The impact of autonymy on the lexicon

In this paper, I wish to examine the relations between the English lexicon and the discourse phenomenon that consists in quoting or mentioning a linguistic object, as in The word „why‟ was spelt incorrectly (henceforth, ‗autonymy‘). Autonymy turns out to underlie the formation of a substantial number of lexemes, all of which may be assumed to derive from a quoted word or sequence rather than from an ‗inert ‘ item. This is a word-formation process that has been largely overlooked by lexicographers, who have tended to consider that the input to word formation was provided by lexemes, stems and affixes, in other words, by elements in the system rather than by occurrences in discourse. I will start by defining autonymy, go on to consider the best way to tackle the question how autonyms can find a way into the lexicon, then quickly go over the existing literature (mainly devoted to so-called ‗delocutives‘). When that has been done, I will review what I regard as the four major guises in which autonyms can serve as input to word-formation processes. It will turn out that a general consequence of the incorporation of autonyms into the lexicon is that they lose most of their reflexivity, i.e. they stop denoting something very much like themselves (as in We don't have any time to look into the why of things). Yet, the fourth category examined will raise the intriguing question whether the lexicon does include some autonyms after all. This hypothesis is supported by some lexicographical data, but also, more decisively, by the finding that, in the case of those items that derive from words or phrases whose utterance carries illocutionary force (e.g. a bravo), it is often difficult to draw a precise line between what is a purely reflexive occurrence (a one-off autonym) and what is a minimally reflexive occurrence (a lexeme derived from an autonym). The question also arises whether one and the same lexical item can be instantiated now as an ordinary lexeme, now as an autonym. If my hypothesis is correct, it sheds new light on the relationships between discourse phenomena and the language system: not only do the former, trivially, nourish the later, but they also connect with it in such a way that certain items do not clearly belong to one domain rather than the other