چكيده لاتين :
Each language has a certain canonical word order. In English, for example, the SVO canonical word order mandates that the object follow the verb and, with non-alternating dative verbs (Mazurkewich, 1984; Pinker, 1989) like donate, the direct object (DO) precede the indirect object (IO): 1.I donated the books to the library.
2.*I donated to the library the books.
However, it has been widely documented that in situations where the DO is longer, this unmarked word order undergoes a change. For example, sentence 4 below is much more acceptable than sentence 3: 3.I donated the books that I had bought several years ago to the library.
4.I donated to the library the books that I had bought several years ago.
This phenomenon, known as the short-before-long principle or heavy NP shift (Arnold et al. 2000, Ross, 1967, Stallings et al. 1998, Wasow, 1997), was initially thought to be universal since incremental models of sentence production (e.g., Bock & Levelt, 1994; Garrett, 1980), which were in vogue when this phenomenon was initially observed, claimed that short constituents are easier to retrieve and are thus placed before longer and heavier ones. However, investigating other non-Germanic languages, especially through corpus studies, researchers realized that in OV languages like Japanese (Chang, 2009; Yamashita & Chang 2001) and Korean (Choi, 2007) the inverse of the short-before-long tendency is true, i.e., it is the longer constituent that tends to precede the shorter one in the pre-verbal domain.
The issue of what factors influence the relative order of objects in Persian sentences, and how the length factor can affect this ordering has been tackled only recently in the literature. Faghiri and Samvelian (2014) conducted a corpus analysis of Persian sentences and concluded that the relative order of the DO and the IO in this language corresponds, to a large extent, to the nature of the DO. To corroborate this finding, Faghiri, Samvelian and Hemforth (2014) carried out a web-based sentence completion task and demonstrated that in Persian the relative order of the DO and the IO depends mainly on whether the DO is definite or indefinite and, to a lesser degree, on the DOs length, with higher likelihoods of the DO-IO-V order as the DO gets longer. These two studies, therefore, seem to suggest that in Persian the relative length of the objects is of secondary importance, and conceptual and/or discourse accessibility of the objects determines their order in the pre-verbal domain (Karimi, 2003).
What remains missing in this growing literature on heavy NP shift in Persian, however, is looking into the Persian native speakers intuition or knowledge of language (Newmeyer, 2003, p. 682) through judgment data. Newmeyer (2003) argues that collecting corpus data, which reflect language usage, is by no means enough to draw definitive conclusions in a linguistic analysis, and native speaker judgments should also be solicited to obtain a better picture of the phenomenon under investigation (see also Manning (2003) who used corpus data to provide counterexamples to data obtained by Pollard and Sag (1994) through judgment tests regarding verb subcategorization in English). In this spirit, the present study seeks to examine heavy NP shift in Persian from a new angle; namely, it intends to examine, primarily through a grammaticality judgment test (GJT), how Persian speakers rate DO-IO and IO-DO sentences, and to what extent the length of the DO affects their ratings. The findings of the GJT are also coupled with a prompted sentence recall task (PSRT), which in turn helps us draw better conclusions about the status of the heavy NP shift phenomenon in Persian.
