[Journal of the Simplified Spelling Society, 1989-2 pp25-29 later designated J11]
[Chris Upward: see Journals, Newsletters, Leaflets, Media, Pamflet 15, Book, Papers.]

Edgar Gregersen & Christopher Upward discuss

Morphemes and Cut Spelling.

Edgar Gregersen is Professor of Anthropology at Queens College of the City University of New York. His previous contributions were published in SSS Newsletter, J2 Spring 1986 (pp14-17), and in JSSS J6 1987/3 (pp17-21) and J9 1988/3 (pp11-13). [See also JSSS J27 2000/1 pp16-18.]

1. Gregersen.
As explained in my article in the J9 1988/3 Journal, an unsatisfactory feature of Cut Spelling appears to be its treatment of morfemes. CS claims to be very much concerned about the integrity of the morfeme, but sometimes it is and sometimes it isn't. For example, if one really were troubled by breaking up morfemes, one couldn't possibly write symbl for symbol, since this breaks up the relationship with symbolic.

1. Upward.
Wat exacly is a morfeme then, if th <-bol> in symbol is to be classd as one? Etymolojicly we can distinguish a prefix sym- (also found in symbiosis, symptom etc - actually a modification of syn-, as in syntax, with further variants in syllable, system) and a root -bol (also found in embolism, bolide etc and cognate with ball- in ballistics). But do these elemnts in symbol hav any living significnce for users of th english languaj? Dos anyone but an etymologist make th conections?

If we compare th morfemes in symbol with those in, say fireman, th difrnce is clear; CS wad not cut fireman to firemn, because evryone senses th meaning man in th secnd sylabl (contrast CS womn, jermn wher th final sylabls in th TO forms woman, German do not hav th meaning man). In symbol th user has no mor sense of a seprat meaning in <-bol> than in gambol, nor has th difrnt <-bal> ending in cymbal any distinguishing valu. Howevr th form symbl has th advantaj of distinguishing th stress patrn in symbl from that in, say, extol; symbl must stress th first sylabl, wile extol givs mor weit to th secnd sylabl. (This clarification of stress patrns is a jenrl advantaj of CS over TO wich has not been widely apreciated yet.) Morfemic structur in itself dosnt sujest symbl, cymbl, gambl ar inadequat spelings.

As for th link with symbolic, we hav no quams about riting able rathr than abil, despite th link with ability, so we need be no mor concernd about th loss of <o> in symbl. Th pronunciation and speling of many words ignor th morfemic structur: we dont wory that window, husband no longr sho th structur wind + eye and house + bond. Edward Rondthaler proposes th useful principl that th speling of words shud be determnd by ther own sound, and not by th sound of a quite difrnt word - and that is an importnt reasn for riting symbl.

2. Gregersen.
I see no reason for calling the <-bol> of symbol a morfeme to begin with: it has no independent meaning. The morfeme is (symbol) in its entirety - just as (caterpillar) is a single morfeme, not to be broken up into, say, {cat} + {er} + {pillar}.

If one wants to build the 'integrity of the morfeme' into a spelling system, it simply means that as much as possible a single morfeme is going to have a single representation - no matter how it may actually be pronounced in a given context.

A famous example comes from Russian, where the integrity of the morfeme is a fundamental principle of the present writing system. There is a morfeme we can write as gorod, meaning 'city', as in Novgorod, 'new city'. This morfeme has several realizations górət', -'gərət, garód - and so on, depending on stress and on whether the form comes at the end of a word or not. If the <o>s aren't stressed, they may be pronounced like <a>s; if the <d> is final, it is pronounced /t/. The only form from which all these variants (or allomorfs) can be derived is gorod, which is a purely hypothetical form in modern Russian and pretty close to Chomsky's underlying form. To maintain this single shape is to preserve the integrity of th morfeme {gorod}. To show variants is to go against the integrity.

2. Upward.
Th behavir of russian dosnt hav to be a modl for othr languajs. Indeed, as Frank Knowles demnstrated (JSSS J8 1988/2, p15, §5.1), th closely related slavonic languaj byelorussian difrs from russian in precisely this respect: it dos not preserv th integrity of th morfeme; and it is claimd that litracy in byelorussian actuly benefits as a result.

