You are on page 1of 13

2009 年4 月 中国英语教学 (双月刊) Apr.2009

第 32 卷 第2 期 CELEA Journal(Bim onthly) Vol.32 No.


2

L1 INFLUENCE ON L2 WRITING :
PROCESS AND PRODUCT

He Yanqun
Sanjiang University

Abstract
  By e m ploying a co m bination of research techniques ,such as think- aloud and stim ulated recall ,this study
investigates the cross-linguistic influence in second language writing at both process and product levels am ong
20 Chinese college students.In terms of process , the paper reveals that though the proportion of L1 thinking
decreases with the im prove ment of English proficiency , the extent of decrease varies with different co m posing
activities.In terms of product , learners with higher English proficiency exhibit a decrease in the total nu m ber
of transfer errors ,yet the decrease varies with different categories. What is im portant is that ,contrary to
co m m on belief ,there is no significant correlation between the total volu me of L1 use and the total nu m ber of
transfer errors per 100 w ords.The factisthat L1 use in differentthinking activities has different effects on the
occurrences of transfer errors. While L1 use in content conception has a significant positive correlation with
lexical errors for the sopho m ore group ,L1 use in text production has a significant negative correlation with
syntactic errors for the senior group.

Key w ords
L1 thinking ;think-
aloud ;transfer errors

1. Introduction
Second language acquisition(SLA )is a co m plex process involving m any interrelated factors ,a m ong
w hich the learner s first language(L1)is considered to be a m ajor factor.Actually , the influence of L1
on the acquisition and use of a second language (L2 ) has long been a focus of interest within SLA
research. Much of the research attention was confined to the sentence or utterance level. However , in
the past tw o decades , the study of L2 writing has beco m e part of the m ainstrea m in SLA research.

In previous researches ,effects of L1 on L2 writing have been studied fro m tw o different


perspectives ,na m ely ,the product and process perspectives.In terms of product , studies of co m positions
written by L2 writers focus ,m ost notably ,on those errors w hich appear to be the result of negative
transfer fro m L1(e .g. Zobl 1979 ,1984 ,1986 ;Schachter 1983 ;Gass 1984 ,1987 ;Wenzell 1989).In
terms of process ,the study of co m posing processes through think- aloud protocols has led to tw o
controversial findings.One claims that non- native writers thinking in L1 detracts the production of good
writing(e.g. Wen &Johnson 1997)w hile the other suggests that such cognitive behavior plays a positive
role in L2 writing.L2 writers have been observed to e m ploy their L1 to “get a strong im pression and
association of ideas for essays”and produce essays“of better quality in terms of ideas ,organization and
details” (Lay 1982 :406 ). They have also been found to use their native language as“an im portant
resource in their continual process of decision m aking w hile writing”(Cu m ming 1989 :128).
In China ,the research on the effect of L1 on L2 writing has m ainly been focused at the product
level.Those researches take the form of either contrastive analysis(e
.g. Cai 1998 ;Li &Zhang 2002)or
3
L1 Influence on L2 Writing :Process and Product   He Yanqun

error analysis(e.g. Li &Cai 2001 ;Chen 2002) .The results show that transfer is an im portant factor in
accounting for errors ranging fro m lexical to discourse level(e.g. Lou 2002 ;Han & Zhou 2003 ;Fan
2001 ).The limitations of those researches are of tw o m ajor types.First ,in m ost of the studies ,
little care
has been taken to separate the errors m ade by learners at different stages of develop m ent or learners of
different language proficiency , thus affording only a static view of L2 acquisition.The second proble m is
that the investigation m ethod usually excludes interviews with the subjects ,w hich is an im portant step in
the judg m ent of transfer errors.

Am ong the few researchesinvestigating the influence of L1 thinking on L2 writing , the study by Guo
&Liu(1997)especially warrants notice.In their study ,think- aloud technique is e m ployed to detect w hat
is going on in the participants mind w hile they are writing.The 12 participants are selected through strict
testing fro m a m ong 50 candidates. Based on the findings of the study ,Guo &Liu propose a cognitive
fra m ew ork for explaining L1 influence on L2 Learner s thinking operations during L2 writing.Following
this line ,Wen &Guo(1998)further analyze part of the data gathered by Guo &Liu(1997),and thereby
investigating the relationship between thinking in L1 and L2 writing ability.These tw o studies contribute
a lot to the research into the L1 influence on L2 writing.However , it is noted that in these tw o studies ,
the relationship between the quality of co m positions and L1 thinking in different co m posing activities(see
3.2.1 ) has not been respectively assessed ,though there are reasons to believe that L1 thinking in
different co m posing activities m ay exert different effects on the production( Wang & Wen 2002b) .
2.Research Questions
Considering the above-m entioned limitations ,this study of L1 influence on L2 writing at both process
and product levels e m ploys a co m bination of research techniques ,such as think- aloud and stim ulated
recall.The study addresses the following three research questions :

1) Do learners with higher L2 proficiency exhibit a decrease in the use of L1 in the process of
co m posing in a L2 ?Does the extent of decrease vary with different co m posing activities ?
2)Do transfer errors decrease with the im prove m ent of learners L2 proficiency ?Does the extent of
decrease vary with different language aspects ?
3 )Are quantity and distribution of transfer errorsin students L2 co m positions related to students L1
thinking during the writing process ?Does this vary with learners L2 proficiency ?

