Search results for: contractual language

Authors: Nikita Katyal, Shaurya Uppal

Abstract:

This paper tackles the significant challenge of accurately classifying customers within a retailer’s customer base. Accurate classification is essential for developing targeted marketing strategies that effectively engage this important demographic. To address this issue, we propose a method that utilizes Large Language Models (LLMs). By employing LLMs, we analyze the metadata associated with product purchases derived from historical data to identify key product categories that act as distinguishing factors. These categories, such as baby food, eldercare products, or family-sized packages, offer valuable insights into the likely household composition of customers, including families with babies, families with kids/teenagers, families with pets, households caring for elders, or mixed households. We segment high-confidence customers into distinct categories by integrating historical purchase behavior with LLM-powered product classification. This paper asserts that life stage segmentation can significantly enhance e-commerce businesses’ ability to target the appropriate customers with tailored products and campaigns, thereby augmenting sales and improving customer retention. Additionally, the paper details the data sources, model architecture, and evaluation metrics employed for the segmentation task.

Keywords: LLMs, segmentation, product tags, fine-tuning, target segments, marketing communication

Procedia PDF Downloads 21

Authors: Weijing Zhou, Yuting Lei, Francis Nolan

Abstract:

In the domain of second language acquisition, whenever pronunciation errors or acquisition difficulties are found, researchers habitually attribute them to the negative transfer of the native language or local dialect. To what extent do Mandarin and local dialects affect English phonological acquisition for Chinese learners of English as a foreign language (EFL)? Little evidence, however, has been found via empirical research in China. To address this core issue, the present study conducted phonetic experiments to explore the roles of local dialects and Mandarin in Chinese EFL learners’ acquisition of L2 English consonants. Besides Mandarin, the sole national language in China, Suzhou dialect was selected as the target local dialect because of its distinct phonology from Mandarin. The experimental group consisted of 30 junior English majors at Yangzhou University, who were born and lived in Suzhou, acquired Suzhou Dialect since their early childhood, and were able to communicate freely and fluently with each other in Suzhou Dialect, Mandarin as well as English. The consonantal target segments were all the consonants of English, Mandarin and Suzhou Dialect in typical carrier words embedded in the carrier sentence Say again. The control group consisted of two Suzhou Dialect experts, two Mandarin radio broadcasters, and two British RP phoneticians, who served as the standard speakers of the three languages. The reading corpus was recorded and sampled in the phonetic laboratories at Yangzhou University, Soochow University and Cambridge University, respectively, then transcribed, segmented and analyzed acoustically via Praat software, and finally analyzed statistically via EXCEL and SPSS software. The main findings are as follows: First, in terms of correct acquisition rates (CARs) of all the consonants, Mandarin ranked top (92.83%), English second (74.81%) and Suzhou Dialect last (70.35%), and significant differences were found only between the CARs of Mandarin and English and between the CARs of Mandarin and Suzhou Dialect, demonstrating Mandarin was overwhelmingly more robust than English or Suzhou Dialect in subjects’ multilingual phonological ecology. Second, in terms of typical acoustic features, the average duration of all the consonants plus the voice onset time (VOT) of plosives, fricatives, and affricatives in 3 languages were much longer than those of standard speakers; the intensities of English fricatives and affricatives were higher than RP speakers but lower than Mandarin and Suzhou Dialect standard speakers; the formants of English nasals and approximants were significantly different from those of Mandarin and Suzhou Dialects, illustrating the inconsistent acoustic variations between the 3 languages. Thirdly, in terms of typical pronunciation variations or errors, there were significant interlingual interactions between the 3 consonant systems, in which Mandarin consonants were absolutely dominant, accounting for the strong transfer from L1 Mandarin to L2 English instead of from earlier-acquired L1 local dialect to L2 English. This is largely because the subjects were knowingly exposed to Mandarin since their nursery and were strictly required to speak in Mandarin through all the formal education periods from primary school to university.

Keywords: acquisition of L2 English consonants, role of Mandarin, role of local dialect, Chinese EFL learners from Suzhou Dialect areas

Procedia PDF Downloads 95

Authors: Yang Cao

Abstract:

Analyzing the oral production of Chinese-speaking learners of English as a second language (L2), we can find a large variety of verb inflections – Why does it seem so hard for them to use consistent correct past morphologies in obligatory past contexts? Failed Functional Features Hypothesis (FFFH) attributes the rather non-target-like performance to the absence of [±past] feature in their L1 Chinese, arguing that for post puberty learners, new features in L2 are no more accessible. By contrast, Missing Surface Inflection Hypothesis (MSIH) tends to believe that all features are actually acquirable for late L2 learners, while due to the mapping difficulties from features to forms, it is hard for them to realize the consistent past morphologies on the surface. However, most of the studies are limited to the verb morphologies in finite clauses and few studies have ever attempted to figure out these learners’ performance in non-finite clauses. Additionally, it has been discussed that Chinese learners may be able to tell the finite/infinite distinction (i.e. the [±finite] feature might be selected in Chinese, even though the existence of [±past] is denied). Therefore, adopting a self-paced reading task (SPR), the current study aims to analyze the processing patterns of Chinese-speaking learners of L2 Russian, in order to find out if they are sensitive to misuse of tense morphologies in both finite and non-finite clauses and whether they are sensitive to the finite/infinite distinction presented in Russian. The study targets L2 Russian due to its systematic morphologies in both present and past tenses. A native Russian group, as well as a group of English-speaking learners of Russian, whose L1 has definitely selected both [±finite] and [±past] features, will also be involved. By comparing and contrasting performance of the three language groups, the study is going to further examine and discuss the two theories, FFFH and MSIH. Preliminary hypotheses are: a) Russian native speakers are expected to spend longer time reading the verb forms which violate the grammar; b) it is expected that Chinese participants are, at least, sensitive to the misuse of inflected verbs in non-finite clauses, although no sensitivity to the misuse of infinitives in finite clauses might be found. Therefore, an interaction of finite and grammaticality is expected to be found, which indicate that these learners are able to tell the finite/infinite distinction; and c) having selected [±finite] and [±past], English-speaking learners of Russian are expected to behave target-likely, supporting L1 transfer.

