• Journal of the Korea Society of Computer and Information
• /
• v.27 no.3
• /
• pp.13-23
• /
• 2022

In this study we leveraged a probing method to evaluate how a pre-trained L2 LSTM language model represents sentences with relative and coordinate clauses. The probing experiment employed adapted models based on the pre-trained L2 language models to trace the syntactic properties of sentence embedding vector representations. The dataset for probing was automatically generated using several templates related to different sentence structures. To classify the syntactic properties of sentences for each probing task, we measured the adaptation effects of the language models using syntactic priming. We performed linear mixed-effects model analyses to analyze the relation between adaptation effects in a complex statistical manner and reveal how the L2 language models represent syntactic features for English sentences. When the L2 language models were compared with the baseline L1 Gulordava language models, the analogous results were found for each probing task. In addition, it was confirmed that the L2 language models contain syntactic features of relative and coordinate clauses hierarchically in the sentence embedding representations.

• Journal of Korean language education
• /
• v.29 no.1
• /
• pp.1-23
• /
• 2018

This paper investigates how adult Mandarin- and English- speaking learners of Korean perceive Korean vowels, with focus on the effect of the first language (L1) and the second language (L2) acoustic relationship, as well as the influence of Korean language experience. For this study, native Mandarin and Canadian English speakers who have learned Korean as a foreign language, as well as a control group of native Korean speakers, participated in two experiments. Experiment 1 was designed to examine acoustic similarities between Korean and English vowels, as well as Korean and Mandarin vowels to predict which Korean vowels are relatively easy, or difficult for L2 learners to perceive. The linear discriminant analysis (Klecka, 1980) based on their L1-L2 acoustic similarity predicted that L2 Mandarin learners would have perceptual difficulty rankings for Korean vowels as follows: (the easiest) /i, a, e/ >> /ɨ, ʌ, o, u/ (most difficult), whereas L2 English learners would have perceptual difficulty rankings for Korean vowels as follows: (the easiest) /i, a, e, ɨ, ʌ/ >> /o, u/ (most difficult). The goal of Experiment 2 was to test how accurately L2 Mandarin and English learners perceive Korean vowels /ɨ, ʌ, o, u/ which are considered to be difficult for L2 learners. The results of a mixed-effects logistic model revealed that English listeners showed higher identification accuracy for Korean vowels than Mandarin listeners, indicating that having a larger L1 vowel inventory than the L2 facilitates L2 vowel perception. However, both groups have the same ranking of Korean vowel perceptual difficulty: ɨ > ʌ > u > o. This finding indicates that adult learners of Korean can perceive the new vowel /ɨ/, which does not exist in their L1, more accurately than the vowel /o/, which is acoustically similar to vowels in their L1, suggesting that L2 learners are more likely to establish additional phonetic categories for new vowels. In terms of the influence of experience with L2, it was found that identification accuracy increases as Korean language experience rises. In other words, the more experienced English and Mandarin learners of Korean are, the more likely they are to have better identification accuracy in Korean vowels than less experienced learners of Korean. Moreover, there is no interaction between L1 background and L2 experience, showing that identification accuracy of Korean vowels is higher as Korean language experience increases regardless of their L1 background. Overall, these findings of the two experiments demonstrated that acoustic similarity between L1 and L2 sounds using the LDA model can partially predict perceptual difficulty in L2 acquisition, indicating that other factors such as perceptual similarity between L1 and L2, the merge of Korean /o/ and /u/ may also influence their Korean vowel perception.

