Individual Words Research Articles

Predicting Generalized Anxiety Disorder From Impromptu Speech Transcripts Using Context-Aware Transformer-Based Neural Networks: Model Evaluation Study.

The ability to automatically detect anxiety disorders from speech could be useful as a screening tool for an anxiety disorder. Prior studies have shown that individual words in textual transcripts of speech have an association with anxiety severity. Transformer-based neural networks are models that have been recently shown to have powerful predictive capabilities based on the context of more than one input word. Transformers detect linguistic patterns and can be separately trained to make specific predictions based on these patterns. This study aimed to determine whether a transformer-based language model can be used to screen for generalized anxiety disorder from impromptu speech transcripts. A total of 2000 participants provided an impromptu speech sample in response to a modified version of the Trier Social Stress Test (TSST). They also completed the Generalized Anxiety Disorder 7-item (GAD-7) scale. A transformer-based neural network model (pretrained on large textual corpora) was fine-tuned on the speech transcripts and the GAD-7 to predict whether a participant was above or below a screening threshold of the GAD-7. We reported the area under the receiver operating characteristic curve (AUROC) on the test data and compared the results with a baseline logistic regression model using the Linguistic Inquiry and Word Count (LIWC) features as input. Using the integrated gradient method to determine specific words that strongly affect the predictions, we inferred specific linguistic patterns that influence the predictions. The baseline LIWC-based logistic regression model had an AUROC value of 0.58. The fine-tuned transformer model achieved an AUROC value of 0.64. Specific words that were often implicated in the predictions were also dependent on the context. For example, the first-person singular pronoun "I" influenced toward an anxious prediction 88% of the time and a nonanxious prediction 12% of the time, depending on the context. Silent pauses in speech, also often implicated in predictions, influenced toward an anxious prediction 20% of the time and a nonanxious prediction 80% of the time. There is evidence that a transformer-based neural network model has increased predictive power compared with the single word-based LIWC model. We also showed that the use of specific words in a specific context-a linguistic pattern-is part of the reason for the better prediction. This suggests that such transformer-based models could play a useful role in anxiety screening systems.

The developmental trajectories of L2 lexical-semantic networks

This study explored the developmental trajectories of L2 Lexical-semantic networks by comparing the structure of lexical-semantic networks of Chinese EFL learners at two levels of English: intermediate and advanced. Analyses of data from semantic fluency tasks found that with the improvement of language proficiency, the L2 lexical-semantic network becomes more connected but less centralized. Analyses of community structure revealed that the lexical-semantic network of advanced learners has denser clusters than that of intermediate learners. Examination of individual words in the L2 lexical-semantic network shows that words tend to reorganize themselves with the expanse of the network. In this process, more frequent and connected words tend to remain in the network as central words while less frequent words with fewer connections have the risk of being lost, although a large number of new words keep joining in the network. The results provide empirical evidence for the preferential attachment model of lexical-semantic network growth.

Literacy and early language development: Insights from computational modelling.

Computational models of reading have tended to focus on the cognitive requirements of mapping among written, spoken, and meaning representations of individual words in adult readers. Consequently, the alignment of these computational models with behavioural studies of reading development has to date been limited. Models of reading have provided us with insights into the architecture of the reading system, and these have recently been extended to investigate literacy development, and the early language skills that influence children's reading. These models show us: how learning to read builds on early language skills, why various reading interventions might be more or less effective for different children, and how reading develops across different languages and writing systems. Though there is growing alignment between descriptive models of reading behaviour and computational models, there remains a gap, and I lay out the groundwork for how translation may become increasingly effective through future modelling work.

A natural language processing approach to uncover patterns among online ratings of otolaryngologists.