3. Gregersen.
Possibly in the learning stages, but once the system is mastered, few problems seem to result. For example in English, the possessive morfeme is always spelt <-'s> (or almost always) despite the fact that there are 3 different pronunciation: /-s, -z. -ez/, as in cat's, dog's, Liz's. To my knowledge, this is one part of English spelling that nearly never poses problems.

3. Upward.
I wud hav to disagree with th idea that hardly any problms arise with th posessiv morfeme in english, at least insofar as th apostrofe is countd as an esential part of it (i.e. of th morfografeme). Lernrs and adults alike experience enormus dificlty in placing this apostrofe corectly. Not merely is ther frequent confusion with th plural <s'> variant (Dogs' Home) and th special placing of th apostrofe befor th <s> in such iregulr plurals as people's, children's, but th apostrofe is also comnly misused to mark plurals and th verb inflexion: *apple's, *make's. Even professionl riters face uncertnty with this morfeme wen th noun ends in <s> (Jones', Jones's?). Furthrmor, at least som foren lernrs Adam Brown cites th exampl of malays) ar confused as to th pronunciation of th invariant <s>, since in ther nativ script <s> always represents voiceless /s/, wheras in english th <s> morfeme is usuly, but by no means always, voiced.

These dificltis al seem to point to one conclusion: morfografic regularity may wel not be a desirebl featur for an ideal riting systm if it conflicts unpredictbly with fonografic regularity. U wil notice that I am here implicitly almost arguing against my own case, that morfemic regularity is one of th atractions of CS! In principl it shud be subordnat to fonografic predictbility.

4. Gregersen.
Please note that I am myself not an advocate of the integrity of the morfeme in all cases either. For instance, no one, as far as I know, has ever proposed cleaning up an egregious exception to morfemic writing in English, the indefinite article, <a> or <an>. I personally would not advocate such a clean-up job because it has little bearing on the rest of the system. But I think it would be desirable to deal with productive, widespread alternations in as unified a way as possible - for example, the possessive <-'s>.

Take an example from German: bunt: Bund are both pronounced the same in isolation. In inflected forms, the <t> and <d> are actually differential: bunte: Bunde (of keys), Bünde (bonds). To maintain such distinctions is not particularly onerous because the native speaker has it within his competence to link up one form to other forms. I can't imagine that German spelling should be reformed to show /t/ finally always.

4. Upward.
But this featur too dos cause lernrs a litl dificlty in jermn. In purely theoreticl terms, this rases th question of wethr a riting systm shud idealy aim primarily to reflect abstract principls like morfofonemic regularity, or wethr th function requirements of users, especialy lernrs, shud take precednce.

5. Gregersen.
Perhaps. But again, once people learn the system, morfofonemic regularity is not a serious cause of spelling errors.

Let us consider one example where TO resembles Russian and German. Consider the following words, all pronounced identically by most Americns: idle, idol, idyll (/ájdl/). (The English may pronounce the last as /ídil/, but let's ignore that because it has no bearing on the principles involved.) In a CS-based spelling system these would ultimately all become ydl. But note that in TO they are each a distinct morfeme, which shows up differently with different stress patterns: idólatry (/ajdól'trij/), id'yllic (/ajdílik/) - idle apparently has no such variants.

5. Upward.
As u say, nativ speakrs may wel hav it within ther competnce to cope with such patrns - but foren lernrs find it less esy. We hav alredy mentiond th case of malays ho from ther own languaj ar acustmd to th letr <s> always representing a voiceless siblnt, and ho then hav to lern to interpret th far mor complex patrns of alternativ voiced and voiceless pronunciations of final <s> in english. Morfofonemic regularity is a direct obstacl to mastry of th languaj in ther case. Speling reformrs tend to asume that th ideal orthografy for english shud stil giv th <s> inflexion invariant speling, eithr always with <s> or always with <z> - but if we ar to cater for th needs of foren lernrs, it seems we shud seriusly considr riting kats but dogz (and for that matr ript but ribd too - an extension of th distinction TO alredy makes between for instnce wept: webbed). For lernrs at least ther ar therfor reasns to dout wethr such integrity of th morfeme is useful.

