Table 3.

Criteria and procedures for determining the best possible correction (BPC) for any utterance and any speaker. Adapted from MacKay et al. [24].

Criterion 1: The BPC corresponds to a speaker’s stated intention when questioned or in the case of corrected errors, to their correction, whether self-initiated or in response to listener reactions.

Criterion 2: When criterion 1 is inapplicable, judges suggest as many corrections as possible based on the sentence and pragmatic (or picture) context and rank these alternative error corrections via procedures 1–4. Then the ranks are summed and BPC status is assigned to the candidate with the highest summed rank.

Procedure 1: Assign a higher rank to BPC candidates that retain more words and add fewer words to what the participant actually said.

Procedure 2: Assign a higher rank to BPC candidates that better comport with the pragmatic situation (or picture) and the prosody, syntax, and semantics of the speaker’s utterance.

Procedure 3: Assign a higher rank to BPC candidates that are more coherent, grammatical, and readily understood.

Procedure 4: Assign a higher rank to BPC candidates that better comport with the participant’s use of words, prosody, and syntax in prior studies (see [24] for ways to rule out possible hypothesis-linked coding biases using this procedure).