Book excerpt: Knowing that the so-called voiced and voiceless stops in languages like English and German do not always literally differ in voicing, several linguists — among them Roman Jakobson — have proposed that dichotomies such as fortis/lenis or tense/lax might be more suitable to capture the invariant phonetic core of this distinction. Later it became the dominant view that voice onset time or laryngeal features are more reasonable alternatives. However, based on a number of facts and arguments from current phonetics and phonology this book claims that the Jakobsonian feature tense was rejected prematurely. Among the theoretical aspects addressed, it is argued that an acoustic definition of distinctive features best captures the functional aspects of speech communication, while it is also discussed how the conclusions are relevant for formal accounts, such as feature geometry. The invariant of tense is proposed to be durational, and its 'basic correlate' is proposed to be aspiration duration. It is shown that tense and voice differ in their invariant properties and basic correlates, but that they share a number of other correlates, including F0 onset and closure duration. In their stop systems languages constitute a typology between the selection of voice and tense, but in their fricative systems languages universally tend towards a syncretism involving voicing and tenseness together. Though the proposals made here are intended to have general validity, the emphasis is on German. As part of this focus, an acoustic study and a transillumination study of the realization of /p,t,k,f,s/ vs. /b,d,g,v,z/ in German are presented.