To sum up, one thing that English and Chinese relative clauses have in common is that they are both adjacent to the head NP. However, for the most parts, they are quite different. First, while English relative clauses are head- initial, Chinese ones are head-final. Second, use of English relative pronouns is optional, but use of Chinese relative marker is mandatory. Third, English has a number of relative
pronoun choices whereas Chinese invariably makes use of “De” as the relative marker.
The positions English can relativize include subject, direct object, indirect object, object of preposition, genitive, and object of comparative. On the other hand, Chinese can relativize the positions that English does, but without object of
comparative. The similarities and differences between English and Chinese relative clauses are summarized in table 2.5.
Table 2.5 Relative clause patterns in English and Chinese, adapted from Gass (1980)
English Chinese
Table 2.5 (continued)
Relative clause patterns in English and Chinese, adapted from Gass (1980)
English Chinese
2.4 Previous studies on the acquisition of relative clauses
Quite a number of research has been devoted to seeing whether there is a
universal order of acquisition of relative clauses. The issue that has been frequently explored is whether the universal markedness relationship as predicted by NPAH is adhered to in second language learners’ interlanguage (Gass, 1979, 1980; Pavesi, 1986; Hyltenstam, 1987, 1990). At the same time, transfer issue is taken into account in the investigation of linguistic universals as native language is deemed a significant influencing factor (Odlin, 1989). On the other hand, psycholinguistic factors are also considered in determining the order of acquisition (Ioup and Kruse, 1977; Schumann, 1980).
In this section, literature regarding the acquisition of relative clauses will be subdivided into two subsections: the first subsection concerns the acquisition of
relative clauses in English and non-Chinese languages and the other concerns the acquisition of Chinese relative clauses.
2.4.1 Acquisition of relative clauses in English and in non-Chinese languages
In the attempt to better determine the relationship between transfer and universal factors in the second language acquisition, Gass (1979) investigated the acquisition of English relative clauses by adult second language learners, who had a variety of linguistic backgrounds. The nine native languages of the learners were Arabic, Chinese, French, Italian, Korean, Persian, Portuguese, Japanese, and Thai. Data from 17 high- intermediate and advanced learners were elicited with two elicitation tasks: grammaticality judgment and sentence combination. 12 types of relative clauses were involved. Note that the indirect object and object of preposition were then collapsed as under one category. Gass claims that language transfer is evident in the acceptance or rejection of the resumptive pronouns in the English relative clauses. The groups whose native languages have resumptive pronouns are more likely to accept ungrammatical English sentences that have resumptive pronouns.
The tendency is evident in the three highest positions on the NPAH, but not in the lowest two positions: genitive and object of comparative. The author argues that we cannot yet determine whether the acceptance of resumptive pronoun in the genitive and object of comparative is due to the influence of native languages or language universal for lack of conclusive evidence. In addition, universal orders as predicted by NPAH were found to be accorded to, the easiest position to relativize being subject and the most difficult position object to comparative. However, genitive was the exception. Gass provided two possible explanations for this. First, genitive is
uniquely coded for case/grammatical relation in English, and there are no variants such as that or which. Second, it is possible that learners treat whose son in (9) as a unit, the subject of the verb came. In doing so, learners might perceive genitive like subject, the highest position on the NPAH, thus making genitive easier to relativize.
(9) The man whose son just came home
In conclusion, it was claimed that language universals played the leading role since they were dominant both in assigning relative orders of difficulty and in determining where language transfer occurs.
Figure 2.1
Percentage of sentences correct on combining task (all group)(from Gass, 1979)
In 1980, Gass conducted a similar study. However, as the 1979 study was based only on the accuracy rate with which learners acquired relative clauses, the 1980 one looked also at the frequency of the relative clauses learners produced.
Gass compared the number of the relative clauses used by given subjects with the same subjects’ performance on the combining task. It was hypothesized that if the number of the relative clauses was related to difficulty, then it would positively
correlate with actual difficulty on the combining task. The correlation was however not significant. To further investigate what learners were actually avoiding in producing relative clauses, Gass examined larner ’s relative clauses in combining task and found four types of avoidance strategies learners used as in (10) (From Gass, 1980):
(10)
1. Substitution of one lexical item for another.
Example: The woman danced. The man is fatter than the woman.
