Balinese Pseudo-Noun Incorporation: Licensing under Morphological Merger Keywords: pseudo-noun incorporation, head movement, adjacency, morphological merger Head Movement of a nominal into a verb (Noun Incorporation) has two consequences. The nominal is adjacent to the verb and is licensed without Case (Baker 1988). This raises the question of whether (i) Head Movement or (ii) the head-head adjacency it creates is responsible for licensing. In this talk, I argue for (ii) by demonstrating that properties of Balinese Pseudo-Noun Incorporation (PNI) are incompatible with (string-vacuous) Head Movement (cf. Baker 2012). Additional data demonstrates that Balinese PNI is also incompatible with a bare NP analysis (Massam 2001). Rather, the data is compatible with a scenario in which the nominal head happens to be adjacent to V0. In this configuration, head-head adjacency licenses the nominal via M(orphological)-Merger (Marantz 1984). (1) Morphological Merger (Marantz 1984) V0 [HP H0]  [V0+H0] Specifically, when the derivation converges with a caseless nominal head H0 linearly adjacent to the verbal head V0, M-Merger applies embedding H0 within the verb. The result of such Merger is a verbal element which need not be Case-licensed – vacuously satisfying the Case Filter. Crucially, such merger need not be realized as affixation (Bobaljik 1994), and it is not in Balinese. Background: Voice and Case in Balinese: Balinese exhibits two transitive voices – Objective Voice (2a) and Agentive Voice (2b): (2) a. jaran-e gugut cicing Theme Verb Agent horse-DEF OV.bite dog b. cicing ŋugut jaran-e Agent N-Verb Theme dog AV.bite horse-DEF ‘A dog bit the horse.’ Regardless of thematic role, the preverbal element occupies [Spec, TP] (Wechsler & Arka 1998). Levin (2012) argues that the voice alternation arises from a parameterization of v0. In (1a), v0 assigns no Case but triggers object shift of the Theme above the Agent, where T0 assigns it Case. In this configuration the Agent is left caseless – a Case Filter violation (Chomsky & Lasnik 1977). I posit that M-Merger rescues the derivation by creating a complex verb within which the merged nominal element need not receive Case. (1b) proceeds as in English, with a distinct v0, realized as a verbal prefix, licensing the Theme and T0 licensing the Agent. Strict Head-Head Adjacency in OV: For the proposed analysis to be on the right track, we must find strong evidence that the verb and post-verbal Agent display strict head-head adjacency. This is exactly what we observe: (3) No element can intervene between a post-verbal Agent and the verb. a. siap-e uber cicing ke jalan-e S V OAgent PP chicken-DEF OV.chase dog into street-DEF ‘A dog chased the chicken into the street.’ b. uber cicing ke jalan-e siape-e V OAgent S PP c. *uber siap-e cicing ke jalan-e *V S OAgent PP d. *siap-e uber ke jalan-e cicing *S V PP OAgent [W & A, p. 405] (4) Post-verbal Themes (4a), but not post-verbal Agents (4b), can undergo wh-movement. a. buku cen John maca. b. *anak cerik cen be-e daar book which J. AV.read person small which fish-DEF OV.eat ‘Which book did John read?’ (‘Which boy ate the fish?’) This behavior is expected if the post-verbal Agent can only be licensed under M-Merger. Intervening elements in (3) or further movement operations in (4) break up the necessary head-head adjacency and (1) cannot apply. The definiteness effect in post-verbal Agents can also be view as an adjacency requirement. (5) Post-verbal full nominals must be indefinite. I Wayan gugut cicing/*cicing-e ento ART W. OV.bite dog dog-DEF that ‘A/*the dog bit Wayan’ [W & A, p. 401] M-Merger is, crucially, an operation on linear order. As such it is sensitive to intervening material that may be structurally lower than the nominal head. If this is the case it follows that definite Agents

