Ross (1967) showed that in relative clauses, not only may the WH-phrase be fronted, but an NP or PP containing the WH-phrase may also be fronted, as shown in (1):

1. (1)

   1. a. This is the child [who]i I’ve been hearing stories about ti.

   2. b. This is the child [about whom]i I’ve been hearing stories ti.

   3. c. This is the child [stories about whom]i I’ve been hearing ti.

   Ross called this phenomenon “pied piping”. His statement of the pied piping convention is given in (2).

2. (2) Any transformation which is stated in such a way as to effect the reordering of some specified node NP, where this node is preceded and followed by variables in the structural index of the rule, may apply to this NP or to any non-coordinate NP which dominates it, as long as there are no occurrences of any coordinate node, nor of the node S, on the branch connecting the higher node and the specified node. (1967:114)

Notice that Ross’s statement applies to any transformation moving an element over a variable. Thus, the prediction is that WH-questions and relative clauses should behave similarly with respect to pied piping. This is not the case, as pointed out by Bresnan (1976:37). Questions seem to be much more limited in what can be pied piped than are relative clauses.

A competence grammar of English must account for the English speaker’s ability to predict the shape of the regular past tense morpheme on the basis of the stem-final segment, faced, for example, with a neologistic verb like *rolk or *glag. But whether efficient processing of known words routinely utilizes this creative ability is another matter. Efficient on-line processing may favour selection from a store over computation. To what extent selection and computation are balanced in on-line production may even be different from the balance required in efficient comprehension.

It would appear therefore that not only is behaviour not a criterion for possession of knowledge — it is in fact only one kind of evidence for it — but also that the constructs required in characterizing competence cannot serve as the only criteria for determining the forms of processing. The evidence from both domains can support but not disconfirm the hypotheses of the other. Nevertheless, the grammars being produced within the generative paradigm are beginning to have the requisite richness of detail that must underly the capacity for language use. As such they provide independently motivated and sufficiently elaborate hypotheses regarding the structures and processes underlying production, allowing the production theorist a preliminary distinction between crucial facts and those best considered irrelevant.

Une série d’articles sur les consonnes épenthétiques est apparue dans les pages de cette revue ces dix dernières années: Walker (1978, 1981), Morin (1980), Singh (1980), Wetzels (1985), Piggott et Singh (1985) et Picard (1987), passant progressivement d’un problème très particulier (le statut phonologique en ancien français des consonnes épenthétiques issues du gallo-roman) à des théories générales sur l’épenthèse. Un certain nombre de données et d’analyses antérieures n’ont malheureusement pas été utilisées par le dernier de ces auteurs, conduisant ainsi à des généralisations hâtives et probablement fausses. Il est peut-être temps maintenant de rappeler quelques-unes des analyses antérieures et de revenir sur certaines de ces généralisations.