In the last two years, two very interesting books about the interaction of phonology and syntax, Selkirk (1984a) and Kaisse (1985), have addressed the problem of characterising fast speech processes and made interesting proposals about the status of fast speech rules (FSRs) and their place in the grammar. Essentially, Kaisse's proposal is that fast speech rules belong to a separate subcomponent of the phonology, which is ordered after the sandhi rules subcomponent: while sandhi rules are sensitive to their syntactic environment, fast speech rules have only phonetic motivation. That is, they operate throughout a string, within as well as across words, independently of its structure (cf. also Rotenberg 1978; Hasegawa 1979).

In phonology, one of the generalisations that seems to hold true across most, if not all, languages is that the overall rhythmic pattern tends to be organised such that there is an alternation of strong and weak syllables (cf. among others, Hayes 1980, 1984; Prince 1983; Selkirk 1984). In other words, languages tend to avoid strings of adjacent strong syllables, as well as strings of adjacent weak syllables. These generalisations are expressed by clauses (a) and (b), respectively, of Selkirk's Principle of Rhythmic Alternation (PRA):

(1) Principle of Rhythmic Alternation (Selkirk 1984: 52)

a. Every strong position on a metrical level n should be followed by at least one weak position on that level

b. Any weak position on a metrical level n may be preceded by at most one weak position on that level

Of course, the underlying rhythmic patterns of a language are not always in conformity with the PRA.

The study of language acquisition during the first year of life is reviewed. We identified three areas that have contributed to our understanding of how the infant copes with linguistic signals to attain the most basic properties of its native language. Distributional properties present in the incoming utterances may allow infants to extract word candidates in the speech stream as shown in the impoverished conditions of artificial grammar studies. This procedure is important because it would work well for most natural languages. We also highlight another important mechanism that allows infants to induce structure from very scarce data. In fact, humans tend to project structural conjectures after being presented with only a few utterances. Finally, we illustrate constraints on processing that derive from perceptual and memory functions that arose much earlier during the evolutionary history of the species. We conclude that all of these machanisms are important for the infants to gain access to its native language.