This paper raises the question whether avoidance is an appropriate terrn to describe the linguistic behaviour in which the learner with particular Ll background tries to underproduce or underrepresent a particular L2 structure (Schachter 1974; Kleinmann 1977,1978). From a psychological standpoint, the term is clearly defined as a genuine phenomenon resulting from the organism's realization that specific forms of input data are avoided (on the levels of concrete representation and/ or abstract mentalization) simply because such forms are difficult to restructure or reorganize in the output * , and thus an alternative 'strategy' should be employed to fill in the consequential gap. However, the individual's resorting to this alternative 'strategy' does not imply his /her 'ignorance' of the avoided form, as misinterpreted by James (1980).2 Rather, the individual is said to have sorne sort of knowledge of the avoided forrn because it is obvious that one cannot avoid what one does not know, a truisrn that has repeatedly been emphasized by researchers like Seliger (1989). For this, reason, an attempt will be made to identify the sorts of knowledge that may lead to the nonuse of a given L2 structure on either of the two levels: concrete representation and abstract mentalization, or to its use but with erroneous manifestations from an L2 perspective, sirice recent research has shown that even Schachter's original study provides insufficient insights into the concept of avoidance (cf. Kaminoto et al. 1992). Furthermore, Kleinmann's account particularly of the passive construction in the case of Arabic-speaking learners of English, will be reconsidered in the light of this identification and the potential frequency how certain linguistic representations must be if certain assumptions about the mental operations that predetermine them are true.
how certain linguistic representations must be if certain assumptions about the mental operations that predetermine them are true. Thus, consideration of the natural consequences leads to a scrutiny into one of the major logical connectives that are familiar in the propositional calculus, viz. conditionals or'if... then'statements. Such connectives describe theoretic possibilities and possible worlds which, in turn, appear to provide this kind of description with perceivable substance. But how linguistic representations (or in fact anything) may be, and how they must be, are above all ontological matters, matters concerning what there is, that feed real conundrums.
The study falls into three main sections: the first section seeks to establish the general psycholinguistic principle which determines avoidance in the process of L2 acquisition. It will draw heavily on what the term 'avoidance'precisely means within the framework of behaviourist psychology, whose historical-conceptual antecedents can be traced back to the empirico-rationalist psychology of the seven-teenth century. A circumscription of the same paradigrns (Le. the para-digms of avoidance delineated by the behaviourists) with a cogni-tive aura will help to understand the phenomenon from a deeper perspective through their application to the process of learning or acquiring a particular L2 structure. The second section goes even more deeply into the ontological dimension of cognition that predetermines the general psycholinguistic principle discussed in the first section. It will thus be a conceptual extension of the sort of cognitive ordi-nance that the avoidance of a given L2 structure presupposes. From an epistemological standpoint, the ontology of this ordinance will be explained in terms of Bergson's notion of 'intuitive knowledge' but with some modifications. The third and final section will be an ex-tensive critical survey of an'avoidance classic', namely, Kleinmann's account of the passive construction in the case of Arabic-speaking learners of English and some of Schachter's speculative conclusions. It should be noted, however, that the choice of the passive construc-tion here is a matter of pure coincidence, and that the conclusions drawn in this critical survey are to be taken as a typical exemplifica-tion applicable to any structural device that exists in any two lan-guages instantiating parametric variation in that device.

Exemplaires