Search

This study explores a GPT-2 architecture's capacity to capture monolingual children's comprehension behaviour in Korean, a language underexplored in this context. We examine its performance in processing a suffixal passive construction involving verbal morphology and the interpretive procedures driven by that morphology. Through model fine-tuning via patching and hyperparameter variations, we assess their classification accuracy on test items used in Shin (2022a). Results show discrepancies in simulating children's response patterns, highlighting the limitations of neural networks in capturing child language features. This prompts further investigation into computational models' capacity to elucidate developmental trajectories of child language that have been unveiled through corpus-based or experimental research.

The present study investigates how heritage speakers conduct ‘good-enough' processing at the interface of home-language proficiency, cognitive skills, and task types. For this purpose, we employ two word-order patterns of two clausal constructions in Korean (suffixal passive; morphological causative) which differ in the mapping between thematic roles and case-marking and the interpretive procedures driven by verbal morphology. We find that, while Korean heritage speakers demonstrate the same kind of acceptability-rating behaviour as monolingual Korean speakers do, their reading-time patterns are notably modulated by construction-specific properties, cognitive skills, and proficiency. This suggests a heritage speaker's ability and willingness to conduct both parsing routes, induced by linguistic cues in a non-dominant language, which are proportional to the computational complexity involving these cues.

The present study investigates the role of three structural factors ("word order," "case-marking," and "verbal morphology") in the comprehension of the Korean suffixal passive by Korean-speaking children. To measure the relative impact of each factor on the comprehension of the passive, we devise a novel method where these factors are obscured systematically by employing acoustic masking -- chewing and coughing. Results from three picture selection tasks show that (1) the Agent-First preference is not ubiquitously employed by young children, but is dependent on other grammatical cues; (2) despite mediocre accuracy rates on passive sentences, we nonetheless find evidence that children have some knowledge about the passive construction; and (3) scrambling of passive sentences does not aid in comprehension. These findings suggest that, in order to understand the acquisition of the passive, all these structural factors need to be considered.

This study investigates how neural networks address the properties of children's linguistic knowledge, with a focus on the "Agent-First" strategy in comprehension of an active transitive construction in Korean. We develop various neural-network models and measure their classification performance on the test stimuli used in a behavioural experiment involving scrambling and omission of sentential components at varying degrees. Results show that, despite some compatibility of these models' performance with the children's response patterns, their performance does not fully approximate the children's utilisation of this strategy, demonstrating by-model and by-condition asymmetries. This study's findings suggest that neural networks can utilise information about formal co-occurrences to access the intended message to a certain degree, but the outcome of this process may be substantially different from how a child (as a developing processor) engages in comprehension. This implies some limits of neural networks on revealing the developmental trajectories of child language.

This study investigates how neural networks reveal developmental trajectories of child language, focusing on the Agent-First strategy in comprehension of an active transitive construction in Korean. We develop three models (LSTM; BERT; GPT-2) and measure their classification performance on the test stimuli used in Shin (2021) involving scrambling and omission of constructional components at varying degrees. Results show that, despite some compatibility of these models’ performance with the children’s response patterns, their performance does not fully approximate the children’s utilisation of this strategy, demonstrating by-model and by-condition asymmetries. This study’s findings suggest that neural networks can utilise information about formal co-occurrences to access the intended message to a certain degree, but the outcome of this process may be substantially different from how a child (as a developing processor) engages in comprehension. This implies some limits of neural networks on revealing the developmental trajectories of child language.

This study investigates how L2 learners achieve the ‘good-enough’ comprehension in Korean. We focus on a suffixal passive construction, given the scarcity of this construction in the L2 textbook input. Results from acceptability judgement and self-paced reading tasks suggest two aspects of L2 comprehension. First, L1 and L2 comprehension do not qualitatively differ regarding ‘good-enough’ processing: the L2 processor utilises both heuristic and algorithmic parsing to reduce the burden of work at hand. Second, the divergence of L1 and L2 processing behaviours during comprehension may originate from various factors around L2 learners (e.g., L2 input, L1–L2 interface, task types), which are assumed to anchor the noisier representations of L2 knowledge.

