|
|
MONOTONICITY AND THE SEMANTICS OF WORD FORMATION
Andrew Koontz-Garboden Stanford University Tuesday, May 11, 12 PM, Terrace Room Sponsored by the Stanford Humanities Center/Mellon Foundation Graduate Research Program The point of departure for this talk is the mystery about the lexicon in (1) and (2), due to Kiparsky (1982:8). (1) Attested (derivation from less to more semantically specified) a. pig = sus animal
b. pig+let = baby sus animal (2) Unattested (derivation from more to less specified) a. lig = baby sus animal
b. lig+bung = sus animal Kiparsky observes that while languages have word-formation operations that add diminutive meaning, as in (1), languages do not seem to have derivations that derive a new lexeme with a more general meaning from a lexeme with a more specific diminutive meaning, as in (2). This asymmetry, he observes, is expected if derivational processes "... can *add* but not eliminate some element of meaning in a word." (Kiparsky 1982:8, emphasis in original). This idea, which I call the Monotonicity Hypothesis (MH), has been the subject of surprisingly little systematic investigation. In this talk, I survey prior literature on the MH (Marchand 1964; Kiparsky 1982; Rappaport Hovav and Levin 1998). I then go on to propose the domain of states (e.g. red, long) and changes of state (e.g. inchoative redden, lengthen; causative redden, lengthen) as an ideal testing ground for the hypothesis, examining two problem cases: (a) Ulwa (Misumalpan; Nicaragua) adjectives (e.g. sang-ka 'green'), which appear to be derived from a morphologically simple causative change of state denoting root (e.g. sang-- 'cause become green') and (b) anticausativization, the phenomenon whereby inchoatives (e.g., Spanish quebrar se 'break-intrans') are derived from causatives (e.g., Spanish quebrar 'break-trans') in the causative/inchoative alternation. For Ulwa adjectives, I discuss data from recent fieldwork which suggest that rather than marking a meaning altering derivation, the appearance of the --ka suffix is syntactically conditioned; --ka does not, in fact, seem to effect a meaning changing derivation. For anticausativization, I review recent arguments suggesting that derived inchoatives have causative semantics as part of their lexical representation (Levin and Rappaport Hovav 1995; Chierchia 2004; Alexiadou et al. 2006), consistent with the MH. With these arguments as a backdrop, I suggest the outlines of an analysis of anticausativization that is consistent with the MH. I conclude by discussing further predictions of the MH, which suggest areas for future research. 
|
