Investigation Into Instruments In Chinese EFL Learners’ Relativization

Relative clause (RC), as a universal linguistic phenomenon, has drawnconsiderable attention in both linguistic and psycholinguistic fields due to its uniquesyntactic property and its frequency and usefulness in our everyday use of language.Researchers home and abroad have conducted a large amount of research into theprocessing and acquisition difficulty of RC from various perspectives, proposinghypotheses and doing empirical investigations to test them. However, differentresearches with the help of different instruments yield different results, even counter, onthe difficulty hierarchy of RC acquisition. Although some researchers have pointed outthat this may be partly attributed to different instruments, no research has beenconducted specially on the influence of instruments on different results till now, letalone a profound investigation into the causes for differences from the perspective ofinstruments. Thus, the present study aims to investigate the influences instruments mayhave in the studies of RC acquisition, and probe further into the possible causes fordifferent hierarchies obtained from different instruments. We hope this study coulduncover different characteristics of instruments involved, and could be a reference forstudies hereafter to study language acquisition and acquisition difficulty from theperspective of instruments.The main findings are as follows:The quantitative data shows that different instruments do influence the difficultyhierarchies of SS, OS, SO, OO in a significant way.1) as for the grammaticalityjudgment test, there is no significant difference among the four types of RCs except forOO and OS;2) the difficulty hierarchies of the four types of RCs in the other three testscould be summarized as follows: OS ranks always the highest in all the three tests; SOranks the lowest in both the sentence combination test and the Chinese-Englishtranslation test, while in the English-Chinese translation test the accuracy for it is muchhigher than that for SS; SS ranks the lowest in both the Chinese-English translation testand the English-Chinese translation test, while in the sentence combination test theaccuracy for it is much higher than that for SO and OO; OO ranks relatively higher in both the Chinese-English translation test and the English-Chinese translation test, whilein the sentence combination test it ranks the lowest.Qualitative data, obtained from think-aloud and interview, to a large extent,explains the above inconsistency from two main aspects: instrument features andcognitive processing strategies.The nature of the sentence combination test and the strategies participants adoptedaffect the difficulty hierarchy of RCs. It is found that our participants just gave theexpected relative marker, while taking no other measure to complete a sentence. Thus,errors like nonadjacency and resumptive pronoun occurred in SS, SO, and OO,especially resumptive pronoun in OO. And this is why the accuracy for OS is thehighest while that for OO is the lowest in the sentence combination test.The different RC structures between Chinese and English affect thecomprehension and production of SS and SO in both the Chinese-English translationtest and the English-Chinese translation test, which leads to lower accuracy in SS andSO, especially in SS. Besides, in the Chinese-English translation test, participants tentto use the strategy as “sentence structure parsing” and “head noun identification”,which caused a large quantity of errors such as main clause-relative clause reversal andincorrect modifying of relative clause in SO, thus, the accuracy for SO is the lowest inthe Chinese-English translation test.The design of error distribution affects the difficulty hierarchy of the four types ofRCs. In the grammaticality judgment test, there is no significant difference among thefour types of RCs except for OO and OS. The design of error distribution in each RCtype is the major determinant of the scores obtained for each RC type. That is to say,difficulties set beforehand in the test sentences, which were required to be judged andthen corrected by the participants, may also have a variety of difficulty levels.From the results we can see that instruments do have influences on the difficultyhierarchies of RCs. Different instruments have their own features, which elicit differentcognitive processing strategies; and different cognitive processing strategies in turnproduce their effects on the RC patterns, thus lead to different difficulty hierarchies for the results obtained via these instruments.