2.2 Variables in Acquisition of Relative Clauses by ESL/EFL Learners
2.2.1.2 Previous Empirical Studies on L1 Interference
In the light of these cross-linguistic discrepancies in RC structure and function,
one would not be surprised to find that considerable SLA research on English RCs has
been devoted to investigating the influence of L1 interference on EFL/ESL learners’
RC acquisition. These studies are in and of themselves the exponents of the
Contrastive Analysis Hypothesis (CAH), predominant in the field of SLA. The CAH
claims that the principal barrier to second language acquisition comes from
interference of an L2 learner’s mother tongue with a second language system being
acquired, especially when these two structurally differ to a great extent, and that a
systematic analysis between the two languages in question can yield a taxonomy of
linguistic contrasts which enables one to predict or explain difficulties an L2 learner
will encounter (Ellis, 1996: 23-27; Brown, 2000: 207-210). Emanating from such a
theoretical backdrop, many researchers (e.g. Schachter, 1974; Schachter et al., 1976;
Bley-Vroman & Houng, 1988, cited in Kamimoto et al., 1992; Zhao, 1989, cited in
Kamimoto et al., 1992; Li, 1996; Wei, 1997; Gisborne, 2000; Yin, 2001; Chan, 2004a,
b) have strongly asserted that L1 transfer is the overriding and sole factor that
determines the success of RC acquisition by ESL/EFL learners. Their findings are
discussed in great detail below.
To begin with, Schachter et al. (1976) attested L1 influence on L2 learners’
inter-language by eliciting grammaticality judgments on English sentences containing
RCs from 100 high-intermediate and advanced ESL students of five language
backgrounds─Chinese, Japanese, Arabic, Persian, and Spanish. In this task, the
subjects were presented with both native English RCs and malformed (non-native)
English RCs which corresponded to those RCs in the subjects’ L1. The results
indicated that although the subjects identified sentences with native RCs as
grammatical, they also tended to identify as grammatical sentences with their L1 form
of non-native RCs. For example, the Chinese group tended to identify as
grammatical those English RCs lacking relative pronouns in the subject position.
Schachter explains this phenomenon as the consequence of L1 transfer, citing the
non-occurrence of English relative pronouns in Chinese RCs.
Positive evidence of L1 interference is also provided by Wei’s 1997 study, in
which 131 Taiwanese EFL learners─roughly representing beginning, intermediate,
and advanced levels ─ were tested on English RCs in comprehension and
sentence-combining tasks. She found that contradictory to Flanigan’s findings (1994)
that the influence of L1 background was minimal on relativization, her subjects,
especially those of lower proficiency, were still occupied by their previous linguistic
repertoire, their RC errors indeed exhibiting L1 interference─including the use of
resumptive pronouns and lack of relative pronouns, both of which are typical of
Chinese RCs─ as well as resembling those errors commonly committed by native
speakers of English. Given the existence of negative L1 transfer, she suggests that
L2 learners acquiring English RCs cannot completely reset a new parameter value
(e.g. from head-final in Chinese to head-initial in English) without being influenced
by their L1.
Further landing support for L1 interference are Gisborne (2000), Yin (2001) and
Chan (2004a, b). Drawing on the Hong Kong database for the International Corpus
of English, Gisborne and Yin examined RC variation phenomena in Hong Kong
English (HKE) and identified some idiosyncratic features of RCs in HKE. Among
them, zero subject relatives (i.e. RCs lacking a relative pronoun in the subject position)
and blurring of the restrictive/non-restrictive contrast were claimed to be the
outcomes of Chinese’ influencing RC formation in HKE. In a similar vein, Chan
conducted a contrastive analysis of noun phrases in English and Chinese with 387
tertiary and secondary students of ESL in Hong Kong. Based on a corpus of
authentic data collected in a research project, she concluded that most English RC
errors committed by her subjects resulted largely from L1 transfer, such as
inappropriate relative pronouns (due to the absence of relative pronouns in Chinese),
missing relatives (due to the serial verb construction19 in Chinese), resumptive
19 The serial verb construction is typical of Chinese, referring to two or more verb phrases or clauses
pronouns, (due to this requirement in indirect and prepositional object RCs in
Chinese), and head-last RCs (especially among lower-proficiency learners, who tend
to adopt a word-for-word translation strategy), as illustrated in the following:
(58) *She is my mother which is the most important person in my life.