→The woman that is thinner than the man danced. (Persain and Chinese) 2. Switching the order of the two sentences so as to embed the sentence which
was intended as the matrix.
Example: He saw the cat. The dog jumped on the cat.
→The dog jumped on the cat that he saw. (Arabic) 3. Changing the identical NP.
Example: He saw the woman. The man is older than the woman.
→He saw the man who is older than the woman. (French) 4. Changing the syntactic structure of the second sentence.
Example: He saw the woman. The man kissed the woman.
→He saw the woman who was kissed by the man. (Arabic)
There are other studies that have similar results to Gass’s concerning the
grammatical judgment of resumptive pronoun. Hyltenstam (1987, 1990) found that his subjects who learned Swedish (which, like English, does not manifest resumptive pronoun) as second language tended to use resumptive pronoun in their interlanguage, no matter whether their native languages have resupmtive pronoun or not. For example, although resumptive pronoun does not occur in Finnish and Spanish,
learners in these groups used resumptive pronoun in oral elicitation task. The pattern
is also dependant on the first language structure, as there are more pronoun retentions in the Swedish of those learners whose L1 uses this strategy more frequently. In particular, the universal hierarchy as predicted by NPAH was generally adhered to, except the genitive and object of comparative and indirect object and object of preposition. It was argued that the order between OCOMP and GEN and that between IO and OBL should be inverted to reflect a psychologically real acquisition order rather than the typological one.
The acquisition order obtained by Hyltenstam was further supported by Pavesi (1986). Pavesi (1986) set out to investigate the possible influence learning context could have in the acquisition of English relative clauses. For the learning context, Pavesi means the different type of discourses: formal and informal, which learners were mostly exposed to. Two groups of speakers of Italian were involved: 48 EFL Italian high school students, who received formal English input through instruction;
and 38 Italian workers in Edinburgh, who had only informal English input from daily contact with English speakers. It was presumed that the acquisition order as predicted by NPAH would be yielded by both groups and that the formal groups would exhibit more marked structures than the informal groups. The elicitation technique used was oral task with pictorial cues as in the study by Hyltenstam, in which subjects were expected to orally produce the intended relative clauses when an interviewer asked questions with illustrated pictures. For example, when the interviewer asked who is No.6? the expected answer was No. 6 is the man who is running. A “+” would be assigned when a correct relative clause was produced 80 percent of the time, otherwise a “- ” would be assigned. Two major conclusions were reached. First, the results generally came out as assumed: both formal and informal groups conformed to the implicational order on the NPAH. Second, more formal learners mastered target- like relativization. However, the pattern for
indirect object and object of preposition and also for genitive and object and comparative was not always as expected as on the NPAH. That is, the learners tended to invert the order of indirect object and object of preposition, and genitive and object of comparative.
Tarallo and Myhill (1983) investigated acquisition of relative clauses in Chinese, Portuguese, German, Japanese, and Persian by native speakers of English who were asked to judge the grammaticality of relative clauses. The findings were three- fold.
First, the use of resumptive pronouns in interlanguage was a universal phenomenon.
This generalization was made because while English is a language that does not allow resumptive pronoun, it was found that learners’ accepting resumptive pronouns in L2 languages was a common practice. Second, learners generally followed the order in the NPAH, except for object of preposition and indirect object. The order of
difficulty was SU/DO>OPREP>IO>GEN (> means is more easier than). Third, there was evidence for transfer effect. For example, German and Chinese do not have preposition in indirect object relativization, but English does. Learners of German and Chinese rejected the relativization of indirect object more often than did that of subject and direct object, showing that learners of these two languages were transferring the feature of English to German and Chinese.
