- Ve as a Second-Position Clitic1
Contrary to previous claims in the literature, it is argued that ve, the yes-no particle in Malagasy, is a second-position clitic. Although ve typically appears at the right edge of the predicate phrase, in sentences with nonstandard word order, it surfaces in second position. Ve is contrasted with other particles that are not second-position clitics but pattern with adverbs. The implications of the placement of ve for analyses of Malagasy clause structure are discussed.
1. Introduction to Particles in Malagasy.
Keenan (1976) notes that certain particles in Malagasy appear immediately before the subject, which is clause final. Some examples are given below: the question particle ve (1a), the exclamative particle anie (1b), and the negative polarity item intsony 'any longer' (1c).2
a. Manapaka bozaka ve Rasoa?
AT.cut grass Q Rasoa
'Is Rasoa cutting grass?'
b. Manapaka bozaka anie Rasoa!
AT.cut grass EX Rasoa
'Is Rasoa ever cutting grass!'
c. Tsy manapaka bozaka intsony Rasoa.
NEG AT.cut grass NPI Rasoa
'Rasoa is no longer cutting grass.'
Due to the above distribution, these particles have been used in the literature on Malagasy as a test for subject-hood and, indirectly, clause structure. In what follows, I focus on the question particle ve,3 and show that, contrary to previous [End Page 135] claims, it is not a particle that marks off the right edge of the predicate phrase (roughly, the verb, its arguments, and any verbal adjuncts). Instead, I show that ve is a second-position clitic, an unstressed, closed-class element that always appears after the first constituent of the clause. In this way, ve patterns with other second-position clitics cross-linguistically, notably pronominal clitics in some Romance and Slavic languages (see Halpern and Zwicky  for a collection of papers on this topic). The main source of evidence for this claim comes from clauses with nonstandard word order: clefts and topicalizations. As I discuss in section 4, the exclamative particle anie and the NPIintsony 'any longer' do not have the same distribution as ve.
Clefts in Malagasy consist of a fronted XP (the focus) followed by the particle no and the remainder of the sentence (the presuppositional clause). It has been argued that the focused XP is in fact the matrix predicate and that the presuppositional clause is a headless relative in subject position (see Paul  for a recent discussion and analysis). A typical example of a cleft is given in (2).
(2) [Pred Rasoa] [Subj no manapaka bozaka].
Rasoa NO AT.cut grass
'It is Rasoa who is cutting grass.'
(lit.) 'The one who is cutting grass is Rasoa.'
Consider now the distribution of ve in a clefted clause. Comparing the position of ve in (3) and (1a) appears to support the claim that the clefted element is the predicate and the remainder of the clause is the subject. In other words, if ve marks the right edge of the predicate, the focused XP in (3) must be a predicate.
(3) Rasoa ve no manapaka bozaka?
Rasoa Q NO AT.cut grass
'Is it Rasoa who is cutting grass?'
This is precisely the evidence that Pearson (1996) cites in favor of a headless relative clause structure for clefts.
Malagasy also has apparent multiple clefts, the so-called "bodyguard construction" (Keenan 1976). In a bodyguard construction, there are two XPs that precede the focus marker no: an adjunct and the subject, as seen in (4).
(4) Omaly Rasoa no nanapaka bozaka.
yesterday Rasoa NO PAST.AT.cut grass
'It was yesterday that Rasoa was cutting grass.'
In this case, ve surfaces between the first element and the second.
(5) Omaly ve Rasoa no nanapaka bozaka?
yesterday Q Rasoa NO PAST.AT.cut grass
'Was it yesterday that Rasoa was cutting grass?' [End Page 136]
Following the same reasoning, the data in (5) would suggest that the second element, Rasoa, is contained within the subject position (within the presuppositional clause) and that omaly 'yesterday' is the main predicate. In other words, the meaning of (5) would be something like 'Was Rasoa's cutting of grass yesterday?'. So far...