(59) *You are the first person came to Hong Kong.
(60) *There is one thing which I can remember it very clearly.
(61) *I wear the dress is very cute.
L1 interference manifests itself not only in L2 learners’ knee-jerk error making,
as revealed by the aforesaid studies, but also in their behavior of avoidance.
Kleinmann (1977) has argued for avoidance of a given structure as indicating areas of
difficulty, predictable on the basis of a contrastive analysis of the target and native
languages. In her classic paper on this issue, Schachter (1974) examined English
free compositions written by both native Americans and 50 intermediate and
advanced ESL learners, whose L1s were Chinese, Japanese, Arabic, and Persian (no
detail given for how this task was administered). She found that while the Arabic
and Persian learners produced as many RCs as their native American counterparts, the
Chinese and Japanese learners produced far fewer, though obtaining a significantly
lower error rate than the former two language groups. Schachter attributes this
avoidance phenomenon or strategy to L1 interference: because the structural
differences between Chinese/Japanese (head-final) and English RCs are greater than
those between Arabic/Persian (head-initial) and English RCs, it would be more
difficult for Chinese and Japanese learners to acquire English RCs, and the increased
learning difficulty would in turn lead to their tendency to avoid using English RCs in
their writing, especially when they feel unsure of getting the target structure right.
Based on the premise that L1 interference operates on the discourse as well as
the syntactic levels, other researchers, such as Bley-Vroman and Houng (1988, cited
in Kamimoto et al., 1992), Zhao (1989, cited in Kamimoto et al., 1992) and Li (1996),
have attempted to put forward alternative explanations for the underproduction of
English RCs by the Chinese subjects in Schachter’s (1974) study. They all contend
that the low frequency rate of RCs in the English of Chinese learners results not so
much from structural interference of L1 with L2 as from pragmatic transfer of the RC
frequency, distribution, and function patterns in L1 to L2─namely, L2 learners may
know the grammatical structure of RCs, but may not know when to use it in English.
To challenge Schachter’s (1974) purely structural view of RCs, Bley-Vroman
and Houng (1988, cited in Kamimoto et al., 1992) set out to examine relative
frequencies of English and Chinese RCs by comparing the first five chapters of the
American literary work The Great Catsby and its published Chinese translation.
They found that only one-third (32/93) of the original English RCs were translated
into the Chinese version, as shown in Table 2:
Table 2: Frequency of RCs in the first five chapters of The Great Catsby and the number of those clauses translated as RCs in the Chinese version (adapted from Bley-Vroman & Houng, 1988: 96, cited in Kamimoto et al., 1992)
RC types in The Great Catsby Rendered as RCs in English Chinese translation*
Restrictive 50 21 (42%) Non-restrictive 43 11 (25%) Total 93 32 (34%)
*Figures in this column indicate only the number of English RCs of both types rendered as RCs rather than other structures in Chinese. There is no formal distinction between restrictive and non-restrictive RCs in Chinese.
Based on this lower RC density in Chinese, Bley-Vroman and Houng reason that
Chinese learners are likely to under-produce RCs in their English, inasmuch as they
do not use RCs frequently in their own native language and may transfer the
preference for non-relative structures (e.g. independent clauses or adverbial clauses)
in Chinese over to English for contexts in which RCs are normally used to achieve the
same communicative purposes.
