Studies of interaction in speaking assessment have highlighted problems regarding

the unequal distribution of interaction patterns in interviews versus paired formats (Van Lier,

1989; Young & He, 1998). These studies, however, only looked at verbal interaction

features, and no attempts in these studies were made to investigate both verbal and nonverbal

interaction features elicited in interviews versus paired formats. Therefore, the purpose of

this dissertation is to examine the effects of task types on the elicitation of interaction

features in speaking assessment. The study has three aims: investigate which interaction

features raters noticed when evaluating interaction in the individual and paired discussion

task; investigate if these different tasks elicited similar or different interaction features; and

examine the extent to which these features contributed to variance in interactional

competence scores across task types. To achieve these goals, an individual scripted interview

and a paired discussion task are analyzed using a mixed-methods approach.

A qualitative analysis of 32 verbal reports from four raters judging test takers’

interactional competence showed that raters attended to five nonverbal and 14 verbal

interaction features in both tasks. An interaction ability scale was developed based on those

features. Two raters evaluated 68 test-taker performances both analytically using the scale

and holistically using an interactional competence scale. The analytic scores were used to

conduct an exploratory factor analysis which revealed four factors: body language, topic

management, interactional management, and interactive listening. Logistic regression

analyses showed that while the individual task elicited more topic management features, the

paired discussion task elicited more interactional management features. Then, the holistic and

analytic scores were analyzed using simple regressions, which showed that body language

and topic management features predicted interactional competence scores in the individual

task, whereas body language, topic management, interactional management, and interactive

listening features were predictors of scores in the paired discussion task.

The findings suggest that both nonverbal and verbal interaction features are important

in the interactional competence construct. The paired format provides test takers with more

opportunities to demonstrate their interactional ability. The study also suggests the

importance of rater training in evaluating interactional competence.