Keywords: features, finite clauses, morphosyntax, non-finite clauses, past morphologies, present morphologies, Second Language Acquisition, self-paced reading task, verb inflections

Procedia PDF Downloads 106

Authors: Tayebeh Norouzi

Abstract:

This study aims to investigate the acquisition process of intonation. It examines the intonation structure of Tokyo Japanese and its realization by Iranian learners of Japanese. Seven Iranian learners of Japanese, differing in fluency, and two Japanese speakers participated in the experiment. Two sentences were used to test the phonological and phonetic characteristics of lexical pitch-accent as well as the intonation patterns produced by the speakers. Both sentences consisted of similar words with the same number of syllables and lexical pitch-accents but different syntactic structure. Speakers were asked to read each sentence three times at normal speed, and the data were analyzed by Praat. The results show that lexical pitch-accent, Accentual Phrase (AP) and AP boundary tone realization vary depending on sentence type. For sentences of type XdeYwo, the lexical pitch-accent is realized properly. However, there is a rise in AP boundary tone regardless of speakers’ level of fluency. In contrast, in sentences of type XnoYwo, the lexical pitch-accent and AP boundary tone vary depending on the speakers’ fluency level. Advanced speakers are better at grouping words into phrases and produce more native-like intonation patterns, though they are not able to realize downstep properly. The non-native speakers tried to realize proper intonation patterns by making changes in lexical accent and boundary tone.

Keywords: intonation, Iranian learners, Japanese prosody, lexical accent, second language acquisition.

Procedia PDF Downloads 169

Authors: Francia Martinez

Abstract:

This study documents and demonstrates the profound impact of the high status of American English and culture in Colombian commercial landscape due to the globalization and commodification of English. It also documents and describes how Colombian advertisers make use of various language and visual mechanisms in the commercial linguistic landscape to convey messages, create an image with which the target audience can identify, and build a relationship with that target audience. The data (in the form of pictures) were collected in different cities in Colombia and were classified and organized into different categories for the reliability and validity of the analysis. The research questions were: do the ubiquity and high status of American English and culture play a major role in the Colombian commercial linguistic landscape? If so, how?, what roles do national and local culture and language (Spanish) play in the commercial linguistic landscape?, and what different linguistic and visual strategies do Colombian advertisers employ to reach their target audience? Based on data analysis and results, American and local culture and icons play a major role when Colombian advertisers create and design their commercial logos and ads to get consumers’ attention and establish a rapport with them in a successful way. In order to achieve their objectives, Colombian advertisers rely on creative linguistic and visual techniques in their ads, such as puns, humor, irony, comparisons, metaphors, mocking, exaggeration, parody, personification, sarcasm, satire, allusion, onomatopoeias, and imitation (copycat or cloning).

Keywords: Colombian ads, linguistic landscape, rhetorical devices, sociolinguistics

Procedia PDF Downloads 309

Authors: Mahmood Dehqan, Peyman Peyvasteh

Abstract:

In recent years there has been a growing interest in issues concerning the impact of sociocultural theory (SCT) of learning on different aspects of second/foreign language learning. This study aimed to find the possible effects of sociocultural teaching techniques on reading strategy of EFL learners. Indeed, the present research compared the impact of peer and teacher scaffolding on EFL learners’ reading strategy use across two proficiency levels. To this end, a pre-test post-test quasi-experimental research design was used and two instruments were utilized to collect the data: Nelson English language test and reading strategy questionnaire. Ninety five university students participated in this study were divided into two groups of teacher and peer scaffolding. Teacher scaffolding group received scaffolded help from the teacher based on three mechanisms of effective help within ZPD: graduated, contingent, dialogic. In contrast, learners of peer scaffolding group were unleashed from the teacher-fronted classroom as they were asked to carry out the reading comprehension tasks with the feedback they provided for each other. Results obtained from ANOVA revealed that teacher scaffolding group outperformed the peer scaffolding group in terms of reading strategy use. It means teacher’s scaffolded help provided within the learners’ ZPD led to better reading strategy improvement compared with the peer scaffolded help. However, the interaction effect between proficiency factor and teaching technique was non-significant, leading to the conclusion that strategy use of the learners was not affected by their proficiency level in either teacher or peer scaffolding groups.

Keywords: peer scaffolding, proficiency level, reading strategy, sociocultural theory, teacher scaffolding

Procedia PDF Downloads 380

Authors: Guangyan Chen

Abstract:

This study examined Chinese teachers’ perceptions of assessing U.S. college students’ L2 (second language) Chinese oral performances at different levels. Ten oral performances were videotaped from which three were chosen as samples to represent three different proficiency levels based on professionals’ judgments according to the ACTFL proficiency guidelines. The three samples were shown to L2 Chinese teachers who completed questionnaires about their assessments for each speech sample. In total, 104 L2 Chinese teachers responded to each of the three samples. The Exploratory Factor Analyses (EFA) of the teachers’ responses revealed three similar rating criteria patterns for assessing the three levels of oral performances. The teachers’ responses to Samples 2 and 3 revealed five rating criteria: Global proficiency, Chinese conceptual framework, content richness, communication appropriateness, and communication clarity. The teachers’ responses to Sample 1 revealed four rating criteria: global proficiency, Chinese conceptual framework, communication appropriateness/content richness, and communication clarity. However, the analyses of variance (ANOVAs) revealed that the proficiency levels of the three oral performances differed significantly across all rating criteria. Therefore, the data suggests that L2 classroom teachers could use the similar rating criteria pattern to assess college-level L2 Chinese students’ oral performances at different proficiency levels.

Keywords: language assessment, L2 Chinese, oral performance, rating criteria

Procedia PDF Downloads 538

Authors: Noah Brandon

Abstract:

Grammatical gender can create a restrictive roadblock for the usage of gender-inclusive language. This research describes grammatical gender structures used in English, Dutch, and German and considers how these structures restrict the implementation of gender inclusivity in spoken and written discourse. This restriction is measured by the frequency with which gender-inclusive & generic masculine forms are used and by the morphosyntactic complexity of the gender-inclusive forms available in these languages. These languages form a continuum of grammatical gender structures, with English having the least articulated structures and German having the most. This leads to a comparative analysis intended to establish a correlation between the complexity of gender structure and the difficulty of using gender-inclusive forms. English, on one side of the continuum, maintains only remnants of a formal grammatical gender system and imposes the fewest restrictions on the creation of neo-pronouns and the use of gender-inclusive alternatives to gendered agentive nouns. Next, the Dutch have a functionally two-gender system with less freedom using gender-neutral forms. Lastly, German, on the other end, has a three-gender system requiring a plethora of morphosyntactic and orthographic alternatives to avoid using generic masculine. The paper argues that the complexity of grammatical gender structures correlates with hindered use of gender-inclusive forms. Going forward, efforts will focus on gathering further data on the usage of gender-inclusive and generic masculine forms within these languages. The end goal of this research is to establish a definitive objective correlation between grammatical gender complexity and impediments in expressing socially constructed gender.

