The Garden-Path Theory of Sentence Processing.

이승아 한국영어학회 2006 영어학 Vol.6 No.3

Lee, Seung-Ah. 2006. The Garden-Path Theory of Sentence Processing. Korean Journal of English Language and Linguistics 6-3, 615-636. Sentences like The horse raced past the barn fell are temporarily ambiguous between a main clause and reduced relative clause analysis. This phenomenon is known as being 'led down the garden-path'. In psycholinguistics, garden-path sentences are used to reveal the extent to which sentence processing is distinct or modular. According to the garden-path theory of sentence processing (e.g., Frazier 1978), syntactic ambiguity is initially resolved on the basis of purely structural principles. There is empirical evidence both for and against the garden-path theory. Yet, capitalising on the fact that the garden-path theorists' assumptions about phrase structure are problematic, this paper takes the position that the garden-path theory, which is a modular model, cannot be upheld.

Grammaticality Judgment of Garden Path Sentences in Persian

Hamideh Marefat,Ali Arabmofrad 서울대학교 인지과학연구소 2008 Journal of Cognitive Science Vol.9 No.1

Garden path sentences have been shown to predict difficulty in language processing, but no study has been conducted on Persian language. This study tested the prediction for Persian ambiguous sentences having two DPs in which the second DP functions as either the modifier of the first DP or the subject/object of the next clause. No significant effects were obtained for grammaticality judgment times but results of the analysis of accuracy of judgments for garden path sentences from 41 Persian native speakers by the use of RSVP technique revealed that there is a garden path effect in such sentences in Persian, and the distance of the critical word from the verb as the disambiguating word affects the comprehension of Persian garden path sentences. Moreover, the results showed that the location of the disambiguating word has an influential impact on processing garden path sentences. The results are discussed in the framework of the late closure principle.

Aging Effects and Working Memory in Garden-Path Recovery

Hyunsoo Yoo,Michael Walsh Dickey 한국언어재활사협회 2017 Clinical Archives of Communication Disorders Vol.2 No.2

Purpose: The relationship between working memory (WM) and sentence comprehension (SC) has drawn many researchers’ attention. The current study examined whether healthy older adults have difficulty comprehending potentially WM-demanding garden-path sentences involving syntactic ambiguities, and whether their comprehension is predicted by their working memory capacity or inhibitory control. Methods: Older adults (n=35, ages 60–89) and younger adults (n=36, ages 19–33) completed two self-paced reading experiments and a battery of cognitive measures (working memory and inhibition tasks). Participants read the sentences containing a Minimal Attachment (MA) ambiguity in Experiment 1 and a Late Closure (LC) ambiguity in Experiment 2. Results: The older adults’ garden-path effect was larger than younger adults’ in on-line measures. However, older adults exhibited higher off-line acceptability judgments for garden-path sentences than younger adults, for both LC and MA sentences. Working memory predicted off-line performance of both younger and older groups in Experiment 2 (LC). Conclusions: The results showed age-related differences in the comprehension of sentences with temporary syntactic ambiguities. Successful recovery from garden paths (revealed by off-line measures) was predicted by working memory. However, the source of age-related differences in real-time processing remains unclear. These results indicate that healthy aging can affect comprehension of challenging, syntactically ambiguous material.

The Effect of Present Activity Verbs on Processing Structural Ambiguity in Japanese Garden-Path Sentences

( Yoshie Yamamori ) 서울대학교 인지과학연구소 2016 Journal of Cognitive Science Vol.17 No.3

This paper addresses the semantics of the present form (known as the -ru form) of activity verbs in Japanese and examines the effect of these verbs in contrast to that of the inflected form (the ta form). Garden-path sentences involving an ambiguity between a simple sentential reading and a relative clause reading generally show a preference for the former reading; when the preferred reading proves to be inconsistent with the correct reading of the sentence, the ensuing processing difficulty is known as the ‘garden-path effect.’ Interestingly, it has been observed that the effect is reduced in sentences that contain activity verbs in the present -ru form in the adnominal clause. One major problem that arises in the interpretation of the -ru form in subordinate clauses is that it is temporally ambiguous, and may be interpreted as belonging to the matrix clause instead. To date, no uniform analysis has been developed to characterize the semantic nature of the -ru form in subordinate clauses. This study will develop a semantics of the -ru form. The data revealed some interesting findings suggesting some logical characteristics of the -ru form. The results help to clarify how the semantic nature of the -ru form exerts an effect on the processing of garden path sentences, and to show that, contrary to what had been assumed in the absence of semantic research into the embedding verb, the activity verbs ending in the -ru form extend the worlds against which a given proposition is evaluated, which has the effect of reducing the GP effect.

한국인 영어 학습자의 오도 문장에 대한 이해와 읽기 과제의 영향

임정현(Lim, Jung Hyun) 한국언어학회 2021 언어 Vol.46 No.2

The purpose of the present study is to examine how L2 learners interpret locally ambiguous ‘garden-path’ sentences, and whether and in what way different reading tasks affect the comprehension of such garden-path sentences. Using an offline experiment, participants were given two types of reading conditions. In one condition, they were asked to read sentences and answer comprehension questions, while in the other , they were told that they might have to perform grammatical judgments on some of the sentences at random, before then answering comprehension questions. The rationale of differentiating between the tasks was to manipulate reading goals . Results showed that participants responded incorrectly more often to the garden-path sentences than when not in the garden-path conditions . This indicated that L2 learners were quite likely to arrive at a final interpretation while lacking full syntactic analysis when reading garden-path sentences, which is consistent with previous studies. More importantly, reading tasks yielded different patterns of sentence comprehension. Specifically, the accuracy of comprehension was higher when the reading goal was to randomly make grammatical judgments on parts of the sentence than when the goal was solely to answer comprehension questions. Results were discussed in terms of ‘good-enough (GE)’ representations in L2 language comprehension.

