The results of the passage-rewriting test indicate that in addition to years of
learning, another variable exerting influence on the extent to which the subjects
utilized RCs in written discourse to background supportive materials is the tendency
to employ either independent and-clauses or adverbial because-clauses for the same
information as should preferably be packaged with RCs. This tendency in actuality
reflects a common overuse of two clause-chaining strategies adopted by L2 learners:
that is, it is typical of L2 learners to draw heavily on the two basic conjunctions, and
and because, in structuring their stream of thoughts in writing.
The wide use of and in learners’ writing can be identified as a consequence of
transfer of strategies for clause linkage characteristic of spoken English into written
English. In English, the conjunction and is a common grammatical resource for
linking two propositions, inasmuch as it plays a multiplicity of ideational roles in
establishing semantic relationships between clauses; it can supply, for example,
additive, adversative, causal, and temporal meanings, depending on contextual
information (McCarthy, 1991: 48), as shown in (88):
(88) Additive: She’s intelligent and she’s very reliable.
Adversative: I’ve lived here ten years and I’ve never heard of that pub.
Causal: He fell in the river and caught a chill.
Temporal: I got up and made my breakfast.
However, the use frequency of and differs to varying degrees between academic
writing and conversational speech: comparatively, it is overwhelmingly frequent in
the latter, where beyond its usual ideational connective uses, the conjunction further
serves as a discourse marker signaling the speaker’s continuation of a conversational
turn (Celce-Murcia & Larsen-Freeman, 1999: 474-478). Without taking heed to
such register differences, most learners unconsciously carry the high frequency of and
in speech over to writing, making heavy use of the conjunction to indicate
inter-clausal relationships in their essays. This overuse of the conjunction and not
only produces odd-sounding prose with an oral tone, but also renders the flow of
information harder for the reader to follow due to some non-essential idea units not
being properly backgrounded in such subordinate clauses as RCs.
Also arising from modality transfer of clause-combining strategies is the
prevalent use of because by learners. Generally speaking, the conjunction because
occurs much more frequently in spoken English than in written English.; in speech,
apart from the cause-effect relationship, it is commonly used to express the
assertion-reason relationship, i.e. as a speech act of stating that “this is the reason I am
saying this” (McCarthy, 1991: 49; Givon, 1993: 299-301; Celce-Murcia &
Larsen-Freeman, 1999: 530). As an illustration, consider (89):
(89) My friend was probably fired, because I don’t see him anywhere.
In (89), apparently, the because-clause does not signal the speaker’s inability to find
the friend as the cause of his being fired; rather, it simply provides a justification or
support for the speaker’s assertion about the friend’s being fired. Typically, learners
are inclined to apply to writing the wide distribution and pragmatic function of
because in spoken discourse; as such, they frequently avail themselves of the
conjunction in their essays, not just to literally convey causal meaning, but also to
colloquially present “the knowledge base” for an assertion (Schleppegrell, 1996:
275-277). When learners overuse the conjunction because in writing, they give their
prose an inappropriately discursive oral quality and more importantly, they fail to
consider employing more specific and more writing-oriented cohesive devices like
embedding with RCs to effectively get their ideas across.
The above discussion on infelicitous transfer of the oral conjunction strategies
with and and because into written genres, which partially leads to one’s failure to
exploit RCs as an useful backgrounding device, may offer an alternative explanation
for RC underproduction in L2 learners’ English writing, a phenomenon commonly
ascribed by SLA researchers to L1 interference, either at syntactic (e.g. Schachter,
1974) or pragmatic levels (e.g. Bley-Vroman and Houng, 1988, cited in Kamimoto et
al., 1992; Zhao, 1989, cited in Kamimoto et al., 1992; Li, 1996).
4.3 Summary
In this chapter, the researcher has reported the main findings from the three
elicitation instruments and elucidated them in relation to the research questions.
Firstly, the results from the RC judgment test indicate that overall, the subjects
did not yet achieve full mastery of English NRRCs. Their accuracy rate of NRRCs
in the test was distinctly low (50.30%) in comparison with that of RRCs (85.60%).