Studies on the role of input in L2 acquisition often estimate L2 input properties through L1 corpora and focus on L2-English. This study probes the initial stage of L2-Korean learning for adult English-speaking beginners of Korean to investigate input-output relations in the acquisition of L2 that is typologically different from English in a more direct manner. We specifically ask how L2 beginner input affects L2 beginner production with respect to Korean postpositions. For this purpose, we investigate how the beginners receive input regarding Korean postpositions from a textbook and to what extent the input characteristics are manifested in learner writing. We found that, whereas the presentation of certain postpositions in the textbook was generally reflected in learner writing, individual postpositions showed disparity in their use between the textbook and the writing. Implications of the findings are discussed in light of L1-L2 differences and how the textbook presents form-function pairings of these postpositions.

We explore how morphological cue occurrence and frequency-based heuristics interplay during sentence comprehension in Korean, a lesser-studied language in this respect. Two self-paced reading experiments with a suffixal passive construction (verb-final vs. verb-initial) and a morphological causative construction (verb-initial in comparison to the same word-order pattern in the suffixal passive) revealed that the heuristics (canonicity of word order; typicality of form-function parings involving case-marking) affected processing behaviours more strongly than the expected advantage of an early-arriving morphological cue in comprehension. Our findings support the heuristic-before-algorithm processing architecture, which is driven by the general property of human cognition that continuously seeks to reduce the burden of work at hand at the earliest opportunities. This appeals to the online cognitive equilibrium hypothesis that argues for the processor’s propensity to enter and remain in the state of cognitive equilibrium as early and long as possible during processing.

It has long been believed across languages that the "Agent-First" strategy, a comprehension heuristic that maps the first noun onto the agent role, is a general cognitive bias which applies automatically and faithfully to children's comprehension. The present study asks how this strategy interplays with such grammatical cues as the number of overt arguments and the presence of case-marking in Korean, an SOV language with case-marking by dedicated markers. To investigate whether and how these cues affect the operation of this strategy, we measure children's comprehension of a transitive construction (with scrambling and omission of sentential components) in a novel experimental setting where arguments and case markers were obscured to varying degrees through acoustic masking. We find that children do not demonstrate the agent-first interpretation strongly in the noun-verb pattern without case-marking, showing their uncertainty about the thematic role of the nominal when it is both the only argument in the sentence and lacks case-marking. They perform significantly better in the patterns with additional cues, the impact of which is asymmetric by age and by the nature of alignment between cues from word order and case-marking. These findings suggest that, for Korean-speaking children's comprehension of a transitive construction, the "Agent-First" strategy is activated properly only in conjunction with other types of interpretive cues.

Attraction effects arise when a comprehender erroneously retrieves a distractor instead of a target item during memory retrieval operations. In Korean, considerable processing difficulties occur in the agreement relation checking between a subject and an honorific-marked predicate when an intervening distractor carries a non-honorific feature. We investigate how attraction effects are managed during the processing of Korean subject-predicate honorific agreement by two Korean-speaking groups with different language use experience backgrounds: college students and airline workers. Results showed that both groups demonstrated stable knowledge of the honorific agreement in the acceptability judgment task. In the self-paced reading task, the airline group, who used honorifics extensively in their workplace, was less affected by the attraction effect than the student group. Our findings suggest that long-term language use experience can modulate how language users manage potential influence from attraction effects in real-time sentence processing.

This study employs a computational method, together with qualitative analysis, to explore the metapragmatics of politeness lexemes that both scholars and laypersons fail to provide effective articulations for. We specifically choose two bewildering politeness lexemes – kyemson ('modesty') and kongson ('politeness') – in Korean, and use the computational approach to identify their metapragmatic differences by analyzing over 27,000 texts that feature their usage. Results show that, while kyemson is associated with speakers' personal and interpersonal qualities, kongson is manifested in a range of normative semiotic practices. They both feature the concepts of lowering oneself and elevating others, but realize them at different levels. In addition, kyemson helps to achieve relational harmony regardless of social hierarchies, and kongson reinforces social hierarchies but speaks little of relational harmony. The findings make two important contributions – namely, providing a novel methodology to address the pitfall in analysing first-order and second-order politeness, and offering an insight into the underexplored politeness lexemes in Korean. [ABSTRACT FROM AUTHOR]