Keywords: sociolinguistics, language and gender, gender, Germanic linguistics, grammatical gender, German, Dutch, English

Procedia PDF Downloads 76

Authors: Samuel Alaba Akinwotu

Abstract:

Over the years, Nigeria’s educational system has greatly suffered from the menace of industrial conflict. The smooth running of the nation’s public educational institutions has been hampered by incessant strikes embarked upon by workers of these institutions. Even though industrial conflicts in Nigeria have enjoyed wide reportage in the media, there has been a dearth of critical examination of the language use that index the conflict’s discourse in the media. This study which is driven by a combination of Critical Discourse Analysis (CDA) and Conceptual Metaphor (CM) examines the discursive and ideological features of language indexing the industrial conflict between the Academic Staff Union of Universities (ASUU) and the Federal Government of Nigeria (FGN) in 2013. It aims to identify and assess the conceptual and cognitive motivations of the stances expressed by the parties and the public and the role of the media in the management and resolution of the conflict. For data, media reports and readers’ comments were purposively sampled from six print and online news sources (The Punch, This Day, Vanguard, The Nation, Osun Defender and AITonline) published between July and December 2013. The study provides further insight into industrial conflict and proves to be useful for the management and resolution of industrial conflicts especially in our public educational institutions.

Keywords: industrial conflict, critical discourse analysis, conceptual metaphor, federal government of Nigeria, academic staff union of universities

Procedia PDF Downloads 142

Authors: Nino Gogatishvili, Davit Kvernadze, Giorgi Japharidze

Abstract:

Background and aims: Landau Kleffner syndrome (LKS) is a rare disorder with epileptic seizures and acquired aphasia. It usually starts in initially healthy children. The first symptoms are language regression and behavioral disturbances, and the sleep EEG reveals abnormal epileptiform activity. The aim was to discuss the efficacy of Clobazam for Landau Kleffner syndrome. Case report: We report a case of an 11-year-old boy with an uneventful pregnancy and delivery. He began to walk at 11 months and speak with simple phrases at the age of 2,5 years. At the age of 18 months, he had febrile convulsions; at the age of 5 years, the parents noticed language regression, stuttering, and serious behavioral dysfunction, including hyperactivity, temper outbursts. The epileptic seizure was not noticed. MRI was without any abnormality. Neuropsychological testing revealed verbal auditory agnosia. Sleep EEG showed abundant left fronto-temporal spikes, reaching over 85% during non-rapid eye movement sleep (non-REM sleep). Treatment was started with Clobazam. After ten weeks, EEG was improved. Stuttering and behavior also improved. Results: Since the start of Clobazam treatment, stuttering and behavior improved. Now, he is 11 years old, without antiseizure medication. Sleep EEG shows fronto-temporal spikes on the left side, over 10-49 % of non-REM sleep, bioccipital spikes, and slow-wave discharges and spike-waves. Conclusions: This case provides further support for the efficacy of Clobazam in patients with LKS.

Keywords: Landau-Kleffner syndrome, antiseizure medication, stuttering, aphasia

Procedia PDF Downloads 64

Authors: L. R. Valencia Pérez, A. Morita Alexander, Peña A. Juan Manuel, A. Lamadrid Álvarez

Abstract:

This research shows the reflection and identification of explanatory variables and their relationships between different variables that are involved with educational technology, all of them encompassed in macro-concepts which are: cognitive inequality, economy, food and language; These will give the guideline to have a more detailed knowledge of educational systems, the communication and equipment, the physical space and the teachers; All of them interacting with each other give rise to what is called educational technology management. These elements contribute to have a very specific knowledge of the equipment of communications, networks and computer equipment, systems and content repositories. This is intended to establish the importance of knowing a global environment in the transfer of knowledge in poor countries, so that it does not diminish the capacity to be authentic and preserve their cultures, their languages or dialects, their hierarchies and real needs; In short, to respect the customs of different towns, villages or cities that are intended to be reached through the use of internationally agreed professional educational technologies. The methodology used in this research is the analytical - descriptive, which allows to explain each of the variables, which in our opinion must be taken into account, in order to achieve an optimal incorporation of the educational technology in a model that gives results in a medium term. The idea is that in an encompassing way the concepts will be integrated to others with greater coverage until reaching macro concepts that are of national coverage in the countries and that are elements of conciliation in the different federal and international reforms. At the center of the model is the educational technology which is directly related to the concepts that are contained in factors such as the educational system, communication and equipment, spaces and teachers, which are globally immersed in macro concepts Cognitive inequality, economics, food and language. One of the major contributions of this article is to leave this idea under an algorithm that allows to be as unbiased as possible when evaluating this indicator, since other indicators that are to be taken from international preference entities like the OECD in the area of education systems studied, so that they are not influenced by particular political or interest pressures. This work opens the way for a relationship between involved entities, both conceptual, procedural and human activity, to clearly identify the convergence of their impact on the problem of education and how the relationship can contribute to an improvement, but also shows possibilities of being able to reach a comprehensive education reform for all.

Keywords: relationships macro-concepts, cognitive inequality, economics, alimentation and language

Procedia PDF Downloads 198

Authors: Ayse Tarhan, Mustafa Ilkan, Mohammad Karimzadeh

Abstract:

Science was considered as part of philosophy in ancient Greece. By the nineteenth century, it was understood that philosophy was very inclusive and that social and human sciences such as literature, history, and psychology should be separated and perceived as an autonomous branch of science. The computer was also first seen as a tool of mathematical science. Over time, computer science has grown by encompassing every area in which technology exists, and its growth compelled the division of computer science into different disciplines, just as philosophy had been divided into different branches of science. Now there is almost no branch of science in which computers are not used. One of the newer autonomous disciplines of computer science is digital humanities, and one of the areas of digital humanities is literature. The material of literature is words, and thanks to the software tools created using computer programming languages, data that a literature researcher would need months to complete, can be achieved quickly and objectively. In this article, three different tools that literary researchers can use in their work will be introduced. These studies were created with the computer programming languages Python and R and brought to the world of literature. The purpose of introducing the aforementioned studies is to set an example for the development of special tools or programs on Ottoman language and literature in the future and to support such initiatives. The first example to be introduced is the Stylometry tool developed with the R language. The other is The Metrical Tool, which is used to measure data in poems and was developed with Python. The latest literature analysis tool in this article is Voyant Tools, which is a multifunctional and easy-to-use tool.

