Now you see it, now you don't: Middle English Lengthening in Closed Syllables. (1).

Ritt, Nikolaus

Es racht sich ganz einfach, sagte Reger, wenn wir uns

dazu hergeben, em Objekt einfach blind zu akzeptieren

noch dazu uber Jahre und Jahrzehnte und moglicherweise

ein ganzes Leben lang, ja gar verehren und lieben, ohne

es immer wieder auf die Probe gestellt zu haben.

Thomas Bernhard, Alte Meister

"[A copy of] the ACHILLES symbol and [a copy of] the TORTOISE symbol encounter each other inside the author's cranium

ACHILLES: Fancy meeting you here! I'd thought that our dialogue in [... Hofstadter's cranium] was the last one we'd ever have.

TORTOISE: You can never tell with [... humans]. Just when you think [... one]'s done with you, [... another one] drags you out again to perform for his readers" (Hofstadter 1985: 604). It's out of the frying pan and into the fire, I'm afraid. I mean we didn't fare too badly with Hofstadter, did we? With Ritt on the other hand I have my doubts. I don't know what he's up to, but if it's not more original than having us re-born just for the fun of it, I fear the worst.

ACHILLES: Don't be such a pessimist. After all, this is the only life we have, and it's certainly good to see you. So, tell me, what have you been up to, all these years?

TORTOISE: I don't know if this will surprise you, but I've read into the history of our mother tongue a bit.

ACHILLES: Now, that is indeed surprising, because I also happen to have developed an interest in that subject, though I wouldn't really refer to English as 'my mother tongue' ...

TORTOISE: ... and you're still calling me a pessimist, are you? (2)

ACHILLES: I beg your pardon?

TORTOISE: Never mind, you were saying?

ACHILLES: ... that English wasn't my mother tongue, but that I still enjoy reading into its history, because it is full of rather mind boggling problems. Take Middle English Closed Syllable Lengthening, for example ...

TORTOISE: My dear ACHILLES, I suppose you mean OPEN Syllable Lengthening. Have you not studied the Handbooks carefully enough?

ACHILLES: I'm sorry, of course I meant 'Open'. But you see, I have become a bit confused, because I have not only read the handbooks but also some more specialised literature on the subject ... (3)

TORTOISE: ... just to make sure we're talking about the same thing. You ARE referring to the change that explains why we say [meik] rather than [maek], [wi:v] rather than [wev] and [houp] rather than [hop]?

ACHILLES: The very same, but, you see, not all linguists agree that the long vowels, most of them now showing up as diphthongs, were indeed brought about by events and processes that deserve to be called 'Open Syllable Lengthening'.

TORTOISE: But certainly you don't deny that the ancestors of make, weave and hope were, at some stage, maken, weven and hopen, and that these word forms syllabified [ma][ken], (we][ven] and [ho][pen) so that one is justified in calling their first syllables 'open', or 'unchecked'?

ACHILLES: Actually, I could, because the principles of syllabication that you seem to be applying are not uncontested either, (4) but for the present argument I won't. Carry on, please.

TORTOISE: My point is just that if vowels in open syllables came to be lengthened by some processes, then 'Open Syllable Lengthening' strikes me as quite an appropriate term, don't you think?

ACHILLES: No, I don't. I would accept the term only, if being in an open syllable was sufficient for 'triggering' the processes that amounted to the 'lengthening', but it wasn't. And here I'm not just reminding you that only NON-HIGH and STRESSED vowels were lengthened in open syllables, and only if it was followed by exactly ONE more syllable. What I mean is that not even every non-high stressed vowel in the penultimate syllable of a word form seems to have been lengthened, if that syllable was open. Had it been, then, we ought to say ['kl[LANGUAGE NOT REPRODUCIBLE IN ASCII]] rather than [[LANGUAGE NOT REPRODUCIBLE IN ASCII]] [[LANGUAGE NOT REPRODUCIBLE IN ASCII]] rather than [[LANGUAGE NOT REPRODUCIBLE IN ASCII]] and ['di:z[LANGUAGE NOT REPRODUCIBLE IN ASCII]t] rather than ['des[LANGUAGE NOT REPRODUCIBLE IN ASCII]t], which we don't. (5) That's why I don't like the term 'Open Syllable Lengthening', you see?

TORTOISE: I do. A vowel that was lengthened in an open syllable was not necessarily lengthened BECAUSE it was in an open syllable, and you would prefer a more explanatory label, right?

ACHILLES: Yes, only that I would say 'descriptively more adequate' instead of 'more explanatory'. Explanation in historical disciplines is a rather complicated issue, (6) I hear, and it seems, anyway, that linguists are still disagreeing about what exactly happened rather than why it happened.

TORTOISE: Amazing.

ACHILLES: What?

TORTOISE: Well, don't you find it rather strange that this should still be an issue, that for almost hundred years scholars should have been unable to establish what happened? That they should have got their facts wrong for such a long time? What about Karl Luick, what about Henry Sweet, the old masters of English Historical Linguistics? Weren't they careful collectors of evidence, thoroughbred philologists, lovers of little details? I can't believe they simply overlooked the fact that nowadays we don't say ['kle[LANGUAGE NOT REPRODUCIBLE IN ASCII]], ['b[LANGUAGE NOT REPRODUCIBLE IN ASCII]tm] or ['di:z[LANGUAGE NOT REPRODUCIBLE IN ASCII]t]. Were they not worried by that?

