In an earlier article (Comrie 1978) I argued, in my role as devil’s advocate, that the go - went alternation should not be treated as suppletive, but rather that there is a single underlying morpheme /g^wɔ̃/ from which both allomorphs can be derived by independently motivated rules. Now, as is well known, the devil will always find work for idle hands, so that his advocate has now turned up some further evidence in favor of this analysis.

In Comrie (1978), the g - w alternation in word-initial position was effectively considered only for go - went, since no underlying initial /g^w/’s were posited for other lexical items (other than transparent recent loans where phonetic [g^w] remains, e.g. Gwen, guano). However, there are several other instances of initial g - w alternation, e.g. guard - ward, guarantee - warranty, war - guerrilla; in addition, there are other velar-labial alternations, e.g. cow - bovine, come - anabasis, although I shall not here be concerned with these—they serve only to show how general the phenomenon is. This calls into question the phonetic environment posited for g^w Development in Comrie (1978:60) (giving [g] before a rounded vowel [w] elsewhere), and indeed the possibility of giving a phonetic environment, but does indicate that the analysis is not an ad hoc response to the go - went alternation.

In many regional varieties of English, for instance in the area around Sunderland and Newcastle-upon-Tyne, the citation form for go is gan, so that for such dialects one has an alternation gan - went. Speakers of such dialects have no feeling that gan - went is more or less suppletive than standard go - went (with which they are also familiar). Yet in these dialects the derivation of gan and went is much simpler than for standard go - went: while the standard language requires four rules for go and seven for went (Comrie 1978:62), such dialects require only one rule for gan and three rules for went, as can be seen in the derivations in (1):

(1)	gwæn	gwæn#d
		gwæn+t	Boundary Readjustment
		gwen+t	Vowel Shift
	gæn	wen+t	gw Development

(For Boundary Readjustment with verbs ending lexically in a lax vowel followed by a resonant, compare spell - spelled/spelt, spill - spilled/spilt, in the dialect also tell - [telt].) while such facts do not necessarily tell us anything about the standard language, they do indicate that there are dialects for which the nonsuppletive analysis is even more strongly justified.

Reference

Comrie, Bernard. 1978. On the go - went alternation: a contribution (?) to the Generative Phonology of English. In T. Ernst & E. Smith (eds.), Lingua Pranca. Bloomington, Indiana: Indiana University Linguistics Club, pp. 59-63.

	The Abstract Abstract—Edith Maxwell
	Spore Theory: A Contribution to the Study of the Biological Foundations of Natural Language—David Athletesfoot
	Son of Lingua Pranca Contents