On the semantics of Cantonese pre-predicate jau5 ‘have’: an assertive existential quantifier

The verb jau5 ‘have’ in Cantonese can be placed immediately before the verbal or adjectival predicate, forming “jau5 + predicate” construction (henceforth pre-predicate jau5), and pervious analyses consider the pre-predicate jau5 as the preposed perfective marker –zo2. The same construction “you ‘have’ + VP” renders ungrammaticality in Mandarin, with the post-verbal perfective –le being its only form.
This paper studies the semantics of Cantonese pre-predicate jau5 ‘have’. I argue that Cantonese pre-predicate jau5 and the perfective marker -zo2 cannot be equivalents because: (1) sentences vary in meaning and selection of predicate if -zo2 is replaced by jau5 ‘have’; (2) jau5 ‘have’ has no inherent aspectual meaning, which is why it can co-occur naturally with aspectual markers indicating imperfectivity, experiential meaning and habituality; and (3) jau5 ‘have’ is quantificational in nature.
The analysis I propose draws heavily on ideas proposed by S. F. Huang (1981), who divided Chinese existential quantifiers into two categories, namely the assertive category and the non-assertive category, depending on the affectivity of the sentence. I argue that Cantonese pre-predicate jau5 ‘have’ is an assertive existential quantifier, with existential closure being part of the assertion. The existential nature of jau5 ‘have’ is inherently borne out in the lexical semantics of “have”, which conveys both quantification and eventuality. There are two possible readings: (i) for non-specific events, jau5 ‘have’ asserts the existence of the event denoted by the predicate, and (ii) for specific situations, jau5 asserts that the final endpoint or the bounded endpoint of the situation has been actualized.
The proposed analysis of Cantonese pre-predicate jau5 ‘have’ adds further evidence that Chinese does have a type of existential quantifier intertwined with assertion, as can be gleaned from S. F. Huang’s analysis on pre-NP “have”. Moreover, the relevant analysis results in two theoretical implications. Firstly, it supports Dahl’s proposal (1981) that, when we are dealing with natural final endpoints or bounded endpoints of situations, we should draw a distinction between actualized and potential final or bounded endpoints (cf. T-property and P-property in Dahl 1981). Secondly, if one considers whether a situation encodes potential or intended bounded endpoints, individual-level statives and stage-level statives are not alike: while individual-level statives have no intrinsic endpoints, stage-level statives are closer to events in having potential, though not actualized, bounded endpoints. Therefore, stage-level statives form a natural class with activities, semelfactives, accomplishments and achievements rather than with individual-level statives, suggesting a clearer remapping of stage-level statives to the event group rather than the state group.