ITL - International Journal of Applied Linguistics - Volume 149, Issue 1, 2005

This paper explores ‘period of residence’ as a factor in the maintenance of an immigrant language, based on the example of Hungarian in Australia. Hungarian speakers arrived in Australia from several different source countries including Hungary, Romania (Transylvania), and areas of the formers Yugoslavia (Vojvodina) and Czechoslovakia (Slovakia). The distinct waves of Hungarian speaking migrants to Australia - 1938-40; 1947-54; 1956-57; 1960s, 70s and 80s; and 1990s - reflect the close connection between sociopolitical events and immigrant source countries for speakers of Hungarian.

The data for the study comprises interviews with 22 families, encompassing all vintages and source countries noted above, supplemented by two focus groups. The study demonstrates that ‘period of residence’interacts with a number of other factors, notably country of origin, reasons for migration, and the prevailing attitudes and policies towards the reception and integration of immigrants in the host country at the time. It highlights the contradictory and ambivalent effects on language maintenance of situations of multiple identity and individual responses to conflict situations.

Though scholars interested in ESL writing pedagogy would like to believe otherwise, most students in high school and college English language courses are not, in fact, required to write academic prose. Rather they are, by and large, asked to write the personal narrative, a genre of prose similar in very fundamental ways to the prototype story. The personal narrative is a challenge because, like the prototype story, it is more than a mere report of events in past time. It is a report of events in past time infused with, what has been identified as the defining component of the story, the evaluative component. The evaluative component enables the writer to represent the dissimilar impact of successive narrative events to the developing theme. In this paper, I will address one significant means of signaling the dissimilar impact of successive narrative events, and the problems learners have with its proper use. In the prototype personal narrative/ story, the temporal movement effected by successive narrative events is not the same. There is non-uniformity in the pace and quality, of the temporal movement, from one narrative event to the next. I will provide a description of the sentence-level mode of implementing this particular means of evaluation in the prose of skilled writers, and examine a representative sample of ESL data to show the difficulty learners face with its implementation. This understanding will be used to generate suggestions for pedagogy which specifically address this area of difficulty.

The situation in many composition classes in Singapore is characterized by scaffolding (Vygot-sky, 1962; Bruner, 1985), where students are given explicit guidance, particularly concerning the organizational structure of the essay. With scaffolding, there is a concern that the students will produce similarly-structured essays. How then does one differentiate a good essay from a less well-written one? Using Halliday's clause-complex framework (Halliday and Matthiessen, 2004), our analysis shows that while the better writers display a greater mastery of clause-complex structures, the weaker writers rely heavily on simplex constructions. In addition to scaffolding, we argue that the appropriate use of clause-complex structures should be incorporated as part of the writing lesson.

The present study focuses on spoken and written data in the British National Corpus (BNC). Based on a review of recent studies on English relative clauses, we formulated a Universal Processing Hypothesis (OS >OO>SS> SO) as target hypothesis to be validated using a corpus data approach. A computer program was designed to calculate the frequency of appearance of the four types of relative clauses (OS, OO, SS, and SO). The results indicated this hypothesis to be a valid predictor of frequency of appearance of relative clauses in the domain for written corpus texts. However, it is not supported in context-governed spoken material. Limitations of the present investigation and the direction of future research are also discussed.