The purpose of this study was to explore how syntactic skill was maintained or lost by L2 Mandarin Chinese learners over time. In addition, this study endeavored to discover how a learner's L1 affects the attrition process of word order in Mandarin Chinese. To find out how certain Chinese syntactic structures were subject to attrition over time and how syntactic errors could be attributed to L1 transfer, five types of Chinese syntax that either resembled English, were very different from English, or had no counterpart in English were selected. They included subject-verb-object sentences, modifiers before modified, time and other adverbial clauses, and object-raising in Chinese specific ba construction. Twenty-four university students of Chinese-as-a-second-language speakers, who intensively learned and used Mandarin Chinese in a host culture setting for 16-22 months, participated in this study. By the time participants were tested a second time, they had discontinued regular usage of the L2 for 12 to 17 years. To find out how L2 syntactic attrition developed over time, participants were divided into three groups according to the year of departure from the L2 environment. They were also grouped into two groups by the length of time in the L2 setting to examine whether exposure time to the L2 affected the maintenance of overall L2 syntactic skill. The results indicated that the subjects retain a fair amount of their language education within the first couple years of discontinued regular L2 usage. In the meantime, it was found that the extra six months exposed to the L2 does not extend the long-term maintenance of overall L2 syntactic skill. The results did not show that the distance of structural properties between the learner's L1 and L2 necessarily predicted patterns of regression towards L1 syntactic ordering. Instead, the frequency of use, how often the structure appears in daily interaction with the target society and how well the syntactic structure was acquired in the first place, played a greater role in predicting whether the structure will likely be forgotten.

The purposes of this study were (a) to examine two groups of healthy older Dutch English bilingual migrants in a New Zealand setting to investigate whether they were showing signs of L2 attrition with accompanying L1 reversion post-retirement, and (b) to identify possible factors which might play a role in the incidence of any L2 attrition and concomitant L1 reversion. Previous research has focused on similar groups of migrants in the Australian context, while New Zealand based research has focused on language shift and maintenance amongst three generations of migrants. The research design involved an analysis of sociolinguistic life, using questionnaires. These included self-assessments of L1 and L2 proficiency at three key times: on arrival in New Zealand, at time of ultimate attainment and post-retirement. Further, an analysis of assessments of respondents’ L1 and L2 proficiency pre-and post-retirement completed by interviewees’ adult children moderated respondents’ self-reports. The findings revealed a considerable overlap between participants’ self-reports and assessments by their adult children. The study also revealed a relationship between participants’ level of prior education and their ultimate attainment in the L2, with those who had come to New Zealand having learned English at Secondary School English very likely to have achieved a “good” or “very good” level of L2 proficiency. Conversely, those who had not learned the L2 at secondary school prior to arriving in New Zealand, were less likely to have achieved a “very good” level of ultimate attainment as evident both from self-reports and assessments by adult children. The design also included a linguistic analysis of elicited free speech. Data focused on key indicators of age, gender, social class, prior education, occupation and predominant linguistic environment pre- and post-retirement. Free speech was examined for code-switching, response latency and L1 structure in respondents’ spoken L2. Results indicated that a majority of respondents showed minimal if any signs of L2 attrition with concomitant L1 reversion, both as evidenced by their spoken L2 and as indicated by self-reports and assessments by adult children. Any signs of L2 attrition which were found appeared linked to respondents’ level of prior education and L2 proficiency on arrival in New Zealand. Being exposed to a predominantly L1 social environment post-retirement also appeared to result in a lifting of the threshold for L2 lexical items, resulting in a slightly increased response latency in the spoken L2. Three participants said they experienced some problems expressing their healthcare needs to medical professionals, to the extent that they were searching for words. All stated they “got there in the end” but needed more time to paraphrase their health needs. Two subjects avoided the use of the L2 during the interview, even when prompted in English. Three respondents engaged in significant codeswitching from L2 to L1 and vice versa, with two engaging in what Muysken (2000) terms “congruent lexicalisation”. Adult childrens’ reports indicated that the respondents in question had always spoken in this manner, but to a greater extent now, post-retirement. Overall, a number of the healthy older subjects interviewed for the study were showing some signs of increased response latency and lexical retrieval problems when expressing themselves in the L2, but none to the degree that they were no longer able to communicate in that language.