Patients increasingly use physician rating websites to evaluate and choose potential healthcare providers. A sentiment analysis and machine learning approach can uniquely analyse written prose to quantitatively describe patients' perspectives from interactions with their physicians. Online written reviews and star scores were analysed from Healthgrades.com using a natural language processing sentiment analysis package. Demographics of otolaryngologists were compared and a multivariable regression for individual words was performed. This study analysed 18 546 online reviews of 1240 otolaryngologists across the USA. Younger otolaryngologists (aged less than 40 years) had higher sentiment and star scores compared with older otolaryngologists (p < 0.001). Male otolaryngologists had higher sentiment and star scores compared with female otolaryngologists (p < 0.001). 'Confident', 'kind', 'recommend' and 'comfortable' were words associated with positive reviews (p < 0.001). Positive bedside manner was strongly reflected in better reviews, and younger age and male gender of the otolaryngologist were associated with better sentiment and star scores.

Nouns endings and finals -ie, -ii in prepositions (from materials of the academic description of Russian orthography)

The article presents a fragment of academic rules, namely those for spelling unstressed vowels in noun endings. Tables and teaching materials accompany the rules. This section of orthography is well-studied and described in both scientific and methodological literature. The provided characterisation is generalising: it systematically presents all orthographically significant types of noun declensions. The main rule formulation does not result in the usual vicious cycle, i. e. when the ending of the original form determines declension type (words ending with -a(-ya), words ending with -o(-ye)), and the ending of the motivating word is determined by the declension type (lake/ozero as window/ okno). The rules address all groups of words and individual lexemes that diverge from standard paradigms and present spelling difficulties. These include bolgarin – bolgary/a Bulgarian – Bulgarians, but anglichanin – anglichane/an Englishman – the English; Pushkin – Pushkinym/Pushkin – by Pushkin, but Chaplin – Chaplinom/Chaplin – by Chaplin; Sofia – o Sofii/Sofia – about Sofia, but Liya – o Lie/Leah – about Leah, Viy – o Vie/Viy – about Vii; j-stem words (Maya)), as well as grammatical variants (usishcha and usishchi/a very big moustache). We also mention individual words with stressed endings which do not correspond to this declension type (put’/path, kiy/cue). The rules prevent errors in dividing word forms into morphemes incorrectly. Such mistakes are common in reference and educational literature. One note discusses the problem of differentiating noun forms ending with -ie, -ii from derivative prepositions and lexicalised (adverbial) prepositional-case combinations associated with them. Their pronunciation coincides with that of prepositions (v zaklyuchenii soderzhitsya oshibka/there is an error in the conclusion – v zaklyuchenie seminara blagodaryu vsekh uchastnikov/in conclusion of the seminar, I thank all participants, skazal v zaklyuchenie/in conclusion, he said). Teaching materials enable educators to consistently form the skills necessary to correctly spell prepositions ending with -ie, -ii, as well as adverbial combinations and prepositional noun phrases associated with them. A sufficient number of examples, their grouping and arrangement enable learners to observe the semantic, syntactic, spelling differences of prepositions, adverbial combinations, and nouns. The materials permit wide variation in the volume and complexity of assignments. This allows teachers to adapt them focusing on the age and the preparation level of learners.

From Word Types to Tokens and Back: A Survey of Approaches to Word Meaning Representation and Interpretation

Abstract Vector-based word representation paradigms situate lexical meaning at different levels of abstraction. Distributional and static embedding models generate a single vector per word type, which is an aggregate across the instances of the word in a corpus. Contextual language models, on the contrary, directly capture the meaning of individual word instances. The goal of this survey is to provide an overview of word meaning representation methods, and of the strategies that have been proposed for improving the quality of the generated vectors. These often involve injecting external knowledge about lexical semantic relationships, or refining the vectors to describe different senses. The survey also covers recent approaches for obtaining word type-level representations from token-level ones, and for combining static and contextualized representations. Special focus is given to probing and interpretation studies aimed at discovering the lexical semantic knowledge that is encoded in contextualized representations. The challenges posed by this exploration have motivated the interest towards static embedding derivation from contextualized embeddings, and for methods aimed at improving the similarity estimates that can be drawn from the space of contextual language models.

