Discussion Forum—A Way with Words, a fun radio show and podcast about language

Well, I can say that there has been a lot of scientific research on resolving homophone and homograph meaning. I have found a raft of very cool Psycho-Linguistics and Neuro-Linguistics experiments built around this very process. There does not appear to be universal agreement on the terminology, since the terminology used in a particular paper relates to the particular focus of the experiment.

In your case, I think you construct a strong bias for the imperative by omitting the subject: contextually, you are forcing folks through word order to finalize their understanding the verb as an imperative, a decision they will soon regret. Only when you reach the I pronoun are you forced to reinterpret the sentence, and re-resolve the verb as first person, past tense.

Most language models posit three processing levels: lexical access; lexical selection; lexical integration.

With homographs (and homophones in listening) we first "look up" all known meanings (lexical access), throw out the ones that can't fit (lexical selection) and, if required, juggle all the remaining as we proceed. In most contexts, lexical selection is very quick, especially since the preceeding context can well contain enough information to conclude the lexical selection process. Finally, we put it all together (lexical integration) into an aggregate meaning.

In your case, you trick us, and we conclude the lexical selction (resolved to the imperative) but then the lexical selection fails at the pronoun I, and we start over. I haven't found any term for this, possibly because it is a contrived situation. If the context didn't trick us by oddly omitting the pronoun, we would keep the ambiguous meanings open in our minds till we could resolve it properly.

If I run into a term for your phenomenon, I'll revisit.