The originality of this study lies in the third objective m entioned above. A careful review of the
previous researches reveals that few of the m have tried to investigate the relationship between the L1 used
in the writing process and the distribution and proportion of transfer errorsin the co m positions ,
though it
is a popular belief a m ong L2 teachers that thinking in L1 willincrease the nu m ber of transfer errors.

3. Methodology
3.1 Data Collection
3.1.1 Participants
A total of 20 university students fro m USTC and Anhui University participated in this cross-
sectional
study.The group consists of 10 second- year English m ajors and 10 senior English m ajors.

The 20 subjects eventually involved in this study were chosen through three stages.At the first stage ,
51 students(24 sopho m ores and 27 seniors)were reco m m ended by their teachers as students w ho were
extrovert and talkative and m ore likely to be co- operative in acco m plishing the task of think-aloud.It is
assu m ed by L2 writing researchers that only those talkative and extrovert students can acco m plish the task
of think- aloud ( Wen 2001 ) . Then the subjects were provided well- constructed training. The training
started with the researcher s de m onstration in w hich we verbalized w hat was going through our minds as
we inferred a missing nu m ber a m ong a list of regular nu m bers.Then all the selected subjects were asked
to perform tw o tasks by think- aloud. The first one was an anagra m task (following Ericsson &Sim on
1987 :49 ) . The subjects were presented with an English w ord w hose constituent parts had been
rearranged.They were required to think aloud as they unscra m bled the w ord.The second one was an
arith m etic proble m (following Ericsson &Sim on 1987 : 34) . With that preparation , the subjects knew the
m eaning of think- aloud and how to carry it out.The second selection was m ade after the training.The
4
CELEA Journal 84

criterion for the second selection was a positive attitude towards think- aloud. The second selection
reduced 22 subjects because they did not feel good with think- aloud.The third selection was m ade after
we collected all the data by think- aloud. This stage eliminated 9 subjects w hose noticeable pause
co m prised m ore than 20 % of the total recorded tim e(following Wen &Guo 1998) .

3.1.2 Data Collection Procedures


The data collection included tw o steps :a think-
aloud task and a retrospective interview

In the first step ,each participant was asked to write a paragraph in English of m ore than 120 w ords
giving their opinions on the issue of undergraduates living in rented houses off ca m pus. Mean w hile ,they
were required to verbalize w hatever was going through their minds. The experim ent was carried out
individually so that we could offer help w hen the participants had any misunderstanding about the think-
aloud technique.They were asked that their think- aloud should be natural in the sense that w hat they
uttered should be the sa m e language as the one they were thinking in.In other w ords , if the inner speech
was conducted in Chinese ,then they should speak in Chinese ;if in English ,then English should be
e m ployed.A digital recorder was used to record each subject s think- aloud process and the co m positions
written by the subjects were also collected as the data of product.

A retrospective interview was conducted im m ediately after the think- aloud task ,in w hich the
subjects were asked to reflect on their thoughts to see w hether they could recall any unreported thinking
processes.As stim ulation , the recorded think- aloud protocol was played back to the subjects. When the
subjects felt necessary to add so m ething , the researcher stopped the m achine and wrote dow n the newly
elicited inform ation.For exa m ple , in one of the sopho m ore s recorded think- aloud protocols ,there is an
utterance like“ 怎么来组织呢?嗯……” w hich indicates the following utterances should be about the
conception of structure ,yet w hat followsis“First , first ,
living with ,well ,probably classm ates ,provides
convenient life.You are going to be late for class and your classm ates m ay just wake you up. .
.”.It m ay
well be that so m e of the thinking activities about conception of structure get unreported. With the pro m pt
of the recorded think- aloud protocol , the participant recalled that he was thinking of“就是原来的first ,
second , third 的结构,这样写比较有层次感”after he asked himself“怎么来组织呢? ”

3.2 Data Analysis


3.2.1 Transcription of the Think-
aloud Protocols
Once the data collection was finished ,the verbal reports were transcribed. The transcriptions
include the participants verbalization of the w hole writing process and the“unreported m ental activities”
(see 3.1.2)that were recalled in the retrospective interview conducted right after the think- aloud task.
Following Cu m ming (1989 )and Wang & Wen (2002a),the transcriptions atte m pted to represent the
participants tape- recorded speech as it w ould usually appear in expository prose ,although several special
conventions were used.A capitalletter indicates the beginning of the spoken equivalent of a sentence and
a period indicates its closure.Suspension points indicate a pause.A question m ark indicates a questioning
intonation. Words in double quotation m arks indicate participants were reading fro m a source text(such
as the directions for writing) .

