Low-Level Lengthening
There seems to be a consensus of opinion among phoneticians that vowels lengthen progressively according to the hierarchy of following consonants shown in (1), with even greater duration pre-pausally (House and Fairbanks 1953, Peterson and Lehiste 1960, House 1961, Delattre 1962, Chen 1970).
(1) 

This `voicing effect' process certainly seems to apply in most varieties of English; it may not be characteristic of Northern English dialects (Roger Lass, personal communication), although again experimental evidence is lacking here. Peterson and Lehiste (1960) give measurements for American English which show that, while preceding consonants appear to have a negligible effect on the duration of following vowels, average vowel durations can more than double between the shortest context ([-p]) and the longest ([-z] for short vowels, [-Ʒ] for long). Wiik (1965) gives similar results for RP. The most salient effect seems to be that of voicing: the vowel durations of Peterson and Lehiste's informants before voiced as opposed to voiceless consonants, in otherwise identical environments, formed a ratio of 3:2.

There are likely to be universal phonetic factors underlying the variable lengthening effect of following consonants (Zimmerman and Sapon 1958), perhaps reflecting the operation of a type of compensatory lengthening: if roughly the same time is allotted to each VC sequence in an utterance, and voiceless consonants are longer than voiced, vowels before voiced consonants may lengthen to maintain a quasi-constant duration for the VC sequence. Whatever the physiological or articulatory motivation for this lengthening process, evidence from Gandour, Weinberg and Rutkowski (1980) indicates that it operates as a language specific phonological rule of English. Gandour et al. argue that, if the voicing effect is purely physiological and due to laryngeal adjustment of some kind, it would not be expected in oesophageal speech. They tested three normal adult males and three laryngectomized patients, all in their fifties. All had hearing in the normal range for their age-group, and none had speech impediments. The laryngectomees were recommended by speech pathologists as having fluent, highly intelligible oesophageal speech. Gandour et al. found that the duration of vowels before voiced consonants was significantly longer than before voiceless consonants for both groups, at the level of p 0.01. There was no significant difference across speaker groups for duration before voiceless consonants, but vowels were significantly longer in voiced contexts for the oesophageal as opposed to the normal speakers. This extra length might be attributed to the slower average speaking speed for the laryngectomized patients, at 2.01 as opposed to 2.97 syllables per second; but if this were the only determining factor, we would expect the relative length difference to carry over to voiceless contexts. It would therefore appear that oesophageal speakers are signalling the voicing effect lengthening, and perhaps even exaggerating it, indicating that `natural phonetic tendencies have apparently been expanded into a phonological rule of the grammar' (Gandour, Weinberg and Rutkowski 1980: 150). I shall call this lengthening process, which is dependent on the voicing effect, Low-Level Lengthening (LLL), and will argue below that it applies postlexically, while SVLR is a lexical phonological rule in Scots and SSE.