turkishenglish.com - Comparison of relative clauses in ESL students



            Having reviewed much of the literature related to the acquisition of relative clauses by L2 learners, it is appropriate to make explicit why more research is warranted and in particular what the goals and limitations of the present study are.  To do this, I will first discuss the issue of transfer with respect to the problem of relative clause acquisition, and then I will outline some of the areas ignored by previous research.  Finally, I will present four hypotheses.


The Question of Transfer

            The question of whether the avoidance evidenced by the Chinese and Japanese students is due to the influence of their native language is hardly at issue in the literature, nor should it be since language background is the one parameter that all of the students with low production rates seem to share.  The question is what sort of influence does the native language have on the acquisition of a second language?  Is it a case of processing difficulty due to the difference in syntactic configuration, or is the functional load of RCs in Japanese and Chinese so different from that of the function of RCs in English that the learners have a difficult time adjusting to the new discourse level functions of RCs?  If it is a case of different processing styles, then we would expect all native speakers of languages with prenominal RCs to have this difficulty when learning a language with postnominal RCs.  If it is a case of discourse transfer (or interference), then we would expect to see at least some negative transfer on this level, and some areas where the L2 learner has not mastered the function of RCs in English, resulting in gaps in their usage.  There is, however, one more possibility which is that syntax, particularly constituent order, affects discourse function in such a way that these two are inextricably bound together (cf. Hwang 1990).  For example, in Korean, subsequent events are not placed in an RC because its prenominal position disrupts the iconicity of events.  It is unclear at this point how much this ordering of constituents influences the function of a particular form, especially cross-linguistically, but it is a possibility.  There are, then, three possible explanations for the low production: syntactic processing (word order difference), discourse function, or a combination of both (i.e. the form-function relationship).


Justification for Present Study

            Although previous research has turned over a number of stones, it seems that many were searched in an attempt to answer different questions than the one that is being asked in this study, and some very obvious possibilities have been left unexplored.

            Several studies have looked at the grammatical functions of both the Head Noun and the Relative NP (e.g. Gass 1980), but these studies were devoted almost entirely to questions related to surface form errors, such as pronominal retention and/or in the testing of how the NP Accessibility Hierarchy or the Markedness Differential Hypothesis is related to RC acquisition.  Gass (1980) makes a general comment about the tendency of Romance students to use a wider range of relative clause types than students from non-Romance languages, stating that they tended to relativize on more positions lower on the NP Accessibility Hierarchy than the other students.  But, Arabic and Persian are by no means Romance languages, so there is no explanation for Schacter’s avoidance hiding in her statement.  Gass does, however, make an important contribution, in that she does a test to determine if there is any correlation between difficulty with positions on the AH and the production of relative clauses.  There was no significant correlation, nor was there a correlation between the number of surface form errors and production (i.e. students who had difficulty with the surface structure of the English RC did not show any inclination towards producing fewer of them for this reason).  The implication is, of course, that syntactic processing is probably not to blame for the low production of RCs by the Chinese and Japanese students. 

            Among the studies of RC acquisition, there are no studies that give detailed descriptions of the distribution of RCs with respect to grammatical function in the IL of L2 learners.  However, if studies like Fox and Thompson (1990) and De Haan (1987) are correct, and if there are correlations between the type of RC and its function on a discourse level, then this is just the sort of thing that ought to be investigated.  If students were using RCs in different ways this should be reflected in the distribution of the grammatical relations of the Head Noun and the coreferential NP in the RC.  No studies have addressed this issue of information flow in discourse.  Do learners from different languages use the same type of relatives with the same distributions?   

            Add to this the findings of the few discourse level contrastive analyses that have been done (Collier-Sanuki 1991, Hwang 1990, and Zhao 1989), and we begin to see the importance of a more detailed investigation of the types of RCs students are using.  These studies indicate that the informational status of the RC is of crucial concern when considering the question of form/function transfer.  For example, in Chinese asserted new information cannot be put in an RC, but is rather encoded as an independent clause.  This difference in form/function relationship may be a key in understanding how L2 learners are using RCs, and it has not been fully investigated.   

