Causative alternation: a little-known alternative, new questions and nontrivial minimal pairs

MANGIALAVORI RASIA, MARÍA EUGENIA; MARIOS MAVROGIORGOS

Congreso; Cambridge Comparative Syntax; 2018

Causative alternation: a little-known alternative, new questions and nontrivial minimal pairsThe causative-inchoative alternation comprises a third variant logically allowed by the combinatorial system, which appears to be available in some (though not all) languages. Such an alternative remains virtually undiscussed in the literature, for various reasons: theoretically, it poses significant challenges for current theories on transitivity alternation, which do not readily cater for its properties; empirically, the construction is not systematically possible in languages normally considered in the discussion (e.g. English), although it is, as argued here, fully productive in languages like Romance (e.g. Spanish, Italian) and Greek. By featuring a unique (external) argument interpreted by default not as undergoer but as cause/initiator (‎(1)c) and stative behavior, this Stative Causative Construction [SCC] is unpredicted under standard accounts of change-of-state [COS] and crucially challenge widely-accepted generalizations on COS verbs. ❶ SCCs challenge the general notion that the internal argument?which seems a stable argument in the alternation if the analysis limits to the variants commonly analyzed ((1)a|b)?is a default constituent in the argument structure of COS verbs (Hale & Keyser [HK] 2002, Rappaport & Levin [RL] 2011).❷ By suggesting that the causative component may be independently realized, SCCs are a problem for a basic principle of event composition according to which the event structure of COS verbs combines two components (cause|process); and the former, if present, causally implicates the latter (2). ❸ Absence of internal-argument-introducing VO would mean that interpretation of Spec,v as init/cause cannot be purely structural (obtained by merge of an unaccusative structure (i.e., [v[Vo], HK 2002, Chomsky 1995, Zubizarreta&Oh 2007).(1)a. La pasta engorda a los niños. ?Pasta fattens the kids?CAUSATIVE/TRANSITIVE (Spanish)b. Los niños engordan. ?The kids fatten [up]? INCHOATIVE/UNACCUSATIVEc. La pasta engorda. (lit. Pasta fattens.)?Pasta is fattening?SCC/UNERGATIVE(2)a. e1 → e2 ( [V1[V2] HK 1993:69, 2002) b.[xCAUSE[BECOME[ySTATE]]] (RL1998:108)Proposal: Romance/Greek differ from English/German by systematically allowing the ext.-argument-introducing CAUS/INITvo to combine with the Root (cf. direct composing with INITv, McIntyre 2004, intransitive incoporation Rosen 1996). This consequently produces an event and argument structure simpler than the causative (dyadic) structure (1)a, but at the same time, semantically and syntactically different from the monoargumental variant (1)b. Key contrast: default interpretation of unique argument as initiator (cf. English: default (undergoer) reading of DP((1)c)). To support our claim, we show that: (A) SCCs do not involve an unaccusative structure; rather, presence|absence of an internal argument, along with the corresponding Vo, gives rise to non-trivial minimal pairs (COS|SCC). (B) This alternative extends to other classes of verbs with similar transitivity alternation. (C) The option correlates with a nontrivial derivational alternative, where morphological marking can be related to realization of a nondefective Vo. Importantly, our results are consistent with the general premise that syntactic projection of argument structure strictly correlates with event structure (Ramchand 2008 i.a)(A) Monadicity, unergativity & stativity. Null/implicit objects [NO] seem a compelling solution for SCCs, especially under the view that COS verbs are bona fide unaccusatives or transitives (RL 1995 a.m.o.). There are, however, contrasts: Namely, unlike NOs (Rizzi 1986), SCCs do not bind reflexives (3), nor allow NO quantification (4). Further, if it is true that bare quantifiers behave as NOs (Cattaneo 2008), (5) shows that perfective inflection is crucially dependent on NOs; otherwise, the construction fails due to the incompatibility of NO-less construction (SCC) with perfect tense (cf. *Questo ha infuriato.). Resultatives like (6), generally allowed by unaccusative/transitive verbs (Rappaport & Levin 1995), are not possible with SCCs. Inasmuch as a focal post-V quantificational expression is missing (7), ne-cliticization, a common test for NOs and unaccusatives in Italian (Russi 2008, Borer 2005) predictably fails. Note also that SCCs are fully productive with unpassivizable verbs (e.g. Sp. enfurecer/It. infuriare) and verbs not allowing NOs (Sp. llenar, It. riempire) (e.g. Esta sopa llena. [(lit.) This soup fills.] ?This soup is satiating?). Data also discourages an account based on generic (null) internal arguments [GA] (Dobrovie 1994, Bhatt & Pancheva 2006). Again, GAs shift the interpretation to an eventive (COS) type, yielding telicity independent of object quantification, as in (8), and licensing endpoint modifiers not admitted by SCCs. This agrees with (5): apparently, an internal argument cannot be invoked without