SEQUENCING IN SLA Phonological Memory , Chunking , and Points of Order

ion processes that tune the phonological system, at a higher level of representational unit, tune the grammar system: As learning progresses, as the L2 lexical categories are acquired, and as the system becomes tuned to L2 word sequences, so the L2 lexical auto-associative net begins to have a structure that is largely independent of the LI net. These processes of abstraction are just the same as those already described for LI—they involve positional analysis of word order. As the network grows, we see the emergence of grammatical regularities and even of overgeneralizations that resemble those of LI learners. Pretty much the same developmental sequences are found in L2 interlanguage as in LI acquisition (Johnston, 1985; the Natural Order Hypothesis in Krashen & Terrell, 1983). For example, just as a young child goes through No + X ("no is happy"), before no/not/don't K("they not working"), before analyzed don't ("she doesn't live there") in the acquisition of English negation, so also do Spanish, Japanese, and other ESL learners (Schumann, 1978). Long (1991) summarizes the general point as follows: The same developmental sequences are observed in the interlanguage of children and adults, of naturalistic, instructed, and mixed learners, of learners from different LI backgrounds, and of learners performing on different tasks Passage through each stage, in order, appears to be unavoidable As would be predicted if this definition is accurate, it also seems that developmental sequences are impervious to instruction, (p. 42) With increasing competence, so mean length of utterance and structural complexity increases. The natural development sequence is well charted and remarkably consistent across native learners: There is a fixed sequence of overlapping stages, each characterized by the relative frequencies of structures, which learners apparently have to traverse on the way to complete mastery of language (Crystal, 1987). Many skills are like this, indeed so much so that the phenomenon is crystallized in the English language: Trying to break a natural order is like "trying to run before you can walk." Chomsky (1988) used this argument to support the idea that language is an independent faculty separate from nonlinguistic cognitive abilities: As far as we know, the development of human mental capacity is largely determined by our inner biological nature. Now in the case of a natural capacity like language, it just happens, the way you learn to walk. In other words language is not really something you learn. Acquisition of language is something that happens to you; it's not something that you do. Language learning is something like Sequencing in SLA 101 undergoing puberty. You don't learn to do it; you don't do it because you see other people doing it; you are designed to do it at a certain time. (pp. 173-174) However, we need to clarify the "natural" in Natural Order: Does it refer to human biological nature or the nature of the world? It is too easy to slip into the erroneous belief that invariance of sequence of development is a necessary and sufficient index of innately given skills such as walking. It is indeed a characteristic of innate skills, but so also does it apply to a wide range of learned abilities. "There is no monopoly on tight chronologies and developmental sequences" (Winter & Reber, 1994, p. 138). For example, we are neither innately preprogrammed to read nor to do arithmetic—both have appeared too late in our cultural development to be evolutionarily given—yet there are characteristic stages of reading development (logographic then alphabetic then orthographic [see N. C. Ellis, 1994e; Frith, 1985]); and in mathematics (counting precedes addition precedes multiplication precedes integration, etc.). Sequences of development can as much, or even more, stem from epistemology, the structure of knowledge in the relevant problem-space, as they can learners' biological processing capacity and neural development. Dewey (1916) proposed the general thesis that the empiricist position is the proper default position and that nativism should only be adopted when the evidence against empiricism becomes overwhelming. If we adopt this good advice for the particular case of language acquisition research, then the default is that invariant developmental sequences of language acquisition inform us about the informational content of language and how more complicated structures arise from simpler, more basic forms when analyzed by general human learning abilities. It is no small understatement to say that both the data and the learning capabilities are multifaceted. There are many sources of constraint on word order in interlanguage: the structures assimilated from the learner's LI experience, the structures assimilated from the learner's L2 experience, and the attentional sequence in semantics that underlies canonical order strategies (Clahsen, 1984). We need first to look to what these rich sources of information in competition can produce in terms of acquisition sequences (MacWhinney, 1987) before we resort to explanations in terms of innateness. This is, of course, a highly contentious issue that is certainly not going to be resolved here. I do not wish to deny the problem cases for such an empiricist stance (e.g., Hylstenstam [1977] as discussed by Larsen-Freeman and Long [1991, p. 95], and the evidence for Universal Grammar in SLA as reviewed in White [1989], Cook [1994a, 1994b], and Eubank [1991]). I merely claim that a large part of L2 grammar acquisition is consistent with constructivist accounts and that there is merit in pushing these to see just how far they will go. Further discussion of the relations between constructivist and UG accounts of SLA is deferred until the concluding section of this paper. In interim conclusion, this section has presented arguments that suggest that ability to learn phonological sequences is at the center of vocabulary learning, idiom learning, and the acquisition of grammar. Humans differ in a wide range of core cognitive abilities, as they differ in ability to learn second and foreign languages. The next section considers whether individual differences in phonological sequence learning do in fact limit language learning skill.