cannot be licensed in post-verbal position – the NP intervenes blocking M-Merger of D0 into the verbal complex. Under commonly held theories of (P)NI, the definiteness effect is attributed to the NP status of the nominal (Massam 2001) or the fact that only lexical heads (N0s) can incorporate (Baker 1988, 2012). Balinese illustrates that head-head adjacency is the correct restriction, because some definite DPs (pronouns and Proper Names) can incorporate. (6) a. be-e daar ida b. be-e daar Nyoman fish-DEF OV.eat 3rd.sg. be-DEF OV.eat N. ‘(S)he ate the fish.’ ‘Nyoman ate the fish.’ This observation is at odds with both Baker and Massam’s proposals. We can be sure that these DPs are licensed under adjacency, because they show the same linearity requirements as NPs (4). (7) a. *be-e daar keras-keras ida b. *be-e daar keras-keras Nyoman fish-DEF OV.eat quickly 3rd.sg fish-DEF OV.eat quickly N. (‘She ate the fish quickly.’) (‘Nyoman ate the fish quickly.’) M-Merger provides a solution to why pronouns and proper names can incorporate but definite descriptions cannot. Definite DPs cannot be licensed under M-Merger as the NP intervenes between D0 and V0. However, pronouns, which occupy D0 and lack a NP complement (e.g Postal 1966, Elbourne 2001), can be realized adjacent to the verb. Similarly, if proper names move to D0 (Longobardi 1994), PF adjacency with the verb holds, as there is no intervening NP material. NP-internal Word Order: If head-head adjacency is required for licensing post-verbal Agents, we expect to see ungrammaticality due to NP-internal elements as well as external elements seen in (4). Balinese – like Tongan (Ball 2005) – has a small class of adjectives that can appear pre- or postnominally (8), but must appear post-nominally when modifying post-verbal Agents (9). (8) (liu) cicing (liu) ŋugut Nyoman (9) Nyoman gugut (*liu) cicing (liu) many dog AV.bite N. N. OV.bite dog many ‘Many dogs bit Nyoman.’ ‘Many dog bit Nyoman. This behavior follows from the head-head adjacency requirement. Only when the head of the nominal is adjacent to V0 can M-Merger occur and license the nominal. Intervening adjectives, just like intervening NPs (3c,5) and PPs (3d) break up the linear adjacency needed for M-Merger. If M-Merger cannot take place the derivation crashes. Massam’s analysis cannot account for NP-internal word order rigidity without further modification. An inert NP should not display any NP-internal word order effects. One might wonder if a DP-layer is necessary to host pre-nominal liu. If this were the case, we would predict preverbal liu to force a specific interpretation, but it does not. Incompatibility with Head Movement: Baker suggests that only NPs generated in [Compl,V] can undergo PNI. DPs also undergo PNI. Furthermore, Themes never undergo PNI in Balinese. Pre-verbal Themes do not display NP-internal word order restrictions (5), and post-verbal Themes can be whmoved (4). (10-11) further confirms that Themes never PNI. (10) Themes can be definite. a. Wayan ŋ-uber cicing-e b. cicing-e uber Wayan W. AV-chase dog-DEF dog-DEF OV.chase W. ‘Wayan chased the dog.’ ‘Wayan chased the dog.’ (11) Themes can be separated from the verb. a.Wayan ŋ-uber keras-keras cicing-e b. cicing-e keras-keras uber Wayan W. AV-chase quickly dog-DEF dog-DEF quickly OV.chase W. ‘Wayan chased the dog quickly.’ ‘Wayan chased the dog quickly. Assuming that Balinese PNI prevents a Case Filter violation, it follows that only the post-verbal Agent, which does not receive Case in OV, incorporates to be licensed. Agent Incorporation is incompatible with Head Movement, as it requires movement from [Spec, vP]. Such movement illicit under the Head Movement Constraint (Travis 1984, Hale & Keyser 1993). If M-Merger is responsible for linear adjacency, then the HMC is never violated. Selected References: Baker, M. 2012. Pseudo-Noun Incorporation as Covert Noun Incorporation: Linearization and Crosslinguistic Variation. Ms. Massam, D. 2001. Pseudo-Noun Incorporation in Niuean. NLLT. Wechsler, S & W. Arka. 1998. Syntactic Ergativity in Balinese: An Argument Structure Based Theory. NLLT.