Schlesinger argues that language acquisition starts of with semantic assimilation, according to which the relation between an agent and an action verb is the prototypical core that triggers semantic and formal assimilation of input into this relation. This semantic relation then develops into the formal subject-predicate relation.
There are several drawbacks of this semantic assimilation. First, the basic relational categories are ambiguous in that some verbs cause categary overlap and some belong to none of the categories. Second, this theory misses out lexical gaps between semantic and formal similarities. No overgeneralizations were observed across the relational categories. For instance, the child never uses stative verbs in progressives, e.g., *I was waiting ice cream although he heard formally similar pattems with action verbs, e.g. I was eating a cake. Third, early production of it + weather verb pairs called into question how the child distinguishes such semantically empty subjects from agents. Finally, different perspectives in the same event can alter verb and agent identification. As not all verbs denote action, not all nouns denote object or person. The child refers to the same concept both by verb and noun, e.g. a lick/licking, at two years of age.
In conclusion, semantic assimilation ignores the child's early awareness of lexical and grammatical requirement on the verbs. Basic semantic information must cluster not in the agent-action core ralation, but in individual verb meanings, on the basis of which the concepts of agent and object should develop.