yielding the consequent eventivity and (variable) telicity. Telicity and progression are only possible under this condition (9). Event modifiers, progressive and perfective tenses are only natural on an undergoer reading of the DP (10) (conceptually odd). Moreover, if for-x-time occurs at all in SCCs (9), it is interpreted as a temporal bound to an individual-level (IL) property, not to an event evolving over time, just like gradability: note the scalar/property degree reading of quantifiers in (11). IL Predication fits well with: generic reading of the subject (9) (not required by INCH frame); predication over lifetime of the subject (Ese chocolate engordaba ?That chocolate was fattening?); incompatibility with perception reports (*He visto al chocolate engordar.?I saw the chocolate fatten?), location in space, etc. Insofar as potentially ambiguous structures (e.g. ((1)c),(11)) get SCC readings, SCCs challenge (2); but also default unaccusativity predicted for monoargumental variants in COS verbs (cf. default linking rule, RL2010)(3)La injusticia {entristece/enoja} (*consigo mismo). ?Injustice saddens/maddens with oneself? (4)El sol calienta *{todos/algunos}.cf. {Ví/Compré/Calenté} {todos/algunos}?The sun heats up all/some?.?I saw/bought/heated all/some [of them]? (5)Questo ha {infuriato/impoverito/indignato}*(molti) ?This infuriates/impoverishes/outrages many?(6)a. El grafito calienta *(hasta quedar incandescente/fundido). SCC?Graphite heats [causes heat] (until [becoming] incandescent/fluid)? b. El grafito (se) calienta (hasta quedar incandescente/fundido). INCHOATIVE?Graphite heats (itself) up until incandescent/fluid?(7)La radiazione infrarossa ne riscalda/brucia *(la metà). ?Infrared radiation burns (half) [of them]?(8)El sol quemó (en un minuto/completamente) *(algo/alguno).?The sun burned *(some) in a minute/completely? (9)Estos payasos *(te) asustan(abruptamente/gradualmente/por un tiempo).?These clowns {*are scary [SCC]/scare you [INCH]} (suddenly/gradually/for some time)? (10)#El chocolate casi engorda/está engordando/engordó. (cf. English default reading of DP in (1)c)?Chocolate almost gets fat/is fattening up/fattened?(11)SPAEsta estufa calienta mucho/ITQuesta stufa scalda troppo.?This radiator causes too much heat?(B) Subject-[SE] and?Object?-Experiencer[OE] verbs are key as they: ❶ show a transitivity alternation paralleling (1); ❷ are analyzed as vINIT verbs with no vPROC complement (Ramchand 2008); ❸ yield stative causative constructions in English/German (Rothmayr 2009, Arad 1998). Spanish SCCs have a structural configuration with the precise semantic and syntactic properties expected from direct composition with vINIT ([vP[voCAUS/INIT,√]]): semantically, the subject is interpreted by default as a cause rather than as holder of result state (vs. English.; Ramchand 2008). Syntactically, unergativity naturally follows under the assumption that vINIT/CAUS introduces ext. arguments (e.g. Harley 2014) and contrasts with the structure seen in languages where SCCs are not free/systematic (English: INIT frighten UND|UND frightens easily|*INIT frightens (be-ing to equate SCC)). The proposed configuration correctly predicts systematic productivity for OE (Subject-Init), as opposed to verbs with opposite distribution (SE): (12)a. El corte de pelo de Juan molesta. (lit., Juan?s haircut annoys) ?Juan?s haircut is annoying? OEb. *Katherine teme.(lit., Katherine fears.) Intended: ?Katherine is frightening?SE(C) Greek also has SCCs behaving consistently: ❶ While the construction is ambiguous in verbs with unmarked INCH form (o ilijos kokinizi (the sun redden.3sg)?the sky becomes red/causes redness), verbs with special (NAct) morphology yield a consequent nontrivial alternation (13) mirroring se-marking in Romance (Haspelmath 1993). ❷ SCCs are free with verbs not allowing NO, showing ILP behavior (To stil pulai (*stin platia) ?The style has selling potential (*in the square)?. ❸ Object realization correlates with eventivity (incompatibility with stop and manner adverbials, (no) habitual reading in present), also giving minimal pairs (I kloun tromazun *(se) me sadhistiko/aschimo tropo. ?Clowns (scare you/*are scary) sadistically/rudely?; I sokolata stamatise na *(se) pacheni ?Chocolate stopped making you fat/*being fattening?; I sokolata pacheni ligho ?Chocolates fattens moderately/ *momentarily?; I koka-kola *(se) fuskoni kathe mera ?Coke makes you bloat/*is bloat(en)ing every day?.❹ Again, perfect tense yields INCH reading (Afti i somba thermanthike.?This radiator INCHheated up/SCC*had heating capacity? (note lifetime effect)). ❺ Greek also forms SCCs from OE, but not from SE (Ta malja tu Kosta enoxlun.(the hair.PL the Kosta.GEN annoy.3P)?Kosta's hair is annoying? cf(12)).(13)a. I somba zestenete. (the radiator heat.NAct.3s) ?The radiator is heating up? INCHb. I somba zesteni. (the radiator heat.3s) ?The radiator heats (≈ has heating capacities)SCCIn sum: If correct, SCCs provide evidence on the optionality of constituents largely seen as default in the literature. Given (A)(B)(C), variation is key to a deeper discussion on phrase structure of COS verbs.