We speak of 'clitics' as if they existed, i.e., as if there were a clitic property per se. Toman has characterized Czech clitics as possessing "specialized, highly marked paradigms" (Toman 1999: 224). In this paper I claim that exactly the opposite is the case, i.e., that special clitics constitute the nexus of all that is unmarked in Czech with respect to productivity. This is perhaps not unexpected if clitics are taken to be bundles of purely formal features and nothing more.
For example, with respect to morphology, LudvĖkov· (in Teshitilov· et al. 1987: 107) reports that -ho is the third most frequent word-final syllable in spoken Czech. This suffix is found on masculine animate singular nominals in the accusative case, as in (1a), on wh-expressions as the unmarked exponent for accusativity across the board, as in (1b), and as the productive ending for JirĖ-class loans, as in (1c).
(1) a. jeho 'him', toho 'that', sveho 'one's own', jejiho 'her', noveho 'new', prvniho 'first'
b. koho? 'whom?'
c. JirĖho 'George', Jackieho 'Jackie', Charlieho 'Charlie'
Not surprisingly, this same ho recirculates as the masculine accusative clitic (cf. Jakobson 1971). With respect to information structure, clitics are neither Topic nor Focus, but their very neutralization, as seen in (2a), which answers either (2b) or (2c) (Avgustinova & Oliva 1995: 22).
(2) a. Dal jsem jim je uzh nekdy minuly tyden.
b. Kdy jsi dal pet ot·zek svym trem prĖtelkynĖm?
c. Kdy jsi dal svym trem prĖtelkynĖm pet ot·zek?
Two consequences follow. First, underspecified non-clitics are predicted to pattern with clitics in 2P. Second, "overspecified" clitics will be forced out of 2P. Both predictions are borne out empirically, as seen with je and sebou in (3a-b), and ho in (4), respectively.
(3) a. Kolik je ti let?
b. Nech·v· sebou snadno manipulovat.
(4) Vchera jsem videl jenom ho.
'Yesterday I saw only him/*it.' (Veselovsk· 1995)
The same principle can be seen at play on a smaller scale in clitic inventories themselves. The more highly marked a clitic's lexical entry, the freer it is to pattern with non-clitics outside of 2P. For example, [+fem.] and [+plur.] clitics (the marked members of their opposition) lack the specialized morphology of 'core' clitics (mi, ti, te, se, si, mu, ho) and enjoy a wider distribution. Conversely, markedness explains why precisely these forms are core clitics, and not others.
The paper is of general significance becaues, if membership in the class of clitics is motivated, formally motivated accounts of how clitics interact with other parts of the grammar may follow.