05Syntax-prosody interface: Evidence from wh-movement in Jordanian Arabic and Egyptian Arabic
Richards (2006, 2010) suggests that wh-movement is prosodically driven. Based on the position of the Comp(lementizer) and the marking of prosodic phrase edges, he claims that if Comp is on one side and the language marks the opposite side of prosodic phrases, then the wh-phrase does not move since Comp and wh-phrase can create a prosodic wh-domain (e.g. Japanese). However, if the language prosodically marks one side and Comp is on the same side, a single wh-domain cannot be created; thus the wh-phrase needs to move closer to Comp to minimize minor phrases (e.g. Tagalog). This study tests Richards' claim as well as the question of what it means for a language to `mark' a prosodic phrase edge. ^ The study addresses wh-movement in Jordanian Arabic (JA) and Egyptian Arabic (EA): the former moves the wh-phrase whereas the latter allows both in situ and movement . JA and EA are a strong testing ground for Richards' theory since it is expected that they would behave alike given that they both descended from standard Arabic and have Comp on the left. However, since JA moves the wh-phrase and EA mostly leaves it in-situ, we would expect to see a prosodic difference between them. In particular, and in light of Richards' theory, it is expected that JA marks the left-edge and EA marks both edges of prosodic phrases. In this study, I present evidence from syntax-prosody interface to support Richards. In particular, I show how certain phonological processes take place either at the right or at the left edges of syntactic constituents. For instance, in JA both resyllabification and epenthesis blockage take place at the left edges of syntactic XPs. By contrast, epenthesis and vowel reduction are taken as right edge markings in EA. ^ To test Richards' theory thoroughly, three experiments were constructed, with two groups of 16 adult participants (8 M, 8 F) from each dialect. The first task elicited the prosodic patterns of 4 sentence structures. The second task elicited productions for wh-questions. The third was an acceptability judgment task where participants were asked to rate wh-movement and wh in-situ sentences. ^ Productions were analyzed acoustically to determine edge-marking in each dialect. In JA, percent time elapsed since onset, downstepping and resetting indicated a left edge marking. However, contour direction and the relatively high pitch range in semitones in non-final phrases pointed to right edge marking, raising an issue because from Richards it was inferred that JA should mark left edges. The analysis of the Egyptian data did not present any challenge since, as Richards predicts, the dialect marks both edges of prosodic phrases. In addition, since pausing behavior marks boundaries clearly, pause locations and duration were measured. Pause analysis provided evidence for relation-based mapping in EA, and end-based mapping in JA which preferred to pause at the end of XPs. The analysis of the rating task revealed that the EA group always leaves wh-phrase in-situ except when it is an adjunct, where they either move it or leave it in-situ. Furthermore, the JA group always moved the wh-phrase except when the sentence is embedded, where a few responses opted for the in-situ strategy. ^ Although the phonological evidence presented here strongly supports Richards' premise, the acoustic analysis and the effect of syntactic factors such as phrase type (ARGUMENT or ADJUNCT) impose a challenge for the theory. Based on our findings, the theory works fairly well but requires modification to account for the JA and EA data, and possibly for other languages as well. Specifically, I suggest that all languages mark both edges with different degrees of strength. However, the cues for edge marking, especially the acoustic ones, are not all equally valued for prosodic marking purposes. Applying this, we conclude that JA by and large marks left edges whereas EA marks the right one.^
Ronnie Wilbur, Purdue University.