            The few contrastive analyses that have been done may prove extremely useful and may hold the key to understanding the avoidance phenomenon reported by Schacter, but a simple contrastive analysis does not provide an explanation.  These studies need to be followed up with detailed examinations of how the learners are actually using RCs in L2 production, specifically what discourse functions these constructions perform.  These studies have not been done, yet when they are done, they should be able to test the claims of those who favor a discourse explanation for avoidance by showing deviant RCs, that is, RCs that function in a way that English RCs cannot, and that correspond to typical RCs in the NL.  In lieu of this, it should at least be shown that certain functions of the RC in English are being avoided.  If the distribution of RCs used by L2 learners with prenominal RCs in their NL mirrors the distribution of RCs used by native speakers except with lower frequency, then the explanation provided by the proponents of discourse transfer becomes extremely vulnerable.   If those who favor a discourse explanation want to prove that Schacter's data is not the result of avoidance, but rather is the result of discourse level transfer, then nothing short of a detailed description of L2 data will suffice to substantiate their claims.   Additionally, without this type of study, it is difficult to judge the merits of the syntactic processing theory. 

            In addition to the above points, no studies have been done to demonstrate what sort of correlation exists between TL proficiency and RC production.  Do learners significantly improve in their ability to use RCs in a variety of ways as their proficiency improves or is this an area in which the students' competency seems to stagnate?  An answer to this question might help classroom teachers develop pedagogy which specifically addresses the question of RC production and targets problem areas. 




Goals and Limitations

            L2 learners whose native language was either Korean or Spanish were chosen for this study.  There were several factors that went into selecting these two groups.  First, I wanted to expand the original study done by Schacter by including learners whose native language differed from those in her group.  I wanted to see if speakers from other languages still evidenced the disparity in number of relative clauses produced.   By broadening the scope of the original investigation, I will provide more data with which to determine if syntactic processing is still a valid explanation for Schacter’s avoidance.  Another reason for choosing these two groups was the availability of data.  This choice gave me access to the largest group of students which would allow me to expand the original study and still maintain a fairly significant database.  

            One drawback to this choice is that I know of no detailed discourse level contrastive analyses of either of these languages.  As a matter of fact,  I do not know of any discourse level contrastive analyses between English RCs and any language with postnominal RCs. This is probably because learners with this type of language background are not generally perceived as having much difficulty with this construction.  This does not, however, affect the present study too adversely since

the goal of this study is not to prove transfer because there is virtually no one who questions the influence of the L1 with respect to this problem.  The goal of this study is description.  I want to see if it is possible to characterize the IL competence of these learners in a way that might shed some light on what they perceive as the functions of English RCs.  Though the goal is not to prove transfer, if two groups can be shown to have significant systematic differences in usage, then this is tantamount to demonstrating L1 influence, even though the source and nature of the influence will not be apparent. 

            Another limitation of this study is that I will not be able to make any firm statements about how well the two groups of learners approximate native usage of the RC because no detailed work has been done on how native speakers use RCs in expository text.  However, the large amount of work that has been done on relative clauses in general should allow for the drawing of some tentative conclusions. 


Hypotheses : There are four general hypotheses that will be tested by this study. 

            (1)  Because of the prenominal position of relative clauses in the NL, the Korean subjects will produce fewer relative clauses than the Spanish students.  The Spanish production will be similar to that of English because their NL has postnominal relative clauses.

            (2)  Production and distribution of relative clauses will become greater and more varied for both groups as TL proficiency increases.

            (3)  There will be clear differences in the pattern of usage and the distribution of types of relative clauses between the two groups.

            (4)  Because of the relationship between information flow and discourse function and due to the SOV word order of Korean,  these students will have more instances of negative transfer on the discourse level than the Spanish students.

Next: Methods and Procedures