ACHILLES: Not particularly, I think. In fact, they were prepared to accept that the effects of what they called sound 'laws' might be partially undone, so that after some time their reflexes would not be as uniform as they might. Take bottom, for instance. Some of the case forms it could assume in Old and Middle English were either not disyllabic or had no open first syllable, as the examples show:

(1) The credyl bothume turnyd on hyghe. (1425 Seven Sag. (P.) 809) A step is not to finde, ne a path of his [a ship's] botme in the flodis (1382 Wyclif Wisd. v. 10) (7)

In these forms Open Syllable Lengthening did not apply, of course, because the vowels were no more in open penultimate syllables, and eventually speakers would find themselves in the awkward situation that the morpheme for {BOTTOM} would get a short vowel in one word form and a long vowel in another ...

TORTOISE: ... like the morpheme for {FOOT} gets an /u/ in the singular and an /i:/ the plural in our language?

ACHILLES: ... yes, a bit like that. Anyway, children, when acquiring the language, would have found such meaningless alternations uneconomic, and preferred to level things out again. Eventually, either one form or the other got generalised over the whole paradigm. In the case of bottom it happened to be the form with the short vowel. (8)

Also, it might have been the case that Open Syllable Lengthening did not affect all dialects of English when it happened. Afterwards, people would have moved about and the unlengthened dialects might have mixed with the lengthened ones.

The bottom line is that both analogical levelling and dialect mixture, none of which can ever really be ruled out, if you really think about it, will always distort the effects of sound laws such as Open Syllable Lengthening, so that apparent counter-examples like bottom don't really falsify the Open Syllable Lengthening hypothesis if there's not too many of them ...

TORTOISE: So, what's the clamour all about then? OSL is fine, isn't it?

ACHILLES: No, not anymore. You see, I was only explaining why Sweet, Luick and their followers didn't seem to loose any sleep over exceptions in the reflexes of OSL. I didn't say that they were right not to. As a matter of fact, the general agreement seems to be that they were wrong, or at least that they made things a bit too easy for themselves. You know, analogy normally works from base forms to inflected cases rather than the other way round, and before you invoke dialect mixture to explain a certain form, you should have some positive evidence that it was really relevant. You're not allowed to dismiss 'exceptions' in the reflexes of any sound law by vaguely invoking analogy (9) and dialect mixture, because if you were you could explain almost anything, and who explains everything explains nothing, does she?

Anyway, although analogy and dialect mixture had long got out of favour in the linguistic community as omnipotent explanatory gap fillers, it was only in the eighties that Minkova realised that this may warrant a revision of the established lore concerning OSL. Taking its Modern English reflexes more seriously than they had been, she discovered, if you grant me that term, that Open Syllable Lengthening wasn't that after all. She concluded that Lengthening really only happened in open syllables that were followed by one that ended in a schwa, which was being lost. Make, weave and hope have all become monosyllabic, you know, and only items like these reflect lengthening with any consistency.

So, I just don't want to speak of Open Syllable Lengthening anymore. (10) If one looks at Modern English reflexes, and if one is not allowed to invoke analogical levelling or dialect mixture, then it is simply wrong to say that stressed vowels were lengthened in open penultimates, because roughly half of them weren't.

TORTOISE: While half of them were, of course.

ACHILLES: Yes, but ...

TORTOISE: But what? That's something, isn't it. I thought we were just trying to look for a way to describe what happened, and saying that half of the stressed vowels in open penultimate syllables got lengthened amounts to an adequate statement, doesn't it?

ACHILLES: Sure, but being true doesn't make it very impressive, not a very powerful law really.

TORTOISE: 'Impressive', 'powerful', these terms do mean a lot to you, don't they? But then, they would, I guess ..., after all, you're ACHILLES.

ACHILLES: Listen, I know this is not the battlefield nor a heroic epic. All I wanted to say was that I prefer statements like Minkova's because they have PREDICTIVE power. After all more than ninety percent of potential inputs to MEOSLS (Middle English Open Syllable Lengthening before Schwa) -- for lack of a better term -- were really affected. That's what I meant, and for me, this IS impressive.

TORTOISE: were it not for the little set of odd words that did get lengthened, although they must have been immune to MEOSLS, like beaver, acorn or moment. They do go back to bever, akern, and moment, don't they? The rhymes of their final syllables were not schwa, nor have they reduced to nothing in Modem English, but still their stressed vowels have been lengthened. Don't they deserve some attention as well? There are more than forty of them, in Minkova's list of potential OSL inputs, if I remember correctly. Powerful and impressive as MEOSLS may be, it doesn't throw any light on those.

ACHILLES: Of course they do, but, listen, I'm beginning to fear, my dear friend, that you were right with your pessimism about our author. He's using you as a mere mouthpiece for his views, (11) and I get to play the role of the dummy. I know what's going to come next. You're going to suggest that the lengthenings are best described in terms of a tendency statement. That there were a couple of conditions that favoured rather than 'triggered' vowel lengthening and that the more of these conditions a single word met, the greater were its chances of getting a reflex with a lengthened vowels. I can foresee it all: lengthening was more likely in words that would typically be stressed in actual utterances than in words that would not (hope vs. have); lengthening was the more likely the lighter the stressed syllable was ([ma[k]e, (CV[C], i.e. lightest) vs. ne[st]/ha[st]e (CV[CC], i.e. heavier), plan[t]e (CVC[C], i.e. heaviest) -- though on the naivete of Ritt's weight calculations see Bermudez-Otero - McCully 1997); lengthening was the more likely the lighter the unstressed part of the foot they constituted was, i.e. most likely in make/crack (unstressed syllable C+un-stable schwa), less likely in beaver/weather (unstressed syllable = CVC, becoming CV), still less likely in capon/bottom (unstressed syllable = stable CVC), least likely in patient/warrant (unstressed syllable = CVCC). Want to see a table? Here it is: lengthening was the less likely the higher a vowel was, (hardly any lengthening of high vowels, more lengthening of low vowels than of mid vowels), lengthening was the less likely the more front a vowel was and the more likely the more sonorous the coda...