Keywords: DH, literature, information technologies, stylometry, the metrical tool, voyant tools

Procedia PDF Downloads 150

Authors: Qin Long, Li Xiaoge

Abstract:

The Knowledge Graphs have now become a new form of knowledge representation. However, there is no consensus in regard to a plausible and definition of entities and relationships in the domain-specific knowledge graph. Further, in conjunction with several limitations and deficiencies, various domain-specific entities and relationships recognition approaches are far from perfect. Specifically, named entity recognition in Chinese domain is a critical task for the natural language process applications. However, a bottleneck problem with Chinese named entity recognition in new domains is the lack of annotated data. To address this challenge, a domain distant supervised named entity recognition framework is proposed. The framework is divided into two stages: first, the distant supervised corpus is generated based on the entity linking model of graph attention neural network; secondly, the generated corpus is trained as the input of the distant supervised named entity recognition model to train to obtain named entities. The link model is verified in the ccks2019 entity link corpus, and the F1 value is 2% higher than that of the benchmark method. The re-pre-trained BERT language model is added to the benchmark method, and the results show that it is more suitable for distant supervised named entity recognition tasks. Finally, it is applied in the computer field, and the results show that this framework can obtain domain named entities.

Keywords: distant named entity recognition, entity linking, knowledge graph, graph attention neural network

Procedia PDF Downloads 90

Authors: Ngoc-Nguyen Nguyen, Hsiu-Ling Chen, Thanh-Truc Lai Huynh

Abstract:

In foreign language learning, vocabulary acquisition has consistently posed challenges to learners, especially for those at lower levels. Conventional approaches often fail to promote vocabulary learning and ensure engaging experiences alike. The emergence of mobile learning, particularly the integration of chatbot systems, has offered alternative ways to facilitate this practice. Chatbots have proven effective in educational contexts by offering interactive learning experiences in a constructivist manner. These tools have caught attention in the field of mobile-assisted language learning (MALL) in recent years. This research is conducted in an English for Specific Purposes (ESP) course at the A2 level of the CEFR, designed for non-English majors. Participants are first-year Vietnamese students aged 18 to 20 at a university. This quasi-experimental study follows a pretest-posttest control group design over five weeks, with two classes randomly assigned as the experimental and control groups. The experimental group engages in chatbot-assisted spaced practice with SRL components, while the control group uses the same spaced practice without SRL. The two classes are taught by the same lecturer. Data are collected through pre- and post-tests, cognitive load surveys, and semi-structured interviews. The combination of self-regulated learning (SRL) and distributed practice, grounded in the spacing effect, forms the basis of the present study. SRL elements, which concern goal setting and strategy planning, are integrated into the system. The spaced practice method, similar to those used in widely recognized learning platforms like Duolingo and Anki flashcards, spreads out learning over multiple sessions. This study’s design features quizzes progressively increasing in difficulty. These quizzes are aimed at targeting both the Recognition-Recall and Comprehension-Use dimensions for a comprehensive acquisition of vocabulary. The mobile-based chatbot system is built using Golang, an open-source programming language developed by Google. It follows a structured flow that guides learners through a series of 4 quizzes in each week of teacher-led learning. The quizzes start with less cognitively demanding tasks, such as multiple-choice questions, before moving on to more complex exercises. The integration of SRL elements allows students to self-evaluate the difficulty level of vocabulary items, predict scores achieved, and choose appropriate strategy. This research is part one of a two-part project. The initial findings will determine the development of an upgraded chatbot system in part two, where adaptive features in response to the integration of SRL components will be introduced. The research objectives are to assess the effectiveness of the chatbot-assisted approach, based on the combination of spaced practice and SRL, in improving vocabulary acquisition and managing cognitive load, as well as to understand students' perceptions of this learning tool. The insights from this study will contribute to the growing body of research on mobile-assisted language learning and offer practical implications for integrating chatbot systems with spaced practice into educational settings to enhance vocabulary learning.

Keywords: mobile learning, mobile-assisted language learning, MALL, chatbots, vocabulary learning, spaced practice, spacing effect, self-regulated learning, SRL, self-regulation, EFL, cognitive load

Procedia PDF Downloads 17

Authors: Asena Biber, Ates Gul Ergun, Seda Bulut

Abstract:

Based on the literature, there are large numbers of studies investigating the relationship between culture and leadership effectiveness. Although giving effective speeches is vital characteristic for a leader to be perceived as effective, to our knowledge, there is no research study the determinants of perceived effective leader speech. The aim of this study is to find the effects of both culture and voice pitch on perceptions of leader's speech effectiveness. Our hypothesis is that people from high power distance countries will perceive leaders' speech effective when the leader's voice pitch is high, comparing with people from relatively low power distance countries. The participants of the study were 36 undergraduate students (12 Pakistanis, 12 Nigerians, and 12 Turks) who are studying in Turkey. National power distance scores of Nigerians ranked as first, Turks ranked as second and Pakistanis ranked as third. There are two independent variables in this study; three nationality groups that representing three levels of power distance and voice pitch of the leader which is manipulated as high and low levels. Researchers prepared an audio to manipulate high and low conditions of voice pitch. A professional whose native language is English read the predetermined speech in high and low voice pitch conditions. Voice pitch was measured using Hertz (Hz) and Decibel (dB). Each nationality group (Pakistan, Nigeria, and Turkey) were divided into groups of six students who listened to either the low or high pitch conditions in the cubicles of the laboratory. It was expected from participants to listen to the audio and fill in the questionnaire which was measuring the leadership effectiveness on a response scale ranging from 1 to 5. To determine the effects of nationality and voice pitch on perceived effectiveness of leader' voice pitch, 3 (Pakistani, Nigerian, and Turk) x 2 (low voice pitch and high voice pitch) two way between subjects analysis of variances was carried out. The results indicated that there was no significant main effect of voice pitch and interaction effect on perceived effectiveness of the leader’s voice pitch. However, there was a significant main effect of nationality on perceived effectiveness of the leader's voice pitch. Based on the results of Turkey’s HSD post-hoc test, only the perceived effectiveness of the leader's speech difference between Pakistanis and Nigerians was statistically significant. The results show that the hypothesis of this study was not supported. As limitations of the study, it is of importance to mention that the sample size should be bigger. Also, the language of the questionnaire and speech should be in the participant’s native language in further studies.

Keywords: culture, leadership effectiveness, power distance, voice pitch

Procedia PDF Downloads 181

Authors: Leticia Souza Guirra, Márcia Eduarda Vieira Ramos, Edlaine Souza Pereira, Leticia Correa Celeste

Abstract:

Introduction: Studies indicate that equine-assisted therapy enables improvements in several areas of functioning that are impaired in children with autism spectrum disorder (ASD), such as social interaction and communication. Objective: The study proposes to analyze the development of dialogic skills of a verbal child with ASD after participating in the equine-assisted therapy Step By Step in Communication. Method: This is quantitative and qualitative research through a case study. It refers to a 6 years old child diagnosed with ASD belonging to a group of practitioners of the Brazilian National Equine-Assited-Therapy Association. The Behavioral Observation Protocol (PROC) was used to evaluate communicative skills before and after the intervention, which consisted of 24 sessions once a week. Results: All conversational skills increased their frequency, with participation in dialogue and initiation of interaction. The child also increases the habit of waiting for his turn and answering the interlocutor. The emission of topics not related to conversation and echolalia showed a significant decrease after the intervention. Conclusion: The studied child showed improvement in communicative skills after participating in the equine-assisted therapy Step By Step in Communication. Contributions: This study contributes to a greater understanding of the impact of equine-assisted therapy on the communicative abilities of children with ASD.