3.2.2 L1 Use in Different Co m posing Activities


Based on the research of Wang & Wen (2002a),the writing process is divided into five different
co m posing activities ,na m ely topic analysis ,content conception ,structure conception ,text production
and control of process.The definitions and exa m ples of the five co m posing activities are show n in Table
3-1.
In each co m posing activity , the nu m ber of English w ords and Chinese w ords were counted ,
including
the repeated utterances. Then the percentage that Chinese w ords constitute of the total w ords was
calculated as an index of the a m ount of L1 thinking(following Wen &Guo 1998 ;Wang & Wen 2002a) .
To determine the changes in the proportion and distribution of L1 use in the writing process ,m eans and
standard deviations of the a m ount of L1 thinking of the tw o groups were respectively calculated.

5
L1 Influence on L2 Writing :Process and Product   He Yanqun

Table 3-
1.Definitions and exa m ples of co m posing activities
Definition Exa m ples
Topic Analysis :an analysis of the “ Write an English paragraph of m ore than 120 w ords giving
topic and the re m arks on the topic your opinions on undergraduates”,undergraduates ,本科生,
“living in rented houses off ca m pus”.本科生出去租房子住是吧。
My opinions , ..
.u m ,a bit difficult.
Content Conception : the conception 租房子有利也有弊。It s quiet but living off ca m pus m ay cause
of content and the re m arks on the so m e kind of proble ms.It will.
..减少交流,减少与同学和室友之
content 间的交流。……租房子也会增加经济负担。很多同学出去住并不
是征得父母的允许,是自己偷偷的在外面,这一点要写进去吗?不
说这个了。
Structure Conception : the 我想把这篇文章放在专四、专四考试那种三段、三段论那种形式
conception of structure and the 写。第一段肯定是陈述自己的观点,就是 advantages and
re m arks on the structure disadvantages 。第二段就是分别陈述 w hat are those advantages
and disadvantages 。第三段就是restate ,e m phasize m aybe a
point 。这个部分是不是写得比较少, 是否应该总结一下?
Text Production : thinking activities So m e students ,students , so m e students like ,这个可以用like ,
related to the production of the m ain like to m ake noise unconsciously.It will also disturb ,disturb the
body of the text other student s rest ,disturb the other students rest , the other ,
...u m ,so w hen the students ,w hen this kind of students ,w hen
this kind of students m ove out of ca m pus , it will be beneficial
for the other m e m bers of the dorm ,oh no ,no ,应该是in the
dorm 吧, in the dorm .
Control of Process : thinking Thirty minutes ,三十分钟,应该来得及吧。一百二十个字应该不
activities related to the m anage m ent 多吧。Ok ,好,开始写。
of the writing process

3.2.3 Qualitative and Quantitative Analysis of Transfer Errors


3.2.3.1 Detection of Transfer Errors
As is popularly acknowledged ,besides production errors , there are a nu m ber of other m anifestations
of negative transfer , such as underproduction(or avoidance), and overproduction(or over- use).Yet the
focus of this research is placed on production errors.An error is defined as a deviation fro m the norms of
British and/ or Am erican standard written English.The m easures of errors are expressed in terms of tw o
categories ,na m ely ungra m m aticity and infelicity.Gra m m aticity is a theoretical notion w hich appeals to
the judg m ent of errors at the syntactic level. On the contrary ,infelicity is a practical notion w hich
appeals to the judg m ent of errors at the lexicallevels.

Tw o steps were e m ployed in the judg m ent of transfer errors.First ,


transfer errors were distinguished
fro m errors of other sources on the basis of Transfer Analysis ,or in other w ords ,by co m paring
interlanguage with m other tongue to see w hether there were any similarities between the errors and the
m other tongue structure. The following are exa m ples of how we applied Transfer Analysis to the
judg m ent of transfer errors(errors italicized):

Ex.1. Those who live out of ca m pus ,they tend to lose close contact with their classm ates or
roo m m ates. (fro m sopho m ore group)
The double no minative construction “those w ho live out of ca m pus ,they. .
.”is considered an
interference fro m the m other tongue ,because in Chinese ,the double no minative construction is co m m on
w hile itis unusualin English.As show n in“那人他爸病了”, the first no minative“那人”is used as a topic
and the second no minative“他爸”as a subject.

In order to validate the results of the first step ,one day after the think-
aloud task ,the subjects were
interviewed about the sources of the errors. This retrospective interview is an essential step in the
6
CELEA Journal 84

judg m ent of transfer errors ,in the sense thatit eliminates those specious transfer errors ,as show n in the
following exa m ple :

Ex.2.So m eone has proble ms in getting along with his roo m m ates and feel( )like living alone.
(fro m senior group)
The error in exa m ple 2 concerning the subject- predicate concord had been presu m ed to be the
interference of m other tongue ,because Chinese does not m ark verbs m orphologically to agree with the
noun classinto w hich the subject or direct object falls.By contraries , the participantinterviewed asserted
that it was sim ply a slip of pen ,and he justified himself by referring to the recorded think-aloud protocol.
Hence ,retrospective interviews can help distinguish mistakes fro m errors.

According to Corder ,an error(in its technical sense)takes place w hen the deviation arises as a
result of lack of knowledge.It represents a lack of co m petence.A mistake occurs w hen learners fail to
perform the co m petence.In other w ords ,w hile a mistake is a rando m perform ance slip caused by
fatigue ,excite m ent ,etc.,and therefore can be readily self-corrected ,an error is a syste m atic deviation
m ade by learners w ho have not yet m astered the rules of the L2.If the learner can self- correct the
incorrect form , it is a mistake(see exa m ple 2).Otherwise ,it is an error(see exa m ple 5).

