Is the considerable overlap between lexemes and lexical items that is a feature of English found in all languages? This question is really twofold. Firstly, are there languages where the proportion of lexical items that are not lexemes is much higher than in English? We might call these ‘idiom-heavy languages’, because relatively many of their lexical items would be phrases rather than words. Secondly, are there languages where the proportion of lexemes that are not lexical items is much higher than English? We might call these ‘neologism-heavy languages’, because relatively many of their words would be items constructed and interpreted ‘on-line’, like the English sentences at (1) and (2), rather than through identification with remembered items.

A possible example of a language of the first kind is Vietnamese, which has no inflectional morphology and almost no bound morphemes (roots or affixes), and where any distinction between morphological compounds and syntactic phrases is dubious. In Vietnamese, therefore, nearly all polymorphemic lexical items must be analyzed as phrasal idioms rather than lexemes (either compound or derived). Among languages that are likely to be more familiar to readers, French too is relatively ‘idiom-heavy’. Many concepts that are expressed by compound nouns in English are expressed by phrases in French:

It is not that French lacks compounds: for example, rouge-gorge ‘robin’ (literally ‘red-throat’), gratte-ciel ‘skyscraper’ (literally ‘scrape-sky’), and essuie-glace ‘windscreen-wiper’ (literally ‘wipe-screen’). But it is notable that these compounds are all exocentric (a robin is not a kind of throat, and a skyscraper is certainly not a kind of sky). In French, endocentric nominal compounds are relatively scarce by comparison with English; in their place, French makes greater use of phrasal idioms.

Examples of languages of the second kind are the varieties of Inuit, or Eskimo, in which many items whose meaning must be glossed by means of a sentence in English have the characteristics of a morphologically complex lexeme (or a word form belonging to such a lexeme) rather than of a larger syntactic unit. In Eskimo, many more lexemes than in English have the entirely predictable and therefore unlisted character that I ascribed to adverb lexemes such as dioeciously. It is as if Eskimo chooses to exploit the morphological route in forming many complex ex-pressions, where many languages would opt for the syntactic route.

Vietnamese and Eskimo represent, if I am right, minimizing and maximizing tendencies in the grammatical and lexical exploitation of morphology, with English somewhere in the middle. Moreover, most linguists would probably agree that the aspects of Vietnamese and Eskimo that I have emphasized render them rather untypical of human languages in general. Does that mean that, other things being equal, languages exhibit a tendency for lexical items and lexemes to converge? If so, why? Are the factors of brevity and versatility sufficient to explain it? These questions have scarcely been raised in linguistic theory, let alone answered. To pose them in an introductory textbook may seem surprising. I hope that a few readers, encountering them at the outset of university-level language study, may take them as a challenge for serious investigation!