Introduction: A central question in psycholinguistic research is how words are combined into sentences during speaking. This question is complicated because words do not act independently; the form of certain words depends critically on the forms of prior words in the utterance. An example of this is agreement, a type of linguistic dependency. For example, in the sentence The bill from the accountants was late, the number specification of the verb (was vs. were) is dependent on the plurality or non-plurality of the subject.
There is a large body of work concerning subject-verb agreement processes (see Bock, 1995 for a summary) . A main finding of Bockís work is that participants often make agreement errors in sentences where an intervening plural noun follows a singular noun, for example, The bill from the accountants were late. The question, then, is what causes these errors and when in the production process do they occur?
Research has shown that in English, errors are less likely to occur when an intervening plural is a pronoun, suggesting that case-marking may be an important variable in error production. The present study examined agreement processes in Russian, where both nouns and pronouns are case-marked. If case-marking is the crucial factor, then plural intervening pronouns should elicit as many errors as do intervening plural nouns. If pronouns elicit fewer errors, then the noun/pronoun difference in English has a different explanation.
Participants: 32 native Russian speakers
Design and Materials: Thirty-two stimulus quadruplets were created. Each stimulus item consisted of a noun phrase in isolation, followed by a complex phrase consisting of a singular subject noun followed by a prepositional phrase modifier. This modifier contained a noun that was either singular or plural and either a common noun or pronoun. Here is an example:
Noun SS condition: Buxgalter. Schet ot buxgaltera...
Results: Analyses of variance showed a significant difference in performance on the number condition (SS vs. SP conditions (p<.006), with more errors in the SP condition). Error rates in the SP conditions are comparable to the rate in English when the intervening plural is a pronoun. There was no effect of type of intervening noun (pronoun vs. noun), and no interaction of number with type
Conclusions: Thus, nouns and pronouns were equally likely to cause errors in agreement in Russian. This suggests that case-marking provides some insulation against errors.