The functional hypothesis of language, based on the assumption that the referential function of language is paramount, is discussed as it applies to interlanguage, the second language spoken by less than proficient native speakers of another language. The presentation includes: (1) a review of the evidence of previous empirical investigations of the functional hypothesis in various different forms of language; (2) a description of two studies of English interlanguage in two learner groups with very different first language backgrounds (Chinese and Czechoslovak) that show how little functional constraints affect the form of interlanguage; and (3) a discussion of the consequences for learners if interlanguage is indeed a highly inefficient means of communicating referential information, and the options available to learners and interlocutors to repair misunderstandings. It is concluded that interlanguage is a poor vehicle for the communication of representational information, and that this weakness is not peculiar to any one group of learners. Rather it is a feature of all interlanguage, including dialects, languages of wider communication, and pidgins and creoles. This referential function tends to require a significant amount of repair through interaction if information is not to be lost altogether. Study data are appended. (MSE)