3.2.3.2 Tw o Levels of Transfer Errors


The analysis of transfer error frequency was based on counts of errors of tw o levels ,na m ely ,lexical
errors ,and syntactic errors.The discourse errors were not counted because it is insignificant to calculate
coherence errors(delay of key point)in one paragraph. No m atter how long the paragraph is ,the
m axim u m of coherence errorsis one.Errors at the substance level ,like punctuation errors were ignored.
Lexical errors include aw k ward collocations and confusion of sense relations(using the wrong one fro m a
set of near- synony ms),as exe m plified below :

Ex.3.Undergraduates have no independent economy ability ,so they m ay not rent a house though
they prefer to do so. (fro m senior group)
Ex.4. We will be m ore independent and m uch riper if we live alone out of ca m pus.
(fro m sopho m ore group)
In exa m ple 3 ,“independent econo m y ability”is a w ord-
for-w ord translation of“独立的经济能力” .
In exa m ple 4 ,the error arises because“m ature”(of a person)and“ripe” (of fruit ,grain ,etc.)are both
“成熟”in Chinese.
With regard to the errors of syntactic level ,there are m any instances under the subsets of o mission ,
misselection ,and misordering ,as show n in the following exa m ples :

Ex.5.People live in rented houses lack the experience of living in a group. (fro m senior group)

The o mission of“- ing”is assu m ed to be the result of the fact that Chinese verbs have no inflectional
m orphe m es.Yet it is noticed that the participant did not o mit the“- ing”in“living in a group”with the
trigger of“of” .The reason is that he has not learnt all the contexts in w hich“- ing”should be used.

Ex.6.If a girl wants to rent( )house out of ca m pus ,safety is another very im portant factor she
should take into consideration. (fro m sopho m ore group)
The error in exa m ple 6 is so m ew hat challenging to analyze. At first glance ,it see ms quite self-
evident that the source of the error is the non- existence of articles in Chinese.The question is w hy the
participant had difficulty in using“a”in“to rent a house”but succeeded in the use of“a”in“a girl” .The
key to the question lies in an analysis of the different functions of the article“a” .As we know , “a”can
be used to denote nu m ber ,w hen it is equated to “one” . The sa m e occurs in Chinese.If you want to
express“一个女孩”, the nu m eral cannot be o mitted.In this circu mstance , there is no difference between
the Chinese and English structure.In addition , “a”can also function as a type indicator ,as in“Ia m a
student”, in w hich “a ”cannot be o mitted.In Chinese ,however ,nouns indicating “type”are often
m arked with bare nouns (nouns with zero specificity m arking )as in “ 我 是 学 生” . The differences
7
L1 Influence on L2 Writing :Process and Product   He Yanqun

between the English and Chinese syste m of denoting a certain type created the learning difficulty ,hence
resulting in the error.The analysis was justified in the retrospective interview with the participant.

Ex.7.One of m y best friends just lives outside the ca m pus.He has to spend m ore than 3000 yuan a
year in renting a house ,this can cause so m e burden to his poor fa mily.
(fro m senior group)
The error in exa m ple 7 results fro m shunning away fro m using relative clauses.Learners m ay avoid
using linguistic structures w hich they find difficult because of differences between their m other tongue and
the targetlanguage(Schachter 1974 ;Klein m ann 1978 ;Kellerm an 1977 ;Dagut &Laufer 1985) .The fact
that relative clauses are absent in Chinese m ay create difficulties for Chinese learners.The participant
began a coordinate clause by using“this”rather than co m pressing the inform ation into a relative clause.

Ex.8.But out of ca m pus ,it is easy to contact with other walks of people.
(fro m sopho m ore group)
The error in w ord order lies in the fact that w hile Chinese is left- branching ,i
.e. nouns are pre-
m odified ,in English if the m odifier is a phrase or a clause ,it appears after the noun( Halliday &Hasan
1976) .That is w hy the participant used“other walks of people”(其他行业的人)instead of“people of
other walks” .
Ex.9. Because you m ust deal with all the things without any help fro m others ,you will be
independent. (fro m sopho m ore group)
The mispositioned clause beginning with“because”is interference fro m Chinese ,for in Chinese the
cause is usually stated before the effect.

3.2.3.3 Quantitative Analysis of Transfer Errors


For quantitative m easures ,total w ords of each co m position were tallied ,and the tw o levels of errors
were respectively counted in all the papers. Then the nu m ber of occurrences per 100 w ords was
calculated. When counting the errors ,we found a proble m , that is ,w hether we should count the tokens
or the types.In other w ords , should we count the repeated occurrences of the sa m e error ?In resolving
this proble m ,we referred to Lennon(1991)in w hich he proposes thatif the repeatis a lexical replica of
a prior error ,it is not to be counted as a distinct error.Thus there is one error in“he play tennis very
well but football he play badly”(Ja m es 1998 :117)and there are tw o errors in“She play tennis very well
and also she swim very well” (ibid.) To determine the changes in the proportion and distribution of
transfer errors m ade by different groups of students , the tw o groups m eans and standard deviations of the
errors per 100 w ords were respectively calculated.

