Testing the agreement/tense omission model using an elicited imitation paradigm

Area of Science:

• Child Language Acquisition
• Developmental Linguistics
• Psycholinguistics

Background:

• The AGREEMENT/TENSE OMISSION MODEL (ATOM) predicts low rates of non-nominative subjects with agreement-marked verbs in child language.
• Previous naturalistic studies suggest children produce these errors more frequently than ATOM predicts.
• Understanding early grammatical knowledge requires accurate assessment of error types and frequencies.

Purpose of the Study:

• To test the ATOM's prediction regarding the frequency of non-nominative subjects with agreement-marked verbs in children's speech.
• To evaluate the efficacy of the elicited imitation paradigm in assessing early grammatical knowledge.
• To determine if non-nominative subject errors with agreement-marked verbs can be disregarded as noise in the data.

Main Methods:

• An elicited imitation paradigm was employed, asking children to repeat sentences.
• A screening procedure identified children producing non-nominative subject errors (e.g., 'her' for 'she').
• Twenty-four sentences with 3sg feminine pronoun subjects and agreement-marked verbs were used.

Main Results:

• All five participating children produced non-nominative subjects with agreement-marked verbs.
• Error rates were not significantly different from chance for most children.
• Three children with a high chance-expected error rate produced errors significantly above 10%, contradicting ATOM predictions.

Conclusions:

• The findings challenge the ATOM's prediction that non-nominative subject errors with agreement-marked verbs are negligible noise.
• Elicited imitation is a viable method for assessing children's early grammatical knowledge, complementing naturalistic data.
• Child language acquisition may involve more complex grammatical computations than proposed by ATOM.