Comprehension, Irritation and Error Hierarchies

to comprehensibility of interlanguage. "Irritation" is also a topic of considerable interest. An assumption underlying the study of irritation is that even if nonnative speech is comprehensible, the form of the message may be associated with a negative affective response from the native speaker. Certain formal or mechanical errors may be considered more important than others by native speakers. The ranking of errors by comprehensibility, irritation or other criteria is referred to as an error hierarchy. An interesting, detailed example of an error hierarchy based on both comprehensibility and irritation may be found in an article by Chastain.2 The data to be discussed is part of a larger study which compared attitudes of U.S. bilinguals and Spanish-speaking learners of English toward native and nonnative speech samples.,' Factor analysis of the data from this study demonstrates that the concept of irritation needs clarification. Native speaker response to interlanguage is not solely the result of irritation but rather of evaluation as well. In the present paper, the response of a native speaker toward language is shown to be largely evaluative and only slightly affective. Such a response is referred to here as a language attitude. Data will be presented below which demonstrate that error hierarchies based on language attitudes are tentatively valid with reference to very low language ability. Comprehensibility is related to phonological characteristics of primitive interlanguage more than to the morphosyntactic ones, but not sufficiently so to warrant a hierarchy in which phonology is ranked first because of its impact on comprehensibility. The findings show that problems with morphology and syntax (referred to simply as morphosyntax throughout this paper) are more salient to native speakers than those with phonology, despite the fact that comprehensibility is apparently somewhat related to phonological accuracy. The practical implications of these findings are that the internalization of morphosyntactic rules should produce a certain level of accuracy in order not to be salient to native speakers and that teachers should encourage accuracy in pronunciation not simply by drilling points of crucial contrast,4 but also by encouraging the student to view pronunciation in a global way and thereby to develop a general 'feel' for what sounds socially appropriate. Many language teachers, and perhaps even most, use textbooks, teaching techniques and testing methods which emphasize the formal accuracy of student speech. Errors in form generally are not considered acceptable by the teacher. Studies which compare teacher attitudes and native speaker attitudes show that some teachers are considerably more severe in their ratings of nonnative speech on formal aspects.' If these ratings are representative of teacher attitudes, then 'irritation' and proof of the existence of native speaker irritation is an issue of considerable importance. Studies of irritation generally involve an attempt to develop an error hierarchy which ranks errors in terms of the degree to which native speakers are annoyed by them. If native speakers are annoyed by certain errors, then perhaps the teacher should strive to eliminate those errors in the classroom. If, on the other hand, natives are not annoyed by any particular error, then another approach may be more appropriate. Oller has theorized the existence of a global language factor which appears to explain common variance in different tests of language ability." Recently, this global language factor has been challenged on the grounds that a general or 'g' factor can be extracted by analysis of any data set.7 Although the general factor demonstrated by Oller may be an artefact of the type of factor analysis used, Oller still