КАК СИНТАКТИЧЕСКИЕ СТРУКТУРЫ ЗАБЫВАЮТСЯ СО ВРЕМЕНЕМ: ЭМПИРИЧЕСКОЕ ИССЛЕДОВАНИЕ ПО ЗАБЫВАНИЮ СТРУКТУРЫ ВОПРОСА В АНГЛИЙСКОМ ЯЗЫКЕ

1. Introduction

Language attrition is described as the phenomenon that learners’ ability of using an acquired language regresses with time due to cease or reduction of use. Ni [8] explored eight factors namely, Proficiency level prior to attrition (PLPA), Time length since onset of attrition (TOA), exposure to target language, age at onset of language attrition, foreign language instruction, affective factors, literacy and gender affecting foreign language attrition based on a large body of review of relevant literature. Among them, TOA has been found to influence the process and rate of language attrition and to be a determinate factor. However, how will TOA pose its influence on English syntactic structures? It is of great significance and convenience to explore the interrogation attrition of participants with different TOAs, and thus make clear how TOA functions in the attrition process of English interrogations.

Before such an investigation is made, however, at least two issues should be taken into consideration. First, how long a period is the valid spectrum for language attrition? Second, how do we plot the valid spectrum into several intervals when assessing different degrees of attrition for participants with different intervals of language attrition?

The valid spectrums and intervals that the previous empirical studies adopted varied with research designs. Generally speaking, the TOAs that the previous studies involved cover four kinds of periods including 3 months, 1 to 4 years, 25 to 35 years and 50 years [7]. Cohen [3] set ‘3 months’ as the valid spectrum for attrition since the European students’ summer vacation lasts 3 months. However, since the spectrum is too short, intervals are not subdivided further and the results of these studies thus diverge to a great extent. Allendorff [1] set ‘18 months’ as valid spectrum when investigating three children’s attrition, and found their language proficiency were attrited after 18-month attrition. Weltens et al. [10] set ‘4 years’ as the valid spectrum, with ‘2 year’ and ‘4 years’ as intervals respectively. However, there were also some empirical studies with attrition spectrum being above 5 years. Hansen [4] adopted ‘35 years’ as the valid spectrum with ‘5 years’ and ’25 years’ as intervals while Bahrick (1984) adopted ‘50 years’, also with ‘5 years’ and ’25 years’ as boundaries. Ni [8], however, divided the ‘30 years’ valid spectrum into three intervals with the adoption of the model of decision tree. Exactly put, interval one covers 1 to 4 years; interval two ranges from 4 to 8 years and interval three is above 8 years. Just following this kind of division, the present study assorted 243 participants into three groups with ‘4 years’ and ‘8 years’ as boundaries. To be exact, the first group consisted of 149 participants whose TOAs range from 2 to 4 years. The TOAs of second group ranged from 5 to 8 years and those of the third group were all above 8 years.

The research questions of this study are proposed as shown below.

What are the differences of attrition on people with different TOAs? How about the correlation between TOAs and interrogation attrition? Can significant difference be detected in the attrition of English interrogations for subjects with different TOAs?

2. Research Method

A cross-sectional paradigm and the technique-accuracy order were involved in the present study.

2.1 Questionnaire

In the questionnaire, a grammaticality judgment task was involved. The task included 40 sentences altogether, grammatically correct or incorrect. Among the 40 items, there were 4 groups of items involving 10 interrogations with "be" verbs, 10 with auxiliary or Modal verbs, 10 with negative words and 10 with embedded subjects. In order to minimize the interference and maximize the reliability of the research, these 40 items were randomly arranged in the questionnaire. Besides, a five-scale Likert assessing system was adopted so that the participants’ attrition can be examined more precisely.

2.2 Participants

The target participants of this study were those who had graduated from non-English majors of a college or a university and have worked at least for more than 2 years. They hardly used English in their work and never received any English training or master program study after graduations. What’s more, we consulted Chinese National English Teaching Syllabus and found that the target participants had been taught all the interrogative structures involved in the questionnaire even in their junior middle schools.

2.3 Reliability and validity of the study

31 sophomore college students majoring in laws in Jiangsu Police College were chosen as the participants of the pilot study. According to the pilot study, the Cronbach’s alpha coefficients of the items in every group range from 0.70 to 0.81, indicating there is strong consistency of results across items in every groups of the questionnaire.

3. Result

3.1 Comparison of the attrition of participants with different TOAs

Table 3-1: Mean Scores of 3 Groups in terms of TOA in Final Survey

Note: Group 1: 2 years≤TOAs of Participants≤4 years

Group 2: 4 years＜TOAs of Participants≤8 years

Group 3: TOAs of Participants＞8 years

Table 3-2: One-way ANOVA for 3 Groups in terms of TOA in Final Survey

Table 3-3: Multiple Comparisons between Groups in terms of TOA in Final Survey

In table 3-1, the mean scores and other descriptive statistics of the three groups are exhibited. The mean score of group 1 (participants whose TOAs range from 2 to 4 years) is 3.91. The mean score of group 2 (participants whose TOAs range from 5 to 8 years) is 3.89, and that of group 3 (participants whose TOAs are more than 8 years) is 3.83. It manifests the participants’ performance on interrogative structures declines with increasing TOAs, or, lengths of disuse.

With a one-way ANOVA, it is found in table 3-2 above, sig. value is 0.522, which is higher than 0.05. It leads to the conclusion that there is no significance difference among the three groups in a general sense. To put it in another way, participants with different TOAs have roughly enough mastery of the interrogation items in the survey. Meanwhile, As indicated in table 3-3, the differences between group 1 and group 2, group 1 and group 3 as well as group 2 and group 3 are not significant in a statistical sense (p=0.667>0.05, p=0.261>0.05 p=0.462>0.05). Roughly speaking, participants with TOAs being longer than 8 years still own the same mastery of the interrogative structures as the participants with TOAs being shorter than 4 years.