TORTOISE: ... and you expect me to suggest, I assume, that Ritt's (1994) way of saying that what looks like 'OSL' simply reflects the combined impact of a couple of lengthening factors, is indeed the most adequate description of the lengthenings. You think I'm going to say that his Quantity Adjustment Rule, or 'QUAR' (Ritt 1994: 96) is all you really need to cover the OSL facts, that it furthermore covers all the other Late Old English changes of vowel quantity, such as Homorganic Lengthening (bindan > bindan), Trisyllabic Shortening (aerende > errand), and Pre-cluster Shortening (dust > dust), and that finally, because all the factors it contains can be related to plausibly universal phonetic constraints (cf. Ritt 1994: 107ff), QUAR is not only descriptively adequate but has some truly explanatory potential as well. Do you really think I've let myself be reborn just in order to say that Ritt has solved it all?

Honestly ACHILLES, you should know me better than believing I would lend myself to playing the role of a hired eulogist? Do you really still underestimate my critical faculties after all we've been through together?

ACHILLES: But I thought that maybe since Ritt was the author ...

TORTOISE: Author, shmauthor! I guess, you'll both be in for a bit of a surprise.

ACHILLES: There's my friend as I know him. I'm all ears.

TORTOISE: OK. So back to our, or MY problem with Minkova's revised OSL. You see, she highlighted that the lengthenings reflected in words like make seem to have been prompted by the fact that the schwa in the final syllable got lost. Originally, make had constituted a trochee consisting of two light syllables. This type of foot is considered to be universally preferred. (12) When the schwa in the second syllable was about to get lost, make was about to become rhythmically underweight, so to speak, and to make up for this, the vowel in the first syllable got lengthened. In the end, a [[CV].sub.s][[CV].sub.w] foot would find itself replaced by [CVV[C].sub.s] foot. Although the latter wasn't really a trochee anymore, at least it would weigh as much and could fill the slot of one, because, putting it simply, it would last about as long as a trochee in actual pronunciation.

ACHILLES: Well, actually Minkova doesn't say exactly that in her article.

TORTOISE: Admitted, but you will allow me to read between the lines a bit won't you? -- What I really only wanted to say is that Minkova's decision to cut MEOSL down to MEOSLS -- or compensatory OSL, if you prefer that -- implies that words of the beaver type should not have lengthened at all. Ritt's account, on the other hand, suggests merely that compensatory OSL should have been more likely and affect more potential inputs than non-compensatory OSL. Which is undeniably truep, of course. Only that 'compensation' doesn't really play a role for Rift.

ACHILLES: It doesn't? Well, it's kind of contradictory to say that Rift predicts compensatory OSL to be more likely than non-compensatory OSL while he doesn't employ the concept of 'compensation' at all. I wonder how you're going to talk him out of THIS.

TORTOISE: You'll be surprised what I'm going to talk him INTO. Look. He handles the fact that OSL happened so often in words that had final schwa as a last syllable by saying that lengthening in a stressed syllable was the likelier the lighter, summarily speaking, the unstressed syllables that followed it were. And there is hardly any syllable that can be lighter than a C-schwa one, particularly if that schwa is kind-of unstable, as in make, or is there?

ACHILLES: Hmmm, I'm beginning to see your point. Make was likelier to be lengthened than beaver, and beaver was likelier to be lengthened than warrant, and warrant was more likely to be lengthened than errand, which was even shortened though the stressed vowel was in an open syllables. In other words, the heavier the weak syllables in a foot were, the less likely was its head to be lengthened. This relates make, beaver, warrant and errand in a cute way, I admit. At the same time it doesn't deny, nor actually even background, the relevance of what used to be described as the 'open syllable' parameter. This is handled by the statement that the weight of the stressed syllable itself influenced the likelihood of lengthening independently, so make got lengthened while lamp wasn't, the first syllable of the latter being too heavy to warrant that. Cute indeed, I'm beginning to find, but I'm afraid I'll withhold my sympathy for QUAR for a while, because you said you were going to talk Ritt INTO something after all, and from my experiences with you, I know that whenever one thinks one has brought a story to a happy conclusion. Bang! There you go and it's good riddance to false idylls. So, I'm prepared for another twist.

TORTOISE: So, you mean you haven't noticed?

ACHILLES: Noticed what? Don't tell me I've overlooked one small but crucial fact once again?

TORTOISE: Your desire for a happy ending may have played a trick on you, and I may even have misled you a bit, but you must forgive me. I had been envious of the playful way in which you made such a big discovery ...

ACHILLES: Discovery?! Me?! I came to you with a PROBLEM.

TORTOISE: And a solution, though you might have been unaware of it, but how do we get good ideas anyway? -- I'm not going to deprive you of your honour, though. The 'discovery' shall be yours and yours alone. -- So let me jump back in our little argument a few paragraphs: ... there is hardly any syllable that can be lighter than a C-schwa one, particularly if that schwa is kind-of unstable, as in make, or is there?

ACHILLES: Well, you asked me already. I don't see how ... but hmmm, let me think? What's less than very little, except nothing at all? except nothing at all? Nothing at all? Nothing at all!!! But, you must be joking, you can't possibly mean

TORTOISE: Lengthening in monosyllabic CVC items, or 'CLOSED Syllable Lengthening'. Exactly. Your very own idea, if I may say so!

ACHILLES: But, by God! There wasn't ever anything like Closed Syllable Lengthening! I see the flaw in Pitt's account now, and what a serious one it is. His notorious Quantity Adjustment Rule would predict that, wouldn't it? It's false as can be then. Ha, serves him right, if only for letting me play such a dumb part!