Keywords: equine-assisted-therapy, autism spectrum disorder, language, communication, language and hearing sciences

Procedia PDF Downloads 79

Authors: Mohsine Jebbour

Abstract:

Motivation and classroom interaction are of prime significance for second/foreign language learning to take place effectively. Thus, a considerable amount of motivation and classroom interaction helps ensure students’ success in and continuation of learning the TL. One way to enhance students’ motivation and classroom interaction in the Moroccan EFL classroom then is through the use of self-disclosure. For the purposes of this study, self-disclosure has been defined as the verbal communication of positive personal information including opinions, feelings, experiences, family and friendship stories to classmates and teachers. This paper is meant to demonstrate that positive self-disclosure can serve as an effective tool for helping students develop favorable attitudes toward the EFL classroom (i.e., English courses, teacher of English, and classroom activities) and promoting their intrinsic motivation (IM to know and IM toward stimulation). A further objective is that since self-disclosure is reciprocal, when teachers of English reveal their personal information, students will uncover their personal matters in return. This will help ensure effective classroom participation, foster teacher-student communication, and encourage students to practice and hence improve their oral proficiency (i.e., the speaking skill). A questionnaire was used to collect data in this study. 164 undergraduate students (99 females and 65 males) from the department of English at the faculty of letters and humanities, Dher el Mehraz, Sidi Mohammed Ben Abd Allah University completed a questionnaire that assessed self-disclosure in relation to motivation (i.e., attitudes toward the learning situation and intrinsic motivation) and classroom interaction (i.e., teacher-student interaction, participation, and out-of-class communication) on a 1 to 5 scale with (1) Strongly Disagree and (5) Strongly Agree. The level of agreement on the positive dimension of self-disclosure was ranked first by the respondents. The hypothesis set at the very beginning of the study, which posited that positive self-disclosure is essential to enhancing motivation and classroom interaction in the EFL context, was confirmed. In this regard, the findings suggest that implementing self-disclosure in the Moroccan EFL classroom may serve as an effective tool to have positive affect of teacher, class and classroom activities. This in turn will encourage the learners to attend classes, enjoy the language learning activity, complete classroom assignments, participate in class discussions, and interact with their teachers and classmates. It is hoped that teachers benefit from the results of this study and hence encourage the use of positive self-disclosure to develop English language learning in the Moroccan context where opportunities of using English outside the classroom are limited.

Keywords: EFL classroom, classroom interaction, motivation, self-disclosure

Procedia PDF Downloads 312

Authors: Christine Demaecker

Abstract:

Many researchers have underlined the importance of metaphors in specialised language. Their use of specific domains helps us understand the conceptualisations used to communicate new ideas or difficult topics. Within the wide area of specialised discourse, wine tasting is a very specific example because it is almost exclusively metaphoric. Wine tasting metaphors express various conceptualisations. They are not linguistic but rather conceptual, as defined by Lakoff & Johnson. They correspond to the linguistic expression of a mental projection from a well-known or more concrete source domain onto the target domain, which is the taste of wine. But unlike most specialised terminologies, the vocabulary is never clearly defined. When metaphorical terms are listed in dictionaries, their definitions remain vague, unclear, and circular. They cannot be replaced by literal linguistic expressions. This makes it impossible to transfer them into another language with the traditional linguistic translation methods. Qualitative research investigates whether wine tasting metaphors could rather be translated with the cognitive translation process, as well described by Nili Mandelblit (1995). The research is based on a corpus compiled from two high-profile wine guides; the Parker’s Wine Buyer’s Guide and its translation into French and the Guide Hachette des Vins and its translation into English. In this small corpus with a total of 68,826 words, 170 metaphoric expressions have been identified in the original English text and 180 in the original French text. They have been selected with the MIPVU Metaphor Identification Procedure developed at the Vrije Universiteit Amsterdam. The selection demonstrates that both languages use the same set of conceptualisations, which are often combined in wine tasting notes, creating conceptual integrations or blends. The comparison of expressions in the source and target texts also demonstrates the use of the cognitive translation approach. In accordance with the principle of relevance, the translation always uses target language conceptualisations, but compared to the original, the highlighting of the projection is often different. Also, when original metaphors are complex with a combination of conceptualisations, at least one element of the original metaphor underlies the target expression. This approach perfectly integrates into Lederer’s interpretative model of translation (2006). In this triangular model, the transfer of conceptualisation could be included at the level of ‘deverbalisation/reverbalisation’, the crucial stage of the model, where the extraction of meaning combines with the encyclopedic background to generate the target text.

Keywords: cognitive translation, conceptual integration, conceptual metaphor, interpretative model of translation, wine tasting metaphor

Procedia PDF Downloads 130

Authors: Nargiz Ismayilova

Abstract:

The issue of modernism and the concept of postmodernism has been the focus of world researchers for many years, and there are very few researchers who have come to a common denominator about this term. During the independence period, the expansion of the relations of Azerbaijani literature with the world has led to the spread of many currents and tendencies formed in the West to the literary environment in our country. In this context, the works created in our environment are distinguished by their extreme richness in terms of subject matter and diversity in terms of genre. As an interesting example of contemporary postmodern prose in Azerbaijan, Parviz Jabrayil's novel "In a Foreign Language" pays attention with its more different plotline. The disagreement exists among the critics about the novel. Some are looking for high artistry in work; others are satisfied with the elements of postmodernism in work. Delimitation of the border between modernism and postmodernism can serve to carry out a deep scientific study of the novel. The novel depicts the world in the author's consciousness against the background of water shortage (thirst) in the Old City (Icharishahar). The author deconstructs today's Ichari Shahar mould. Along with modernism, elements of postmodernism occupy a large place in the work. When we look at the general tendencies of postmodernist art, we see that science and individuality are questioned, criticizing the sharp boundaries of modernism and the negativity of these restrictions, and modernism offers alternatives to artistic production by identifying its negatives and shortcomings in the areas of artistic freedom. The novel is extremely interesting in this point of view.