6. Gregersen.
Two points. First, I don't think your characterization of spelling reform is historically accurate. Nue Speling has always shown an <s: z> distinction in the plural - and in the possessive, and in the third person singular present ending of verbs - and most other reformers to my knowledge do the same. Henry Sweet in the 19th century certainly did. Harry Lindgren in the 20th certainly does. As I pointed out in my earlier discussion of the possessive, I would not, however. I have proposed in a previous issue of this Journal that <-z> be the invariant indicator of all three morfemes - but with different 'boundary' markers: <'z> for the possessive, but <.z> for the other two.

Secondly, I am amazed that you bring in the problems of foreigners learning English. When I criticized the use of <th> in late versions of Nue Speling for both the voiced and voiceless pronunciations/ð, θ/, Valerie Yule countered that maintaining the older Nue Speling digraph <dh> for the voiced pronunciation made sense only for foreigners. And Cut Spelling would use <th> for both as well. I don't think you can remain on both sides of the fence.

6. Upward.
Cut Spelling has no choice on this point, and dosnt in fact imply that eithr alternativ invariant <s, th> or difrential <s: z, th: dh>, is in principl th ryt aproach for an ideal orthografy. With 3 minor exeptions, CS as curently used adheres fairly strictly to th fonografotactic corespondnces alredy found in TO; therfor it canot start using <z> for inflexions, let alone a grafeme like <dh> wich dos not ocur in TO at al. Like any first staje reform, CS must be subject to severe limitations as to how inovativ its speling patrns can aford to be, and it draws th line at using <z> as an inflexion and <dh> in any circmstnces - watevr ther theoreficl merits.

7. Gregersen.
Yet another issue has to be discussed. In the United States at present, the leading school of linguistics is that associated with Chomsky. Whether one is a member of that school or not, one must take account of the fact that he has made a tremendous dent in the way people think about language in the USA. CS doesn't take this reality into account, and may be confronted by objections from educators and the intelligentsia that its proposals are unscientific (read: anti-Chomskyan). I do not take this view myself, and I think that altho it would be a good thing to accommodate principles like 'the integrity of the morfeme' if it doesn't prove to be too difficult or awkward, ultimately something approaching a fonemic writing (with suitable modifications to take into account widespread dialect variations) must be the basis for an orthography.

7. Upward.
Yr advice about how to confront th linguistic establishmnt in th USA is, I am sure, very wise. Nevrthless, I cant resist th oportunity here to giv my vews on Chomksys conception of english speling. I am very scepticl about it, even down to his implication that languaj as we speak, hear, rite and read it is merely a surface manifestation of undrlying, deep structurs. As I se it, any structurs such as morfemes wich we identify in languaj ar merely notions we impose on languaj in ordr to describe and explain th patrns we observ; but, as John Downing noted, ther is no evidnce that they hav any sycolojicl reality. In particulr, I canot se that Chomskys aproach justifys keeping th <o> in symbol, wen th post-accentul shwa it represents causes lernrs and even mature users such apaling dificltis.

It has always seemd to me (and Valerie Yule provided plenty of evidnce) that empiricly Chomsky is just plain rong to sujest that th integrity of th morfeme is an esential featur of ritn english. He quoted a few exampls to demnstrate his point, such as th fact that th speling of courage is preservd in courageous, despite th radicl shift in pronunciation. But ther ar numerus contrry exampls, th most fundamentl of al in english being th vowl altrnation in jermanic roots, as between sang, sing, song, sung or deep, depth. In words of romance derivation too ther ar many variations, as between pronounce: pronunciation and maintain: maintenance. Ultimatly, surely, one has to conclude that TO is such a hoch-poch of contradictry patrns that almost any jenrlization one may try to make about it can be disproved with a host of countr-exampls.