TORTOISE: Dumb? I am beginning to think he's casting you quite out of your character this time. Why are you so sure there never was any Closed Syllable Lengthening?

ACHILLES: Now, my dear friend, this time it's you who are underestimating me. I HAVE read the handbooks, after all, and quite thoroughly at that, and they do NOT mention lengthening in CVC monosyllables.

TORTOISE: As a rule, no. You're right, and I bet you'd eat a whale if you found the old masters and the handbooks were wrong there?

ACHILLES: As a rule, no? A whale? Come on, now that's cheap. It may well be that whale goes back to Old English hwoel, but even the OED says that he present form whale represents oblique forms (OE hwalas, etc.)" (OED s.v. whale). I'm not going to make a fool out of myself and posit Closed Syllable Lengthening because of whale. And listen. I do know that whale isn't the only example of its kind. Already Henry Sweet listed quite a number of them, but also he was adamant that "short vowels in final stressed syllables (which were generally monosyllables) could not be lengthened ..." (1888: 167), and that lengthenings in such words must be due to analogy. "Many OE neuters with short root-syllable", he explains, "take a final e in ME, thus OE gebed, 'prayer' appears in ME as bede = MnE bead, the OE pl. gebedu being apparently taken for a masc. Or fem. Sg. Like medu 'mead', caru. Hence the long vowels in LME dale, 3ate (gate), blade, bede, hole, cole = the OE neuters doel, goet, bloed, gebed, hol, col" (1888: 168). That's analogy. Know what? I'll stick with Minkova, and go for Compensatory OSL. There's a law for you that DOES work. No exceptions, well, hardly any, and for the rest I don't care. Otherwise, we might as well say that lengthenings might have occurred anywhere, and that the same was true of shortenings and that's the coincidence of opposites and the end of reason, if you ask me.

TORTOISE: I don't want to appear didactic, but I don't how you can decide to accept Minkova's account of OSL AND the analogical account of hwale. You can't possibly have both without being more self-contradictory than even Ritt has ever managed to be. Don't you remember that the dismissal of analogy as an explanation of exceptions to 'OSL' was what made it possible for her to come up with 'Compensatory OSL' in the first place. If you say that whale has its long vowel from hwalas, then you must equally admit that the short o in bottom may have derived from botme or botome. And whale is an ironic example anyway, because you're actually arguing that the long vowel comes from a form in which Minkova argued that Compensatory OSL should not have occurred in the first place, because schwa was more stable, at least for a time, in checked syllables such as [l[LANGUAGE NOT REPRODUCIBLE IN ASCII]s].

ACHILLES: I see, so if I believe in Compensatory OSL, I would contradict myself, wouldn't I, if accepted analogy as an explanation of the lengthening in coal or whale?

TORTOISE: I'm afraid you would.

ACHILLES: But nobody except you seems to see this as a problem at all.

TORTOISE: I'm notorious, I know, but there it is. Show me that my worries are unfounded, please.

ACHILLES: Well, I'll try my best, and I'm not prepared to accept defeat quite so soon. I think there is a way out, actually. You see, in the case of whale and coal, for example, one need not assume that their length derives from the plural, or necessarily any inflected form. Might one not argue that at a time when final schwa in words like flake [fla:k[LANGUAGE NOT REPRODUCIBLE IN ASCII]] got unstable, it would have been difficult for speakers of English to realise at all that words such as hwal (from oE hwael) and words such as flake (from ON flake) belonged to different classes. The latter would, more often than not, have been pronounced monosyllabic anyway, i.e. [flak], and thus looked pretty much like [Xwal] words. By that rationale, there would have been no reason why hwal [Xwal] could not have been mistaken for /Xwa:l[LANGUAGE NOT REPRODUCIBLE IN ASCII]/ with the schwa deleted. As is well known, words like hwal even came to pronounced with unetymological schwas eventually, which goes to show that humans don't acquire their language from the etymological dictionary. So w ords like whale are not really problematic. They simply moved over into the flake class, where they came to behave exactly like Minkova says they should have. I am pretty sure that this is what she would argue, because she actually even includes whale in her list of potential OSL inputs, just like coal, blade, drake, and a few others, by the way. Now, is that not an elegant solution?

TORTOISE: Only that it leaves us with God as an exception to Compensatory OSL, doesn't it?

ACHILLES: 'Exception', come on. I never said that it was the RULE for CVC items to have merged with CVCV items. And where there is no rule, there can't be any exceptions, or can there?

TORTOISE: And vice versa, I'm afraid. If you say that all lengthenings in CVC items must have happened because they were mistaken for CVCV items with deletable schwa, this leaves you with NO way of determining whether or not Closed Syllable Lengthening EVER took place. The question becomes simply unanswerable, and uninteresting. That we DO find it interesting, however, and that it MAY just be answerable, is something that we owe to Minkova and the way in which she showed how established lore literally demands to be revised, if only you are a bit more disciplined about analogy than our ancestors used to be.

ACHILLES: But analogical levelling does occur, so I don't see why you should be so completely against employing it as an explanatory device occasionally.

TORTOISE: Look, I'm not, but where would it get us? Let us assume that every modern reflex of a CVC item that has a long vowel has that vowel due to analogy, does that not beg a question?

ACHILLES: Well, yes, I think I see what you mean. We could ask for example, why some items had their vowels lengthened by analogy while others did not. Why whale and God, and why not whal and Gode?

TORTOISE: And how do you suggest should one go about answering that question?

ACHILLES: Well, not at all, I guess. Those things may have been matters of chance, more or less, so the question is essentially unanswerable, isn't it?

