Titles like this one baffle me. Actually, not the whole title, just the middle part. I understand conjunction in a normal sort of way, I think. If so, the conjunction we have here is not of equals but of theme and exemplification – i.e. something like ‘My main burden is instances of sound change that can be found in English language history.’ It is further entailed that based on what the author has been able to say about these sound changes, new light will be thrown on the general problem of sound change. That leaves room for growth: I'll read it that way. But before details, a few words of appreciation. The book is fun to read, written in the elegant yet casual style which is so successfully cultivated by the British university system, and in America so rarely achieved outside the New Yorker magazine. Like playing a violin, you have to start young.

I will assess JJS's explanations only of the main sound changes he deals with, except for Grimm's Law, to which he devotes half a chapter (second half of chapter 3). It is certainly an important discussion, but I have nothing to contribute to it. His theory is based on contact between Germanic tribes and the Raeti (a mixed Celtic-Illyrian people who lived in the Alpine passes of Austria and Switzerland) in the five centuries before the birth of Christ. He does not (surprisingly) introduce any comments about the glottalic theory, which had a controversial run in the 1970s. I will also ignore a small number of changes that are well known and transparent on universal phonetic grounds, such as [e > i /–NC] (e.g. bindan). All of the latter are familiar to most readers of this journal.

The sound changes to which I shall give serious attention, then, are Old English breaking, Middle English lengthening of stressed vowels in open syllables, and Early Modern English vowel shifting. I should point out that JJS is very generous in setting the scene for these changes: he provides an appendix 1 entitled ‘The principal sound changes from proto-Germanic to Early Modern English’ in which he lays out examples and contextual details for the intensive discussions of vowel breaking, lengthening, and shifting – those remain the true focus of the book. There is a very short appendix 2, dedicated entirely to his explanation of Middle English open syllable lengthening of i and u, which is very welcome – given that i and u have been controversial since the early 1950s. How did they – in words like weevil, beetle, sweep, evil, wood (OE wudu), door (OE duru) – how did they get unaligned with the matching short vowel? Do we have to posit a sound change (‘lowering of ME short high vowels before lengthening’, or the like)? I won't give away his secret, except to say that it depends on etymological twists rather than structural generalizations that we're familiar with.

I begin, then, with Old English breaking. It took place in environments that have usually been phonetically classified as ‘retracting’ – eald ‘old’, heard ‘hard’, eahta ‘eight’, seolh ‘seal’, eorl ‘noble’, feoh ‘property’. The most even-handed discussion of these is to be found in Hogg's Grammar (Reference Hogg1992: 16–25), which concludes by accepting the traditional position, but with a significant reservation:

For reasons which are not clear to me, JJS puts great weight on the existence (which has never been denied) of the minimal pairs in –rC by metathesis vs. –rC by breaking: cf. aern ‘house’ vs. earn ‘eagle’. In section 2.5 JJS takes the position that ‘breaking would seem to be a realizational or phonotactic development’ (p. 94). It certainly is: therefore the mystery is why aern and others survived without breaking; forms like earn are the natural development, forms like aern are the ‘sports’ which should have suffered an early demise because they violate canons of phonetic rationality. In a phonetically based theory, they are less favored than the corresponding (apparently contrastive) broken forms. As far as I can find, once these forms were pointed out (by me and Barritt in 1951), no one tracked them further. Mea culpa. In 1976, however, they were finally conscientiously tracked by Andrei Danchev and published in a Faculty of Letters Annual publication, University of Sofia. Danchev found exactly a dozen pairs of the aern–earn type (p. 62). While the small number doesn't dismiss the argument altogether, it does diminish the value of it substantially.

Real diphthongs (bi-moric) are a natural source for another type of ‘breaking’, namely those with on-glide development in the onset of the syllable. This is a powerful argument if a nucleus of two morae is the normal weight requirement for this type of diphthong to arise: ‘Particularly apposite, I [JJS] would argue, is the fairly frequent development from these forms, of rising diphthongs in varieties of Middle English, such as Middle Kentish yealde “old”.’ However, the question of weight remains (as admitted in Samuels 1952: 24). Here, in the opinion of JJS, an opinion widely shared but not universal, ‘David White's 2004 arguments concerning the existence of short diphthongs seem very strong, namely, that “short diphthongs” are rare in languages of the world; thus it seems eccentric . . . to assume them for Old English’ (p. 94 fn.).