• English Language & Literature Teaching
• /
• v.18 no.1
• /
• pp.115-132
• /
• 2012

In order to confirm ecological validity of D$\ddot{o}$rnyei's second language motivational self, the present study investigated 495 South Korean secondary school students' L2 learning motivation and motivated behavior by using a questionnaire survey. The participants' ideal L2 self, ought-to L2 self, integrativeness, and instrumentality were examined and compared to identify which motivational factor had the most effect on their motivated L2 learning behavior. Among Korean secondary school students, the concept of integrativeness was replaced with the ideal L2 self as a more appropriate concept for understanding Korean secondary school students' L2 learning motivation. As for instrumentality, promotional aspects demonstrated a significant correlation with both the ideal and ought-to L2 selves, whereas preventional aspects showed a relatively high correlation with the ought-to L2 self. The ideal L2 self had the most impact on the motivated behavior, and the ideal L2 self and ought-to L2 self reflected the students' different motivational characteristics. It is suggested that D$\ddot{o}$rnyei's (2005, 2009) L2 motivational self system can be a better psychological model than Gardner's (1985) socio-educational model for understanding Korean secondary school students' L2 learning motivation.

• Korean Journal of English Language and Linguistics
• /
• v.4 no.1
• /
• pp.33-60
• /
• 2004

This study was an attempt to present an appropriate way of interpreting L2 learners' linguistic behavior within Universal Grammar (UG) framework. Based on the Korean EFL adult learners' performance on the Subjacency violation sentences, the study suggested that the EFL learners are able to acquire subtle knowledge of target grammar and their linguistic behavior should be interpreted with the most recent version of UG theory, the Minimalist Program (MP) notion. The MP notion seems more plausible to accommodate incomplete L2 grammar while acknowledging UG-constrained interlanguage which the previous version, Principles and Parameters (P&P) approach, could not explain very well. The study observed no age-effects among the Korean EFL learners in their linguistic competence measured by the performance on the UG-constraint violation sentences. Having suggested that the MP notion can be a more reasonable tool to explain the EFL learners' linguistic behavior, the study introduced comprehensive hypotheses such as Constructionist Model (CM) and the Ontogeny Phylogeny Model (OPM).

• Journal of Korean language education
• /
• v.29 no.3
• /
• pp.25-53
• /
• 2018

This study examined the influence of lexical factors on verbal Eojeol recognition. To meet the goal, forty-five L2 Korean learners and twenty-two Korean native speakers took Eojeol decision tasks measured with the lexical factors such as 'number of strokes', 'number of consonants and vowels', 'number of syllables', 'number of morphemes', 'whole Eojeol frequency', 'root frequency', 'first-syllable-sharing frequency', and 'number of dictionary meanings.' As a result, 'whole Eojeol frequency' was the most effective factor to predict Eojeol recognition reaction time for native speakers and L2 learners, which supports the full-list model. Other lexical factors influencing Eojeol recognition reaction time in L2 learners were different following their proficiency level.

• The Journal of the Convergence on Culture Technology
• /
• v.8 no.5
• /
• pp.489-495
• /
• 2022

It makes difference to LSTM D/L(Deep Learning) results for language model construction as the corpus preprocess changes. An LSTM model was trained with a famouse literaure poems(Ki Hyung-do's work) for training corpus in the study. You get the two wordvector sets for two corpus sets of the original text and eraised word ending text each once D/L training completed. It's been inspected of the similarity/analogy operation results, the positions of the wordvectors in 2D plane and the generated texts by the language models for the two different corpus sets. The suggested words by the silmilarity/analogy operations are changed for the corpus sets but they are related well considering the corpus characteristics as a literature work. The positions of the wordvectors are different for each corpus sets but the words sustained the basic meanings and the generated texts are different for each corpus sets also but they have the taste of the original style. It's supposed that the D/L language model can be a useful tool to enjoy the literature in object and in diverse with the analysis results shown in the study.

• Korean Journal of English Language and Linguistics
• /
• v.3 no.3
• /
• pp.375-410
• /
• 2003

As Snow (1989) and Sternberg (1985) have long argued, learning, and adaptation to the learning environment or classroom context (at the levels of instructional treatment, interventionist focus on form technique, or pedagogic task) is a result of the interaction of context at each of these levels of description with learners' patterns of abilities. In this paper I argue that this is an important area of research for SLA pedagogy, as well as SLA theory development, and I review recent developments in the study of L2 learning conditions; of the abilities contributing to L2 aptitude; and of their interaction with the processes involved in successful classroom learning and practice, and propose a model of ‘multiple aptitudes’ for classroom learning based on these findings.

