In dealing with the elimination of Slavic liquid diphthongs (art, tart, turt), Slavic linguistic scholarship (Jakobson’s 1952 article, Bethin’s recent book, Slavic Prosody, and many other studies) has often used modern reflexes of a or o as evidence of either a generalization of length or shortening at the time of the elimination of these diphthongs. However, the behavior of such words as Polish król is a strong counterargument to that claim, since the o acts like a long vowel. A short o would have given modern /u/ only in pre-jer position, yet here we have /u/ in all environments.
My paper argues that in the tart groups, a moraic r was always changed to a non-moraic liquid, and this always caused a lengthening of the vowel as compensation. The only reason for the Northern Slavic o is the fact that the rounding of short a > o occurred before the vowel lengthening. In other words, the relative chronology of the a > o change determines the basic difference between the trot/torot reflexes and the trat type.
However, the issue is not quite as simple as that. The behavior of North Slavic anlaut art groups does indeed allow us to use modern o/a reflexes as evidence of the original quantity. (Northern art reflexes are split not on the basis of “intonation,” as commonly stated, but vowel quantity.) North Slavic (including Czech) demonstrates the original inherited quantity in anlaut position and the fact that the r never became moraic in this anlaut environment, in contrast to inlaut tart, where it did. The non-moraic r made art an early candidate for metathesis in the North, but this rule reached the South later, together with the rule for the elimination of moraic r in the inlaut tart diphthongs. Therefore, the North conservatively retains quantity in art (rot and rat), but lengthens the vowel in tart, while the South treats both alike (rat and trat).
The Northern chronological difference in the art and tart metatheses also explains East Slavic polnoglasie. In the earlier period, the original bi-moraic a of rat could be treated as a long vowel, due to the presence of phonemic quantity. Later, after the lengthening of tort to bi-moraic toort, East Slavic was losing the quantitative opposition, so that toort was accommodated as a two-syllable sequence [to-ort]. Metathesis then affects only the second of these syllables, giving the familiar polnoglasie torot.
In virtually all Slavic zones, trat reflexes go with the presence of syllabic r, while trot/torot reflexes go with an absence of syllabic r. In both cases, the reason is that North Slavic had an earlier change of non-mid short vowel to a mid vowel. The first of these cases is the a > o change. The second is the Northern change of high i,u > mid, in so-called “strong” position, which included the position before the moraic r of the turt diphthongs. North Slavic experienced this change in time for the moraic r to be eliminated. The mid vowel created a greater sonority difference between the new vowel and liquid than had been the case in turt. Following a regular pattern of rules for monophthongization, greater sonority differences caused a non-high first component to prevail as moraic, (er/or/ar, etc.).
In Czecho-Slovak and South Slavic, though, high vowels did not change to mid in time for this to occur. Lesser sonority differences caused the first component to assimilate to the second one (two r moras, i.e., a long syllabic r). In other words, Czecho-Slovak and South Slavic experienced both “Short > Mid I” (i.e. a > o) and “Short > Mid II” (i/u > mid) too late to affect tart/turt, in contrast to North Slavic, which obtains its trot/tart or torot/tort type reflexes as a result of early Short > Mid I and II.
The set of changes enumerated above presents a rather different picture of the Slavic liquid diphthong evolution. This paper will argue that it is justified by the facts.