3.2.4 Correlational Analysis of Transfer Errors and L1 Use


For the analysis of the correlation between those transfer errors in the production and L1 use in the
writing process ,Pearson correlations were calculated between the error frequency at the tw o different
levels and the m ean percentage that Chinese w ords constitute of the total w ords in each co m posing
activity.

4.Findings
4.1 L1 Use in the Writing Process
In general less L1 is used by the seniors than the sopho m ores in the co m posing process(26 % vs.
41 %) .Table 4- 1 reveals the m ean percentage and standard deviations of L1 use in different co m posing
activities.To a certain degree , the result deviates fro m the findings of Wang & Wen(2002a) (29 % and
13 % respectively for senior and sopho m ore subjects) . Tw o reasons m ay account for the discrepancy.
First ,our experim ent was conducted at the very beginning of a se m ester ,thus the sopho m ores had only
received one year of college education.Second , in their study ,tw o out of the four senior subjects(thatis
50 %)use few (0.2 %)or no Chinese in the reported writing process ,w hich greatly reduces the m ean
volu m e of Chinese thinking.Though in our study there is also one senior participant w ho spoke English
through the w hole writing process , the influence is m uch less re m arkable ,because she accounts for only
1/10 of the total.

8
CELEA Journal 84

The distribution of L1 thinking is uneven.Resultsin Figure 4- 1 revealthat for both groups ,L1 use in
the control of process and structure conception clearly exceeds L1 use in the other co m posing activities
(86 % and 83 % for sopho m ores and 71 % and 72 % for seniors).The subjects also tend to rely on L1 in
content conception(68 % for sopho m ores and 46 % for seniors) .The m ean percentage of L1 thinking is
sm aller in topic analysis and text production (21 % and 30 % for sopho m ores and 11 % and 14 % for
seniors) .This finding suggests that the participants are m ore likely to use L1 as their co m posing language
in the co m posing activities that are not directly related to text production (control of process , structure
conception and content conception) .The result conforms to the study of Wang & Wen s(2002a).

  Table 4-
1.Means and standard deviations of L1 use
Co m posing Activities
Proficiency
TA CC SC TP CP
Sopho m ore
Mean 21 % 68 % 83 % 30 % 86 %
SD 0.16 0. 31 0.18 0.15 0. 18
Senior
Mean 11 % 46 % 72 % 14 % 71 %
SD 0.12 0. 31 0.32 0.12 0. 40
TA =Topic Analysis ;CC =Content Conception ;SC =Structure Conception ;TP =Text Production ;CP
=Control of Process

Figure 4-
1.Uneven distribution of L1 use

Though the overall L1 use decreases with the im prove m ent of English proficiency ,the extent of
decrease varies with different co m posing activities ,as show n in Figure 4-
2.Fro m Figure 4-
2 we can see
that the tendency of decrease is strongerin text production and content conception.In fact ,if we re m ove
the data obtained fro m the only one participant(out of ten seniors)w ho used no Chinese at all in the
co m posing process ,the unevenness of decrease will be m ore re m arkable(see Figure 4-
3).The sopho m ore
group s L1 use in the control of process is only seven percentage points higher than that of the senior
group(86 % vs.79 %)and the sopho m ore group s L1 use in structure conception is three percentage
points higher than that of the senior group (83 % vs.80 %),w hereas the form er s L1 use in text
production and content conception is respectively 14 and 17 percentage points higher than that of the
latter group(30 % vs.16 % and 68 % vs.51 %) .

9
L1 Influence on L2 Writing :Process and Product   He Yanqun

Figure 4-
2.Decrease of L1 use in different co m posing activities(1)

Figure 4-
3.Decrease of L1 use in different co m posing activities(2)1

4.2 Error Frequency


Table 4- 2 lists the m eans and standard deviations of errors of tw o levels.For both groups , syntactic
errors are m ore frequent than lexical errors(as show n in Figure 4- 4). Generally speaking ,the total
nu m ber of errors at the tw o levels decreases with the im prove m ent of English proficiency(2. 83 vs.1. 42
per 100 w ords),yet the decrease varies with different categories. As we can see in Figure 4- 5 ,the
tendency of decrease is stronger with syntactic errors(2. 17 vs.1. 32),and less strong with lexical errors
(0.66 vs.0.40).

  Table 4-
2. Means and standard deviations of errors of tw o levels
Level
Proficiency
Lexical Syntactic
Sopho m ore
Mean 0.66 2.17
SD 0.55 1.08
Senior
Mean 0.40 1.32
SD 0.23 0.66
  Values show the nu m ber of occurrence per 100 w ords.