In Hogg's Grammar we find summarized his position on the ‘short diphthongs’ (Reference Hogg1992: 17):

Unfortunately this elegant argument presupposes that functionally contrastive length did exist or could exist from the beginning, but that is what is at issue and cannot be presupposed. See the summation of David White's position, below (quoted from JJS, 94fn. with enthusiastic approval expressed by Smith). Whether you are persuaded by White's massive accumulation of evidence, fully published in SHEL-2 and showing that diphthongs of this form do not exist in natural languages –

(i.e. a nuclear vowel followed by a glide vowel, and the two somehow counting as just one mora) – or by his summation of relevant data, you have to wonder why there are no cases that come close to giving us a basis for counting this question as ‘solved’. I don't think real cases actually exist, but don't believe me; believe David White, who has worked very hard at finding real cases. I raised (with Westbrook Barritt) this version of ‘the digraph problem’ in 1951, though this bibliographic fact seems to have escaped Smith's attention as, unhappily, Daunt's 1939 proposal about Old Irish spelling practices escaped my own attention in 1951. But the theoretical point – crucially, that the digraphs that emerged by ‘breaking’ represented single-mora units in Old English phonology – did not get overlooked. Practically everything of interest under this general heading, including in particular Anglo-Saxon scansion, is linked to the condition that ‘short diphthongs’ count as single-mora phonological units. In CHEL-1, published in the same year as Hogg's Grammar, he accepted the mora-based argument:

It seems to me that this was, in 1992, very close to full agreement on fundamental issues in ‘the digraph controversy’. But before leaving this topic, I want to review Marjorie Daunt's analysis of 1939, though I cannot develop it in the detail it deserves. It has been short-changed by almost everyone who has opened their computers on the subject.

Miss Daunt's critical insight is this: ‘A supporting reason in favour of regarding the writings of eo and ea as only variants of [e] and [æ], in certain conditions, is that in the overwhelming majority of cases the forms to be traced in Middle English show exactly the development to be expected of [e] and [æ], e.g. OE earm, ME erm, arm’ (Daunt Reference Daunt1939: 128). Miss Daunt did not deal with the technical questions of conditioning of allophones, which were extensively (and to some extent incorrectly) dealt with by Stockwell & Barritt, Hockett, and others; but it is doubtful whether that issue is very important, since the conditioning has always been obvious in the mainline examples. They suffice to show that the so-called retracted allophones, at least, were, by and large, not in contrast with [e] and [æ], and these examples can be multiplied by a factor of 10–20. Also, they had a short life – they didn't survive into Middle English. If we take them as the norms in retracting environments, as I do, and as she did, there is really not much left to be controversial about. Neither JJS nor I have an interpretation of OE breaking that stands up to all the questions that have been raised; and different answers always leak, in some respect. I'm content to leave the matter there, since I no longer believe there is a coherently specifiable environment in which to dump the retracted allophones.

Quantity changes from OE to ME include homorganic lengthening (child – lC, climb – NC, comb – NC) – and Middle English Open Syllable Lengthening, MEOSL – nama, macian, throte, duru > name, make, throat, door. Both processes are argued to be efforts to establish approximately isochronous intervals between stressed syllables. I don't see how anyone can prove it, reasonable though it sounds. Smith's discussion is clear and persuasive, though the topic has worn thin in recent years. Minkova had MEOSL basically right in 1982. Ideally, a schwa drops off the end of the word and the stressed syllable becomes bi-moraic. But on that scenario, isochrony is irrelevant to the explanation, since MEOSL is a compensatory process which maintains length at the foot level, whereas isochrony is a rhythmic adjustment across several feet.

