Grammaticalization and 'lateral' grammaticalization: new perspectives on linguistic interfaces and functional categories

Simpson and Wu (2002) analyze Chinese shi-de constructions and propose a new type of grammaticalization in the Minimalist framework called 'lateral' grammaticalization (LG). It is compared with Roberts and Roussou (R&R) (2003) and van Gelderen's (2004, 2011) Minimalist analyses of grammaticalization (henceforth 'standard' grammaticalization (SG)) and while both display 'structural simplification', as defined by R&R (2003:198-201) and van Gelderen (2011:13-21), LG does not display 'upward feature analysis', which is a diagnostic trait of SG (R&R (2003:200)). Rather, there is 'lateral feature analysis' where one functional category (e.g. D) is re-analysed as another (e.g. T) (S&W (2002:201-202)). Furthermore, while examples of SG regularly display semantic and morphophonological weakening (R&R (2003:218-229)), those of LG do not, which seems to suggest that 'upward feature analysis', not 'lateral feature analysis', is the cause behind weakening in grammaticalization. Bybee (2003, 2011) proposes that frequency is the main driving force behind weakening in grammaticalization, since as the grammaticalizing element undergoes weakening in semantics, it becomes compatible with a wider range of complements and hence rises in frequency which causes it to become automatised and lose morphophonological substance. There is therefore a causal relationship between semantic and morphophonological weakening ('parallel reduction hypothesis'/'co-evolution of meaning and substance' (Bybee et al (1994:17-21)). In my comparison between the Romance future (Latin infinitive + habere) (SG) and the Chinese copula shi (LG), it is discovered that 'upward feature analysis' in SG does indeed give rise to 'context expansion', namely a widening in the range of complementation (cf Bybee (2003)), which leads to rise in frequency and morphophonological weakening, while 'lateral feature analysis' in LG crucially leads to 'context reduction' which lowers the frequency of the grammaticalizing element and hence pre-empts its morphophonological weakening. Cross-linguistic patterns of weakening are examined and there seems to be a correlation between frequency effects in syntactic change and the rate of morphophonological weakening of functional elements in grammaticalization. This is significant, since it is widely assumed that functional elements are necessarily weak (cf Selkirk (1984:335-337)) when the evidence from a formal analysis of grammaticalization suggests that the empirical properties of functional elements, namely their differential rates of weakening, can (and should) be derived from their genesis, namely grammaticalization (SG/LG). This opens up new research questions which are left for future investigations.