Ioup and Kruse (1977) tasted three hypotheses, PDH (Kuno, 1974), NPAH and parallel function hypothesis (PFH) (Sheldon, 1973) to see whether the native language interference or language universals best describe second language acquisition of
English relative clauses. Learners of divergent linguistic backgrounds were involved:
Chinese, Japanese, Arabic, Persian, and Spanish. They were to judge grammaticality of 36 written sentences which were comprised of OS, OO, SS, and SO types of
relative clauses. Recall that PDH predicts the order of difficulty of OS, OO>SS, SO,
more difficult than right embedding; and that NPAH predicts the order of acquisition of OS, SS> OO, SO, on the basis that relative clauses on subjects will be easier than those formed on objects. The PFH Sheldon proposed maintains that relative clauses which are most easily to acquire are those in which the relative pronoun has the same function as the head noun, thus having the difficulty order of SS, OO>SO, OS. In answering the research question, Ioup and Kruse analyzed the variables of language group, sentence type, and the combination of the two, and found that contrary to the CA hypothesis, sentence type rather than native language background is the most reliable predictor of error. Besides, the data showed that Kuno’s hypothesis: PDH received the most support, establishing that center-embedding as the parameter of perceptual complexity is the most significant variable in relative clause processing.
Table 2.6 The order of acquisition in the three hypotheses Hypothesis Order of acquisition/difficulty
PDH OS, OO>SS, SO
NPAH OS, SS> OO, SO
PFH SS, OO>SO, OS
Schumann (1980) analyzed relative clauses in the production data of
international subjects who lived in the U.S. to determine the frequency with which the various relative clause types were used. An analysis of the frequency with which relative pronouns are present or absent in both obligatory and optional contexts was also made. One subject named Jorge was a 13 year-old Colombian boy who
attended an all English school where he had only very minimal instruction in English as a second language. He mostly produced OS and OO types (with the OS
predominating) and relatively few SO and SS. As for relative pronoun, Jorge always supplied a relative pronoun where required and supplied it over 60 percent of the time
where it was optional. 10- year-old Juan was Jorge ’s younger brother. He had the same kind of exposure to English as Jorge did. The relative clauses he produced were mostly OS and OO with SO and SS occurring less frequently. He also supplied a relative pronoun where required and did so almost 80 percent of the time where it was optional. Marta was a five- year-old girl who attended all- English nursery school in Cambridge. OS and OO were more frequent in her spoken data than were SO and SS, and relative pronouns occurred 80 percent of the time in obligatory contexts but only 21 percent in optional contexts. Cheo, the five-year-old boy who attended an all-English kindergarten in Boston, produced more OS and OO types than SO and SS ones. Relative pronouns were supplied by him 45 percent of the time where required and 12 percent where optional. Alberto was a 33-year-old man who worked in a frame manufacturing factory. While receiving little instruction in English when he was in Costa Rita, he essentially used OS relative clauses and virtually supplied no relative pronouns at all. Angela was 78-year-old woman who had lived in the U.S. since she was 37, and received no English instruction. She used few relative clauses, and when she did, OO and OS were more frequent than SO and SS. Like Alberto, She did not use relative pronouns. Guiseppe was Angela’s 84-year-old husband. His exposure to English was greater than his wife due to work.
The relative clauses he used were more OS and OO types than SO and OO ones. He used more relative pronouns than his wife did, especially in obligatory contexts. In conclusion, equaling low frequency with difficulty and high frequency with ease, it was claimed that OO and OS sentence types were preferred to SS and SO ones, in support of Kuno’s (1974) PDH that center-embedding is the most crucial factor in processing relative clauses. This finding also confirms the research results in Ioup and Kruse’s study (1977). Table 2.7 presents the
Table 2.7 Summary of Schumann’s study
Name Age
Frequency of RC types (>
appears more frequently than)
Use of relative pronoun
Jorge 13 OS, OO>SO, SS 100%
Juan 10 OS, OO>SO, SS 98%
Marta 5 OS, OO>SO, SS 80%
Cheo 5 OS, OO>SO, SS 45%
Alberto 33 OS only Rarely
Angela 78 OS, OO>SO, SS None
Guiseppe 84 OS, OO>SO, SS More frequent
than Alberto
Flanigan (1995) tested the comprehension and production of restrictive relative clauses (SS, SO, OO, and SO). The comprehension task involved asking subjects to identify the referent in the relative clauses and the production task is sentence
combination. The subjects were young elementary students who were speakers of nine languages: Chinese, Indo/Malay, Korean, Icelandic, Arabic, Spanish, Sinhalese, and Hebrew. The data revealed that the learner proficiency and task type
significantly affect the performance, with high proficiency performing better than low proficiency and with performance on the production section worse than on the
comprehension section. The order of accuracy obtained for the four relative clause types were OS>OO>SS>SO. The most important constraint on relative clause interpretation and production was claimed to be center-embedding that interrupt the flow of the matrix sentence interpretation. Besides, an analysis of the use of resumptive pronoun did not show evidence of transfer for lack of correlation.
