Language-based Model Research Articles

Syntax, morphosyntax, and serial recall: How language supports short-term memory

In the classic view of verbal short-term memory, immediate recall is achieved by maintaining phonological representations, while the influence of other linguistic information is negligible. According to language-based accounts, short-term retention of verbal material is inherently bound to language production and comprehension, thus also influenced by semantic or syntactic factors. In line with this, serial recall is better when lists are presented in a canonical word order for English rather than in a noncanonical order (e.g., when adjectives precede nouns rather than vice versa; Perham et al., 2009, Quarterly Journal of Experimental Psychology, 62[7], 1285–1293). However, in many languages, grammaticality is not exclusively determined by word order. In German, an adjective–noun sequence is grammatical only if the adjective is inflected in congruence with the noun’s person, number, and grammatical gender. Therefore, we investigated whether similar effects of syntactic word order occur in German. In two modified replications of Perham et al.’s study, we presented lists of three pairs of adjectives and nouns, presented in adjective–noun or in noun–adjective order. In addition, we manipulated morphosyntactic congruence between nouns and adjectives within pairs (Exp. 1: congruently inflected vs. uninflected adjectives; Exp. 2: congruently inflected vs. incongruently inflected adjectives). Both experiments show an interaction: Word order affected recall performance only when adjectives were inflected in congruence with the corresponding noun. These findings are in line with language-based models and indicate that, in a language that determines grammaticality in an interplay of syntactic and morphosyntactic factors, word order alone is not sufficient to improve verbal short-term memory.

Fraction detectives: bilingual students investigate the hidden identities of equivalent fractions

This paper conceptualizes children’s mathematical thinking from a materialist perspective on language and mathematics. This perspective considers human and non-human bodies as ontologically equivalent; that is, as both being agentive, vibrant, and animated, thus resisting static representations. This conceptualization is an alternative to the interactionist and language-based models that have dominated research in language and mathematics. This conceptual approach informs the paper’s non-hierarchical analysis of a bilingual classroom studying the concept of equivalent fractions. The non-hierarchical nature of this analysis permits a dynamic shift of focus from learners to materials to mathematical concepts. This non-traditional analysis informs the paper’s final discussion regarding the nature of children’s mathematical thinking. The discussion highlights how working outside language-based interactionist models provides an alternative view of children and artefacts as equal partners working within and as part of the vibrant boundaries between matter and meaning.

Hard for humans, hard for machines: predicting readmission after psychiatric hospitalization using narrative notes

Machine learning has been suggested as a means of identifying individuals at greatest risk for hospital readmission, including psychiatric readmission. We sought to compare the performance of predictive models that use interpretable representations derived via topic modeling to the performance of human experts and nonexperts. We examined all 5076 admissions to a general psychiatry inpatient unit between 2009 and 2016 using electronic health records. We developed multiple models to predict 180-day readmission for these admissions based on features derived from narrative discharge summaries, augmented by baseline sociodemographic and clinical features. We developed models using a training set comprising 70% of the cohort and evaluated on the remaining 30%. Baseline models using demographic features for prediction achieved an area under the curve (AUC) of 0.675 [95% CI 0.674–0.676] on an independent testing set, while language-based models also incorporating bag-of-words features, discharge summaries topics identified by Latent Dirichlet allocation (LDA), and prior psychiatric admissions achieved AUC of 0.726 [95% CI 0.725–0.727]. To characterize the difficulty of the task, we also compared the performance of these classifiers to both expert and nonexpert human raters, with and without feedback, on a subset of 75 test cases. These models outperformed humans on average, including predictions by experienced psychiatrists. Typical note tokens or topics associated with readmission risk were related to pregnancy/postpartum state, family relationships, and psychosis.

On the Legitimacy of Language-Based Accounting Standards

Reporting schemes have been criticized as being socially constructed, based on norms, and lacking an objective basis to legitimize them. While respecting the norm-like nature of accounting standards, I develop a language-based model in order to gain insight into the roles of the standard setter and supporting infrastructure in promoting legitimacy. I model two threats to legitimacy: managerial rent seeking and a lack of objectivity (disagreement on the meaning of phrases). The results indicate that while increasing penalties (infrastructure) reduces rent seeking and improves objectivity, the standard setter can also reduce rent seeking, but with no effect on objectivity; the arbiter’s conservatism (also infrastructure) reduces rent seeking but lessens objectivity as well; more generic terms increase rent seeking but improve objectivity; and closing loopholes reduces rent seeking but with no affect on objectivity. Moreover, the language-based model affirms the traditional role that conservatism plays in establishing legitimacy.

