Lexicalization of negative senses

The goal of this project is to identify which negated concepts, connected mainly to states and events, are expressed lexically across languages, cf. (i) English dunno < I don't know or (ii) Tundra Nenets jexerasj 'not know'. Both (i) and (ii) can be semantically decomposed into a negative component and a positive sense. Following (Brinton and Traugott 2005, Moreno-Cabrera 1998), such forms are considered instances of lexicalization. The term is used here in a synchronic sense. Lexicalizations of negation are mentioned in numerous works, some examples include Jespersen (1917: 13, in passim), Croft (1991), Payne (1985), van Gelderen (2008). De Haan (1997), Palmer (1995), van der Auwera (2001) cover lexicalizations of modal senses such as 'not be able to', 'need not', etc. Eriksen (2011) discusses negation strategies in non-verbal sentences which also include lexicalized expressions meaning 'not be'. However, a systematic cross-linguistic survey of lexicalized negative senses is missing both in the literature on negation as well as in work on lexical typology, cf. (Evans 2010, Goddard 2001, Koch 2001, Koptjevskaja-Tamm 2008). Zeshan (2004), on the other hand, offers a detailed discussion of irregular negative senses in sign languages. These expressions can be also viewed as lexicalizations and are compared with the results of the current study.

          The data used here come from a sample of 97 geographically and genealogically distinct languages and also from three family-based samples which cover Slavic, Uralic and Polynesian.

          Formally, lexicalizations of negation appear to be of two kinds. The first one is illustrated by (i) above; it represents a fused form where a phrase containing the negator don't has fused with the verb form know. The grammatical morpheme in the fused form is frequently, though not always, a negator. In the second kind, cf. (ii), the form jexerasj cannot be further segmented into separate morphemes and shows no formal relation to the negative morpheme in Tundra Nenets or the expression for the corresponding positive sense, cf. (1) below. In the current data set, lexicalizations of the second kind prevail. It should also be noted that these words are sometimes the only negators of the corresponding positive senses but there are also cases where they co-exist with a regularly negated positive verb cf. (2) below. There are also cases where they simply lack an affirmative correspondent.

          My database contains lexical expressions for 65 negative senses which can be grouped into broader semantic domains. There are also a handful of senses which are regularly lexicalized cf. (3) below. Perhaps not surprisingly, all of the semantic and grammatical domains identified by Zeshan (2004: 50) as being coded by irregular negatives in sign languages are also observed as lexicalizations in spoken languages. These domains are: cognition (not know, not understand), emotional attitude (not want, not like), modals (cannot, need not), possession/existential (not have, not exist), tense/aspect (did not, not finished), evaluative judgment (not right, not enough). One domain that tends to be lexicalized in spoken languages but is not reported by Zeshan for sign languages is labeled here non-utterance; it is represented by senses such as 'not talk', 'not tell/inform', 'not mention', cf. (4) below for an example.

          Lexicalized expressions for 'not exist' are so common, that it is easier to identify areas where they do not occur. In the current sample such areas are Western Europe, South East Asia and southern and central parts of South America. Lexicalized expressions of the remaining senses appear to occur less commonly in Africa than in the other macro-areas. As regards the micro-samples, Slavic and Polynesian languages show a preference for lexicalization of 'not want' while 'not know' is commonly lexicalized in Uralic. Further on, within each family, these lexicalizations can be correlated with smaller genealogical and areal clusters. 

          The cross-linguistic data collected here do not allow for the postulation of an implicational hierarchy; that is, it is currently not possible to predict the order of lexicalization of negative senses. However, it is clear that negative lexicalizations are organized around a limited number of cognitively salient categories. As Zeshan (2004: 51) points out "events and states such as not liking, not knowing, not having are all identifiable human experiences". This is why these concepts are often expressed by lexicalized expressions cross-linguistically regardless of language medium.