Keywords: concept of postmodernism, modernism, delimitation, political postmodernism, modern postmodern prose, Azerbaijani literature, novel, comparison, world literature, analysis

Procedia PDF Downloads 136

Authors: Elcin Timur Cakmak, Ayse Oguzlar

Abstract:

This study presents the results of the tweets sent by Twitter users on social media about the Russia-Ukraine war by bigram and trigram methods. On February 24, 2022, Russian President Vladimir Putin declared a military operation against Ukraine, and all eyes were turned to this war. Many people living in Russia and Ukraine reacted to this war and protested and also expressed their deep concern about this war as they felt the safety of their families and their futures were at stake. Most people, especially those living in Russia and Ukraine, express their views on the war in different ways. The most popular way to do this is through social media. Many people prefer to convey their feelings using Twitter, one of the most frequently used social media tools. Since the beginning of the war, it is seen that there have been thousands of tweets about the war from many countries of the world on Twitter. These tweets accumulated in data sources are extracted using various codes for analysis through Twitter API and analysed by Python programming language. The aim of the study is to find the word sequences in these tweets by the n-gram method, which is known for its widespread use in computational linguistics and natural language processing. The tweet language used in the study is English. The data set consists of the data obtained from Twitter between February 24, 2022, and April 24, 2022. The tweets obtained from Twitter using the #ukraine, #russia, #war, #putin, #zelensky hashtags together were captured as raw data, and the remaining tweets were included in the analysis stage after they were cleaned through the preprocessing stage. In the data analysis part, the sentiments are found to present what people send as a message about the war on Twitter. Regarding this, negative messages make up the majority of all the tweets as a ratio of %63,6. Furthermore, the most frequently used bigram and trigram word groups are found. Regarding the results, the most frequently used word groups are “he, is”, “I, do”, “I, am” for bigrams. Also, the most frequently used word groups are “I, do, not”, “I, am, not”, “I, can, not” for trigrams. In the machine learning phase, the accuracy of classifications is measured by Classification and Regression Trees (CART) and Naïve Bayes (NB) algorithms. The algorithms are used separately for bigrams and trigrams. We gained the highest accuracy and F-measure values by the NB algorithm and the highest precision and recall values by the CART algorithm for bigrams. On the other hand, the highest values for accuracy, precision, and F-measure values are achieved by the CART algorithm, and the highest value for the recall is gained by NB for trigrams.

Keywords: classification algorithms, machine learning, sentiment analysis, Twitter

Procedia PDF Downloads 73

Authors: Sora Suh, Catherine Michener

Abstract:

This paper examines the collaborative forms of advocacy that a group of four public school teachers took for their emergent bilingual students in one public school district. While teacher advocacy takes many forms in and out of the classroom, much advocacy work is done by individuals and less by collective action. As a result, individual teachers risk isolation or marginalization in their school contexts when they advocate for immigrant youth. This paper is intended to contribute to the documentation and understanding of teachers’ advocacy work as a collaborative act in teacher education research. The increase of ELs in US classrooms and a corresponding lack of teacher preparation to meet the needs of ELs has motivated the training of educators in linguistically responsive education (e.g., ESL, sheltered English instruction [SEI], bilingual education). Drawing from educational theories of linguistically responsive teaching for preparing educators, we trace the linguistically responsive advocacy work of the teachers. The paper is a multiple case study that tracks how teachers’ discussions on advocacy during a teacher preparation program leading to collaborative actions in their daily teaching lives in and out of school. Data collected includes online discussion forums on the topic of advocacy, course assignments on the topic of advocacy, video-audio recordings of classroom teaching observations, and video-audio recordings of individual and focus group interviews. The findings demonstrate that the teachers’ understanding of advocacy developed through collaborative partnerships formed in the teacher preparation program and grew into active forms of collaborative advocacy in their teaching practice in and out of school. The teachers formed multi-level and collaborative partnerships with teachers, families, community members, policymakers from the local government, and educational researchers to advocate for their emergent bilingual students by planning advocacy events such as new family orientations for emergent bilinguals, professional development for general education teachers on the topic of linguistically responsive instruction, and family nights hosted by the district. The paper’s findings present types of advocacy work in which teachers engage (pedagogical, curricular, out-of-school work) and provide evidence of collaborative advocacy work by a group of engaged educators. The paper highlights the increased agency and effective advocacy of teachers through teacher education and collaborative partnerships and suggests a need for more research on collaborative forms of teacher advocacy for emergent bilinguals.

Keywords: language education, teacher advocacy, language instruction, teacher education

Procedia PDF Downloads 115

Authors: Eleni Karantzola, Athanasios Karasimos, Vasiliki Makri, Ioanna Skouvara

Abstract:

Historical linguistics unveils the historical depth of languages and traces variation and change by analyzing linguistic variables over time. This field of linguistics usually deals with a closed data set that can only be expanded by the (re)discovery of previously unknown manuscripts or editions. In some cases, it is possible to use (almost) the entire closed corpus of a language for research, as is the case with the Thesaurus Linguae Graecae digital library for Ancient Greek, which contains most of the extant ancient Greek literature. However, concerning ‘dynamic’ periods when the production and circulation of texts in printed as well as manuscript form have not been fully mapped, representative samples and corpora of texts are needed. Such material and tools are utterly lacking for Early Modern Greek (16th-18th c.). In this study, the principles of the creation of EMoGReC, a pilot representative corpus of Early Modern Greek (16th-18th c.) are presented. Its design follows the fundamental principles of historical corpora. The selection of texts aims to create a representative and balanced corpus that gives insight into diachronic, diatopic and diaphasic variation. The pilot sample includes data derived from fully machine-readable vernacular texts, which belong to 4-5 different textual genres and come from different geographical areas. We develop a hierarchical linguistic annotation scheme, further customized to fit the characteristics of our text corpus. Regarding variables and their variants, we use as a point of departure the bundle of twenty-four features (or categories of features) for prose demotic texts of the 16th c. Tags are introduced bearing the variants [+old/archaic] or [+novel/vernacular]. On the other hand, further phenomena that are underway (cf. The Cambridge Grammar of Medieval and Early Modern Greek) are selected for tagging. The annotated texts are enriched with metalinguistic and sociolinguistic metadata to provide a testbed for the development of the first comprehensive set of tools for the Greek language of that period. Based on a relational management system with interconnection of data, annotations, and their metadata, the EMoGReC database aspires to join a state-of-the-art technological ecosystem for the research of observed language variation and change using advanced computational approaches.

Keywords: early modern Greek, variation and change, representative corpus, diachronic variables.

