The idea of reducing the redundancy of a verbal message in a statistical way is presented as a practiced technique of language testing. Considering the temporality of speech comprehension, and the necessarily sequential intake of information, these cues may include the serial order of elements and transitional probability. To give the background of reduced redundancy tests, the contructivist view of listening comprehension is outlined as a creative, active cognitive operation with several implications: (1) it means that processing is facilitated by the linguistic and pragmatic organization of the message, together with its presentation in context; (2) it implies that memory, especially short-term storage, is an essential part of comprehension; and (3) it implies that the native listener is only partly bound by the properties of the signals he receives. Two reduced redundancy techniques of listening comprehension testing are reviewed that present the learner with messages that do not contain all the information they normally carry. In the noise test and the aural cloze test, the learner needs to mobilize his total awareness of the linguistic and pragmatic structure. Preliminary observations on an experimental test of aural cloze with Finnish learners imply that the tests have instructional value and may have a stronger theoretical basis than the completion tasks traditionally used to measure listening for details. (NCR)