Syntactic priming is well-established as a source of influence on language production (e.g., Bock, 1986), with stronger effects occuring with lexical overlap between prime and target sentences (Pickering & Cleland, 2003). Much less evidence exists to document similar syntactic priming effects in comprehension. In this talk, I will present the findings from a series of eye-movement monitoring experiments focusing on reduced relative clauses, as in sentence (1):

(1) The defendant examined by the lawyer was unreliable.

The first pair of experiments manipulated lexical overlap, using sentences like (2), in which the critical past participle examined appears in both the prime and target sentence,

(2) The doctor examined by the specialist had a large mole.

and (3), in which a different participle appears.

(3) The doctor visited by the specialist had a large mole.

Robust priming effects were observed for pairs like 1 & 2, but not for pairs like 1 & 3. A third experiment used sentences like (4) containing unreduced relative clauses.

(4) The defendant who was examined by the lawyer turned out to be unreliable.

Unreduced relative primes also facilitated processing of sentences like (1).

A further pair of eye-tracking experiments investigated the source(s) of facilitation in syntactic priming. The first experiment tested whether passive constructions with similar local structure in the critical region, but different overall syntactic structure as the target, facilitate processing of reduced-relative targets. This experiment used three types of primes (as in 5-7):

(5) The contestant selected by the judge.... (Reduced Relative) (6) The contestant was selected by the judge... (Passive) (7) The contestant selected the wrong answer... (Main Clause)

The results replicated the priming effects for reduced relative primes, and provided some evidence for faciltation resulting from passive primes.

The final experiment tested whether short relatives (like 8) would also prime reduced relative target sentences.

(8) The applicant who was considered had a good track record. (Short Relative)

Three other prime types were also tested, for comparison with previous experiments. The results showed that all three types of relative clause prime sentences facilitated processing of the reduced relative target sentences.

The pattern of results across the full set of experiments indicates that

1. priming does depend on lexical overlap between primes and targets.
2. priming does not depend on suppression of preferred structures.
3. priming does not result from a template-matching process.
4. priming does not result from low-level repetition priming.
5. priming does not require explicit presentation of the critical component in the prime sentence.

Thus, the source of priming is most likely in syntactic structures, or possibly verb-argument frames, accessed when the critical repeated past participle is accessed.