10
CELEA Journal 84

Figure 4-
4.Uneven distribution of errors

Figure 4-
5.Decrease of errors of tw o levels

4.3 Correlation Between Error Frequency and L1 Use


With respect to the correlation between error frequency and L1 use during the writing process ,Table
4-3 and Table 4- 4 respectively show the Pearson product-m o m ent correlation coefficients.For both the
sopho m ore and senior groups , there is no significant correlation between the total volu m e of L1 use and
total nu m ber of errors per 100 w ords(r =. 341 ,p =. 336 for sopho m ores and r =-. 459 ,p = .182 for
seniors) .This result suggests that generally speaking ,thinking in L1 w hile co m posing in L2 does not
encourage the occurrence of transfer errors ,contrary to the popular belief that such cognitive behavior
will result in transfer errors.

Yet there exist considerable differences between the tw o groups.As show n in Table 4- 3 and Table 4-
4, for the sopho m ore subjects ,L1 used in content conception is positively correlated with the nu m ber of
errors at the lexical level(r =. 686 ,p <.05),w hile for the senior group ,there exists no such kind of
correlations.In an atte m pt to account for the correlation between L1 used in content conception and the
occurrences of lexical errors ,we studied the sopho m ores think- aloud protocols and their co m positions
and found that m ost of the lexical errors are the results of a w ord-for-w ord translation fro m the Chinese
w ords used in content conception.A study of the seniors think- aloud protocols and co m positions reveals
that the process of translation also exists within this group ,yet the seniors are m ore likely to edit out
un wanted transfer during translation

A peculiar finding is that for the senior group ,L1 used in text production is negatively correlated
with the nu m ber of syntactic errors(r = -. 610 ,p =.061) .Though owing to the sm allsize of our corpus
the correlation is not significant at.05 level ,the finding warrants notice because it m ay so m ew hat
indicate a way in w hich thinking in Chinese has a positive influence on reducing transfer errors.
According to Wen &Guo(1998),L1 thinking in L2 writing process has five types of functions(as show n
in Appendix 2 ) . The three im portant functions of L1 during text production are search m otor for
11
L1 Influence on L2 Writing :Process and Product   He Yanqun

content , search m otor for forms , and confirm ation.In an atte m ptto find out w hich function is beneficial
for reducing transfer errors ,we carefully studied the think- aloud protocols of the tw o groups and found
that only w hen L1 is used as a search m otor for forms ,itis possible to reduce transfer errors.The factis
that the senior group used L1 as the search m otor for forms m ore frequently than the sopho m ore group.
Moreover ,w hen the sopho m ore group did use L1 as the search m otor for forms , in m ost of the cases ,it
failed to edit out the transfer errors.

  Table 4-
3.Correlations between error frequency and L1 use(sopho m ore)
Level
Lexical Errors Syntactic Errors Total Nu m ber
L1 Use
Topic Analysis .361 .297 .371
Content Conception .686 * .272 .479
Structure Conception .155 .070 .115
Text Production .216 .203 .242
Control of Process .317 .238 .308
Total Volu m e .511 .182 .341
  *Correlation is significant at the 0.05 level(2-
tailed)

  Table 4-
4.Correlations between error frequency and L1 use(senior)
Level
Lexical Errors Syntactic Errors Total Nu m ber
L1 Use
Topic Analysis .088 .133 .131
Content Conception -
.274 -
.469 -
.452
Structure Conception -
.071 -
.159 -
.147
Text Production -
.282 -
.610^ -
.565
Control of Process -
.141 -
.115 -
.131
Total Volu m e -
.282 -
.476 -
.459
  *Correlation is significant at the 0.05 level(2-
tailed);
^p =.061

5.Conclusions
Based on the m ajor findings ,we conclude that the distribution of L1 thinking is uneven ,with L1 use
in content conception and structure conception clearly exceeding that in text production.The reason lies
in the different patterns of storage of the three types of knowledge (topical knowledge ,discoursal
knowledge and linguistic knowledge)L2 writers retrieve during the process of writing(Cu m ming 1989) .
The proportion of L1 thinking decreases with the im prove m ent of L2 proficiency ,yet the extent of
decrease varies with different thinking activities ,with the tendency of decrease stronger in text
production and content conception.

In terms of the frequency of transfer errors , syntactic errors are m ore frequent ,w hile lexical errors
are relatively infrequent.Learners at a higher proficiency level exhibit a decrease in the occurrences of
transfer errors ,yet the extent of decrease varies with different categories ,with the tendency of decrease
stronger in syntactic errors.Tw o reasons m ay account for the difference in the extent of decrease.One is
that owing to the relative infrequency of lexical errors , they have little roo m to decrease.The other is
that the im prove m ent of the learners English proficiency in different aspects is not synchronic ,with the
develop m ent at syntactic level outpacing lexical.

Generally speaking , there does not exist significant correlation between L1 thinking and the transfer
errors in the text resulting fro m the writing process. However ,L1 use in different thinking activities
exerts different effects on the production of transfer errors.For the lower- proficiency students ,L1 use in
content conception has a significant positive correlation with lexical errors ,w hich can probably be
explained by the factor that so m e of the errors are a result of a w ord- for- w ord translation fro m the
Chinese w ords concepted during content conception.For the higher- proficiency students ,L1 use in text
production has a significant negative correlation with syntactic errors.A possible reason is that w hen this
group of students use m other tongue as the search m otor for forms , it m ay help reducing syntactic errors.
12
CELEA Journal 84

Findings in this study m ay help reshape the stereotyped view of so m e L2 teachers on writing at the
undergraduate level. Since L1 use during the writing process ,generally speaking ,has no significant
correlation with the occurrences of transfer errors , teachers do not need to forbid students fro m thinking
in their m other tongue.Contrary to co m m on belief , thinking in L1 m ay even help the students with their
writing ,as we have discovered in text production for the senior group.