Although the third graders were found to perform significantly better than the first
graders, their accuracy rate of NRRCs still remained extremely low (53.67%). A
further exploration into the results revealed two possible causes for the subjects’ poor
performance in properly using NRRCs. One was the tendency to overuse RRCs.
As shown in their RRC-prevalent responses, the subjects had a strong tendency to use
RRCs in most contexts, even where the use of their non-restrictive counterparts would
be more felicitous. The other cause was the tendency to use NRRCs only with
uniqueness-referring NPs and dismiss other referentially accessible NPs. As attested
by their great performance differences among different NP contexts, the subjects’ use
of NRRCs seemed to be correlated with personal pronouns (78.95%), one-of-a-kind
NPs (85.65%), and proper NPs (74.47%), without being generalized to other equally
applicable NP contexts, such as linguistically definite NPs (17.76%), situationally
definite NPs (19.44%), narrowly specified NPs (19.91%), whole-referring NPs
(15.25%), and generic NPs (54.37%). This rather limited distribution of NRRC use
was displayed by both the first- and third-graders: among the eight non-restrictive NP
contexts, the two groups performed equally well in lexically accessible NP contexts
(75.17% and 84.27%, respectively) and equally poorly in non-lexically accessible NP
contexts (25.59% and 35.17%, respectively), except in generic NPs, where the third
graders significantly outperformed the first-graders. The subjects’ tendency to
restrict their use of NRRCs to particular contexts was then confirmed by their
retrospective explanations for rendering non-restrictive judgments to have resulted
from their inappropriately associating NRRCs with uniqueness-referring, as opposed
to referential accessibility.
The subjects’ tendency to overuse RRCs in any given context and propensity to
limit their use of NRRCs to uniqueness-referring NPs are respectively indicative of
two deficiencies in their acquisition of NRRCs: (1) a lack of a definite distinction
between RRCs and NRRCs in their mental grammar of English RCs; and (2) a lack of
comprehensive knowledge of when to use NRRCs. The subjects’ flawed acquisition
of NRRCs is first of all accounted for by L1 transfer. Not acutely aware of the
difference in the classification of RCs between Chinese and English, the subjects may
have unconsciously transferred the purely restrictive nature of RCs in Chinese into
English and thus failed to make a dichotomous RC differentiation in their mental
grammar of English RCs. Also postulated to be at the root of the subjects’
unsuccessful attempt at acquiring NRRCs is cognitive complexity. To completely
understand the context-based notion of referential accessibility, associated with
NRRCs, could place great cognitive loads on the subjects, the majority of whom were
so used to learning English at a context-vacuumed, sentential level; as such, they may
still have encountered considerable difficulties drawing a clear-cut line between RRCs
and NRRCs in their mental grammar of English RCs. Furthermore, the subjects’
failed acquisition of NRRCs is attributable to input frequency. From a survey of RC
input in English textbooks, the amount of NRRC input was found to be scanty and
this low frequency of NRRC input in text may have contributed to the subjects’ failure
to readily consolidate and internalize NRRCs into their abuidling mental grammar of
English RCs. Last but not least, instructional effects never fail to be irrelevant in the
subjects’ acquisition of NRRCs. Through an examination of instructional materials,
some inadequacies in their presentation of NRRCs were identified as likely causes of
the subjects’ inability to well distinguish between RRCs and NRRCs in their mental
grammar of English RCs and to gain a full understanding of when to use NRRCs,
including overemphasis on RRCs at the expense of usage of NRRCs, few attempts to
raise awareness of the relationship between NRRCs and referential accessibility, and a
shortage of context-rich exercises for NRRCs. In addition, instructional effects may
even help explain the significant group difference in generic NPs: familiarity with the
rule stressing the co-occurrence of NRRCs with uniqueness-referring NPs may have
enabled the third-graders to generalize their use of NRRCs by deductive reasoning to
generic NPs, which share the same semantic feature of “particularness.”