Similarity is closeness: Using distributional semantic spaces to model similarity in visual and linguistic metaphors

Abstract The semantic similarity that characterizes two terms aligned in a metaphor is here analysed through a corpus-based distributional semantic space. We compare and contrast two samples of metaphors, representative of visual and linguistic modality of expressions respectively. Popular theories of metaphor claim that metaphors transcend their modality to influence conceptual structures, thus suggesting that different modalities of expression would typically express the same conceptual metaphors. However, we show substantial differences in the degree of similarity captured by the distributional semantic space with regard to the modality of expression (higher similarity for linguistic metaphors than for visual ones). We argue that this is due to two possible variables: Conventionality (linguistic metaphors are typically conventional, while visual are not) and Complexity (visual metaphors have modality-specific inner complexities that penalize the degree of similarity between metaphor terms captured by a language-based model). Finally, we compare the similarity scores of our original formulations with those obtained from different possible verbalizations of the same metaphors (acquired by replacing the metaphor terms with their semantic neighbours). We show that while this operation does not affect the average similarity between metaphor terms for visual metaphors, the similarity changes significantly in linguistic metaphors. These results are discussed here.

From a Smoking Gun to Spent Fuel: Principled Subsampling Methods for Building Big Language Data Corpora from Monitor Corpora

With the influence of Big Data culture on qualitative data collection, acquisition, and processing, it is becoming increasingly important that social scientists understand the complexity underlying data collection and the resulting models and analyses. Systematic approaches for creating computationally tractable models need to be employed in order to create representative, specialized reference corpora subsampled from Big Language Data sources. Even more importantly, any such method must be tested and vetted for its reproducibility and consistency in generating a representative model of a particular population in question. This article considers and tests one such method for Big Language Data downsampling of digitally accessible language data to determine both how to operationalize this form of corpus model creation, as well as testing whether the method is reproducible. Using the U.S. Nuclear Regulatory Commission’s public documentation database as a test source, the sampling method’s procedure was evaluated to assess variation in the rate of which documents were deemed fit for inclusion or exclusion from the corpus across four iterations. After performing multiple sampling iterations, the approach pioneered by the Tobacco Documents Corpus creators was deemed to be reproducible and valid using a two-proportion z-test at a 99% confidence interval at each stage of the evaluation process–leading to a final mean rejection ratio of 23.5875 and variance of 0.891 for the documents sampled and evaluated for inclusion into the final text-based model. The findings of this study indicate that such a principled sampling method is viable, thus necessitating the need for an approach for creating language-based models that account for extralinguistic factors and linguistic characteristics of documents.

LANGUAGE IMPACT TOWARDS SOCIAL BEHAVIOR OF AT-RISK CHILDREN

Current study examines the use of language towards behavior of at-risk children. At-risk youth living separated from their family are deemed as most vulnerable children with highest risk having high level of anxiety, emotional disorders, aggression and problematic behavior. Children deprived of parental care living across Bulgarian residential homes (N=40) divided into 3 age groups (aged 7-10; aged 11-14; and aged 15-17) participated in 12 experimental activities and has been subjected surveillance of changes in their behavior according to certain use of language and voice (tone). Children adhere to or violate the rules in activities was the main criteria that have been defined for observation. The use of language indicators were divided into certain words with encouraging and reprobation meaning combined with louder or normal voice. ANOVA measures of surveillance assessed the impact of language and speech on behavior of children and indicated significant results for different age groups. The main hypothesis of the survey traces the impact of encouraging words with positive meaning and normal voice for better effect on socially acceptable behavior. Scientifically justified findings of this study indicates existence of a pattern supported the positive impact of language and speech that specialists and teachers can use helping the child absorbing pro-social behavior and gives directions of an effective language-based model of consulting and educating children. The results in current research can help social workers, social educators, teachers and psychologists to find suitable strategy in communication and proper use of language for better results on socialization and adaptation of the child with behavior problems. Article DOI: https://dx.doi.org/10.20319/pijtel.2019.31.0121 This work is licensed under the Creative Commons Attribution-Non-commercial 4.0 International License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc/4.0/ or send a letter to Creative Commons, PO Box 1866, Mountain View, CA 94042, USA.