TORTOISE: It is indeed, I'm afraid, and that's my whole point. As soon as you invoke analogy to explain what happened in a certain set of word forms, you might as well say right away that you don't consider the events to be rule-governed at all. 'Analogy' is thus a euphemism for saying 'some things just happen, that's all there is to that'. -- Fortunately, Minkova showed that one can do better.

ACHILLES: But I thought you were not too fond of her paper?

TORTOISE: On the contrary. I cannot think of many papers that were as groundbreaking as hers. This doesn't mean that I have to accept all her conclusions, of course, or do I?

ACHILLES: No, of course not, but let me do a little summary for myself at this point. What have we agreed on, so far? First, we do not take recourse to analogy, because if we did, we might as well live with the first version of Open Syllable Lengthening, no matter how many exceptions we find in Modern English reflexes. From that, however, it follows that we cannot simply say that lengthening in CVC items was the result of analogy, because we need to be consistent ... unless, however,...

TORTOISE: Unless what?

ACHILLES: Unless the number of lengthened CVC items is so ridiculously small that recourse to analogy is less awkward than making a big fuss about two or three words. Or would you establish a sound law on account of a few isolated cases?

TORTOISE: No, of course not, and I'm afraid we'll have to get empirical at some stage. I fully agree that if whale, coal and blade were the only examples of lengthened CVC items, we would probably not want to claim that anything by the way of a sound change took place there. But let's get down to it, shall we. You don't happen to have a Old English dictionary on you?

ACHILLES: Now, this is a fortunate coincidence. Look here. There's Holthausen for you, plus, ta-ta-ta-taaaa, the OED on CD-ROM. I suggest we go through Holthausen first, search for inputs to Closed Syllable Lengthening, see which of them have Modern English reflexes, and how many of the latter are long. Then we counter-check in the OED. That should do for a start, shouldn't it?

TORTOISE: Of course. So let's see.

TORTOISE: and ACHILLES get down on the ground, and for a while you hear nothing but the rustling of paper, the whirr of the CD drive, and a few muttered "Now, look here!"s, and "Now, who would have..."s. Eventually you see ACHILLES frantically typing something into his notebook, cursing at the machine as he goes along. Suddenly his face assumes a worried expression, and he says:

ACHILLES: But isn't this childish? What are we doing here? Do you really think that with the help of two dictionaries we shall revise the insights of more than a hundred years of philological and linguistic scholarship. That's as arrogant as it is naive, isn't it?

TORTOISE: A hero being frightened of his own courage? Now, you're an infinite source of surprise, dear friend. But you should know me better than that. Who says we're revising anything? All we'll do is question things, or rather highlight a problem so that greater minds may set to work on it. We might give them something to be puzzled about, or maybe not even that, we'll see. I'm sure, most of them will like that, and if for nothing else, they might be grateful for having saved them the tedious task of going through dictionaries themselves. So, let's get on with it, shall we. Let's see what we've got.

ACHILLES: Well, in my first search I may have been a bit over-inclusive. Before we start counting numbers, could you please remind me of what exactly we are looking for, please, so that we may get rid of all the useless stuff I may have come up with?

TORTOISE: All right. So we are looking for Late Old English words with a CVC structure ...

ACHILLES: ... wait, what about words with final geminates, such as bat(t) 'bat', becc 'beck', bed(d) 'bed', bil(l) 'bill', byt(t) 'bit', cat(te) 'cat', cinn 'chin', clamm 'clam', clodd 'clod', cocc 'cock', denn 'den', gncet(t) 'gnat', hcet(t) 'hat', henn 'hen', cnoll 'knoll', net () 'net', pott 'pot', toll 'toll', wed(d)? Whatever their codas were really like in Old English, it's pretty safe to assume that by the 13th century, their final consonant clusters had simplified, isn't it, so do we include them?

TORTOISE: Well, only if we can be sure that Quantity adjustment did not start before, and there does not seem to be general agreement on that question. In particular, Ritt suggests that it may have started much earlier (Ritt 1994: 93), and there's a paper by Richard Hogg (1996) just come out, where he presents evidence for what he calls a phonetic forerunner of Open Syllable Lengthening. So, to be on the safe side, I would suggest to leave them aside for the time being. -- But anyway, I was saying that we are looking for OE CVC items that have modern English reflexes

ACHILLES: ... so that means I'll have to get rid of bod 'commandment', bor 'drill', brec 'sound', bred 'board', drif 'drift', foec 'shelf', goed companionship', gin 'depth', gram 'angry', git 'you two', hoef yeast', hroed 'quick ' active', hrif 'belly', lof 'praise', lot 'trick 'dole', met 'measurement', rop 'broth', soel 'hall', scroec 'thrush', spoec 'pole, stick', spic 'bacon', sproec 'shoot', doec 'roof', woed sea ' ford', woel 'place of pilgrimage', wit 'we two' and wroec 'revenge', which don't, right?

TORTOISE: Right. So we're looking for CVC items with Modern English reflexes, as I was saying, from which we can tell whether or not we have cause to assume that their vowels were lengthened during the period in which the other processes of quantity adjustment seem to have occurred.

ACHILLES: OK. So, as far as I can see it we have quite an impressive corpus. I'm counting 122 items. It's not as big as the set which Minkova examined for her OSL paper, but ...

TORTOISE: Let me see ... hmm, now, but these won't do: you can't include lake, pace, case, estate or despite. They are French loans, and were probably taken over with a long vowel in the first place. They don't tell us anything.

ACHILLES: 117, then.

TORTOISE: And I won't admit door (just possibly related to OE dor), gore and lore either. In those, the modern length may be a later development due to the influence of the post-vocalic /r/. There's no way to tell. Bare (OE boer), aware (OE woer) and weir (OE wer) are different, because the quality of the Modern vowels indicates that hey must have been long before the Great Vowel Shift. I'll accept those.