Procedia PDF Downloads 65

Authors: Adelina Castelo

Abstract:

The increasing number of learners of Portuguese as Foreign Language in China justifies the need to define the phonetic profile of these learners and to design didactic materials that are adjusted to their specific problems in pronunciation. Different aspects of this topic have been studied, but the production of oral vowels still needs to be investigated. This study aims: (i) to identify the problems the Chinese learners of Portuguese experience in the pronunciation of oral vowels; (ii) to discuss the didactic implications drawn from those problems. The participants were eight native speakers of Mandarin Chinese that had been learning Portuguese in College for almost a year. They named pictured objects and their oral productions were recorded and phonetically transcribed. The selection of the objects to name took into account some linguistic variables (e.g. stress pattern, syllable structure, presence of the Portuguese oral vowels in different word positions according to stress location). The results are analysed in two ways: the impact of linguistic variables on the success rate in the vowels' production; the replacement strategies used in the non-target productions. Both analyses show that the Chinese learners of Portuguese (i) have significantly more difficulties with the mid vowels as well as the high central vowel and (ii) do not master the vowel height feature. These findings contribute to define the phonetic profile of these learners in terms of oral vowel production. Besides, they have important didactic implications for the pronunciation teaching to these specific learners. Those implications are discussed and exemplified.

Keywords: Chinese learners, learners’ phonetic profile, linguistic variables, Portuguese as foreign language, production data, pronunciation teaching, oral vowels

Procedia PDF Downloads 221

Authors: Kimberly M. Cacciato, Steven J. Singer, Allison R. Shapiro, Julianna F. Kamenakis

Abstract:

This study examines the dynamics of communication among researchers and participants who have various levels of hearing, use multiple languages, have various disabilities, and who come from different social strata. This qualitative methodological study focuses on the strategies employed in an ethnographic research study examining the communication choices of six sets of parents who have Deaf-Disabled children. The participating families varied in their communication strategies and preferences including the use of American Sign Language (ASL), visual-gestural communication, multiple spoken languages, and pidgin forms of each of these. The research team consisted of two undergraduate students proficient in ASL and a Deaf principal investigator (PI) who uses ASL and speech as his main modes of communication. A third Hard-of-Hearing undergraduate student fluent in ASL served as an objective facilitator of the data analysis. The team created reflexive journals by audio recording, free writing, and responding to team-generated prompts. They discussed interactions between the members of the research team, their evolving relationships, and various social and linguistic power differentials. The researchers reflected on communication during data collection, their experiences with one another, and their experiences with the participating families. Reflexive journals totaled over 150 pages. The outside research assistant reviewed the journals and developed follow up open-ended questions and prods to further enrich the data. The PI and outside research assistant used NVivo qualitative research software to conduct open inductive coding of the data. They chunked the data individually into broad categories through multiple readings and recognized recurring concepts. They compared their categories, discussed them, and decided which they would develop. The researchers continued to read, reduce, and define the categories until they were able to develop themes from the data. The research team found that the various communication backgrounds and skills present greatly influenced the dynamics between the members of the research team and with the participants of the study. Specifically, the following themes emerged: (1) students as communication facilitators and interpreters as barriers to natural interaction, (2) varied language use simultaneously complicated and enriched data collection, and (3) ASL proficiency and professional position resulted in a social hierarchy among researchers and participants. In the discussion, the researchers reflected on their backgrounds and internal biases of analyzing the data found and how social norms or expectations affected the perceptions of the researchers in writing their journals. Through this study, the research team found that communication and language skills require significant consideration when working with multiple and complex communication modes. The researchers had to continually assess and adjust their data collection methods to meet the communication needs of the team members and participants. In doing so, the researchers aimed to create an accessible research setting that yielded rich data but learned that this often required compromises from one or more of the research constituents.

Keywords: American Sign Language, complex communication, deaf-disabled, methodology

Procedia PDF Downloads 116

Authors: Heejin Song

Abstract:

Employing a critical action research approach that rejects essentialist onto-epistemological orientations to research in English language teaching (ELT) and interrogates the hegemonic relations in the knowledge construction and reconstruction processes, this study illuminates how an action research-informed pedagogical practice can transform the English for academic purposes (EAP) teaching to be more culturally and linguistically inclusive and critically oriented for English language learners’ advancement in academic literacies skills. More specifically, this paper aims to showcase the action research-informed pedagogical innovations that emphasize multilingual learners’ multiliteracies engagement and experiential education-oriented learning to facilitate the development of learners’ academic literacies, intercultural communicative competence, and inclusive global citizenship in the context of Canadian university EAP classrooms. The pedagogical innovations through action research embarked in response to growing discussions surrounding pedagogical possibilities of plurilingualism in ELT and synchronous online teaching. The paper is based on two iterations of action research over the pandemic years between 2020 and 2022. The data includes student work samples, focus group interviews, anonymous surveys, teacher feedback and comments on student work and teaching reflections. The first iteration of the action research focused on the affordances of multimodal expressions in individual learners’ academic endeavors for their literacy skills development through individual online activities such as ‘my language autobiography,’ ‘multimodal expression corner’ and public speeches. While these activities help English language learners enhance their knowledge and skills of English-spoken discourses, these tasks did not necessarily require learners’ team-based collaborative endeavors to complete the assigned tasks. Identifying this area for improvement in the instructional design, the second action research cycle/iteration emphasized collaborative performativity through newly added performance/action-based innovative learning tasks, including ‘situational role-playing’, ‘my cooking show & interview’, and group debates in order to provide learners increased opportunities to communicate with peers who joined the class virtually from different parts of the world and enhance learners’ intercultural competence through various strategic and pragmatic communicative skills to collaboratively achieve their shared goals (i.e., successful completion of the given group tasks). The paper exemplifies instances wherein learners’ unique and diverse linguistic and cultural strengths were amplified, and critical literacies were further developed through learners’ performance-oriented multiliteracies engagement. The study suggests that the action research-informed teaching practice that advocates for collaborative multiliteracies engagement serves to facilitate learners’ activation of their existing linguistic and cultural knowledge and contributes to the development of learners’ academic literacy skills. Importantly, the study illuminates that such action research-informed pedagogical initiatives create an inclusive space for learners to build a strong sense of connectedness as global citizens with increased intercultural awareness in their community of language and cultural practices, and further allow learners to actively participate in the construction of ‘collaborative relations of power’ with their peers.

