This experiment continues the GUME experiments investigating methods for teaching English to Swedish school pupils. The aims of this particular study are twofold: (1) to analyze and classify the errors made by Swedish pupils in an oral English test in an effort to establish error patterns; and (2) to describe how the twelve most frequent errors in a systematic classification of errors are used to form the basis of an acceptability and an intelligibility test, which is subsequently submitted to native Englishmen. Such an experiment raises the possibility that the objectives of intelligibility and communicative comprehension may be more important than grammatical correctness. Included here are reports on the oral tests administered to Swedish students, the analysis of error patterns, the intelligibility experiment with English informants, conclusions, and discussion of implications. Appendixes provide additional details. (VM)