Font Size: a A A

A Multifactorial Analysis And Computational Cognitive Investigation Of Chinese Ba Sentences

Posted on:2020-10-08Degree:DoctorType:Dissertation
Country:ChinaCandidate:Y FangFull Text:PDF
GTID:1365330596473591Subject:Foreign Language and Literature
Abstract/Summary:PDF Full Text Request
Ba sentences,as an important syntactic structure in modern Chinese,have triggered heated discussions in linguistic studies.Most of previous studies focused on constiutents,grammatical meanings as well as construction mechanisms of ba sentences,while few explored factors that affect syntactic choice and constructions of ba sentences and their synonymous sentences.In this study,ba sentences that can be converted into SVO sentences were extracted from five genres of CCL corpus.Using a multifactorial analysis and computational cognitive methods,we investigated the following three questions:(1)what factors affect syntactic choice of ba sentences and SVO sentences?(2)what are roles of dependency distance and surprisal in constructing ba sentences and SVO sentences?(3)what factors can affect longdistance dependencies in ba sentences? We were also concerned about the impact of genres during this process.In the first question,we focused on six lexical and syntactic factors,i.e.,definiteness,animacy,length,syntactic parallelism,givenness and pronominality,as well as two indicators from computational cognitive science,namely,dependency distance and surprisal.Surprisal was divided into N-gram surprisal and DG surprisal according to the estimation either based on N-gram models or probabilistic dependency grammars.The results showed that speakers are more likely to choose ba sentences instead of SVO sentences when objects are known,when lengths of objects are short and when there is no syntactic parallelism between sentences.Dependency distance and N-gram surprisal can also affect syntactic choice: speakers prefer sentences with shorter dependency distances and smaller N-gram surprisals.Genres have no impact on syntactic choice.As for the second research question,the results showed that dependency distances interact with N-gram surprisals in ba sentences and the converted SVO sentences,and the results were consistent among genres.In news and practical writings,interactions were also found between dependency distances and DG surprisals in SVO sentences and the converted ba sentences,but in other cases,interations exist only when dependency distances of ba sentences are beyond a certain value.No difference was found between naturally occurring sentences and the converted sentences concerning both dependency distances and surprisals.In ba sentences and the converted ba sentences,there is an interaction between dependency distances and DG surprisals,but there is no interaction between dependency distances and N-gram surprisals.In SVO sentences and the converted SVO sentences,there is no interaction between dependency distances and surprisals.In the last research question,we focused on relations between long-distance dependencies in ba sentences and the following factors: givenness,dependency relations,word frequencies,as well as main syntactic components and word classes leading to increased dependency distances.Long-distance dependencies may occur between ba verbs and subjects or between ba verbs and their syntactically related verbs.The results showed that givenness of subjects and dependency types affect dependency distances,thus may reduce cognitive loads caused by long-distance dependencies.Givenness of objects also affects dependency distances,but in an opposite direction to the former two factors.Word frequencies do not affect dependency distances.Increased dependency distances between subjects and ba verbs are mainly due to the increase of object lengths and lengths of word classes other than nouns and verbs.Words except nouns and verbs may reduce cognitive loads because they contain no new discourse referents.Our findings showed that syntactic choice and constructions of ba sentences and SVO sentences are influenced by various factors.We can understand relations among those factors better through the multifactorial analysis.In addition,we used two indicators from computational cognitive science,i.e.,dependency distance and surprisal,and the results verified their functions in sentence constructions.In short,language is not an autonomous cognitive faculty,so we can carry out linguistic research from cognitive perspectives in order to get more interesting results.
Keywords/Search Tags:ba sentences, SVO sentences, computational cognitive science, multifactorial analysis
PDF Full Text Request
Related items