8. Gregersen.
It seems to me that CS could at least consider the ending furious, impetuous, generous, callous, monstrous, pompous, viscous, etc for which the corresponding abstract nouns have <-osity>: curiosity, etc. If CS writes curius, it breaks up the relationship with the noun curiosity. To preserve the integrity of these morfemes, CS would have to write curios, etc.

8. Upward.
Apart from th question of how curios wud then be distinguishd from th plural of curio, TO itself breks th integrity of th morfeme in these cases by droping th <u> in th noun; if th concept of 'morfeme integrity' had any real meaning here, TO wud rite curiousity. (Also compare french curieux: curiosité - not much morfemic integrity there.) And th word vacuum, by this argumnt, actuly justifys th CS speling vacuus; similry th pair picture: pictorial provides a TO modl for th <u: o> altrnation in curius: curiosity. Th reasn that CS drops th <o> and not th <u> in curius, jenerus, etc, howevr is that <-us> is a comn ending in english (curius then alyns with radius, terminus, etc), so that th CS forms alyn with wider regularitis; but th <-os> ending is rare (rhinoceros is unusul), and to giv mor words th rarer ending wud make th systrn less predictbl.

9. Gregersen.
One point at a time. I don't concede that curios should be the plural of curio to begin with, since the ending <-os> is fonologically misleading and could be confused with that of asbestos no matter how curious were written.

9. Upward.
If I may just intrject here: we seem to be talking about two difrnt kinds of speling reform - long-term and short-term. Th objection to curios as a plural form is a matr of long-term planning for an ideal english orthografy. But th CS objection to th letr <o> in curious is a practicl observation as to th problms that peple hav today in distinguishing such endings as in curious: radius. Th <o> in curious cud be dropd tomoro without any dificlty; but th <s> in curios cud only be chanjed to <z> as part of a fundmentl reorgnization of th riting systm. CS dos not atemt a fundamentl reorgnization; it merely atemts to streamline som of th uglir and mor trublsm excresnces of TO.

10. Gregersen.
But secondly, TO is not consistently morfofonemic. As a matr of record, it is probably the case that the Chomskyans who were interested in spelling reform would try to make the system more consistent in this regard (rather than in showing actual pronunciation). They would probably suggest spelling the suffix <-able> as <-abil>, which you discussed earlier in a totally different way. The point I'm trying to make is that pointing out that TO - or French orthography for that matter - breaks the integrity of the rnorfeme is not a case for or against maintaining such integrity.

The arguments you give for writing <-us> rather than <-ous> are more cogent when you play up counter-examples from English itself. It is an empirical issue whether the counter-patterns are more common than the one I proposed.

Another problem with CS comes to mind: I see no rule in CS that prevents cutting both sweety and sweaty to swety. The spelling swety would seem to parallel weevil>wevil, and easily>esily.

10. Upward.
Th CS respect for morfemes (and especialy morfofonemes) is particulrly aparent here. Wile th <a> in sweat and al its derivativs is fonograficly redundnt (swet shud mach th speling of set), th secnd <e> in sweet is not fonograficly redundnt (sweet ryms with feet), and therfor it is kept in al th derivafivs too.

Th same principl aplys to easily, wich is based on th root morfeme TO ease. Th <a> is fonograficly redundnt, as we se from th final sylabl in Chinese; but if we can cut ease to ese, th integrity of th morfeme demands that easy, easily, disease etc shud be cut likewise to esy, esily, disese. Ther ar even intriging posbilitis here that th confusion between TO forms like tease, lease, cheese, geese cud be somwat reduced if CS adoptd forms like tese, lease, chese, geese, wher readrs and riters wud becom acustmd to th patrn of a voiced <s> preceded by a singl vowl-letr, but a voiceless <s> preceded by a vowl-digraf. (One wud howevr need to ask wethr such a distinction is worth making, or wethr an across-th-bord cut to tese, chese, lese, gese wudnt be betr.)

With TO weevil, alternativ CS forms ar posbl, and a decision is needd as to wich is prefrbl; CS cud eithr say th <i> is redundnt, and so rite weevl in line with anvl; or else th <ee> cud be simplifyd, producing th paralel spelings wevil, evil.