Vowel shifting – ‘the Great Vowel Shift’: Smith's discussion of this topic faces some difficulties, but on the whole it is a remarkably courageous effort to draw a complex picture linking social matters with the on-going complexities of the vowel shift. I do not have enough space to sort them all out, but I shall present briefly a different way of looking at vowel shifts: and my warrant for this is that it constructs a substantially differentiated linguistic reality within which a shift exists. I am taking the word shift as referring to a type of phonological change in which at least two stages of the change may coexist in a relationship of adjacency, and speakers may use more than one stage in their own speech; and the variation can be functionally loaded (marking class accents, etc.). I shall assign names to the relevant shifts, taking them each as a unifying dialect/accent marker. Every variety of English must make at least one choice (they may make more – see below) as their prototypical candidate for the vowel (for example) of mine, night, bind, find, slight, tide, but they don't have to stop with just one: people who have traveled a lot may be comfortable with several of the diaphones in the current lower range, others may be uncomfortable with any diaphone other than the one found in their Webster's:

1. (1) is intended to be read (iconically within a vowel triangle) from upper left to lower right, and chronologically from top to bottom. The phonetic values are to be read as choices available for the same lexical set in different areas of the world where English is spoken. Each of the members of the set is a diaphone, i.e. a dialectally selectable allophone. They are not mutually exclusive, though one diaphone or another is generally expected to be associated with the usual determinants: class, locale, sex, education, age, history, breadth of travel, and variegation of exposure to available diaphonic options.

2. (2) Organized on the same principles, we have (2), the back correspondents of the front antiperipheral shift, read chronologically from top to bottom and from right to left:

   

Other shifts we need to formulate as units snaking their way through the vowel space on pre-assigned rails (maybe; I'm not certain about how the complicated guidance systems get into the sophisticated speaker's awareness) include at least these:

1. (3) The low front peripheral shift for LOE {æ: > ey > iy} PDE great, mean

2. (4) The low back peripheral shift for LOE {a: > ɔ: > ɔw > ow} PDE know, bone

3. (5) The mid-front peripheral shift for EME {e: > i:/iy} EME specan

4. (6) The mid-back peripheral shift for EME {o: > ow} EME throte

5. (7) The back dissimilation shift for EMdE {uw, ow, ɔw > iw, ɛw, ɑh} you, boat, cause. (7) is unstable in current AmE usage.

For any one accent at any given time, there is a normative value, or several, recognizable for every shift; in particular, different diaphones can coexist, unlike in the standard view of the shifting elements, in which they must count as contrastive units. On p. 128 Smith presents what he calls the ‘traditional model’ of the vowel shift, whereby each member of the shifting set must displace its predecessor in any given position. In figure 6.1 he collapses shifts (1) and (3) into a single continuum. (2) is the shift which goes from high back, down through center, to low front, with early cut-off at the center level along much of the US east coast (inland to the Piedmont). (1) and (2) are recognized as ‘Canadian raising’ in the dialect literature. (7) is a recent comer in America and is recognized as standard only in the middle value (as in boat). The other two values are norms throughout the Midlands and the South but they are heard as nonstandard in North-Central America.

What I have tried to outline above is a new way of looking at vowel shifting. In this view, different stages of the various shifts can coexist, which is a fact that matches the reality of which Virginians, for example, are generally aware, namely that house and houses have different nuclei in the stressed syllable, though if you ask them to produce it you get something like the vowel of the boot words in the singular form, but in the plural the vowel sounds more like {æw}, the vowel generally heard in America for the count words. The crucial claim that this formatting makes is that multiple members of a diaphone may, and usually do, coexist as historic stages in the shift (they don't displace one another, necessarily, which they must do in the standard picture); at the same time they can coexist as contemporaries, becoming living complementary allo-forms in geographical, social, and stylistic distribution. This information is lost in the familiar diagrams along the lines drawn by Jespersen down to Smith.

My discussion has been mainly by way of suggesting alternative ways of looking at some of the phenomena discussed by JJS. His book is stimulating and challenging, surely a fine addition to the excellent literature on English sound change.