The purposes of this study were (a) to examine two groups of healthy older Dutch English bilingual migrants in a New Zealand setting to investigate whether they were showing signs of L2 attrition with accompanying L1 reversion post-retirement, and (b) to identify possible factors which might play a role in the incidence of any L2 attrition and concomitant L1 reversion. Previous research has focused on similar groups of migrants in the Australian context, while New Zealand based research has focused on language shift and maintenance amongst three generations of migrants. The research design involved an analysis of sociolinguistic life, using questionnaires. These included self-assessments of L1 and L2 proficiency at three key times: on arrival in New Zealand, at time of ultimate attainment and post-retirement. Further, an analysis of assessments of respondents’ L1 and L2 proficiency pre-and post-retirement completed by interviewees’ adult children moderated respondents’ self-reports. The findings revealed a considerable overlap between participants’ self-reports and assessments by their adult children. The study also revealed a relationship between participants’ level of prior education and their ultimate attainment in the L2, with those who had come to New Zealand having learned English at Secondary School English very likely to have achieved a “good” or “very good” level of L2 proficiency. Conversely, those who had not learned the L2 at secondary school prior to arriving in New Zealand, were less likely to have achieved a “very good” level of ultimate attainment as evident both from self-reports and assessments by adult children. The design also included a linguistic analysis of elicited free speech. Data focused on key indicators of age, gender, social class, prior education, occupation and predominant linguistic environment pre- and post-retirement. Free speech was examined for code-switching, response latency and L1 structure in respondents’ spoken L2. Results indicated that a majority of respondents showed minimal if any signs of L2 attrition with concomitant L1 reversion, both as evidenced by their spoken L2 and as indicated by self-reports and assessments by adult children. Any signs of L2 attrition which were found appeared linked to respondents’ level of prior education and L2 proficiency on arrival in New Zealand. Being exposed to a predominantly L1 social environment post-retirement also appeared to result in a lifting of the threshold for L2 lexical items, resulting in a slightly increased response latency in the spoken L2. Three participants said they experienced some problems expressing their healthcare needs to medical professionals, to the extent that they were searching for words. All stated they “got there in the end” but needed more time to paraphrase their health needs. Two subjects avoided the use of the L2 during the interview, even when prompted in English. Three respondents engaged in significant codeswitching from L2 to L1 and vice versa, with two engaging in what Muysken (2000) terms “congruent lexicalisation”. Adult childrens’ reports indicated that the respondents in question had always spoken in this manner, but to a greater extent now, post-retirement. Overall, a number of the healthy older subjects interviewed for the study were showing some signs of increased response latency and lexical retrieval problems when expressing themselves in the L2, but none to the degree that they were no longer able to communicate in that language.

This report addresses the lack of progress in the field of Second Language Attrition (L2A). Review of L2A history and literature show this to be cause by lack of appropriate data. Five criteria for appropriate data are suggested and a corpus of L2A data (57,000 words, spoken Spanish) which meets the criteria is presented. The history of the corpus is explained in detail, including subject selection, instruments and methods of collection, and markup -- XML was used to annotate the corpus with nineteen categories of speech errors, adapted from Nation's (2001) "Learning Vocabulary in Another Language." An example analysis of how the corpus can be used for L2A research is provided with step-by-step instructions on writing scripts for data extraction and post-processing in the Perl language. Source code is included in the text. Complete beginners tutorials on the XML and Perl languages are included in the appendices. The report also introduces a website, developed specifically to host the corpus, where researchers may register, download the corpus and share work they have done with the corpus. All files used in the example project, as well as this report, are available for download at the website. Findings from the example analysis support Plateau Phases, the Regression Hypothesis and suggest the Threshold Hypothesis does not apply to marked forms. This shows the corpus to be of great value to the L2A research community.