11. Gregersen.
Well, even accepting that CS would not merge the spelling of sweety, sweaty, the same problem can arise in other cases. For instance, in CS as currently practised, it does seem that who: hoe are both spelt ho. Even if Klasik Nue Speling were not adopted, I can't imagine any final stage of decent rational spelling that did not distinguish them. (I myself prefer hu for who.)

11. Upward.
This pair is indeed problmatic. We hav to ask wat alternativ ther myt be in CS to riting ho for both words: one myt refrain from cuting eithr one or both of them; or one myt go furthr, and respel who as hu, as u sujest. But th consequences wich that then implys for th re-speling of whom, whose ar not clear: presumably hum, hus ar not feasbl, tho huse myt be alryt. Howevr such drasticly chanjed forms incur problms of compatbility, especialy bakwrds compatility: if u had been taut hu, cud u read who? (Ther is of corse also a problm of bakwrds compatbility with ho for who, but it is less serius.)

12. Gregersen.
I must say, I don't for a moment admit your criterion of "backwards compatibility" to begin with.

12. Upward.
Do u therfor not think that at least for a transitionl period peple wud need to be able to read both old and new spelings? Surely that must be an indispensbl practicl requiremnt for th implementation of any speling reform. Othrwise evryone in th world ho used ritn english wud hav to be reeducated for th day of reform - hardly a practicbl proposition.

13. Gregersen.
But apart from that, there is a further issue beyond the danger of collapsing pairs of differently pronounced words into the same form. There is also a danger that good spellings would be abandoned, only to be restored later. If the final spelling of weevil were to be weevil, as in Klasik Nue Speling, then wevil would be an unfortunate intermediate stage. If CS also contemplates cutting needle to nedle, that form too has profound disadvantages, since I presume needl (or somthing like it) is what the final stage is most likely to be.

13. Upward.
These ar valid points. It is indeed posbl that th long <e> vowl (/i:/) wud in som ideal orthografy one day be spelt <ee> (tho many reformrs beleve that basicly th letr <i>, not <e>, shud be used to represent that sound, as that is its intrnationl valu). If <ee> wer envisajd, it wud admitdly be stratejicly absurd for a Staje I reform, such as CS, to move away from, rathr than closer towards, th final speling. So perhaps weevl, needl wud be prefrbl to wevil, nedle.

14. Gregersen.
I still feel that these questions show that it is not clear what direction CS is going in. Where does it lead on to next?

14. Upward.
Wile not denying that ther is uncertnty of this kind with a few speling patrns in CS, I dont think ther is any risk that CS wud preemt furthr reforms of any particulr kind on a larj scale. Aftr al, considr th basic rationale of CS: it aims to remove redundnt letrs, and if th letrs it removes ar truly redundnt, no later reform wud evr think of restoring them.

As for specific reforms that myt folo on from CS, one of th simplst cud be to respel hard <c> as as <k>-, at the same time <k> cud also replace <q> if that wer thot desirebl. Aftr that one cud atak th <c, s, z> problm.

But perhaps we cud now try and reach a conclusion about th extent to wich CS dos or dos not respect th integrity of th morfeme.

15. Gregersen.
In a system of writing concerned with preserving the integrity of the morfeme, all three words idle, idol, idyll would have to be written differently. Klasik Nue Speling does so (the only thing that is missing there is a stress mark). Th CS aproach howevr wud not distinguish them, and therefore you cannot say that CS is for the preservation of the morfeme: it simply isn't. In fact the spellings u, yr for you, your go out of the way to ignore morfeme units.

15. Upward.
In point of fact, CS (initialy at least) leves idle, idol as they ar, since to cut them to idl wud fail to indicate th preceding long vowl, and sujests a rym with riddle (CS ridl). If idyll wer deemd to rym (as in british pronunciations with ridl, on th othr hand, then it cud be cut acordingly to just idl. Th form yr for your is a tru abreviation, or a word-syn, used as a compromise solution between th english perceved pronunciation of your as a homofone of yore and th americn perceved pronunciation as a homofone of ewer. But perhaps th consensus wil be that in these circmstances your must be kept, altho it is a blatant non-rym with our, sour etc.

