Learning Japanese as a Second Language: A Processability Perspect ...

‘black box theory’ (Ellis, 1985, p. 265), due to insufficient evidence and the untestable characteristics of the hypotheses. Furthermore, it has many theoretical shortcomings in the explanation of its hypotheses.

The next model, the multidimensional model, and its successors(i.e., predictive framework and PT), provide clear definition of the stages, unlike ‘i +1’. They are also open to validity testing.

2.1.4. From Multidimensional Model to Predictive Framework

Multidimensional Model

The ZISA Group’s multidimensional model (Clahsen et al., 1983; Meisel et al., 1981; Pienemann, 1980) was one of the first SLA studies to explain IL in terms of the learners’ ‘processing’ skills, abandoning the target language orientation that was dominant in the 1970s. This is the model from which PT originated, a development that will be discussed later.

The ZISA project started in the late 1970s with a cross-sectional study of 45 adults and a two-year longitudinal study of 12 adults. It focused on the study of German as a second language (GSL) by immigrants from a Romance language background. The model introduced two distinct dimensions of learners’ IL: the developmental feature axis and the variational feature axis.

Developmental Features

The ZISA Group proposed that learners’ development should be explained in terms of the features observed in their production because all L2 learners undergo developmental stages in a fixed order. This idea is supported by the GSL empirical study of word order acquisition rules: Both adults and children showed five distinct stages of developmental sequence, and they accumulated new rules over each of five stages (Clahsen, 1980; Meisel et al., 1981; Pienemann, 1980).

What is different from Krashen’s ‘natural order hypothesis’ is that there is an explanation for each stage and a reason for the IL developmental sequence. Each of the stages is realised by the combination of