Validity of Task Choice on Global and Specific Pronunciation

Second language pronunciation tasks are the means to eliciting speech performance; however, sparse research has considered task choice from a validity and methodological perspective. Drawing on validity (Harding, 2018) and task theoretical frameworks (Saito & Plonsky, 2019), this study investigates the most common tasks (i.e., read-aloud, spontaneous speech, elicited imitation, and picture description) in second language pronunciation research (see Thomson & Derwing, 2015) and triangulates their effect on specific pronunciation (i.e., segmental and suprasegmental properties) and global pronunciation (i.e., listeners’ judgments of comprehensibility, accentedness, and proficiency). Fifteen speakers participated, and fifty-six untrained listeners rated speech across four tasks (two controlled and two spontaneous). One-way repeated measures ANOVAs and regression analyses were conducted. Speakers had more vowel and consonant deviations and a faster speech rate on the read-aloud task and more sentence prominence on both of the controlled tasks. Overall, speakers were rated to be more comprehensible and proficient and have less of an accent on controlled tasks as opposed to spontaneous task types, where they were rated more severely. Task-associated pronunciation properties differentially predicted listeners’ ratings, especially pitch range. Recommendations are made to strengthen validity claims by aligning interpretations of pronunciation ability with task types.