The presence of the phonemes /ü(:) ö(:)/ in OE and their interpretation as front rounded monophthongs are universally accepted. Yet the rise of this subsystem remains one of the most controversial problems of Old English phonology. The brunt of the problem lies in the interpretation of the immediate results of i-umlaut and, concurrently, the vowel system immediately before i-umlaut. In our analysis of PrOE vowels we follow W. F. Twaddell and M. I. Steblin-Kamenskij, interpreting the oppositions /u(:) o(:)/ – /i(:) e (:)/ as rounded versus unrounded, and the i-umlaut of /u(:) o(:)/ as resulting directly in the front rounded monophthongs /ü(:) ö(:)/. Yet the results of the i-umlaut of /a(:)/, viz., the front unrounded monophthongs /æ(:)/, suggest the interpretation of the PrOE oppositions /a(:)/ – /æ(:)/ as back versus front. The occurrence of /a(:)/ in velar positions points to the same. For example, in WS /æ1:/ was replaced by /a:/ in such forms as māgas, māga (beside migǣg kinsman’), tālum (beside tǣl ’calumny’), etc. It is to be noted that in Anglian dialects such substitution occurs only before the labial /w/. We may infer from this that in Anglian dialects at least the opposition of the long /a :/-/æ:/ was reinterpreted as rounded versus unrounded. But already before breaking the Anglian /æ1:/ was replaced by /e:/, so that in the period immediately before i-umlaut Anglian dialects must have possessed the only low vowel /a:/ with the back and front allophones [a:] – [æ:] ([æ:] before /ij/). Consequently, the i-umlaut of the long /a:/ in Anglian (and Kentish) means the phonemicisation of its allophones into the opposition /a:/ – /æ:/, whereas the i-umlaut of the short /a(:)/ is but the syntagmatic replacement of this phoneme by /æ/. In WS the i-umlaut of both the long and the short /a(:)/ consists in the syntagmatic replacement of the back /a(:)/ by the front /æ(:)/.
In Kentish the phonemes /ü(:)ö(:)/ were replaced by /e(:)/ already in the late 9th-early 10th centuries. Approximately at the same time or even earlier the purely syntagmatic replacements /ü(:)/ < /i(:)/, /i/ < /ü/, /e/ < /ö/ took place. The most important of these changes is the replacement of /ü/ by /i/ in the neighbourhood of palatal consonants, viz., before ht([x’t]), c(/č/ < /k’/), cg (/dž/ < < /gg’/), g([γ’], /j/), sc (/š/ < /sk’/), and after c(/k/ = [k’]), sc(/š/). This change is most widely attested in WS writings, yet it must have taken place in other dialects as well, especially West Mercian. In many non-WS dialects /i/ from /ü/ in the neighbourhood of palatal consonants may also be due to dialect borrowing; cf. the wide spread of the forms cing and drihten. In the 9th-10th centuries /ö(:)/ was unrounded to /e(:)/ in WS and the greater part of Northumbrian and Mercian. In these dialects the front rounded series was restored with the monophthongization or /eo(:)/ to /ö(:)/. Yet already in the 11th – 12th centuries in Northern, East Midland, and part or Southwestern this series was completely lost; in the rest of the dialects (the greater part of Southwestern and West Midland) the front rounded vowels survived until the 15th century. Therefore the English dialects of the 12th – 15th centuries ran into two major areas, (1) the area or West Midland and Southwestern dialects, preserving rront rounded vowels, and (2) the area of Northern, East Midland and South- eastern dialects, without front rounded vowels; the latter area may be subdivided into the i- and e- areas, reflecting the unrounding of /ü(:)/ to /i(:)/ and /e(:)/, respectively.
The instability of front rounded vowels may be explained to a certain extent by their peripherial status in the system. Their final loss may be connected with the replacement of the correlation short versus long by the correlation checked versus free.