay to categorize type of taboo words and a theory from Coates and Lakoff to examine how men and women speak differently. This study used the Euphoria series as a main source of the study due to the popularity of this series among young generation. The findings reveal that out of 229 collected data, ten different categories of taboo words exist in this series which are cursing (17,03%), followed by profanity (8,73%), blasphemy (0,44%), taboo (3,95%), obscenity (8,73%), slang (10,48%), epithet (21,40%), insult and slur (17,47%), and finally scatology (0,87%) and also the difference between women and men in using taboo words. .

Social Media Zero-Day Attack Detection Using TensorFlow

In the current information era, knowledge can pose risks in the online realm. It is imperative to proactively recognize potential threats, as unforeseen dangers cannot be eliminated entirely. Often, malware exploits and other emerging hazards are only identified after they have occurred. These types of risks are referred to as zero-day attacks since no pre-existing anti-malware measures are available to mitigate them. Consequently, significant damages occur when vulnerabilities in systems are exploited. The effectiveness of security systems, such as IPS and IDS, relies heavily on the prompt and efficient response to emerging threats. Failure to address these issues promptly hinders the effectiveness of security system developers. The purpose of this study is to analyze data from the Twitter platform and deploy machine learning techniques, such as word categorization, to identify vulnerabilities and counteract zero-day attacks swiftly. TensorFlow was utilized to handle the processing and conversion of raw Twitter data, resulting in significant efficiency improvements. Moreover, we integrated the Natural Language Toolkit (NLTK) tool to extract targeted words in various languages. Our results indicate that we have achieved an 80% success rate in detecting zero-day attacks by using our tool. By utilizing publicly available information shared by individuals, relevant security providers can be promptly informed. This approach enables companies to patch vulnerabilities more quickly.

Treatment-Induced Recovery Patterns Between Nouns and Verbs in Mandarin-English Bilingual Adults With Aphasia.

This study aimed to investigate treatment effects of naming therapy targeting nouns and verbs in Mandarin-English bilingual adults with aphasia (BWA). Twelve Mandarin-English bilingual adults with chronic aphasia completed a 40-hr semantic-based naming treatment for either nouns or verbs. Eight of these participants completed both noun and verb treatment, and the other four completed either noun or verb treatment. Participants were trained in either Mandarin or English for both treatment cycles. Weekly naming probes were measured to capture the direct treatment gain and within- and cross-language generalizations. Performance on the standardized language assessments was analyzed to examine further generalizations beyond the word level and to standardized naming tasks. Responses in the weekly naming probes showed significant treatment gains in both noun and verb treatment, but the effect was greater in verb treatment. Generalization to semantically related items was captured in noun treatment only. Cross-language generalization was identified in both noun and verb treatment with a larger effect in verb treatment. Additionally, widespread generalizations beyond the word level and to standardized naming tasks were found following both noun and verb treatment, but the effect was larger following noun treatment in discourse and verb naming tasks. Findings from this study suggested robust treatment effects of semantic-based naming treatment targeting nouns and verbs in Mandarin-English BWA. However, patterns of treatment gains and generalizations differed between these word categories. This study provides strong evidence of bilingual aphasia rehabilitation in Mandarin-English BWA. https://doi.org/10.23641/asha.23818299.