This study investigated the effects of construction types on Korean-L1 English-L2 learners' verb-construction integration in online processing by presenting the ditransitive and prepositional dative constructions and manipulating the verb's association strength within these constructions. Results of a self-paced reading experiment showed that the L2 group spent longer times in the verb-construction integration in the postverbal complement region when processing the ditransitive construction, which is less canonical and highly avoided in the learners' L1, than when processing the prepositional dative construction, which is more canonical and shares similar structural features with the L1 counterpart. In the following spillover region, L2 learners showed faster reading times as proficiency increased when the verb was strongly associated with the prepositional dative construction. Our findings expand the scope of current models on L2 sentence processing by suggesting that construction types and L2 proficiency may affect the L2 integration of verbal and constructional information.

While verbs and argument structure constructions are essential for deriving sentence meaning, their roles in sentence processing remains less known. To address this issue, the present study explored how a verb’s lexical properties and the strength of verb–construction associations influence second language (L2) sentence processing. In two self-paced reading experiments, Korean-speaking learners of English and native English speakers read English argument structure constructions containing verbs with varying lexical properties and association strength. In both Experiment 1 (involving the prepositional dative construction) and Experiment 2 (involving the caused-motion construction), only the L2-English learners, and not the native speakers, spent more time integrating the target constructions with verbs of weaker associations than those with stronger associations. Furthermore, the learners exhibited greater difficulty with verb–construction integration when the verbs were less frequent and less concrete; L2 proficiency did not modulate these effects. Our findings support the constructionist approach, suggesting that both verbs and verb–construction associations play an instrumental role in sentence comprehension. [ABSTRACT FROM AUTHOR]

The 'good-enough' processing account argues that, given the parallel activation of two parsing routes—algorithmic and heuristic parsing, the processor prefers heuristics over algorithms when unfolding incoming input. Literature on L2 'good-enough' processing conjoins with this argument, also claiming that various factors may modulate how the L2 processor adjusts its way to heuristic or algorithmic parsing. The present study investigates how L2 learners with contrastive L1 backgrounds (Czech; English) achieve 'good-enough' comprehension in Korean, a popular L2 target but understudied for this topic. We focus on morphological causative and suffixal passive constructions, which differ in terms of the alignment between thematic roles and case-marking and the interpretive computation that verbal morphology invites. Participants joined acceptability judgement and self-paced reading tasks, with manipulation of word order (verb-final vs. verb-initial). Results from these tasks suggest two aspects of L2 comprehension. First, L1 and L2 comprehension do not qualitatively differ regarding 'good-enough' processing: the L2 processor utilises both parsing routes to reduce the burden of work at hand at the earliest opportunity. Second, the divergence of L1 and L2 processing behaviours during comprehension may originate from various factors surrounding L2 learners (e.g., L2 usage, L1–L2 interface, task types), anchoring the noisy representations of L2 knowledge. [ABSTRACT FROM AUTHOR]

This study investigates how L2 learners of Korean in foreign language-learning environments manifest the target knowledge as a function of various factors surrounding the learners (e.g., L2 textbook input, language-specific properties, general language-use experience in class), specifically concerning the learners' written production of the postposition–verb pair in a locative postposition–verb construction. For this purpose, we analyze two textbook types and learner writing from two L1 groups, focusing on postposition/verb use, Type–Token Ratio, and keyness. We find some meaningful relationships, but also inconsistencies, between L2 textbook input and L2 written production regarding this construction. We discuss implications of the findings with respect to how L2 knowledge is shaped at the interface of L2-textbook characteristics, language-specific properties involving the target knowledge, learners' L1, and general language-use experience in class. Based on the results, we also discuss pedagogical implications, with emphasis on (methods and technologies of) utilizing the existing corpora for instructional purposes. [ABSTRACT FROM AUTHOR]