By examining both English writings on impressions of China by Chinese
Americans and Canadians and their Chinese translations, Zhao (1989, cited in
Kamimoto et al., 1992) also concludes that Chinese indeed makes less use of RCs
than does English, as indicated in Table 3:
Table 3: RCs in English text and their Chinese translation (adapted from Zhao, 1989: 107, cited in Kamimoto et al., 1992)
RCs in English RCs in Chinese RCs in English in English in Chinese =RCs in Chinese only only 124 91 59 (48%) 65 32
More importantly, her study revealed that there are two types of English RCs which
have no RC equivalents in Chinese due to their special functions not performed by
Chinese RCs, namely, adding parenthetical assertions and focusing information (e.g.
presentative RCs with extraposition or existential there), as previously noted in
Section 2.2.1.1. Such functional differences between English and Chinese RCs,
Zhao claims, not only provide hard counterevidence for the implicit anglocentric view
in Schachter’s (1974) study that the way RCs function in English also holds true for
those in other languages, but also offer explanations for the relatively low production
rate of English RCs by her Chinese subjects: the subjects may have transferred the
limited range of RC functions in Chinese to English, thereby employing RCs merely
for identification or characterization in their English writing.
Further refining Schachter’s (1974) avoidance theory, Li (1996) propounds a
differentiation between conscious avoidance, as portrayed by Schachter, and
subconscious underproduction, a situation where a learner under-produces a certain
structure mainly because he/she lacks a full understanding of the common contexts for
using it (i.e. he/she may not be aware of all the functions it can serve). Li’s study
involved 11 Chinese ESL learners of Hong Kong (intermediate and advanced) in
definition questions (e.g. What is a clock?), Chinese-to-English translations (some of
which were adapted from Zhao’s examples of English RCs with special
pragmatic/discourse functions not served by Chinese RCs), and individual
retrospection interviews. In line with Schachter’s findings, her Chinese subjects
were indeed observed to employ RCs with low frequency. Nonetheless, in their
individual interviews, most of her subjects denied having deliberately tried to avoid
using English RCs because of the perceived gross structural differences between
English and Chinese RCs. Moreover, her subjects were quite successful in
producing all the English RCs in the translation test, except for those that served to
add parenthetical information or to focus information (i.e. to present a topical
referent). Based on the results, Li contends that Chinese learners’ sporadic use of
RCs in their English writing may be explicable in terms of their failure to discern the
subtle pragmatic differences between English and Chinese in RC function. That is,
under the influence of L1 transfer, Chinese learners often use English RCs as a pure
noun modifier, as they do with Chinese RCs, and tend to lose sight of the additional
functions RCs can serve in English; as a result, they subconsciously produce fewer
RCs in their English writing.
It is worth pointing out that the foregoing RC studies on L1 interference may be
tainted by two weaknesses in their research methodology. One weakness concerns
the nature of elicitation tasks. Most of the studies above employed elicitation tasks
that imposed a high degree of control over L2 learners’ output production, including
grammaticality judgments, comprehension questions, sentence-combining, and
translation. As a result, learners’ true competence in the target structure may not
have been shown. The translation task, in particular, is more problematic in that the
extent of L1 interference may have been aggravated by directly inviting the use of L1
in producing English RCs. Moreover, even though free writing, a more spontaneous
task, was adopted by Schachter (1974), the task suffers from lack of control as well as
lack of methodological detail in the data collection procedure (i.e. how she collected
her writing samples is left unknown). Her use of this unconstrained elicitation
format certainly begs the question as regards whether her subjects may have truly
been forced to reveal facets of their inter-language under investigation (Corder, 1973).
This is so because Schachter gave little consideration for the nature and topic of her
writing task (no detail given for what topic was chosen), not placing limitations on her
subjects (e.g. requiring them to follow specific instructions or pictorial prompts) that
would prompt the use of English RCs in their compositions.
The other weakness relates to the failure to consider the effect of L1 interference
observed in relation to L2 proficiency. In these previous studies, with the exception
of Wei (1997), little is said about the role of L2 proficiency in their subsequent data
analysis. Instead, they tend to make a general conclusion without making allowance
for the very fact that L2 proficiency may well play a part in intervening and
determining the degree to which L2 learners’ RC performance is influenced by L1
transfer.