Knowledge Organization, Data and Algorithms: The New Era of Visual Representations

This article shows how visual representations have progressively taken the lead over classical language-based models of knowledge organization (KO). The paper adopts a theoretical and historical perspective and focuses on the consequences of the changes in the volume of data generated by data production on the KO models. Until now, data visualization tools have been used mainly by researchers with expertise in textual data processing or in computational linguistics. But now, these tools are accessible to a greater number of users. Thus, there are new issues at stake for KO, other professions and institutions for gathering data that contribute to defining new standards and KO representations.

Efficient design optimization of complex system through an integrated interface using symbolic computation

A complex system is composed of several subsystems and numerous lower level components. It is inefficient to design the complex system at the system level by using high-fidelity models. A system level approach using a language-based model is required to design such a highly complex system before implementing a detailed design. However, as commercial process integration and design optimization software packages focus on detailed design, which uses the high-fidelity models, it is difficult to perform the design optimization of a highly complex system, such as a combat vehicle. Moreover, as commercial optimization software packages use numerical computation, the gradient calculation cost can increase, and the matrix computation process can be inefficient in terms of optimization time. Therefore, in this study, an integrated interface for efficient design optimization was developed to focus on concept design using a MODELICA language-based model. Additionally, the design variable screening by using analysis of variance, surrogate modeling through sequential design of experiments, and symbolic computation were used to solve the aforementioned problems. These were applied to the design optimization of the combat vehicle system to demonstrate the effectiveness of the integrated interface and symbolic computation. In conclusion, a concept design utilizing a MODELICA language-based model was achieved, and the optimization time achieved by symbolic computation was largely reduced in comparison to the optimization time achieved by numerical computation.

Valence in Organizational Culture: Cultural Alignment, Employee Satisfaction, and Market Performan

Academy of Management Annual Meeting Proceedings includes abstracts of all papers and symposia presented at the annual conference, plus 6-page abridged versions of the “Best Papers” accepted for inclusion in the program (approximately 10%). Papers published in the Proceedings are abridged because presenting papers at their full length could preclude subsequent journal publication. Please contact the author(s) directly for the full papers. Valence in Organizational Culture: Cultural Alignment, Employee Satisfaction, and Market PerformanMelody Chang, Balazs Kovacs and Amanda SharkeyMelody ChangYale School of ManagementYale School of ManagementU. Of Chicago, Balazs KovacsYale School of ManagementYale School of ManagementU. Of Chicago and Amanda SharkeyYale School of ManagementYale School of ManagementU. Of ChicagoPublished Online:9 Jul 2018https://doi.org/10.5465/AMBPP.2018.10844abstractAboutSections ToolsDownload CitationsAdd to favoritesTrack Citations ShareShare onFacebookTwitterLinkedInRedditEmail AbstractPrior research has emphasized the importance of a ""strong culture"" for firm performance, defining cultural strength on the basis of employee agreement about the beliefs, norms and values that are prized by a firm. We refine the established notion of ""strong culture"" by proposing that ""strong"" cultures encompass two possibilities: (i) having a high level of alignment in what employees describe as positive aspects of culture and (ii) having a high level of alignment in what employees describe as negative aspects of culture. We argue that these two types of cultural strength differently impact how employees evaluate their organizations and how the market anticipates the future performance of these organizations. We adopt a language-based model to measure cultural alignment among employees using over 100,000 unique employee reviews of 816 publicly traded firms across different industries on the Glassdoor website, an online career intelligence platform. Consistent with our theoretical arguments, we find that greater agreement on what employees consider as positive aspects of culture is positively associated with employee satisfaction and market performance. We also find that a higher level of agreement on what employees deem negative aspects of culture portends devaluation of an organization by its employees and by the stock market. Additionally, we find that agreement on negative cultural attributes, compared to agreement on positive cultural attributes, has a greater impact on employees' and the market's evaluation of organizations. We offer insights for managerial practice by suggesting that focusing on eliminating negative aspects of culture that employees agree on will have a greater payoff than focusing on boosting agreement about positive aspects. Contributions to the organizational culture literature are discussed.FiguresReferencesRelatedDetails Vol. 2018, No. 1 Permissions Metrics in the past 12 months History Published online 9 July 2018 Published in print 1 August 2018 InformationCopyright of Academy of Management Proceedings is the property of Academy of Management and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. The Academy of Management does not maintain the full text version of papers presented at the Academy's meetings. Please contact the author(s) directly to obtain access to the full text content.KeywordsAOM Annual Meeting Proceedings 2018AOM Chicago 2018

