An Experimental Look at Interactions between Passive and Japanese Aspectual Verbs

Across languages, aspectual verbs commonly select for a clausal complement, and they often exhibit transparency to normally clause-bound phenomena, known as ‘restructuring’ behaviors. Language families in which aspectual verbs exhibit restructuring behaviors include Romance (e.g. Aissen and Perlmutter 1976, 1983, Rizzi 1982, Burzio 1986, Cinque 1999, 2003, 2004), Germanic (e.g. Wurmbrand 2001), Austronesian (e.g. Chung 2004), Dravidian (e.g. Agbayani and Shekar 2008), and language isolates such as Basque (e.g. Arregi and Molina-Azaola 2004) and Japanese (Shibatani 1973, 1978, Kageyama 1993, 1999, Nishigauchi 1993, Koizumi 1994, 1995, 1998, and Matsumoto 1996). One of the restructuring phenomena that has been frequently discussed in previous studies on Japanese aspectual verbs (JAVs) is passive. 1 Previous studies claim that some JAVs allow for long passive, or passivization of an embedded object with the passive morpheme attached to the matrix aspectual verbs, in addition to passive in embedded clauses, or embedded passive. Other JAVs, however, allow for only one of them (Shibatani 1973, 1978, Kageyama 1993, 1999, Nishigauchi 1993, and Matsumoto 1996). The different behaviors of JAVs with the two passive constructions are some of the reasons that led previous studies to propose a three-way classification of JAVs: (i) optionally restructuring (e.g. allowing for both long and embedded passive), (ii) obligatorily restructuring (e.g. allowing only for long passive), and (iii) non-restructuring (e.g. allowing only for embedded passive). More recently, an analysis of JAVs as functional heads has been proposed based on the same three-way classification (Fukuda 2007, in press). However, there are at least two potential problems for this classification of JAVs. First, reliability of long passive as syntactic diagnostic has been questioned for other languages. For instance, long passive in German has been deemed too marginal to be a reliable diagnostic in some studies (e.g. Reis and Sternefeld 2004). Second, there are disagreements among previous studies about acceptability judgments of some of the combinations between JAVs and the two passive constructions. While Shibatani (1973, 1978), Nishigauchi (1993), Matsumoto (1996) claim that one of the JAVs, oe ‘finish1’, only allows for long passive, Kageyama (1993) claims that it also allows for embedded passive at least for some speakers. Because of these factors, a more systematic examination of interactions between JAVs and passive is desirable to find out whether (i) long passive with JAVs is a robust phenomenon and (ii) the contrasts among JAVs with the two passive constructions reported in previous studies can be replicated. In this study, I report the results of two acceptability judgment experiments that examined the interactions between the two passive constructions, long and embedded, and four JAVs: hajime ‘begin’, tsuzuke ‘continue’, oe ‘finish1’, and owar ‘finish2’. These four JAVs were chosen because they represent the core aspectual distinctions (i.e. inception, continuation, and termination) and were frequently discussed in previous studies. The results of these experiments show that long passive is a robust phenomenon in Japanese and confirm the contrasts among JAVs with the two passive constructions with hajime ‘begin’, tsuzuke ‘continue’ and oe ‘finish1’. However, our findings also show that owar ‘finish2’ is compatible with neither of the two passive constructions, despite the claim in previous studies that it is compatible with embedded passive (Matsumoto 1996). I discuss implications of our findings for an analysis of JAVs and proposed an analysis of the gap between previous studies and our experiments with owar ‘finish2’. This paper is structured as follows. Section 2 introduces the observations reported in previous studies about interactions between the four JAVs and the two passive constructions. Section 3 briefly introduces the functional head analysis of JAVs (Fukuda 2007, in prep.), which has been motivated based on the three-