This study is an experimental phonetic investigation of phonological voicing oppositions - specifically, those involving preaspiration as found in Icelandic, Scottish Gaelic, and Irish. Three main aspects of these oppositions are dealt with. - The first involves the production 8f these contrasts. Productions by native speakers of the three languages were monitored using techniques including electroaerometry and photoelectric glottography. Specific attention is directed to the durational correlates of the opposition, and how these vary across languages and for different phonetic environments, including stress variation. The differences and similarities between contrasts which involve preaspiration and those which involve postaspiration are discussed in some detail. In considering production aspects of these oppositions, interest focusses also on the laryngeal mechanisms which control voicelessness and aspiration (pre- and post-). Glottographic data presented suggests that the nature and the amplitude of glottal gesture for a voiceless or aspirated stop is very precisely tailored to the prevailing aerodynamic conditions. The second aspect considered is that of the historical development of these contrasts. Past hypotheses are discussed in some detail, and a more phonetically-based alternative is proposed. The general tendencies of change which affect voicing oppositions and which have traditionally been termed lenition processes are also considered. Suggestions are made regarding the phonetic motivation of such changes, suggestions based on existing research as well as on certain results of this study. Thirdly, the perception of these oppositions is considered. Perceptual experiments are repotted using synthetic and computeredited natural speech stimuli. In the first instance, these tests bear on specific questions regarding the perception of preaspirated stops, which arise out of the production data in the earlier chapters. In addition, the broader question of the perception of voicing contrasts is discussed. A serious problem facing the researcher in this area (and in the area of.linguistic contrasts generally) is that of how the simple binary linguistic percept is arrived at from the large number of potentially relevant cues. The question arises as to whether there is a single dominant cue or whether voicing detection involves separate monitoring of the various cues which have been postulated in the literature. A hypothesis is proposed to the effect that the linguistic percept depends on a judgement regarding the relative ratio of two properties of a larger, Vowel+Consonant unit. The multiplicity of postulated cues may be an artifact of an excessively segmental approach; these seemingly disparate cues may be seen instead to be feeding into a cumulative decision based on a syllable-type unit.