According to the observed data above, there were no significant differences among groups with different TOAs while participants’ performance on interrogative structures declined with increasing years after graduation, i.e., TOAs.

Whereas, how, and to which extent do participants’ TOAs and their attrition relate to each other? A Correlation Analysis is thus run to figure out this issue.

3.2 Correlation between participants’ TOAs and their attrition

It was worth noting that only 169 participants who passed CET-4 or/and CET-6 were taken into the analysis in order to minimize the interference caused by the variable, i.e., different PLPAs.

Table 3-4: The Distribution of Participants and Their TOAs

In table 3-4, the distribution of the 169 participants, their TOAs as well as their mean scores in final survey are presented, with 41 participants who didn’t pass CET-4 and 33 participants who didn’t take CET-4 excluded. From the table, there are 46 participants with TOAs being 2 years, 43 with TOAs being 3 years, 18 with TOAs being 4 years, 11 with TOAs being 5 years, and so forth. It shows that the participants’ syntactic proficiency declines with their increasing TOAs.

Table 3-5: PCC for Participants’ TOAs and Their Scores

** Correlation is significant at the 0.01 level (2-tailed).

The statistical data in table 3-5 indicates that the results are significant at the 0.01 level with Pearson Correlation Coefficient (PCC) r being -0.805 and sig. value being 0.002 (2-tailed), which means there is highly inverse correlation rate between participants’ TOAs and their scores. To be exact, the shorter TOA one owns, the higher score he or she will get in interrogation test, and vice versa.

By this token, TOA did play an important role in affecting the process of attrition although there were no significant differences among the 3 groups with different TOAs. To be more specific, participants with short TOAs had a better retention than those with longer ones, to some extent.

4. Discussion

This finding of the present study, on the one hand, is in agreement with Hansen’s [4] results as well as Graham’s (1990) findings, who argue that learners’ performance on target language skills declines with their increasing TOAs, or, lengths of disuse. On the other hand, the finding also differs from those of precedent studies a little. Tomiyama's research [9] reveals ‘4 years’ is a sensitive boundary after which vocabulary and fluency are attrited to a great extent. Ni [8] also maintains that participants with TOAs ranging from 4 to 8 years are most sensitive to attrition. However, no significant difference was spotted between participants with TOAs being longer and shorter than 4 years from the present study.

What cause the abnormal finding of the study?

First, in the circle of language attrition, it is commonly accepted that syntactic knowledge is relatively more invulnerable to attrition than other knowledge, such as lexicon, phonology and fluency [5]. In other words, syntactic attrition is less significant than lexical attrition. Therefore no significant differences are spotted between the performances of participants with TOAs being longer and shorter than 4 years in terms of syntactic level.

Second, the numbers of respondents with different TOAs are disproportionably distributed in the present study. To be more precise, there are 149 participants with TOAs ranging from 1 to 4 years, while there are just 65 participants with TOA ranging from 4 to 8 years, and only 29 participants above 8 years. The disproportionable distribution inevitably poses its influence on the data and thus interferes with the findings of the study.

Third, Ni [8] asserts at least eight factors are identified in affecting foreign language attrition, namely, PLPA, TOA, exposure to target language, age at onset of language attrition, foreign language instruction, affective factors, literacy and gender. What’s more, it is reviewed that the speed of attrition depends on a number of factors such as the difficulty of the learned material, how meaningful the material is to the subject, representation of material, and other physiological factors. The 243 participants involved in the present study are unequally distributed in terms of PLPAs, foreign language instruction, age at onset of language attrition, gender and so forth. Nevertheless, in the study they are classified into 3 groups just with their TOAs as a sole criterion. Factually, participants in the same group vary from person to person in terms of PLPA, foreign language instruction, age at onset of language attrition and etc [6]. These factors are absolutely bound to influence the results. The present study, however, does not take interactions of these different factors into consideration and surely suffers from the fallacy of simplification.

Fourth, according to Ni [8], a critical threshold exists in the EFL learner’s original language levels. The critical threshold is a point, or a level, above which the learner has a stable enough mental representation of some linguistic elements or structures that are stable and resilient to loss, below which attrition is rapid and extensive. In the survey of the research, all of the participants involved are college graduates and almost all of them own an original language level above the critical threshold. That is to say, almost all of them have a stable enough mental representation of English interrogations that are stable and resilient to loss. In this sense, the insignificant difference among three groups with different TOAs is comprehensible.

5. Conclusion

In the present study, no significant difference in interrogation attrition is observed among participants with different TOAs, although their performance on interrogative structures declines with increasing TOAs, i.e., lengths of disuse.

In the light of the analysis and discussion of interrogation attrition, some tentative suggestions are proposed for English learners and teachers to tackle the issues of attrition and improve learning and retaining.

The group of people whose acquired language skills are most vulnerable to attrition was identified as subjects with TOAs being more than 8 years. Accordingly, this group should deserve much more attention. On the one hand, teachers are expected to introduce more efficient mnemonics and more useful strategies that can produce strong memory traces to learners. On the other hand, learners themselves should be encouraged to develop their own methods to strengthen their retention of English skills.

However, methodologically speaking, attrition, by its nature, is rather difficult to be investigated empirically. Hence, a longitudinal paradigm is preferable in language attrition research, where the learner’s dynamic process of acquisition and attrition can be traced. Thus a longitudinal method should be adopted as a best way to examine the learner’s attrition in future research.