Paper short abstract:

Co-construction (defined as joint construction of a sentence) examples in the literature are found to not form grammatical units. The sentence analysis of these examples stems from a bias in that utterances by separate speakers are atemporally patched together in a quest to find sentences.

Paper long abstract:

Studies (e.g., Ono and Thompson 1995; Hayashi 2003) have discussed a phenomenon called co-construction, where ‘two participants engaged in conversation to jointly produce a single syntactic unit such as a sentence’ (Lerner 1991), as in:

A.

Kinoo atami no hoo de ookina jishin ga …. (Yesterday Atami of direction in large earthquake SUB)

Yesterday, in the Atami area, a big earthquake …

B.

Ee, arimashita ne. (Yes existed PTCL)

Yes, occurred (, didn’t it?). (Suzuki and Usami 2006)

However, since B’s utterance begins with ee ‘yes’, combining the two parts results in a strange sounding string:

?kinoo atami no hoo de ookina jishin ga ee arimashita ne

‘Yesterday, in the Atami area, a big earthquake, yes, occurred (,didn’t it?)’

Even more problematic to the co-construction analysis is observed in what the two utterances do in the talk. A’s utterance introduces ‘yesterday’s big earthquake in the Atami area’ into the talk. B accepts it with ee ‘yes’ and continues to talk about the shared referent by confirming its occurrence as ‘(it) occurred (, didn’t it?)’. Japanese has long been known for the frequent non-overt expression of referents (i.e., zero anaphora), and this characterization based on the shared referent seems more realistic than the co-construction analysis which produces a strange sounding string.

We examine Japanese examples previously analyzed as co-construction in order to show that speakers are better understood to be simply engaged in activities such as continuing to talk and asking about the shared referent. We suggest that the co-construction analysis stems from a common bias in that researchers fail to see how each utterance is produced or for what purpose, and patch together the utterances by multiple speakers atemporally in their quest to find sentences.