Syntactic processing under load and noise interference

Gloria Waters,1 Sasha Yampolsky1 & David Caplan2
1
Boston University, 2 Harvard Medical School

gwaters@bu.edu

 

We present two studies of interference effects on on-line syntactic processing.  Both studies used self-paced listening with end-of-sentence plausibility judgments.  Sentences differed in syntactic complexity, with more complex sentences containing object-relativized clauses (subject object (SO) and cleft object (CO) sentences) and simpler sentences containing subject-relativized clauses (subject subject (SS), object subject (OS), and cleft subject (CS) sentences).

In the first study, 36 undergraduate students were tested under concurrent load conditions that were varied between no load, a 3 digit load, and a 5 digit load.  In three comparisons, (SO–SS; SO–OS; CO–CS) there were effects of load and of syntactic complexity in digit recall, in RTs on the plausibility judgments, and in self-paced listening times.  In the self-paced listening data, listening times increased at the embedded verb of object-relativized sentences, reflecting the increase in load at that point.  There were no interactions of load and complexity in any performances.

In the second study, 90 undergraduate students were tested on CO and CS sentences under noise conditions that were varied between no noise, a -3 dB S:N ratio, and a -4.5 dB S:N ratio.  Noise effected word recognition, as measured by repetition, in both sentence types equally.  There were effects of noise and of syntactic complexity in RTs on the plausibility judgments, and in self-paced listening times.  In the self-paced listening data, listening times increased at the embedded verb of object-relativized sentences, reflecting the increase in load at that point.  There were no interactions of load and complexity in the self-paced listening times.

These results indicate that on-line syntactic processing is not affected by either a concurrent memory load or concurrent noise.  They provide evidence that the processes involved in assigning one aspect of syntactic structure are at least partially independent of those that retain items in a general verbal working memory system.