Fro m our study ,we discovered that the m ajority of errors at the lexicallevel resulted fro m a m ental
w ord-for-w ord translation. Therefore ,in the classroo m ,teachers need to e m phasize early on that
learning a foreign language does not m ean finding a w ord- for- w ord equivalence in another language ,and
they should help their students grasp language atlarger chunks , thatis , gradually getting beyond the w ord
level to the sentence level and then the discourse level.

6.Limitations and Future Research Directions


The m ost im portant limitation of this study lies in the unavoidable weakness of think- aloud
technique.Although the participants had been sufficiently trained before the think- aloud task ,it cannot
be guaranteed that the data elicited by think- aloud can reflect natural thinking.Though the participants
were required to verbalize w hatever was going through their minds ,and a retrospective interview was
carried out right after the think- aloud task ,the think- aloud protocols cannot 100 % report the inner
thoughts.In addition ,owing to the de m anding nature of think- aloud ,only those extrovert and talkative
participants are able to perform think- aloud tasks ,thus it is doubtful that the data obtained fro m such a
highly selective group is the sa m e as those w ho are screened out.Therefore , the conclusions deduced here
should be justified by studies based on larger sa m ple size.

A nu m ber of questions e m erging fro m this study have led us to propose the following directions for
further research. Firstly ,our study focuses on transfer errors ,w hile the m anifestations of negative
transfer also include avoidance and overuse. Further study is required to discover to w hat extent the
avoidance and overuse of certain structures are correlated with L1 thinking during the writing process.

Secondly ,w hen interviewing the participants about the source of the errors ,we found that for a few
sopho m ores ,the delay of key point does not result fro m a preference for“indirectness”in the language
and culture of Chinese , rather , they do not see co m posing as the creation of a w hole discourse.They are
even unfa miliar with the conventions of writing in Chinese ,just as there are m any English- speaking
students w ho lack the co m petence at the discourse level of writing (Brow n ,1981 ) .This pheno m enon
introduces the notion of develop m ental factors in L1 learner s writing. Mohan &Lo(1985)propose that
both in L1 and L2 writing ,sentence- level skills such as spelling ,punctuation and gra m m atical accuracy
are established earlier w hile the co m petence in organizing larger units of discourse is a later develop m ent.
Thus both develop m ental and transfer factors should be taken into consideration w hen we explain
learners difficulties in the organization of L2 discourse. Further research should investigate to w hat
extent each of the factors can account for the difficulty.

Notes
1.The participant w ho used no Chinese at all during the w hole writing process is excluded.

References
Bro w n ,R .1981.National assessments of writing ability.In C .Frederiksen &J.Do minic(eds. ). Writing :
Process ,Development and Com munication.New Jersey :Lawrence Erlbau m Associates ,Publishers.
Connor ,U .1996. Contrastive Rhetoric.Cam bridge :Cam bridge University Press.
Corder ,S.1967.The significance of learner s errors.International Review of Applied Linguistics V : 161-169.
Corder ,S.1971.Idiosyntactic dialects and error analysis.International Review of Applied Linguistics IX :149-
159.
Cu m ming ,A .1989. Writing expertise and L2 proficiency. Language Learning 39 : 81- 141.
Ericsson ,K .&H .Sim on.1987.Verbal reports on thinking.In C .F rch &G .Kasper(eds. ).Introspection
in Second Language Research.Clevedon :Multilingual Matters.
Gass ,S.1984. A review of interlanguage syntax :Language transfer and language universals. Language
Learning 34 :115- 132.
13
L1 Influence on L2 Writing :Process and Product   He Yanqun