ACHILLES: ... 114 ...

TORTOISE: ... throw out croet. It became cart by metathesis, and doesn't really count as a CVC item, does it?

ACHILLES: ... 113 ...

TORTOISE: ... forget pronouns, prepositions and other items that are not normally stressed in utterances. In these, shortness is predicted by anybody dealing with he matter. So throw out fram 'from', hit 'it', in 'in', sum 'some', til 'till', ooet 'that', oan 'then, than', ois 'this', ous 'thus', wio 'with', 3et 'yet'. Forget is...

ACHILLES: 101

TORTOISE: stoe3 'stay', doe3 'day', sti3 'sty' and moe3 'may', in which the final consonant got vocalised ...

ACHILLES: 97

TORTOISE: ... forget ecg 'edge', hecg 'hedge' and facg 'fadge', because at least phonetically the final affricate can count as a cluster ...

ACHILLES: 94

TORTOISE: ... then I don't see why you included boest 'bast'. /st/, although special, is a cluster after all.

ACHILLES: 93

TORTOISE: ... then I'm afraid we have to do without broo 'broth', as well as the alternative /br[LANGUAGE NOT REPRODUCIBLE IN ASCII]:[theta]/ pronunciation' (counts as 2 items), pos 'poss' or 'pose' in the meaning of the 'cold' you get (counts as 2 items), swoeo 'swath', mos 'moss', sloo 'sloth' and gloe(:)d 'glad', which might actually go back to a long vowel anyway. The shortness in these items, may just as well have resulted from Early Modern English pre-dental shortening, so the evidence they provide is ambiguous.

ACHILLES: 86

TORTOISE: ... and, finally, we'll have to make do without items that had high vowels. It is more or less generally agreed that high vowels were as good as immune to lengthening, due to their internal phonetic shortness, unless particular factors made up for that. So discard brim 'brim', busc 'bush', clif 'cliff', cwic 'quick', fisc 'fish', grin 'grin', girn 'girn', hild 'lid', lim 'limb', pic 'pitch', scip 'ship', slic 'slick(-stone)', slic 'sleek(-stone)', slid 'dial. slid', smio 'smith', wull 'wool', tin 'tin' ...

ACHILLES: God! This leaves us with a lousy 69 inputs. Not much to build an impressive law on ...

TORTOISE:: Well, a bit more so, I would say than the two or three unambiguous inputs (aerende 'errand', aemette 'ant') on which Trisyllabic Shortening rests (see Ritt 1992) ... but, by Zeus, get on with it, it's me who's getting impatient, for a change ... how many of the items still in have a long vowel in Modem English.

ACHILLES: Now, let me see ... can this be right? I'm counting 36 as against 33 short ones. That's more than half, that's about the same ratio as we get for Classical Open Syllable Lengthening. Who would have expected this?

TORTOISE: Not our author, that's for sure, because otherwise he would surely have made a big deal out of it in his Quantity Adjustment ... but give me more please.

ACHILLES: All right, your wish is my command. We have lengthened boer 'bare', bed 'bead', bloed 'blade', broc 'broke, misery', col 'coal', cran 'crane, heron', doel 'dale', foer 'fare, journey', flot 'float, action of floating (OED)', grot 'groats', ham 'hame' or 'the collar of a draught horse', hol 'hole', hop 'hope', hwoel 'whale', loet 'late', mot 'mote (of dust)', sceat 'sheet', sloed 'slade', sol 'sole, pool', spoer 'spare, chalk', sped 'spade: the gummy or wax-like matter secreted at the corner of the eye (OED)', stoc 'x-stoke (in place names)', toel 'tale', tot 'tote, vault', oel 'theal, plank', woer 'aware', wer 'weir', spak 'spake', brak 'broke', gaf 'gave', bad 'bade', bloec 'Blake (proper name)', groef'grave', stoef 'stave, stoer 'stare', scead 'shade' as opposed to unlengthened boec 'back', boeo 'bath', bet 'better', bloec 'black', broes 'brass', cot 'cot', dol 'dull', foet 'vat', god 'god', groes 'grass', hlot 'lot', loc 'lock', los 'loss', poeo 'path', plot 'plot', soed 'sad', soep 'sap', scead 'shed', sc ot 'shot', set 'set', sloec 'slack', slop 'slop', smoel 'small', soc '(dial.) sock', spaer 'spar', stoef 'staff', swan 'swan', trod 'trod', tro3 'trough', oroc 'throck, share-beam', woesc 'wash', sat 'sat', grot 'grot',

Wow. I can see it already. ACHILLES and Tortoise, 1997: An Historical Grammar of English. Vol. 2. Chapter 2: Vowels. Section 1: Quantity. [section]37(b): Lengthening in Closed Syllables.

V [right arrow] [+long]/__C#

as in hop 'hope', oel 'theal', woer 'aware', wer 'weir', spak 'spake'

TORTOISE: Well, I thought, ever since Minkova this type of sound law has been regarded as unacceptable, or has it?

ACHILLES: Of course, you're right. It was just for a moment, you know, that I was dreaming of days long past

TORTOISE: Never mind. You have still dug something up there, it seems to me. And you have taught Ritt quite a lesson, haven't you?

ACHILLES: What do you mean 'taught him a lesson'? I'm beginning to get quite frustrated, as a matter of fact, because it seems that all we've done is to prove that his Quantity Adjustment Rule is adequate after all, haven't we? The smaller the overall weight of the weak syllables in a word form's foot, the greater is the chance of a word form to have its vowel lengthened. That's what it says, doesn't it? And, well, here we go.