The (lack of) effect of dynamic visual noise on the concreteness effect in short-term memory

ABSTRACTIt has been suggested that the concreteness effect in short-term memory (STM) is a consequence of concrete words having more distinctive and richer semantic representations. The generation and storage of visual codes in STM could also play a crucial role on the effect because concrete words are more imaginable than abstract words. If this were the case, the introduction of a visual interference task would be expected to disrupt recall of concrete words. A Dynamic Visual Noise (DVN) display, which has been proven to eliminate the concreteness effect on long-term memory (LTM), was presented along encoding of concrete and abstract words in a STM serial recall task. Results showed a main effect of word type, with more item errors in abstract words, a main effect of DVN, which impaired global performance due to more order errors, but no interaction, suggesting that DVN did not have any impact on the concreteness effect. These findings are discussed in terms of LTM participation through redintegration processes and in terms of the language-based models of verbal STM.

The non-strategic nature of linguistic long-term memory effects in verbal short-term memory

Abstract The contribution of lexical and semantic knowledge to verbal short-term memory (vSTM) span is explained by language-based models, assuming that vSTM is deeply grounded within the linguistic system with to-be-remembered items being activated in a non-strategic and automatic manner. However, direct evidence for a non-strategic account of lexical and semantic contributions to vSTM span is scarce. In this study, we assessed the influence of several types of long-term linguistic knowledge (lexicality, lexical frequency, semantic similarity and imageability) on vSTM using a fast encoding running span procedure preventing any strategic processes during encoding. We observed reliable effects of lexicality (words vs. nonwords, Experiment 1), lexical frequency (high vs. low frequency words, Experiment 2) and semantic similarity (related vs. unrelated lists, Experiment 3) on running span performance. However, word imageability (high vs. low imageability words, Experiment 4) did not consistently impact running span performance. Experiment 5 showed that the imageability effect only appears in standard immediate serial recall conditions which do not prevent list-strategic encoding. This study provides novel evidence for linguistic accounts of vSTM by demonstrating a robust impact of lexical and surface-level semantic knowledge on vSTM in non-strategic, fast-encoding conditions.

Enculturation Trajectories: Language, Cultural Adaptation, and Individual Outcomes in Organizations

How do people adapt to organizational culture, and what are the consequences for their outcomes in the organization? These fundamental questions about culture have previously been examined using self-report measures, which are subject to reporting bias, rely on coarse cultural categories defined by researchers, and provide only static snapshots of cultural fit. By contrast, we develop an interactional language use model that overcomes these limitations and opens new avenues for theoretical development about the dynamics of organizational culture. We trace the enculturation trajectories of employees in a midsized technology firm based on analyses of 10.24 million internal emails. Our language-based model of changing cultural fit (1) predicts individual attainment; (2) reveals distinct patterns of adaptation for employees who exit voluntarily, exit involuntarily, and remain employed; (3) demonstrates that rapid early cultural adaptation reduces the risk of involuntary, but not voluntary, exit; and (4) finds that a decline in cultural fit for individuals who had successfully enculturated portends voluntary departure. The supplemental material is available at https://doi.org/10.1287/mnsc.2016.2671 . This paper was accepted by Olav Sorenson, organizations.

Improving P300 spelling rate using language models and predictive spelling

The P300 speller brain-computer interface (BCI) provides a means of communication for those suffering from advanced neuromuscular diseases such as amyotrophic lateral sclerosis (ALS). Recent literature has incorporated language-based modeling, which uses previously chosen characters and the structure of natural language to modify the interface and classifier. Two complementary methods of incorporating language models have previously been independently studied: predictive spelling uses language models to generate suggestions of complete words to allow for the selection of multiple characters simultaneously, and language-model-based classifiers have used prior characters to create a prior probability distribution over the characters based on how likely they are to follow. In this study, we propose a combined method which extends a language-based classifier to generate prior probabilities for both individual characters and complete words. In order to gage the efficiency of this new model, results across 12 healthy subjects were measured. Incorporating predictive spelling increased typing speed using the P300 speller, with an average increase of 15.5% in typing rate across subjects, demonstrating that language models can be effectively utilized to create full word suggestions for predictive spelling. When combining predictive spelling with language-model classification, typing speed is significantly improved, resulting in better typing performance.