The effects of typographic enhancement on L2 collocation processing and learning from reading: An eye-tracking study

Collocations have been of considerable interest to applied linguistics and SLA researchers, due to their pervasiveness in language. In corpus research, collocations are broadly defined as pairs of content words like nouns and adjectives that appear more frequently in language than the occurrence frequency of the individual words would predict (Biber et al. 1999). Typical examples in English are heavy rain and strong wind. As conventional forms of expression, collocations may be used to express concepts in an efficient manner, or serve as phrasal terminology in technical, scientific, or academic discourse (e.g. Schmitt and Carter 2004). Research has shown that L2 learners tend to produce many collocations that are deviant or non-nativelike (e.g. *make a photo) in their speech (Nekrasova 2009) and writing (e.g. Durrant and Schmitt 2009; Laufer and Waldman 2011). Although the evidence is far from conclusive, it is generally thought that L2 collocation knowledge develops slowly, and therefore does not always show up at the level of productive use (e.g. Durrant and Schmitt 2010). A better understanding of why this might be the case requires a closer look at the variables that may affect the learnability of collocations.

Disentangling Some Conceptual Knots.

Disentangling Some Conceptual Knots.

Multilingual spoken word recognition: A megastudy approach

Bilingualism research has primarily focused on the perception and processing of individual sounds or word learning. Many studies have investigated how bilingual listeners perceive sound contrasts that don't create lexical distinctions in their native language. There is substantially less research that has investigated how word-level properties impact L2 auditory processing. The present study examines how auditory lexical processing differs between monolingual and bilingual listeners with different language backgrounds. We collected lexical decision accuracies and latencies for 26,793 words and 9600 pseudowords from the Massive Auditory Lexical Decision database from native and non-native listener responses. We compare the language backgrounds of our 1028 listeners and group them into four groups: native speakers, early, early-late, and late bilinguals. We report the findings of an analysis investigating how language background and proficiency modulate lexical effects to understand how language background influences spoken word recognition. We find differences in the effects of lexical frequency, phonological neighborhood density, and phonological uniqueness point between the different listener groups. We discuss the impact of bilingualism in a word recognition task and how these results inform models of spoken word recognition and second language acquisition.

Impact of word properties on list learning: An explanatory item analysis.

A variety of factors affect list learning performance and relatively few studies have examined the impact of word selection on these tests. This study examines the effect of both language and memory processing of individual words on list learning. Item-response data from 1,219 participants, Mage = 74.41 (SD = 7.13), Medu = 13.30 (SD = 2.72), in the Harmonized Cognitive Assessment Protocol were used. A Bayesian generalized (non)linear multilevel modeling framework was used to specify the measurement and explanatory item-response theory models. Explanatory effects on items due to learning over trials, serial position of words, and six word properties obtained through the English Lexicon Project were modeled. A two parameter logistic (2PL) model with trial-specific learning effects produced the best measurement fit. Evidence of the serial position effect on word learning was observed. Robust positive effects on word learning were observed for body-object integration while robust negative effects were observed for word frequency, concreteness, and semantic diversity. A weak negative effect of average age of acquisition and a weak positive effect for the number of phonemes in the word were also observed. Results demonstrate that list learning performance depends on factors beyond the repetition of words. Identification of item factors that predict learning could extend to a range of test development problems including translation, form equating, item revision, and item bias. In data harmonization efforts, these methods can also be used to help link tests via shared item features and testing of whether these features are equally explanatory across samples. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Perception error variation in masking contexts

Masking noise is an important tool in speech and hearing experiments. It provides a means to simulate real world conditions and provide misperceptions that are as naturally elicited as possible. However, the exact characteristics of the variable approach to this parameter is poorly understood. The present investigation explores the types and variability of errors across different noise conditions. The English Consistent Confusion Corpus [Marxer et al., JASA 140, EL458–EL463 (2016)], which provides data on consistently reproducible mispercepts, was used for this investigation. The dataset contains tokens of words in multiple masking conditions: speech shaped noise; three talker babble modulated noise; and four-talker natural babble, correlated with signal to noise ratios that bias a listener towards a specific mispercept. This dataset contains 3200 individual word tokens from four different speakers with 15 listeners. The interaction between types and variability of misperception with the different types of masking noise is explored. As well, the suitability of the different noise conditions for speech perception experiments is discussed.

