24 Alice Faber, Marianna Di Paolo, and Catherine T. Best
shifts described in section 2.2, and, indeed, prior to open syllable lengthening
(OSL). In these dialects (and in other northern dialects), Old English Ɨ STONE
had not backed to /ܧࡃ/. In most of these dialects, NAME and STONE ultimately
merged; that is, æ, when it lengthened to [æࡂ], merged with Ɨ. However, in a
few parts of the West Riding of Yorkshire, STONE merged with HEAP/SPEAK,
as /ԥܼ/, while NAME varies between /ԥܼ/ and /eԥ/. Assuming that OSL affected
all short vowel qualities at the same time, Johnston’s reconstruction involves
an early merger of reÀ exes of OE Ɨ STONE and ܭࡃ HEAP; following OSL, ܭࡂ
SPEAK merged with STONE/HEAP while æࡂ NAME remained distinct.
Minkova (1982) demonstrates that OSL was a concomitant of the loss of
¿ nal -e. According to the authorities cited by Minkova, both OSL and loss of
-e were underway in the north early in the 13th century. It follows then that
the bottom half shifts began, at least in some areas, by the start of the 13th
century. The merger of ۘ NAME and æj DAY did not reach southern Stan-
dard English until the early 17th century. The raising of ܭࡂ/ܭࡃ SPEAK/HEAP
to /i/ (and the consequent merger with Ɲ FEED) did not take place until the
16th century in non-standard (especially northern) speech and late in the 17th
century in the standard (Dobson §106, §107); thus, Stockwell and Minkova
(1988) exclude this ¿ nal raising to merger from the vowel shift proper. How-
ever, in our view, the ultimate merger of ܭࡂ/ܭࡃ HEAP/SPEAK with Ɲ FEED
in /i/ is the last step in a sequence of events set in motion by the late Middle
English lengthening of *ܭ SPEAK in stressed open syllables.
2.3.1 Step 1: The lengthening of *ѓ
Despite the crowded Middle English front vowel space, the lengthening of ܭ
SPEAK did not immediately lead to merger. This claim is supported by two
pieces of evidence. First of all, ܭࡂ SPEAK remained distinct from ܭࡃ HEAP at
least until the middle of the last century in a wide area of northwestern Eng-
land, according to the SED, and more scattered areas elsewhere.^10 This area
(i nd icated i n Fig u re 1.2) i ncludes La ncash i re a nd t he West R id i ng of York sh i re,
as well as Cheshire, Derbyshire, and parts of Staffordshire, Shropshire, Suf-
folk, Cornwall, Devon, Hampshire, Gloucestershire, Wiltshire, Oxfordshire,
and Buckinghamshire. The typical contrast is something like /ԥܼ/ in HEAP
vs. /ܭj/ in SPEAK (Orton and Barry 1969). In much of the area in which ܭࡃ
HEAP and ܭࡂ SPEAK are distinct, they are distinct from Ɲ FEED as well. As
in Standard English, the reÀ ex of Ɲ FEED tends to be /i/. Given the retention
of a distinction between ܭࡃ and *ܭࡂ in some areas, with both patterning as long
vowels, ordinary comparative methodology requires reconstructing a two-
stage process in the ancestors of Southern Standard English (cf. Lass 1988):