• Journal of the Korea Society of Computer and Information
• /
• v.25 no.11
• /
• pp.27-40
• /
• 2020

In this paper, we investigate the correlation between the amount of English sentences that Korean English learners (L2ers) are exposed to and their sentence processing patterns by examining what Long Short-Term Memory (LSTM) language models (LMs) can learn about implicit syntactic relationship: that is, the filler-gap dependency. The filler-gap dependency refers to a relationship between a (wh-)filler, which is a wh-phrase like 'what' or 'who' overtly in clause-peripheral position, and its gap in clause-internal position, which is an invisible, empty syntactic position to be filled by the (wh-)filler for proper interpretation. Here to implement L2ers' English learning, we build LSTM LMs that in turn learn a subset of the known restrictions on the filler-gap dependency from English sentences in the L2 corpus that L2ers can potentially encounter in their English learning. Examining LSTM LMs' behaviors on controlled sentences designed with the filler-gap dependency, we show the characteristics of L2ers' sentence processing using the information-theoretic metric of surprisal that quantifies violations of the filler-gap dependency or wh-licensing interaction effects. Furthermore, comparing L2ers' LMs with native speakers' LM in light of processing the filler-gap dependency, we not only note that in their sentence processing both L2ers' LM and native speakers' LM can track abstract syntactic structures involved in the filler-gap dependency, but also show using linear mixed-effects regression models that there exist significant differences between them in processing such a dependency.

• English Language & Literature Teaching
• /
• v.16 no.4
• /
• pp.39-69
• /
• 2010

This study aims to identify how participants in online-incorporated English learning perceive interaction between achievement and factors of learning and personality. Using grounded theory analysis, this study attempts to generate a theoretical model depicting how the factors work with the L2 learners situated in the learning setting. A total of 231 college freshmen participated in online and offline EFL learning programs for the duration of one semester. In addition, all respondents completed a survey questionnaire on their learning experiences. In the investigation of the differences between low- and high-proficiency groups, audio-taped interviews with 20 selected students, 10 from each group, have revealed differences not only in the types of personal and instructional factors, but also, more importantly, in the interrelationship between these factors in each group's learning model. These models effectively explained the statistically significant differences in four questionnaire items, such as online learning and contributions of offline class sections to their L2 achievement. These findings entail L2 practitioners' shared understandings of their students' perspectives of learning in the specific L2 learning context.

• English Language & Literature Teaching
• /
• v.9 no.2
• /
• pp.213-241
• /
• 2003

As an attempt to characterize the English test section of CSAT (College Scholastic Ability Test) and to get some suggestions, this study raised the research questions, as 'What are the main factors that affect students' English test scores in CSAT, and how big influences do they have?' It has been hypothesized that among main factors are the L1 competence, represented by the Korean test scores in CSAT, background knowledge or intelligence, represented by the "total" scores in CSAT, and the two types of L2 knowledge (vocabulary and grammar on one hand and prosody m the other hand), measured by the test devised specially for this study. The individual effect of the L2 vocabulary and grammar (one kind of L2 knowledge) was 70%, that of background knowledge or intelligence 61%, that of the L1 competence 50%, and that of the L2 prosody knowledge (the other kind of L2 knowledge) 32%. According to the stepwise regression, the whole effect of these four factors was 74%. The findings suggest that first, although CSAT is based on the top-down model of comprehension, the bottom-up model of learning should be more emphasized in our English class. Also, since background knowledge or intelligence is the second most influential factor, the top-down model of learning that helps students learn to understand by activating their various schemata must also be very effective.