A corpus study of English conditionals

Abstract

This study analyzes the two most frequent ways of marking conditionals in English through the use of if and unless. The occurrence of these two ways of marking conditionals in authentic texts contained in two large computerized corpora is the main focus of the study. The Brown corpus is a representative sample of American written English while the LOB corpus is a parallel representative sample of British written English. Both corpora are taken from a wide range of genres. The analysis of these two conditionals in the two corpora identifies 117 if and 20 unless syntactic patterns. The 117 if patterns are classified into four semantic categories, and the 20 unless patterns into three semantic categories. The relative frequencies of both the syntactic forms and the semantic categories of these two conditionals in the Brown and LOB corpora are then established. The study uses the frequency data obtained from the corpus analysis to examine the adequacy of a small selection of course books for the learning of English as a second or foreign language. Analysis of the corpora and of the ESL/EFL texts show the following: 1. Seven out of 117 if syntactic patterns and five out of 20 unless syntactic patterns account for most of the conditional sentences used by native speakers of English. 2. For if, the conditional clause precedes the main clause as by far the most common order, whereas for unless, the reverse is true. A subsequent study among native speakers of English who rated the acceptability of a group of sample sentences from the corpora, however, showed that neither main clause nor if clause first was strongly preferred. 3. Factual if and unless conditionals rank the highest, closely followed by predictive conditionals. 4. If...not and unless have different semantic functions. 5. American and British written English do not differ significantly in the use of if and unless conditionals. 6. Most ESL/EFL texts examined neither provide accurate descriptions of conditionals nor reflect how conditions are typically expressed. The study suggests that the frequency information on if and unless conditionals obtained from the computer corpus analysis can provide a better empirical basis for the development of materials for teaching English.