TORTOISE: So, he gets the credit for having overlooked one of the most obvious implications of his own rule. Nothing I envy him for particularly, if I may say so. And then, it doesn't seem to me that the prediction implied in his Quantity Adjustment Rule is borne out as nicely as he would hope, anyway. After all, his rule would seem to predict that vowel lengthening in CVC monosyllables ought to be much more frequent than it actually seems to be. As frequent, at least, as in disyllables with final schwa, words of the name type, you know. And of those more than 90 percent were lengthened, were they not? So he can't be too pleased, I would think.

ACHILLES: But look, in some other respects our data conform just beautifully to his rule. It says that lengthening should be more frequent among low vowels than among mid vowels, for example. Of the 37 low vowels in our set 21 (that is to say 57%) got lengthened, while of the mid vowels only 15 out of 33 (that is to say 45%) show up as long. Furthermore, Ritt claims that lengthening should be more likely before sonorant codas than before obstruents. And, here it is: of the 19 pre-sonorant vowels in our set, fifteen show up as long in Modem English. That's more than 78%. Of the 50 vowels before obstruents, only 21 are long today, that is 42%. I am beginning to feel pity for Quantity Adjustment. It's too nice, in some respects. But then you are right of course. About half of the items lengthened, where one would expect nearly 100%, doesn't amount to a very adequate prediction ...

TORTOISE: You know what, I am beginning to suspect that there is yet another way in which Ritt stopped pathetically short of where he was about to get in his little treatise. It's not only that he didn't take the obvious step to check the behaviour of CVC monosyllables, it's also his way with dealing with Vennemann's (1986) point about how much information about prosodic structure the lexical phonological representation of a word can be assumed to contain. Ritt (1994: 71ff.) made a perfectly valid observation there, but he didn't see its implications. If he had, then I think he might have even integrated the behaviour of CVC monosyllables into his Quantity Adjustment Rule, had he cared to look at it, that is to say.

ACHILLES: Now I'm afraid I'm beginning to lose you altogether. Could you unfold your ideas a bit more explicitly, please?

TORTOISE: Well, you see, Theo Vennemann (1986: 57-60) argued that the lexical representations of words didn't have foot structure at all. All they contained was information about which of their syllables would get primary stress IF the need should arise in an actual utterance, within an intonation group, so to speak. This is particularly true of monosyllables, of course, because sometime s they might get to play the role of footheads, at other times they might even constitute feet by themselves, and at still other times they might appear within feet headed by syllables outside themselves. Now, Ritt, argued that this was theoretically true, but in many respects not relevant. Thus, the stressed syllable in a trisyllabic item could be sure, so to speak, that, on the average, the feet it would come to head would be one syllable longer than those headed by a disyllabic item, and two syllables longer than those headed by a monosyllable. Therefore, in his sense, some information about utterance prosody is indeed alr eady contained in, or predictable from, the phonological structure of words. It tends to work out fine, particularly when one compares words with one another, to speak of them as if they had foot structure themselves.

ACHILLES: I see. Calling a trisyllabic word a foot amounts to saying that the feet in which it figures will contain two weak syllables more than the feet a monosyllabic word will head. It's sloppy, but convenient shorthand, and works for many purposes, am I getting you?

TORTOISE: You are indeed, but apply this to CVC monosyllables and the way they compare to CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] disyllables, and look where that gets us.

ACHILLES: Well, let me see? The feet in which a CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] disyllable can appear, will always be one syllable longer than those in which a CVC monosyllable may figure, or won't they?

TORTOISE: Well?

ACHILLES: You doubt that? But didn't you say ... Ah, not always, I think I see your point: if a CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] item happens to get followed by a word that begins with a vowel, then the schwa might get deleted, so that in such configurations CVC monosyllables and CVC disyllables will merge, prosodically speaking. All right. But even if all English words began with a vowel, this would only imply hat CVC items should lengthen just AS often as CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] items, which they didn't. And the fact that there have always been words in English that began with a consonant..

TORTOISE: Well, what about this fact?

ACHILLES: Now, it's obvious, is it not? Before consonant-initial words, CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] items would remain disyllabic, and monosyllables monosyllabic.

TORTOISE: ... and heavy!

ACHILLES: Heavy? What do you mean?

TORTOISE: Just what is obvious: before consonant initial words, the vowels in CVC monosyllables would find themselves followed by two consonants, in syllables that nobody would call anything but closed and heavy. If a CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] disyllable, on the other hand, comes to be followed by a consonant, then, while it might retain its second one, its first syllable will still be light, won't it? Lighter, in any case, than a CVC monosyllable in a comparable configuration. Therefore, all Ritt would have needed to do is to assume that the weight of the syllable in which a vowel found itself had a relatively greater impact on its chance to get lengthened than the weight of the weak syllables it was followed by. And this assumption is not implausible, either. After all, words like lamp did not lengthen! Quite on the contrary, words like kepte even shortened to kept. So, here we are: Ritt's Quantity Adjustment rule not only predicted the lengthenings in monosyllables without him caring about it, but it would also have been able to account for the fact that they were less frequent than among CVC[LANGUAGE NOT REPRODUCIBLE IN ASCII] items -- if only he had developed Vennemann's argument a bit further than he did.

ACHILLES: Now, you are being a bit hard upon him, aren't you? He was only trying to find a way to unify the description of the big quantity changes that were recognised at the time he began to concern himself with the problem. He didn't set out to discover a new sound change, and if you don't seek, you will not find.

TORTOISE: It's ironical that you should be defending him now. After all, his Quantity Adjustment rule is swallowing your neat little discovery, isn't it? No more daydreaming of volume 2, section 2, [section]1(b): Lengthening in Closed Monosyllables ... Aren't you sad at all about the way of things? One day they're here, next day they're gone ...

