'Why'-Remnants in Japanese matrix sluicing

Since Hasegawa (2008), it has been argued that matrix sluicing in Japanese is not derived from the same source as embedded sluicing (Hasegawa 2008, Hiraiwa and Ishihara 2012, Abe 2015), which has been rather unanimously believed to be a reduced wh-cleft in current literature due to the optional presence of copula da (Ross 1969, Kuwabara 1996, Kizu 1997, Fukaya and Hoji 1999, Merchant 2001, Hiraiwa and Ishihara 2002, 2012, Saito 2004, Fukaya 2007, Hiraiwa 2021, among many others). Recently, however, Fujiwara (2020) documented matrix sprouting in Japanese and claimed that matrix sluicing should still receive the same cleft-based analysis as embedded sluicing. This study will present some strong pieces of evidence which will be problematic for Fujiwara’s account, with the crucial data coming from a known discrepancy between certain types of ‘why’-phrases in Japanese; it has been observed that whilst phrases like naze/nande/doosite ‘why’ are compatible with a copula/cleft structure, others, namely nandemata/nandatte ‘why’, are not (Oguro 2013, Hiraiwa 2021); and that whilst the former type of ‘why’-phrases can act as a remnant/pivot under embedded sluicing, the latter type of ‘why’-remnants cannot (Hiraiwa 2021). With my novel data, however, it will be shown that nandemata/nandatte is perfectly grammatical as a remnant under matrix sluicing, despite their incompatibility with clefts. Moreover, crucially, it turns out that the grammaticality of such ‘why’-remnants carries over not only to matrix merger, but also to matrix sprouting. These results will be very problematic for Fujiwara, because if matrix sprouting is derived from a reduced wh-cleft, then it should be predicted that sprouting should not be available where clefts are not. I will then turn to the copula puzzle in Japanese matrix sluicing and suggest that matrix merger sluices in Japanese are structurally ambiguous between clefts and wh-question, following Abe (2015).