Gass ,S.1987.The resolution of conflicts am ong co m peting syste ms :A bidirectional perspective. Applied
Psycholinguistics 8 :329- 350.
Halliday ,M .A .K .&R .Hasan.1976. Cohesion in English.London :Long man.
James ,C .1998. Errors in Language Learning and Use :Exploring Error Analysis. Addison Wesley Long man
Limited.
Kaplan ,R .1966.Cultural thought patterns in intercultural education. Language Learning 16 :1- 20.
Lay ,N .1982.Co m posing processes of adult ESLlearners. TESOL Quarterly 16 :406.
Lennon ,P.1991.Error :So me proble ms of definition , identification and distinction. Applied Linguistics 12 :
180- 196.
Li ,C .&S.A .Tho m pson.1981. Mandarin Chinese :A Functional Reference Gram mar.Berkeley ,California :
University of California Press.
Mohan ,B .A .& W . A .Lo.1985. Acade mic writing and Chinese students :Transfer and develop mental
factors. TESOL Quarterly 19 : 515- 534.
Nunan ,D .1992. Research Methodsin Language Learning.Cam bridge :Cam bridge University Press.
Schachter , J.1983.A new account of language transfer.In S.Gass &L.Selinker(eds. ) . Language Transfer
in Language Learning.Ro wley ,MA :New bury House.
Wen ,Q .F.&R .K .Johnson.1997.L2 learner variables and English Achieve ment :Astudy of tertiary- level
English majors in China. Applied Linguistics 18 :27- 48.
Wenzell ,V .E .1989.Transfer of aspects in the English oral narratives of native Russian speaker.In H . W .
Dechert & M .Raupach(eds. ). Transfer in Language Production.Norw ood ,NJ :Ablex.
Zobl ,H . 1979. Syste ms of verb classification and cohesion of verb- co m ple ment relations as structural
conditions on interference in a child s L2 develop ment. Working Papers on Bilingualism 18 : 43-57.
Zobl ,H .1984.Cross- linguistic generalizations and the contrastive dimension of the interlanguage hypothesis.
In A .Davies et al. (eds.).Interlanguage.Edinburgh :Edinburgh University Press.
Zobl ,H .1986. Word order typology , lexical govern ment and the prediction of m ultiple ,graded effects on
L2 w ord order. Language Learning 36 :159- 184.
蔡金亭, 1998 , 汉语主题突出特征对中国学生英语作文的影响, 《 外语教学与研究》第4 期。
陈万霞, 2002 , 英语学习者作文中的搭配错误分析, 《 解放军外国语学院学报》第1 期。
樊养才, 2001 , 非英语专业硕士生英语写作卷面分析, 《 西安外国语学院学报》第2 期。
郭纯洁、刘芳, 1997 , 外语写作中母语影响的动态研究, 《 现代外语》第4 期。
韩玲、周平, 2003 , 中国学生英语写作中的照应偏误分析, 《 外语研究》第2 期。
李德煜、张会欣, 2002 , 从思维差异看中国学生英语写作中的失误, 《 山东外语教学》第4 期。
李景泉、蔡金亭, 2001 , 中国学生英语写作中的冠词误用现象, 《 解放军外国语学院学报》第6 期。
娄宝翠, 2001 , 中国学生英语写作中的造词现象, 《 外语教学与研究》第1 期。
毛荣贵、 D .S.Houston , 1998 《,中国大学生英语作文评改》。上海:上海交通大学出版社。
王文宇、文秋芳, 2002a , 母语思维与二语写作, 《 解放军外国语学院学报》第4 期。
王文宇、文秋芳, 2002b , 母语思维于外语作文分项成绩之间的关系, 《 外语与外语教学》第10 期。
文秋芳, 2001 《,应用语言学研究方法与论文写作》。北京: 外语教学与研究出版社。
文秋芳、郭纯洁, 1998 , 母语思维与外语写作能力的关系:对高中生英语看图作文过程的研究, 《 现代外语》
第4 期。
左岩, 1995 , 汉英部分语篇衔接手段的差异, 《 外语教学与研究》第3 期。

Appendix 1 L1 use in different co m posing activities(20 students)


Co m posing Activities
Proficiency
TA (%) CC (%) SC (%) TP(%) CP(%) Total(%)
Sopho m ore
2.1 8 95 43 31 96 53
2.2 8 60 87 52 89 54
2.3 0 100 100 27 100 46
2.4 32 43 84 8 49 18
2.5 0 16 100 8 67 11

14
CELEA Journal 84

Co m posing Activities
Proficiency
TA (%) CC (%) SC (%) TP(%) CP(%) Total(%)
2.6 29 78 100 30 79 39
2.7 43 94 92 35 100 58
2.8 19 23 65 24 90 24
2.9 38 91 91 49 89 65
2.10 34 77 72 33 100 46
Senior
4.1 15 87 100 33 81 52
4.2 0 23 47 9 0 11
4.3 25 72 83 18 100 41
4.4 0 13 80 0 100 9
4.5 20 69 60 29 77 49
4.6 0 0 0 0 0 0
4.7 0 63 100 12 92 28
4.8 12 11 73 2 100 3
4.9 10 53 83 27 100 38
4.10 32 68 98 11 55 32

Appendix 2 Errors of two levels(20 students)

Lexical Syntactic Total


Sopho m ore
2.1 0.85 2.13 2.98
2.2 0 1.30 1.3
2.3 1.15 1.92 3.07
2.4 0.50 1.00 1.5
2.5 0 1.18 1.18
2.6 0.96 4.81 5.77
2.7 1.45 2.17 3.62
2.8 0.46 2.76 3.22
2.9 1.18 2.35 3.53
2.10 0 2.07 2.07
Senior
4.1 0.63 0.63 1.26
4.2 0.37 1.12 1.49
4.3 0.45 1.35 1.8
4.4 0.61 2.42 3.03
4.5 0 0.80 0.8
4.6 0.56 1.69 2.25
4.7 0 0.39 0.39
4.8 0.46 1.38 1.84
4.9 0.38 1.15 1.53
4.10 0.56 2.26 2.82
Values show the nu m ber of occurrences per 100 w ords.

15

You might also like