Semantic Change in Grammaticalization

In this paper, I want to establish semantic reanalysis as an independent mode of meaning change in language history. My point of departure will be cases of language change that have traditionally been classed as „grammaticalization“, on one hand because we will find very nice instances of semantic reanalysis particuarly in that field, and on the other hand because I think that the notion of semantic reanalysis that I propose can be particularly helpful in elucidating some hitherto poorly understood semantic changes in grammaticalization. In order to achive this double purpose, to characterise semantic reanalysis and to demonstrate its power as a tool of diachronic analysis, the paper is organized in the following way: I will first briefly present some cases of grammaticalization, focussing mainly on the meaning changes they involve. I will next give an overview over previous proposals as to what kind of semantic processes underlie these changes, arguing that neither of them is fully convincing. I will then demonstrate how the meaning side of the changes in question can be analysed as semantic reanalysis in semantic composition. Finally, I will evaluate the emerging picture with respect to the common observation that the meanings of words in grammaticalization just „fade away“ or „bleach“. 1 When words turn into grammar It has long been established that independent lexical „content“ words can change their status in the sentence by being reanalysed as more bound, functional „grammatical“ words. You can find many sophisticated proposals in the literature that aim at making this intuitive description of a development precise and I do not wish to add to them. I will take „grammaticalization“ as a descriptive term that covers about any change of that kind. Let me give you some examples. The Modern French phrase ne pas expressing sentence negation developed from Latin negation ne and the former full Latin noun passum meaning ‘step’. The development must already have taken its start in Vulgar Latin from sentences corresponding to (1) where the noun can still be interpreted in this original sense. In Old French, pas was already in use to express emphatic negation with more than only verbs of movement, and for a long time it was one in a wide field of emphatic negations. Today, it has become the unmarked form of negation. (1) pas il ne va pas he not go step = ‘he does not go a single step’ (2) il ne rit pas he not laughs PAS = ‘he does not laugh’ (3) il n’a pas de femme he not has PAS of-a wife = ‘he does not have a wife’ The development is summarized in Posner (1997) while Lucien (1965) offers a detailed description of the history of negation in French. Similar constructions that express negation by means of reference to a small quantity are quite frequent; other examples are English ‚not a bit‘ or German ‚kein bißchen‘, ‚kein Meter‘ (colloquial). While these constructions retain their emphatic flavour, French ne pas has lost all pragmatic markedness. Tendencies in spoken French to give up ne pas in favour of the simpler pas have been reported since at least 100 years ago but ne pas has survived as the official negation which remains compulsory in many constructions even in spoken French. The raise of the English going to future is often presented as one of the paradigm cases of grammaticalization. Plausibly, the development starts from purely motional uses of go like in (4), passes on to uses for motion-with-a-purpose (5), (6) which are usually ambiguous (in In: Graham Katz, Sabine Reinhard, and Philip Reuter, eds. (2002), Sinn & Bedeutung VI, Proceedings of the Sixth Annual Meeting of the Gesellschaft für Semantik, University of Osnabrück Konstanzer Online-Publikations-System (KOPS) URL: http://nbn-resolving.de/urn:nbn:de:bsz:352-0-396755 54 terms of present-day English) between a motionand a future-reading, and finally allows for unambiguous future uses like (7), (8). (4) John is going to Paris (5) — to get a beer (6) — to be married (7) — to finish the talk at 5 o’clock (8) — to like Paris Another way to express future tense has emerged in Central Eastern Bantu languages. Here, we find that the verb ‚say‘ is the basis for future constructions. Examples are given in (9) and (10): (9) Tumbuka: wa-ti wa-lut-e 3sg-say 3sg-go-FV ‘s/he will go (soon)’ (10) Makonde: aí (a-)sum-é 3sg-say 3sg-buy-FV ‘s/he will buy (soon)’ (FV= future vowel. Quoted from Botne, 1998) An example closer to home is the German (weak) determiner lauter. It corresponds approximately to the English adnominal uses of only yet without being a focus particle. It derives from the adjective lauter (= ‘pure’) via uses where its syntactic position and morphological shape is ambiguous between adjective and determiner, like in (12). The uses in (13) are unambiguously determiners, mostly because the concept of purity, even far-stretched as it was in past centuries, never could extend to things like grades or old mushrooms. (11) Es war lauterer Wein im Glas it was pure wine in the glas = ‘In the glas there was pure wine’ (12) Im Glas war lauter Wein In-the glas was pure/only wine = ‘There was only/pure wine in the glas’ (13) lauter gute Noten, lauter alte Pilze only good grades, only old mushrooms Let me finally mention the case of German selbst where the old intensifier (E. –self) develops a new use as a focus particle (E. even). While all previous examples exhibited a visible cline from „content“ word to „function“ word, it appears difficult here to decide whether an intensifier is more or less grammatical than a focus particle. Yet, the meaning development that led from the former to the latter use of selbst is a clear case of semantic reanalysis (Eckardt, 2001/in press). This shows that semantic reanalysis ranges far beyond examples that have elsewhere been labelled „grammaticalization“. But let me start by looking at these cases. 2 Meaning change in grammaticalization: Previous proposals