The Garden-Path Theory of Sentence Processing

Seung-Ah Lee(이승아) 한국영어학회 2006 영어학 Vol.6 No.3

정원길(garden path) 문장 처리에 관한 GPT-2 신경망 언어 모델과 인간의 비교 연구

김유영(Yoo Young Kim),박명관(Myung-Kwan Park) 한국중원언어학회 2022 언어학연구 Vol.- No.64

안구운동 추적을 이용한 한글 오인문장 처리 과정의 이해

이윤형 한국자료분석학회 2009 Journal of the Korean Data Analysis Society Vol.11 No.6

Readers experience difficulty in understanding when reading a garden-path sentence since incoming phrase does not fit the interpretation of the sentence developed up to that point. The purpose of this study was to investigate the resolution processes of such difficulty in Korean garden-path sentences. Using eye-tracking during reading, the syntactic ambiguity resolution processes of the garden-path sentences were compared with the processes of the various control sentences. Various eye-tracking measures reflecting the initial processes showed that Koreans attach an incoming syntactic elements into the already developed interpretation. However, eye-tracking measures for the reanalysis processes showed that Koreans try to make full and accurate representation of the sentences instead of the just good enough comprehension suggested by Ferreira, Christianson and Hollingworth(2001). This study also showed that eye-tracking method can be effectively used to see the detailed on-line processing mechanism for the Korean sentence comprehension. 본 연구에서는 한국어 문장 정보처리 과정을 살펴보기 위하여 안구운동의 추적을 이용하여 한글 오인문장을 읽고 이해하는 과정을 살펴보았다. 대학생들을 대상으로 문장구조의 재분석이 필요한 오인문장과 재분석이 필요하지 않은 문장들을 읽을 때 나타나는 안구운동의 양상을 비교한 결과 문장이해 과정 중 초기 처리를 반영하는 안구운동의 측정치들에서는 오인문장의 경우와 그렇지 않은 조건들 간에 차이가 없었으나 문장의 복잡성을 이해하고 처리하는 것을 반영하는 측정치들과 문장정보의 재분석 과정을 반영하는 측정치들의 경우에는 오인문장의 경우에 더 오랜 응시시간과 응시빈도를 보였다. 이것은 한국어 오인문장의 이해시 영어의 경우와 유사하게 일단 부분적인 정보들을 이용하여 문장의 구조를 만들어 나가지만 영어와는 달리 적당히 만족할 만한 수준까지만 해석하지 않고 문장을 정확하게 재분석하려고 한다는 것을 의미한다. 본 연구를 통해 안구운동의 측정이 한국어 문장의 이해 과정을 살펴보는데 효과적으로 사용될 수 있다는 것을 보여 주었다.

Garden-Path Sentence Processing of English Textbook LSTM

김유영 현대영미어문학회 2023 현대영미어문학 Vol.41 No.3

This paper aims to evaluate the garden-path sentence processing of LSTM trained on L2 English textbooks published for Korean English learners. It was conducted in order to identify which frequency information these textbooks need to be strengthened or improved. The predicted effects from L2 (textbook)-LSTM was compared to L1 (wiki)-LSTM and reading times of L1 native speakers. The present paper found that English textbooks contain frequency information that affects native speakers, as evident from the garden-path effects displayed by L2-LSTM. However, there were some differences, such as the lack of co-occurrence frequencies that are prominent in native data and native speakers.

Syntactic reanalysis and lingering misinterpretations in L2 sentence processing

Jonghyeon Lee,Jeong-Ah Shin 경희대학교 언어정보연구소 2016 언어연구 Vol.33 No.S

Readers often misinterpret sentences and reanalyze them, but their initial misinterpretations sometimes linger even after they finish reading the sentences. This lingering effect has been argued as evidence that comprehenders create interpretations to some extent that it seems “good enough” for them to understand the meaning (Christianson, Hollingworth, Halliwell, and Ferreira 2001). This study examined L2 learners’ processing of garden path sentences, investigating whether they show a lingering effect as L1 speakers do by employing the gender mismatch paradigm (Sturt 2003). In the self-paced reading experiment, participants read garden path sentences such as While Mary washed the boxer who was tall coughed and warmed herself with a blanket. If participants correctly rule out an initial misinterpretation from the temporary ambiguity, the reflexive herself would create a gender mismatch effect (i.e., reading times are slowed in the gender mismatch condition). After reading the sentences, participants were asked to answer follow-up comprehension questions such as Did Mary wash the boxer? The results showed that L2 learners were able to notice the temporary local ambiguity in a sentence and reanalyze the structure as L1 readers do. However, the rate of accurate responses for comprehension questions was significantly lower in the ambiguous condition than in the unambiguous condition, indicating that the L2 learners’ initial misinterpretations lingered. L2 learners do not seem to lack detailed syntactic representation, and their lingering misinterpretation could be explained by the good enough approach as for L1 readers. Participants eventually recovered from the garden path, showing the gender mismatch effect, but their global comprehension was significantly influenced by the initial misinterpretation, since they gave more incorrect answers when they read garden path sentences.