Contextual Adversarial Representation Learning for Multiword Expressions

Word composition is a promising method to learn the representations of long text. Unfortunately, word composition falls short when inferring representations for non-compositional multiword expressions, such as “go banana.” Presently, many methods treat multiword expressions as single words and learn their representations in a manner similar to individual word representations. However, numerous multiword expressions exhibit ambiguity, expressing distinct meanings, whether literal or idiomatic, depending on the context. In response to these challenges, our paper proposes an adversarial context-aware representation learning method for multiword expressions, which generates representations based on the contexts of their occurrences. An adversarial training framework is introduced to further enhance the representation learning method. Experimental results confirm the benefits of sense disambiguation for multiword expressions in representation learning. Moreover, our proposed method demonstrates competitive performance on idiom token classification and compositionality prediction tasks.

Translation and cultural adaptation of the Health Utilities Preschool to Brazilian Portuguese

IntroductionHealth research is particularly important in low- and middle-income countries (LMICs), where improvements must be achieved with limited resources, and where the great majority of the world's population, especially children, live. Improvements in public health detection in Brazil have resulted in cancer becoming the most prevalent cause of death by disease in the group aged 1 to 19 years, hence, delivering cost-effective care to the group is a priority. Preference-based measures of health status and health-related quality of life (HRQL) integrate morbidity and mortality and provide utility scores for the estimation of quality-adjusted life years to be used in cost-effectiveness analyses and economic evaluation. The generic preference-based instrument Health Utilities - Preschool (HuPS) measures the health status of young children and is applicable to the age group 2 to 5 years, who carry the highest incidence of cancer in childhood. MethodsThe translation of the HuPS classification system followed recommended protocols from published guidelines. Forward and backward translations were performed by a team of six qualified professionals and linguistic validation was undertaken with a sample of parents of preschool children. Main resultsInitial disagreements on individual words occurring in 0.5-1.5% were resolved by consensus. A final version of the instrument was validated by the sample of parents. ConclusionsThe translation and cultural adaptation of the HuPS into Brazilian Portuguese were accomplished as the first step in the validation of the HuPS instrument in Brazil.

The growing gap: A study of sleep, encoding, and consolidation of new words in chronic traumatic brain injury

Word learning is an iterative and dynamic process supported by multiple neural and cognitive systems. Converging evidence from behavioral, cellular, and systems neuroscience highlights sleep as an important support for memory and word learning over time. In many lab-based word learning experiments, participants encode and subsequently retrieve newly learned words in a single session. These designs are inadequate to capture the full dynamic word learning process, making them less ecologically valid. Single timepoint studies also limit investigation of the role of behavioral and lifestyle factors, like sleep, in supporting word learning over time. Adults with a history of traumatic brain injury (TBI), who commonly exhibit deficits in the memory systems that support word learning and report concomitant sleep disturbance, provide a unique opportunity to examine the link between memory, sleep, and word learning. Here we examined word learning over time and the influence of sleep on short- and long-term word recall in 50 adults with chronic moderate-severe TBI and 50 demographically matched neurotypical peers. We used a randomized within-participant crossover design to assess immediate encoding of new words and the consolidation of those words over time across intervals that did or did not involve sleep. Participants completed this study over the course of two weeks in their own homes to capture the iterative, dynamic process of real-world word learning. We also measured sleep in free living conditions using actigraphy throughout the experiment. Participants with TBI exhibited a word learning deficit that began at encoding and persisted across time. Critically, this deficit grew over the course of the week. The performance gap between groups was larger at the 1-week post-test than the immediate post-test, suggesting deficits in both encoding and consolidation of new words in individuals with TBI. Participants with and without TBI remembered more words when they slept after learning. Ecologically valid research designs that examine the relationship between memory, sleep, and word learning over time promise to advance mechanistic accounts of word learning and improve the long-term retention of new words in individuals with and without brain injury.

