Effects of Construction-Grammar-based Instruction on the Sentence Production Ability of Korean College Learners of English

Abstract

The present study explores a grammar instruction which is primarily based on construction grammar. Construction grammar views constructions as the basic linguistic units and posits hierarchical networks among constructions (Goldberg, 1995, 2006, 2013). The current research investigated whether and to what extent construction grammar-based instruction would assist Korean college learners of English with their language production, especially in written tasks. In addition, it examined to what extent the factor of network within the construction grammar framework plays a role in developing sentence production ability. In order to examine instructional effects, the students were divided into three groups: networked construction group (CNG), construction-only group (COG), and non-constructional group (NCG). All groups participated in three testing sessions and eight instructional sessions, targeting six core ASCs (intransitive-motion, intransitive-resultative, transitive, ditransitive, caused-motion, and resultative constructions). Their performance was measured using elicited translation tasks, picture description tasks, and guided writing tasks for sentence production ability, together with grammaticality judgment task for the knowledge of ASCs. Major findings of the study are as follows. First, all the instructional treatments were effective for the learners improvements in sentence production in the immediate and delayed posttests. However, the construction-based groups showed more significant improvements in their sentence production than the non-constructional group. The construction-based instruction was proved to be durable over four weeks of the posttest period. Notably, as revealed in the results of the grammaticality judgment task in the two posttests, the construction-based groups were seen to obtain constructional knowledge through the construction-based instruction. Second, there were no network effects on the learners overall sentence production in the immediate posttest. However, there were significant differences between the network-based group and the non-networked groups, particularly in the double object dative and resultative constructions. In the delayed posttest, in general, the networked-based instruction had lasting effects on the learners improvement in sentence production. Lastly, there were salient features in the productivity and accuracy of ASCs. The ratio of no response decreased sharply. The construction-based groups tried to produce sentences using the light verb get in the test items that were left blank by the learners in the pretest. The error types such as misordering, indirect object-omission, and complement-omission, which are associated with verb argument structures and their grammatical realizations, decreased significantly in the construction-based groups. In addition, the use of the preposition-omission and overgenerated be constructions decreased significantly in the construction-based groups. All in all, the study showed that the construction grammar-based instruction can help Korean college learners of English improve their sentence production ability, suggesting a more promising alternative to traditional grammar instruction of sentence structures.