16. Gregersen
What to do? I suggest you must either give up saying that you are for the integrity of the morfeme, or else change CS drastically. Tertium non datur. The first option isn't all that bad: Greek and Latin certainly didn't maintain morfeme integrity - and as you pointed out earlier, the same is true in the orthography of some Slavic languages, including Byelorussian. There is a lot to be said for how they are written. A truly international spelling for English, where <iy> or <ij> or <ii> represented the sound of <ee> in seen would not be compatible with preserving morfemes and would further obscure dialect variations. For example, obscene: obscenity would be, written obsiyn: obsenity or the like, whereas in Klasik Nue Speling the relationship is at least partially captured: obseen: obsenity. Similarly for dialect variants such as éekonomiks vs énomiks (in non-morfemic spelling, distinguished as something like íkonomiks vs ékonomiks). But one could very well argue that the advantages of having international values for symbols greatly outweigh the value of morfeme integrity.

At present, I tend to be for morfeme-integrity, however, because we thereby ignore many variations in pronunciation which would break up the English-speaking world, since the treatment of unstressed vowels varies dramatically across accents, and these variations would have to be ignored.

16. Upward.
I think u hav shed a lot of lyt on th efect CS has on morfemes. Perhaps we can define that efect as folos: with few exeptions (u, yr being a striking case), CS preservs th integrity of morfemes wen these ar undrstood to be based on th stressd vowls in th roots of words that undrgo no fonolojicl chanje. Thus wen sweet takes a sufix, it keeps th speling of th root as in sweety; similarly CS spels th past tense of need as needd, not as neded. On th othr hand th english languaj comnly varis th speling of roots wen they undrgo fonolojicl chanje, as wen lead forms its past tense as led; and CS extends this patrn wen TO obscures it by means of redundnt letrs, therfor speling th past tense of read as red. Similrly TO obscures th fonolojicl altrnation between symbol: symbolic by means of a redundnt <o>; CS then cuts th <o> to produce symbl, wich then machs th TO altrnation between able: ability. To conclude: CS givs precednce to sound-symbl corespondnce over ful morfemic integrity, but preservs both wen they ar not in conflict.

17. Gregersen.
Again, we disagree slightly about analysis. The form led (and the forms cited earlier - sang, sung etc) have to be considered as containing two morfemes (Verb + {Past}) - but of such irregular shapes that they must be treated in a fairly ad hoc way. The word depth, which you mentioned before, is more in line with the issue of maintaining the integrity of the morfeme, since <dep-> is clearly a variant of deep. The most consistent way to maintain such integrity would be to indicate <ee> as <e> with a macron, or bar above: dēp vs dep-th. But barring the introduction of a macron or some other diacritic - which is typologically not so hot - <deep: depth> is about the best we can do.

17. Upward.
My vew of a good speling systm is far less 'deep' than that - in fact, it is delibratly 'shalo'. It simply beleves that if u can spel words as they ar pronounced, that that is th best posbl speling systm that cud evr be devised, and in practice no one wud hav any trubl with it: such a systm wud be functionly perfect, and that is al that matrs. Th problm is that, in presnt circmstnces, th english languaj just dos not lend itself to a straitforwrd reform of that kind, so we hav to explor roundabout routes towards that end. And that is wher complications and disagreemnts so ofn arise.

18. Gregersen
To sum up my position. I tend to be for preserving the integrity of the morfeme if this just involves designing symbols in a special way (e.g. <ae> instead of <ey> or <ci> or <e> to keep words like insane: insanity together) or if it might also accommodate dialect variation (as not indicating vowel reductions in unstressed syllables often does). But I don't think it would be an absolute disaster if a purely fonemic systm were at the base of an orthography.

All I have suggested is that CS sometimes is and sometimes is not able to keep morfemes intact. That for me is not the major drawback of CS. But what I think that drawback is could be the basis for another dialog.

18. Upward.
And since CS of corse dos keep insane: insanity togethr, perhaps we ar not so very far apart aftr al. Edgar Gregersen, thank u for this discussion.

Back to the top.