Lips Reading Spoken Arabic Word Based on The Geometric Shape Features of The Lip

This research aims to determine how to guess individual Arabic words based on visual cues such as lip movement and form extraction. A three-step process was developed to accomplish this goal, the first of which is the identification of the face. Second, the lip region is targeted, and finally, the movement of the speaking lips is analyzed to determine what is being said. This research proposes an alternative approach that uses biometric engineering features extracted from the changing shapes of the lips to make the best guess at the word being spoken. This method uses a set of 20 landmarks around the mouth's outer and inner edges, the high point of an open mouth, and the point where the upper and lower lips meet. The proposed study took a different approach by looking at the shape of the lips and how the points on the upper and lower lips move, as well as figuring out what was being said (the outer and the inner).

Disentangling word stress and phrasal prosody: A view from Georgian

This paper investigates the interaction of word stress and phrasal prosody in Georgian by studying the distribution of acoustic cues (duration, intensity, F0) in controlled data. The results show that initial syllables in Georgian words are marked by greater duration than all subsequent syllables, regardless of syllable count and phrasal context. After excluding domain-initial strengthening as an alternative explanation, this finding provides evidence in favor of fixed initial stress. Likewise, initial syllables are marked by greatest intensity, but the consistent gradual drop in intensity throughout the word suggests that this effect may not be stress-related. The F0 results align with the existing accounts: individual lexical words form ACCENTUAL PHRASES, marked by a low pitch accent on the initial syllable and a high final boundary tone on the final syllable. Additionally, new evidence for a phrasal accent, aligned with the penult, is presented. F0 targets are shown to be completely absent in the context of post-focal deaccenting, which shows that F0-marking in Georgian is reserved for phrasal prosody and is not intrinsic to stress-marking. These results help account for the facts related to word stress, phrasal intonation, and their interplay in Georgian, the object of debate in the literature.

Predictability and decomposability separately contribute to compositional processing of idiomatic language.

When reading, comprehenders construct a message-level representation and integrate new information as it becomes available. Such compositional processing may differ for idioms, where the meanings of the individual words do not always relate to the figurative meaning. Here, we examined how predictability and idiom decomposability contribute to compositional processing. Participants' EEG was recorded while they read sentences containing idioms that varied in decomposability and phrase-final word cloze probability, or their literal match (break the ice/slip on the ice) with little context prior to the phrase, along with adjective insertion conditions (break the freezing ice/slip on the freezing ice). Cloze probability modulated N400 amplitudes to critical words for both idiomatic and literal phrases, whereas P600 amplitudes only elicited by idiomatic phrases were also modulated. Phrases with adjective insertions reduced acceptability judgments, particularly for idioms, and led to N400 amplitude differences compared to critical words, but only for idioms. N400 differences were also found between idiomatic and literal contexts at the point of adjective insertion. Additionally, both cloze probability and decomposability modulated gamma band activity, with greater gamma activity for more predictable and less decomposable idioms, but at different times and with different scalp topographies, suggesting dissociable components of processing. These results support a hybrid model of processing in which multiple linguistic factors determine the type of processing engaged by the brain to comprehend non-literal language. When prior context is minimal, compositional processing may still be engaged when reading idioms.

Diminished Repetition Suppression Reveals Selective and Systems-Level Face Processing Differences in ASD.

