Le: A procedural marker in Mandarin Chinese

The semantics of the particle le in Mandarin Chinese has been hotly debated in both Chinese and non-Chinese linguistic literature in particular during the past 25 years. Le is a post verbal particle, which also occurs in sentence final position. This syntactic flexibility and the fact that the presence of le seems to render distinct default readings depending on its syntactic position has given rise to the conventional treatment of le as a homophonic and homographic representation of two distinct morphemes with different semantic characteristics. The so called post verbal particle le is generally treated as a perfective marker (Li and Thompson (1981), Smith (1997)) while the so called sentence final particle le by most authors is defined as either a perfect marker (Li et al. (1982)), an inchoative/change-of-state marker (Chao (1968), Chan (1980) and Melchert (1980)) or a mood marker (Fang et al. (1992)). I aim to show that this distinction is unnecessary, that a unified treatment of le is possible, and that the view that le has semantic content should be discarded in favour of a view of le as apragmatic particle. This proposal is primarily based upon empirical evidence of the interpretation of le in post verbal and sentence-final position in modern Chinese. Basing my arguments on the analysis of telephone conversations between native speakers of Mandarin Chinese (CallHome Mandarin Chinese speech corpus, Linguistic Data Consortium) I will show that post verbal le and sentence-final le contribute to similar readings, temporal as well as modal and that its interpretation is highly context dependent. The temporal interpretations of le are formally described within a Reichenbachian (1947) combined tense/aspect system, which defines the aspects time-relationally and not metaphorically (see also Klein (1994)). This paper presents a definition of le asa procedural marker (the relevance-theoretic notion of procedures – words that result in non-translational activation – is discussed in for example Carston (2002)) which activates the concept [boundary] but which contains no encoded semantic content