A New Grammatical Category: Impulsatives

Many languages exhibit a construction which has a meaning akin to “x feels like Ving”, often called a desiderative. This form has been observed in the literature for languages and language families as diverse as Quechua languages (Cusihuaman, 2001, Cole,1981), South Slavic languages (Murasic, 2006, Rivero, 2004, Franks:1995), Albanian (Kallulli, 1999) and Finnish (Pylkkänen, 1999). In Albanian, the desiderative is composed of the non-active form of the verb and a dative argument, as in example (1). In contrast, the Finnish desiderative is composed of an argument with partitive case and a verb with passive morpheme, which is homophonous with the causative morpheme, as in example (2). Previous analyses have accounted for these constructions separately, however, a cross-linguistic view provides a clearer perspective. I argue that we need to recognize a new grammatical category for these constructions that have been assumed identical to desideratives, but have a distinct form and interpretation. These I will call impulsatives. Impulsatives differ from desideratives in that impulsatives semantically are not volitional and are often translated as ’feel like’ whereas desideratives mean ‘want’ or ‘will’. Syntactically, subjects in impulsatives carry experiencer case rather than normal subject case marking and lastly, verbs in impulsatives carry morphology that is non-active and do not agree with their subject. In both languages, there is no dedicated morphological or lexical element that denotes the impulsative meaning. Furthermore, impulsative constructions introduce modal semantics. In the modal world of impulse, the experiencer is the external argument of the verb. Without any dedicated morphology, the source of the modality of impulsative is mysterious. I propose that the modality arises from a covert impulsative head with the semantics in (3). Furthermore, in both Finnish and Albanian, in order to receive the impulsative reading, the verb must be unergative. In Albanian, the impulsative reading cannot be obtained unaccusatives, as in (4). The Finnish impulsative is also sensitive to the same verbal restriction and cannot obtain the impulsative interpretation with unaccusative verbs, as in example (5). In order to account for these selectional restrictions, the impulsative head selects for vDO’. vDO is one flavor of v that introduces animate external arguments, the agent of unergative and consumption verbs (Folli and Harley, 2002). The impulsative head also introduces the experiencer argument which it assigns case to. Because this argument is an experiencer, it gets the normal case assigned to experiencers in the language. dative in Albanian and partitive in Finnish. Another similarity is the use of non-active or passive morphology. In order to account for this, the covert impulsative head attaches at vDO’, before the external argument would be projected. The external argument of the internal predicate is blocked from being projected. The non-active or passive morphology is a reflection of the syntax that lacks a projected external argument (Embick, 2004), as in (6). In Pylkkänen’s analysis of impulsatives, which she calls causative desideratives, in Finnish, she claims that the partitive argument is introduced by an applicative head while simultaneously being the internal argument of the causative head that bears no external argument of its own. Thus, the argument is affected by a caused event, evoking a mental or psychological reaction to it. However, if the causing event has an external argument, the impulsative reading is lost, as in (7). This is unexplained under Pylkkänen’s analysis. In Kallulli (1999)’s analysis of impulsatives in Albanian, she claims that the modality stems from the non-active morphology. She claims that the non-active morphology shifts event types. Specifically, when added to activity verbs, the non-active morphology changes the activity into a state. If Albanian impulsatives were states they would not be bi-eventive. However, Albanian impulsatives appear to be bi-eventive as they appear with two conflicting time adverbs in example (8). This is unexplained under Kallulli’s analysis. While Albanian and Finnish Impulsatives have different morphology, they share the same syntax and semantics. The analysis that I proposed not only unifies Albanian and Finnish impulsatives but it also explains the selectional restrictions and the non-active and passive morphology found in these constructions. This unified analysis suggests that impulsatives be recognized as a new grammatical category.