This study investigated whether different groups of native speakers assess second language learners' language skills differently for three elicitation techniques. Subjects were six learners of college-level Arabic as a second language, tape-recorded performing three tasks: participating in a modified oral proficiency interview, narrating a picture depicting a story, and reading a text aloud. The recordings were rated by three groups: 15 native Arabic-speakers teaching in the United States, 31 non-teaching native Arabic-speakers living in the United States, and 36 non-teaching native Arabic-speakers living in Lebanon. Ratings were given both holistically and on a nine-point scale of proficiency. Three response dimensions were assessed specifically: grammar/pronunciation; creativity in presenting information; and amount of detail provided. Results indicated variability of performance across tasks as well as between individuals. In sum, it was found that oral ability, tasks, and raters all affected students' scores. Further analysis of the effects of different tasks and of different raters on assessment of second-language performance is recommended. A 23-item bibliography and analysis data are appended. (MSE)