Nenad Lovrić
Graduate Center, City University of New York
The inclusion of od, a non-thematic preposition similar to English of, significantly lowers the preferred attachment of a relative clause (RC) in the ambiguous construction N1-(od)-N2[Gen]-RC in silent reading of Croatian (Lovrić et al., 2000). The only theory proposed so far to explain this effect is the Implicit Prosody Hypothesis (IPH) of Fodor (1998) and Quinn et al. (2000). It claims that attachment preferences in silent reading are sensitive to a language's default (typical) patterns of prosodic phrasing, mentally projected onto sentences by the reader. Examination of Croatian RC sentences read aloud has revealed that the presence of od changes the typical prosodic phrasing of the sentence: it increases the likelihood of a prosodic break after N1, grouping N2 more closely with the RC. When od is absent, Croatian has no break after N1, but a more or less obligatory break after N2, separating it from the RC (Lovrić et al., 2001).
For the IPH to be correct, it must be the case that this explicit prosodic difference caused by od should lower RC-attachment preferences for hearers even when od is absent. The present experiment tests this prediction by crossing the presence/absence of od with the two characteristic prosodic patterns observed in the reading-aloud experiment (Prosody 1: a prosodic break after N1 and no break before the RC; Prosody 2: no break after N1 and a break before the RC). Subjects listened to syntactically ambiguous sentences in the four versions; see examples below. (Prosody 1 without od and Prosody 2 with od are both acceptable though less usual.) After each sentence, the subject answered a question checking the attachment of the RC.
The results show an overwhelming main effect of Prosody (p<.001): hearers attach the RC low (5% high attachment) if a sentence has the prosodic phrasing typical of the od construction (Prosody 1), and attach the RC high (70% high attachment) if a sentence has the prosodic phrasing typical of the od-less construction (Prosody 2). Crucially, these preferences hold whether or not od is present in the sentence (F1, F2 < 1 for main effect of +/-od and for +/-od x Prosody interaction), indicating that attachment responses are controlled by the prosody typically associated with +/-od, not by any other properties (syntactic, semantic, etc.) that od might have.
Based on these data for explicit prosody, it is reasonable to conclude that the lowering effect of od on RC-attachment in silent reading is, as the IPH proposes, mediated by prosodic phrasing.
Examples ('/' marks boundary of intermediate or major prosodic phrase)
| PROSODY 1: | N1 / (od) N2 RC | |||||||||||
| Upoznali smo kćerku | / | (od) učiteljice što odlazi na put. | ||||||||||
| Met+1.p.pl. are daughter "We met the daughter |
(of) teacher[Gen] that leaves
on trip of the teacher that is leaving on a trip." |
|||||||||||
| PROSODY 2: | N1 (od) N2 / RC | |||||||||||
| Upoznali smo kćerku (od) učiteljice | / | što odlazi na put. | ||||||||||
| Met+1.p.pl. are daughter (of)
teacher[Gen] "We met the daughter of the teacher |
that leaves on trip that is leaving on a trip." |
|||||||||||
|
|
|||||||||||
References
Fodor, J. D. 1998. Learning to parse? J. of Psycholinguistic Research, 27, 2, 285-319.
Lovrić, N., Bradley, D. & Fodor, J. D. 2000. RC attachment in Croatian with and without preposition. Poster presented at AMLaP Conference, Leiden.
Lovrić, N., Bradley, D. & Fodor, J. D. 2001. Implicit prosody resolves syntactic ambiguities: Evidence from Croatian. CUNY/SUNY/NYU Mini-Conference on Linguistics, Stony Brook.
Quinn, D., Abdelghany, H. & Fodor, J.D. 2000. More evidence of implicit prosody in silent reading: French, English and Arabic relative clauses. Poster presented at the 13th Annual CUNY Conference on Human Sentence Processing, San Diego.