Across languages, a passive construction is known to manifest a misalignment between the typical order of event composition (agent-before-theme) and the actual order of arguments in the constructions (theme-before-agent), dubbed non-isomorphic mapping. This study investigates comprehension of a suffixal passive construction in Korean by Mandarin-speaking learners of Korean, focusing on isomorphism and language-specific devices in the passive. We measured learners' judgment of the acceptability of canonical and scrambled suffixal passives as well as their reaction times (relative to a canonical active transitive). Our analysis generated three major findings. First, learners uniformly preferred the canonical passive to the scrambled passive. Second, as proficiency increased, the judgment gap between the canonical active transitive and the canonical suffixal passive narrowed, but the gap between the canonical active transitive and the scrambled suffixal passive did not. Third, learners (and even native speakers) spent more time in judging the acceptability of the canonical suffixal passive than they did in the other two construction types. Implications of these findings are discussed with respect to the mapping nature involving a passive voice, indicated by language-specific devices (i.e., case-marking and verbal morphology dedicated to Korean passives), in L2 acquisition. [ABSTRACT FROM AUTHOR]

We investigate how Alor Malay–Abui child bilinguals, a non-WEIRD and underrepresented population in the field, comprehend and produce reflexive and non-reflexive possessive constructions in Abui. Their profiles of language development are explored through functionalist accounts that highlight the close connection between language and domain-general learning capacities, with special emphasis on three models: Competition Model, usage-based approach, and processing determinism.A total of 17 children (experimental group) and 17 adults (comparison group) complete a forced-choice sentence comprehension task (where they listen to two sentences following a short video prompt and verbally select the sentence that best describes the clip) and an elicited production task (where they watch a series of video clips and describe what is happening in each clip in a semi-spontaneous way).This study employs mixed-effects modelling for data analysis. The global model per experiment includes Condition (reflexive; non-reflexive) and Group (child; adult) as fixed effects and Participant and Item as random effects, with the maximal random-effects structure allowed by the model.Results show asymmetric performance by bilingual children on the two construction types across both experiments. Our findings provide insight into the developing processor in bilingual children, influenced by various factors such as usage frequency, cognitive skills, and task effects, within the functionalist framework.The findings of this study align with functionalist accounts for language development, illuminating the intricate deployment of linguistic knowledge in the minds of bilingual children interfacing with language-usage experience and domain-general factors. By combining the three models of functionalism, our findings are more properly understood, highlighting their complementary nature in addressing developmental trajectories of linguistic knowledge for child bilinguals. Importantly, our study shifts our attention to a unique bilingual context that has not been explored – lesser-studied languages and learning environments, contributing to offsetting the sampling bias persistent in the field. [ABSTRACT FROM AUTHOR]

We investigate Korean-speaking children's knowledge about clause-level constructions involving a transitive event – active transitive and suffixal passive – through corpus analysis and Bayesian modelling. The analysis of Korean caregiver input and children's production in CHILDES revealed that the rates of constructional patterns produced by the children mirrored those uttered by the caregivers to a considerable degree and that the caregivers' use of case-marking was skewed towards single form-function pairings (despite the multiple form-function associations that the markers manifest). Based on these characteristics, we modelled a Bayesian learner by employing construction-based input (without considering lexical information). This simulation revealed the dominance of several constructional patterns, occupying most of the input, and their inhibitory effects on the development of the other patterns. Our findings illuminate how children shape clause-level constructional knowledge in Korean, an understudied language for this topic, as a function of input properties and domain-general learning capacities, appealing to the usage-based constructionist approach. [ABSTRACT FROM AUTHOR]

The present study investigates how heritage speakers conduct good-enough processing at the interface of home-language proficiency, cognitive skills (inhibitory control; working memory), and task types (acceptability judgement; self-paced reading). For this purpose, we employ two word-order patterns (verb-final vs. verb-initial) of two clausal constructions in Korean-suffixal passive and morphological causative-which contrast pertaining to the mapping between thematic roles and case-marking and the interpretive procedures driven by verbal morphology. We find that, while Korean heritage speakers demonstrate the same kind of acceptability-rating behaviour as monolingual Korean speakers do, their reading-time patterns are notably modulated by construction-specific properties, cognitive skills, and proficiency. This suggests a heritage speaker's ability and willingness to conduct both parsing routes, induced by linguistic cues in a non-dominant language, which are proportional to the computational complexity involving these cues. Implications of this study are expected to advance our understanding of a learner's mind for underrepresented languages and populations in the field., Competing Interests: The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest., (Copyright © 2024 Shin.)