Repeated exposure to a stimulus results in reduced neural response, or repetition suppression, in brain regions responsible for processing that stimulus. This rapid accommodation to repetition is thought to underlie learning, stimulus selectivity, and strengthening of perceptual expectations. Importantly, reduced sensitivity to repetition has been identified in several neurodevelopmental, learning, and psychiatric disorders, including autism spectrum disorder (ASD), a neurodevelopmental disorder characterized by challenges in social communication and repetitive behaviors and restricted interests. Reduced ability to exploit or learn from repetition in ASD is hypothesized to contribute to sensory hypersensitivities, and parallels several theoretical frameworks claiming that ASD individuals show difficulty using regularities in the environment to facilitate behavior. Using fMRI in autistic and neurotypical human adults (females and males), we assessed the status of repetition suppression across two modalities (vision, audition) and with four stimulus categories (faces, objects, printed words, and spoken words). ASD individuals showed domain-specific reductions in repetition suppression for face stimuli only, but not for objects, printed words, or spoken words. Reduced repetition suppression for faces was associated with greater challenges in social communication in ASD. We also found altered functional connectivity between atypically adapting cortical regions and higher-order face recognition regions, and microstructural differences in related white matter tracts in ASD. These results suggest that fundamental neural mechanisms and system-wide circuits are selectively altered for face processing in ASD and enhance our understanding of how disruptions in the formation of stable face representations may relate to higher-order social communication processes.SIGNIFICANCE STATEMENT A common finding in neuroscience is that repetition results in plasticity in stimulus-specific processing regions, reflecting selectivity and adaptation (repetition suppression [RS]). RS is reduced in several neurodevelopmental and psychiatric conditions including autism spectrum disorder (ASD). Theoretical frameworks of ASD posit that reduced adaptation may contribute to associated challenges in social communication and sensory processing. However, the scope of RS differences in ASD is unknown. We examined RS for multiple categories across visual and auditory domains (faces, objects, printed words, spoken words) in autistic and neurotypical individuals. We found reduced RS in ASD for face stimuli only and altered functional connectivity and white matter microstructure between cortical face-recognition areas. RS magnitude correlated with social communication challenges among autistic individuals.

The associations between child and item characteristics, use of vocabulary scaffolds, and reading comprehension in a digital environment: Insights from a big data approach

Scaffolding features that provide multimodal support for the pronunciation and meaning of words are increasingly common in digital reading environments. These vocabulary scaffolds are intended to aid the accurate pronunciation and understanding of individual words in context, thus supporting both vocabulary development and comprehension of text. However, the evidence on their efficacy remains inconclusive. The present study adds to the evidence base by examining: 1) whether child characteristics predict the use of vocabulary scaffolds; 2) whether the use of vocabulary scaffolds is associated with reading comprehension performance; and 3) whether the association between the use of scaffolds and reading comprehension is modulated by child and/or item characteristics. A large cohort (N ∼ 120,000) of 5- to 8-year-old children in the United States interacted with a gamified digital reading environment with embedded vocabulary scaffolds, thereby generating a large observational dataset of user log files. Confirmatory analyses with Generalized Linear Mixed Models (GLMMs) indicated that children with lower literacy skills, beginning readers, girls, and bilingual students were more likely to use the scaffold. Overall, the use of scaffolds was associated with better reading comprehension performance. The association between the use of scaffolds and reading comprehension was modulated by both child and item characteristics. We conclude that vocabulary scaffolds may be promising tools to facilitate reading comprehension and reduce performance differences amongst diverse learners in digital reading environments. Educational implications and recommendations for future research are discussed.

The Effects of Repeated Oral Reading Practice on the Retention of High-Frequency Multiword Items for EFL Learners: Multiple Dimensions

Research has shown that second language (L2) learners generally lack multiword expression knowledge, and L2 researchers and practitioners have tried various techniques to assist L2 learners to acquire it more efficiently. This study adopted an under-researched technique—repeated oral reading—to enhance the retention of high-frequency multiword items by 62 EFL college students divided into experimental (n =38) and control (n =24) groups. Fifteen unfamiliar multiword items comprising only known individual words were selected through a pre-test based on a theme-based text. All students received a formal instruction first, followed by the experimental group orally reading the text six times under a time constraint. A two-week delayed post-test was used to test students’ retention of four dimensions of multiword knowledge: aural forms and aural meanings, and written meanings and use. Except for use, the experimental group significantly outperformed the control group in the other three dimensions. Four fixed factors (oral reading speed, prior vocabulary knowledge, dimensions of multiword knowledge, and the number of words per item) were analyzed via GLMM. Results showed three factors had significant effects on retaining multiword items except oral reading speed. Based on the results, pedagogical implications are discussed and suggestions are made.