Izumi (2003) tested three hypotheses of the acquisition of relative clauses: NPAH, PDH, and SOHH. The subjects were 61 ESL learners who spoke different native languages: Chinese, Arabic, French, Japanese, Kazah, Korean, Persian, Polish, Portuguese, Spanish, Thai, and Turkish. Both productive and receptive tasks were
used as measuring instrume nt since it was believed that tasks in different modalities may present different degrees and/or types of difficulty in learners’ language
processing. The author used sentence combination task to examine subjects’
productive ability, and interpretation task and grammaticality task to measure subjects’ comprehension of relative clauses. The relative clause types that were focused were SO, SS, OO, OS, SOPREP and OOPREP. Note that the SDO and SIO were collapsed into SO, and ODO and OIO into OO. This clearly is a weakness that was also indicated by the author, because the exclusion of SG, OG, SOCOMP, and OCOMP makes the study not so comprehensive. Data were analyzed separately under the three measurements, which yield the order of difficulty as in table 2.8. It was found that the overall order of difficulty can be predicted by the matrix positioning, with the matrix object position being easier than the matrix subject position, and within each matrix position, the difficulty order can be predicted, at least partly, by the RC types as specified in the AH. According to these results, the author came to the conclusion that NPAH and PDH, being based on different rationales, can be seen to be complimentary to each other since NPAH is associated to the effect of canonicity within the RC and PDH is related to the notion of processing interruption in the matrix sentence. As to SOHH, the research results did not lend a full support to it, which the author explained by stating that “a simple statistical counting of the number of discontinuities in a sentence, in any case, may be too coarse to predict acquisition difficulty.”
Table 2.8 Order of difficulty of English relative clauses in Izumi’s three tasks task Order of difficulty in Izumi ’s study
Sentence combination test OS>OO>OOPREP>SS>SO>SOPREP Interpretation test OS>OOPREP>OO>SS>SOPREP>SO Grammaticality judgment OS>SS>OOPREP>OO>SO>SOPREP
Note: the order of difficulty by SO Hierarchy is OS>OO/SS> OOPREP/SO>SOPREP
There is some research done in the EFL Taiwan context. Chen, J. J. (2004) investigated production of English relative clauses in composition by 102 university students who majored in English. The results demonstrated the order of acquisition of OS>OO>SS>SO. Chen, J. J. claimed that this order generally followed the predictions made by NPAH and PDH. Chen, C. S. (2004) conducted an analysis of errors of English relative clauses produced by senior high school students in Taiwan.
Data in his study also showed that Kuno ’s PDH was supported. Liu (1999) focused on the issue as to whether years of learning and native language had any effect on learners’ acquisition of English relative clauses. The author claimed that whereas years of learning English played an important role, L1 had little effect.
2.4.2 Summary
As is evident in the summarization table for the studies on English relative clauses (Table 2.10), there are some discrepancies about the order across the studies.
They exist mostly in the order among indirect object and object of preposition, and genitive. To reiterate, Pavesi (1986) found that there was a reverse relationship between indirect object and object of preposition. Besides, instead of object of comparison being the most difficult one to relativize, genitive was the most difficult