ACHILLES: But how unlike you to say such a thing! Such a melancholy point-of-view ... In Hofstadter's cranium, I would have been the one to make such statements. Your perspective tended to be much more Eastern, then, remember the Zen-koans you used to come up with ...

TORTOISE: ... he's not handling us right, you see. So why should we finish our discussion on a positive note. We might just as well leave with all the mess we've made still lying around all over the place, and leave his business as unfinished as Ritt left his treatment of Quantity Adjustment.

ACHILLES: Unfinished? Well, with the help of you and me, a hypothesis has been refined and corroborated in a rather pleasant way, hasn't it? And although my own nice new sound law hasn't survived for more than a couple of pages, I feel satisfied that we have brought things to at least some kind of positive conclusion.

TORTOISE: Conclusion? You bet!

[FIGURE 1 OMITTED]

(1.) I am grateful to Roger Lass for many other things, but also for suggesting that the genre of the dialogue deserves to be revived, to Douglas R. Hofstadter for showing how to do this well and for supplying the personae of ACHILLES and TORTOISE (whose entertaining and instructive dialogues introduce the main themes in Hofstadter's incomparable volume Godel, Escher, Bach), to Ricardo Bermudez-Otero for making me aware that I actually believed in Closed Syllable Lengthening, and to Donka Minkova for providing many arguments that I would never have been able to think up myself. Of course none of them are in any way responsible for all the faults I will certainly have committed.

(2.) Clearly, TORTOISE finds Ritt's way of introducing the topic not very subtle.

(3.) Such as Minkova (1982), Lass (1985), Libermann (1992), Minkova -- Stockwell (1992), Ritt (1994).

(4.) The allusion is to the disputed status of syllable boundaries. Thus, general maximal syllabication (as proposed, for example, by Anderson 1986, and subscribed to in Lass 1992 and Ritt 1984) would have the intervocalic consonants as ambisyllabic'.

(5.) ACHILLES is referring to Minkova (1982).

(6.) I don't really think it's necessary to point out that we owe much of our awareness of this fact to Lass (1980).

(7.) The examples are from the OED (sv. bottom)

(8.) Very explicit examples of this kind of analogical account are to be found, in Wright (1928: 52) or Mosse (1952: 17).

(9.) See Ritt (1992) for an example of the dangers inherent to unrestricted (and implicit) utilisation of analogy as an explanatory device.

(10.) Neither does Lass (1992: 74).

(11.) Those expressed -- in much greater detail -- in Ritt (1994).

(12.) See Dziubalska-Kolaczyk (1993: 58), and the references given there.

REFERENCES

Anderson, John M.

1986 "Suprasegmental dependecies", in: Jacques Durand (ed.), 55-135.

1997 Review of Ritt 1994. Journal of Linguistics.

Bermudez-Otero, Ricardo -- Christopher B. McCully

Berndt, Rolf

1960 Einfuhrung in das Studium des Mittelenglischen. Halle: Niemeyer.

1984 A history of the English language. (2nd edition.) Leipzig: VEB.

1992 The Cambridge history of the English language. Vol. 2: 1066-1474. Cambridge: CUP.

Blake, Norman (ed.)

Brunner, Karl

1960 Die englische Sprache. (2nd edition.) Tubingen: Niemeyer.

1986 Dependency and non-linear phonology. London: Blackwell.

Durand, Jacques (ed.)

Dziubalska-Kolaczyk, Katarzyna

1995 Phonology without the syllable. Poznan: Motivex.

Hofstadter, Douglas R.

1984 Godel, Escher, Bach: An eternal golden braid. New York: Basic Books.

1985 Metamagical themas: Questing for the essence of mind and pattern. New York: Bantam.

Hogg, Richard

1996 "Old English Open Syllable Lengthening", Transactions of the Philological Society 1994: 57-72.

Holthausen, Friedrich

1974 Altenglisches Etymologisches Worterbuch. (3rd edition.) Heidelberg: Winter.

1980 On explaining language change. Cambridge: CUP.

Lass, Roger

1985 "Minkova noch einmal: MEOSL and the resolved foot", Folia Linguistica Historica 6: 245-265.

1992 "Phonology and morphology", in: Norman Blake (ed.), 23-155.

Liberman, Anatoli

1992 "A bird's-eye view of Open Syllable Lengthening in English and in the other Germanic languages", NOWELE 20: 67-87.

Luick, Karl

1914/1921 Historische Grammatik der Englischen Sprache. Oxford: Blackwell.

[1964] [Reprinted Stuttgart: Tauchnitz]

Minkova, Donka

1982 "The environment for open syllable lengthening in Middle English", Folia Linguistica Historica 3: 29-58.

Minkova, Donka - Robert Stockwell

1992 "Homorganic clusters as moric busters in the history of English: The case of -ld, -nd, -mb, in: Matti Rissanen et al. (eds.), 191-206.

Mosse, Fernand

1952 A handbook of Middle English. (Translated by James A. Walker.) London: Johns Hopkins.

Oxford English Dictionary on CD-ROM

1992 (2nd edition.) Oxford: OUP.

Pinsker, Hans Ernst

1974 Historische Englische Grammatik. Munchen: Hueber.

Rissanen, Matti et al. (eds.)

1992 History of Englishes: New methods and interpretations in historical linguistics. Berlin, New York: Mouton de Gruyter.

Ritt, Nikolaus

1992 "Luick, the theoretician", VIEWS 1: 93-104.

1994 Quantity adjustment. Vowel lengthening and shortening in Early Middle English. Cambridge: CUP.

Sweet, Henry

1888 History of English sounds. Oxford: Clarendon Press.

Vennemann, Theo

1986 Neuere Entwicklungen in der Phonologie. Berlin, New York: Mouton de Gruyter

Wright, Joseph

1928 An elementary Middle English grammar. London OUP.