Keywords: action research, EAP, higher education, multiliteracies

Procedia PDF Downloads 79

Authors: Cynthia Briggs, Kimberly Gerardi

Abstract:

Over the past six years, the authors have been working together on World Language Collaborations in the Middle School French Program at St. Luke's School in New Canaan, Connecticut, USA. Author 2 brings design expertise to the projects, and both teachers have utilized the fabrication lab, emerging technologies, and collaboration with students. Each year, author 1 proposes a project scope, and her students are challenged to design and engineer a signature project. Both partners have improved the iterative process to ensure deeper learning and sustained student inquiry. The projects range from a 1:32 scale model of the Eiffel Tower that was CNC routed to a fully functional jukebox that plays francophone music, lights up, and can hold up to one thousand songs powered by Raspberry Pi. The most recent project is a Fragrance Marketplace, culminating with a pop-up store for the entire community to discover. Each student will learn the history of fragrance and the chemistry behind making essential oils. Students then create a unique brand, marketing strategy, and concept for their signature fragrance. They are further tasked to use the industrial design process (bottling, packaging, and creating a brand name) to finalize their product for the public Marketplace. Sometimes, these dynamic projects require maintenance and updates. For example, our wall-mounted, three-foot francophone clock is constantly changing. The most recent iteration uses Chat GPT to program the Arduino to reconcile the real-time clock shield and keep perfect time as each hour passes. The lights, motors, and sounds from the clock are authentic to each region, represented with laser-cut embellishments. Inspired by Michel Parmigiani, the history of Swiss watch-making, and the precision of time instruments, we aim for perfection with each passing minute. The authors aim to share exemplary work that is possible with students of all ages. We implemented the reverse engineering process to focus on student outcomes to refine our collaborative process. The products that our students create are prime examples of how the design engineering process is applicable across disciplines. The authors firmly believe that the past and present of World cultures inspire innovation.

Keywords: collaboration, design thinking, emerging technologies, world language

Procedia PDF Downloads 43

Authors: Nkereke Mfon Essien

Abstract:

In an ambisyllabic environment, one consonant sound simultaneously assumes both the coda and onset positions of a word due to its structural proclivity to affect two phonological processes or repair two ill-formed sequences in those syllable positions at the same time. This study sets out to examine the structural conditions that trigger this not-so-common phonological privilege for consonant sounds in the English language and Nigerian English and if such constraints could have any correspondence in the language studied. Data for the study were obtained from a native speaker of English who was the control and twenty (20) educated Nigerian speakers of English from the three ethnic/linguistic groups in Nigeria. Preliminary findings from the data show that ambisyllabicity in English is triggered mainly by stress, a condition which causes a consonant in a stressed syllable to become glottalised and simultaneously devoices the nearest voiced consonant in the next syllable. For example, in the word coupler,/'kʌplɜr/ is realized as ['kʌˀpl̥ɜr]. In some Nigerian English, preliminary findings show that ambisyllabicity is triggered by a sequence of intervocalic short, high central vowels and a coda nasal. Since the short vowel may not occur in an open syllable, the nasal serves to close the impermissible open syllable. However, since the Nigerian English foot structure does not permit a CVC.V syllable, the same coda nasal simultaneously repairs the impermissible syllable foot to (CV.CV) by applying the Maximal Onset Principle since this is a preliminary investigation, a conclusion would not suffice yet.

Keywords: ambisyllabicity, nasal, coda, stress, phonological process, syllable, foot

Procedia PDF Downloads 17

Authors: Alireza Behfar, Mohammad Mahdavi

Abstract:

Nowadays, in the light of progress in the world of science, technology and communications, mastery of learning international languages is a sure and needful matter. In learning any language as a second language, progress and achieving a desirable level in speaking is indeed important for approximately all learners. In this research, we find out how preparation can influence L2 learners' oral fluency with respect to individual differences in working memory capacity. The participants consisted of sixty-one advanced L2 learners including MA students of TEFL at Isfahan University as well as instructors teaching English at Sadr Institute in Isfahan. The data collection consisted of two phases: A working memory test (reading span test) and a picture description task, with a one-month interval between the two tasks. Speaking was elicited through speech generation task in which the individuals were asked to discuss four topics emerging in two pairs. The two pairs included one simple and one complex topic and was accompanied by planning time and without any planning time respectively. Each topic was accompanied by several relevant pictures. L2 fluency was assessed based on preparation. The data were then analyzed in terms of the number of syllables, the number of silent pauses, and the mean length of pauses produced per minute. The study offers implications for strategies to improve learners’ both fluency and working memory.

Keywords: two factors, fluency, working memory capacity, preparation, L2 speech production reading span test picture description

Procedia PDF Downloads 228

Authors: Sean W. T. Bayly, Daniel Glover, Don Horrell, Simon Horrocks, Barnes Callum, Stuart Gibson, Mac Misuira

Abstract:

We assess suitability of recent, ∼7B parameter, instruction-tuned Language Models Mistral-v0.3, Llama-3, and Phi-3, for Generative Named Entity Recognition (GNER). Our proposed Multi-Level Information Retrieval method achieves notable improvements over finetuned entity-level and sentence-level methods. We consider recent developments at the cross roads of prompt engineering and Retrieval Augmented Generation (RAG), such as EmotionPrompt. We conclude that language models directed toward this task are highly capable when distinguishing between positive classes (precision). However, smaller models seem to struggle to find all entities (recall). Poorly defined classes such as ”Miscellaneous” exhibit substantial declines in performance, likely due to the ambiguity it introduces to the prompt. This is partially resolved through a self verification method using engineered prompts containing knowledge of the stricter class definitions, particularly in areas where their boundaries are in danger of overlapping, such as the conflation between the location ”Britain” and the nationality ”British”. Finally, we explore correlations between model performance on the GNER task with performance on relevant academic benchmarks.

Keywords: generative named entity recognition, information retrieval, lightweight artificial intelligence, prompt engineering, personal information identification, retrieval augmented generation, self verification

Procedia PDF Downloads 44

Authors: Delphine Vendryes, Dushyanth Sekhar, Baojia Tong, Matthew Theisen, Chester Curme

Abstract:

Much of the data that inform the decisions of governments, corporations and individuals are harvested from unstructured documents. Data extraction is defined here as a process that turns non-machine-readable information into a machine-readable format that can be stored, for instance, in a database. In financial services, introducing more automation in data extraction pipelines is a major challenge. Information sought by financial data consumers is often buried within vast bodies of unstructured documents, which have historically required thorough manual extraction. Automated solutions provide faster access to non-machine-readable datasets, in a context where untimely information quickly becomes irrelevant. Data quality standards cannot be compromised, so automation requires high data integrity. This multifaceted task is broken down into smaller steps: ingestion, table parsing (detection and structure recognition), text analysis (entity detection and disambiguation), schema-based record extraction, user feedback incorporation. Selected intermediary steps are phrased as machine learning problems. Solutions leveraging cutting-edge approaches from the fields of computer vision (e.g. table detection) and natural language processing (e.g. entity detection and disambiguation) are proposed.

Keywords: computer vision, entity recognition, finance, information retrieval, machine learning, natural language processing

Procedia PDF Downloads 109