Secondly, the results of the context translation test suggest that notwithstanding
their syntactic fluency in RCs, the subjects, by and large, lacked a deeper appreciation
of the pragmatic/discourse functions commonly served by English RCs. The
subjects’ frequency rates of RC use exhibited significant differences among the four
contexts: identifying (95.83%), characterizing (91.67%), presentative (48.75%), and
parenthetical (57.92%). Specifically, they typically employed RCs merely to
identify a referent as a known entity or to characterize a referent as a particular type;
rarely did they avail themselves of RCs in presenting a topically important referent or
interpolating parenthetical assertions about a referent. This pattern of RC use was
characteristic of both the first- and third-graders. Besides, no significant difference
was found between the two groups for their RC use in each context, indicating that the
third-graders’ having more exposure to RCs in their English study did not necessarily
translate into their mastery over such advanced uses of RCs as presenting and
interpolating. Further analysis of the subjects’ translation in the test highlighted one
factor in their underuse of RCs: a tendency to consistently employ independent
clauses as a common alternative to RCs in presentative (51.25%) and parenthetical
(42.08%) contexts. Moreover, it was found that the subjects, across groups, were in
the habit of misusing the definite article the with RCs which mainly served to
characterize an NP as a particular type or category instead of identifying it as a
familiar or known entity, though the tendency was not significant (31.34% and
31.21% for the first- and third-graders, respectively).
The subjects’ failure to extend their RC use to presentative and parenthetical
contexts beyond identifying and characterizing ones basically underlies inadequacies
in the teaching of RCs. On the one hand, EFL instruction seldom goes further to
draw learners’ attention to the use of RCs as a topic-presenting strategy. This is so
because often tackling grammar at a sentential level, EFL instruction lacks a more
discourse-oriented approach to sensitizing learners to the presentative use of RCs, a
function better understood only in connected stretches of discourse. On the other
hand, EFL instruction makes few attempts to help learners take heed to the use of RCs
as an information-adding interpolator. With its overemphasis on usage of various
relative markers, EFL instruction typically treats NRRCs merely as an adjunct to the
presentation of RRCs, providing supplementary facts about how the two types of RCs
differ syntactically and semantically; little account is given of usage of NRRCs, let
alone the information-adding function of NRRCs. As regards the subjects’ tendency
to use independent clauses in place of RCs in presentative and parenthetical contexts,
the researcher identified the tendency as reflecting a lack of awareness of the
cross-linguistic difference between Chinese and English in their preferred structures
for the two contexts. That is, on the false premise that both languages utilized the
same grammatical resource for a given communicative function, the subjects still
deemed independent clauses, the same structure used in Chinese, to be appropriate
structural equivalents in English for presenting a topic or interpolating extra
information, and hardly ever considered other alternatives like RCs to achieve such
functions. Lastly, as for the subjects’ misuse of the definite article with RCs in a
characterizing context, possible explanations were propounded in terms of the
influence of false instruction emphasizing the co-occurrence of the with RCs and the
exposure to input typical of identifying RCs, namely, RCs with definite NPs.
Lastly, the results of the passage-rewriting test show that the subjects were quite
able to make good use of RCs in writing as a useful backgrounding device. In
rearranging idea units in the passage, most subjects (56.67% to 80.83%) were capable
of packaging amplifying information of description with RCs. Two variables were
found to influence the extent of their RC use in writing. One was years of leaning:
the third-graders apparently performed better than the first-graders in employing RCs
to background information, with the group difference ranging from 8% to 28%. The
other variable was the tendency to use independent and-clauses or adverbial
because-clauses instead of RCs to code the same piece of information.
The subjects’ prevalent use of independent and-clauses or adverbial
because-clauses for the same information as otherwise packaged with RCs is
explicable in terms of infelicitous transfer of clause-chaining strategies typical of
spoken English into written English. Unaware of register differences between
speaking and writing in their common grammatical resources for integrating
information, the subjects may have transferred the high frequency and pragmatic
function of and and because in speaking into written discourse, thus overusing these
two “colloquial” conjunctions where more writing-oriented and more effective
strategies for clause linkage like RCs were expected. This modality transfer of
conjunction strategies may well serve as another plausible explanation for L2 learners’
RC underproduction in writing, typically attributed to either structural or pragmatic
interference from L1.