The German Who Happened To Be a Jew: Time, History, and Identity in Victor Klemperer's Diary

This article proposes an interpretation of Victor Klemperer's journal of the Nazi period as a text fundamentally concerned with German identity. By highlighting the Diary's binary organization, I attempt to explain Klemperer's outline of a language-based model of Deutschtum, which is structurally bound to Jewish identity and social conditions, and stands in special relation to history, memory, and perception of time. Klemperer's engagement with history, identity, memory, and the past, is especially relevant in Germany, where both conservative and progressive positions have shared the conviction that, in Meier's words, ‘the elusive commonality of being German can be clarified through the representation of the past’. In this sense, a thorough reading of the Diary reveals that the dialectics of Klemperer's writing mirrors the structure of post-war German collective memory, which in turn relates to the Diary's manifold reception.

Student Presentation: A Conceptual Model

This communication suggests a language-based conceptual model on student presentation, as a curriculum learning strategy. Student presentation promotes understanding that can be enhanced if the student is not under stress. In any given medical school, we assume, if the student mother-tongue language is not similar to the school education language, then the student is likely to be stressed. It aimed to contribute ideas towards quality enhancement in medical education. This is a concept study. In this work, we suggest a conceptual model where the student is encouraged to utilize his/her mother-tongue language sources for in-depth understanding of the presentation topic, whenever he/she understands the topic partially. Research has shown that some students, during activities using non-mother-tongue language, were unable to completely understand the subject and/or unable to fully participate in discussions. The present study represents a reaction to a frequent call from medical education experts, around the world, that any given school develops curriculum learning strategies that best meet its mission. Future research exploring on the role of linguistic factors in student presentation, is likely to agree with this work, towards quality enhancement in medical education.

A Web Services Description Language-Based Description Model of Internet of Things Services

A Web Services Description Language-Based Description Model of Internet of Things Services

On ranking relevant entities in heterogeneous networks using a language-based model

A new challenge, accessing multiple relevant entities, arises from the availability of linked heterogeneous data. In this article, we address more specifically the problem of accessing relevant entities, such as publications and authors within a bibliographic network, given an information need. We propose a novel algorithm, called BibRank, that estimates a joint relevance of documents and authors within a bibliographic network. This model ranks each type of entity using a score propagation algorithm with respect to the query topic and the structure of the underlying bi-type information entity network. Evidence sources, namely content-based and network-based scores, are both used to estimate the topical similarity between connected entities. For this purpose, authorship relationships are analyzed through a language model-based score on the one hand and on the other hand, non topically related entities of the same type are detected through marginal citations. The article reports the results of experiments using the Bibrank algorithm for an information retrieval task. The CiteSeerX bibliographic data set forms the basis for the topical query automatic generation and evaluation. We show that a statistically significant improvement over closely related ranking models is achieved.

Natural language-based user interface for mobile devices with limited resources

In this paper, we propose a natural language-based interface model to enable a user to articulate a request without having any specific knowledge about a mobile device. In consideration of the very limited computing and memory capacity of the mobile device and to keep the development cost low, the proposed model does not depend on typical natural language techniques, but on a ranking technique, that is simplified based on the mathematical derivation process with the following assumptions. One assumption is that a device control command consists of a function and its parameters. The other assumption is that the parameter is represented as few predictable patterns, whereas the function can be represented as various sentence patterns. To deal with these various sentence patterns, the proposed model selects the top ranked command candidate with the highest score after generating all possible candidates with their scores. Furthermore, the ranking score function is designed to achieve a high discriminative capability by the simulation of the process of generating every candidate. Experimental results show that the proposed model with 2.9 megabytes performs at 96.27% accuracy, which is slightly lower than 97.06% of the baseline model with 135.2 megabytes.

The interaction of word frequency and concreteness in immediate serial recall

Word frequency and word concreteness are language attributes that have been shown to independently influence the recall of items in verbal short-term memory (STM). It has been argued that such effects are evidence for the action of long-term memory knowledge on STM traces. However, research to date has not investigated whether these variables interact in serial recall. In two experiments, we examined the behavior of these variables under factorial manipulation and demonstrated that the effect of word frequency is dependent on the level of concreteness of items. Serial recall performance is examined with reference to two explanatory approaches: Walker and Hulme's (1999) dual-redintegration account and language-based models of STM. The data indicate that language-based models are more compatible with the observed effects and challenge the view that frequency and concreteness effects in STM are the products of distinct mechanisms.