[Journal of the Simplified Spelling Society, 1988/3 pp17-23 later designated J9]
[Chris Upward: see Journals, Newsletters, Leaflets, Media, Pamflet 15, Book, Papers.]

Conflicting Eficiency Criteria in Cut Spelling - 1.

Christopher Upward.

Most of this paper was presentd at th Societys Fifth Intrnationl Confrnce in july 1987; furthr aspects of th question wil be examnd in a sequel in th 1989/1 J10 isu of th Jurnl. Th Cut Speling used here is fairly radicl, and readrs wil find many of its mor problmatic forms discussd in th articl belo (or in th sequel).

0. ABSTRACT.

With its 3 rules for removing redundnt letrs, th Cut Speling tecniqe for reforming english speling substantialy improves th eficiency of th ritn languaj in respect of econmy, simplicity and fonografic regularity, wile ensuring th new orthografy and th old ar mutuly compatbl. Howevr these criteria of econmy, simplicity, regularity and compatbility conflict with each othr in certn wel-defined orthografic environmnts, and decisions then hav to be made as to wich criteria shud take precednce. Thus: exessiv econmy benefits th riter at th expense of th readr; th visul disturbnce of removing silent initial letrs (as in naw, nee, rong) reduces compatbility between old and new forms; and mecanicl aplication of th 3 cuting rules somtimes blurs crucial distinctions (as between long and short vowls). Sub-rules ar therfor required, to alow exeptions to th main rules. This articl discusss th main circmstnces in wich such conflicts arise and makes som tentativ sujestions as to how they may best be resolvd.

1. INTRODUCTION.

1.1 Th rationale of Cut Speling (CS).
Th CS aproach to english speling reform, as orijnly conceved by Valerie Yule and subsequently systmatised by th presnt authr, primarily involvs th omission of redundnt letrs, rathr than any holesale respeling of words or sounds. This.aproach has sevrl importnt featurs to comend it.
1.2 Th rules of CS.
To establish wich letrs ar redundnt, th definition used is: 'letrs ofring no sycolojicl asistnce to th human readr or riter'. They nearly al fal into one of 3 categris:

1. Som, like <b> in debt, ar totaly irelevnt to pronunciation. Rule I of CS therfor produces th form det.

2. Many, like th alternativ spelings for th 'obscure' vowl shwa wen it precedes final <l, m, n, r> ar hyly unpredictbl; similrly th insertion of <e> in many inflexions givs rise to frequent speling uncertnty. TO itself somtimes omits these vowl-letrs anyway (as in apple, spasm, isn't, centre, hated, rathr than appele, spasem, isen't, centere, hateed), but CS dos so regulrly. Rule 2 of CS thus produces th forms apl: chapl, spasm: fathm, isnt: presnt, centr: entr, hated: hatd, puts: pushs, volubl: valubl.

3. TO dubls consnnts frequently but inconsistntly and usuly unecesrly. Rule 3 of CS says that consnnts ar not normly dubld, so regulrising numerus spelings that difr by singl or dubld consnnts in TO, as copr: propr, rabit: habit, ad: bad, abreviate: abrij, afraid: afray, inoculate: inocuus.

With a litl practice these rules ar soon mastrd, and once lernt can be aplyd straitforwrdly across th languaj. Howevr ther ar circmstnces wher such cuts ar fonograficly misleading, or hav disadvantajs that may outwei th advantajs, and this study wil try and catalog them. Readrs ho hav not atemtd to use CS may form th impression that th hole systm is ridld with problms; but this is not in fact so, th systrn is jenrly clear-cut and simpl to oprate, and it must be remembrd that no speling systm has yet been devised for english that avoids al problms. CS has to be jujd not by th legacy of problms it inherits from TO, but by how much it improves on TO. Th problms ar points of detail, and ar not centrl to CS as a systm; but they do need furthr discussion and reserch. Nevrthless, despite som remaining uncertntis in th detail, th CS systm as a hole has been refined and basicly proved itself thru years of practicl experience, as readrs wil apreciate if they hav folod its developmnt in th pajes of th SSS Newsletter and Journal since 1985.

In adition to th 3 rules for cuting letrs, th authr curently aplys 3 limitd letr-chanjing rules wich also remove serius inconsistncis in TO and at th same time shortn th speling of many words. These letr-chanjing rules ar:
These letr-chanjing rules ar not an esential part of CS, but ar curently included because th coresponding TO spelings create considrbl uncertnty and dificlty, and, unlike most othr letr-chanjing rules, these thre ar simpl and self-containd, and do not giv rise to a chain of complications elsewher in th systm.

2. BREVITY vs READBILITY.

2.1 Brevity as eficiency.
Brevity itself can mean eficiency, altho it dos not necesrly do so. We can se this if we compare th 3 alternativ ritn forms of th names of years: arabic numerals (e.g. 1957), roman numerals (e.g. MDCCCXLVII), and alfabetic letrs (e.g. nineteen-hundred-and-forty-seven). Th arabic numerals take up least space, ar red and ritn fast and acuratly, and do not require a nolej of english. Relativly short and also intrnationly undrstood, but awkwrd both to en- and decode, ar th roman numerals: most readrs wil probbly not imediatly recognise wethr or not th abov exampl represents th same year as th arabic numerals or th alfabetic rendring. Th alfabetic form by contrast requires a nolej of english and is cumbrsm both for readrs, ho require at least two y-fixations, and for riters, ho need nearly 8 times as long to rite it as th arabic numerals. In these exampls, th most economicl form is th most eficient for both readrs and riters.

2.2 Exessiv brevity.
Howevr, a conflict of eficiency criteria can arise from th difrnt needs of riters and readrs. For th riter th shortst posbl representation of words may be th most eficient; but, as with shorthand, exessiv brevity can impede reading.

Th potential problms of exessiv brevity ar seen in Ayb Citrons SPD SPLG, [3] wich acheves much gretr econmy than CS, but at th cost of ful sound-symbl corespondnce. SPD SPLG uses 100 wordsyns (singl letrs, digrafs, trigrafs and som longr forms, each representing a hole word hose TO form is much longr), but th script is hard, if not imposbl, to decifer, unless one lerns th code. Considr th sentnce

D u hav t x tu fays t cmty n c u d t job?
wich shud be red as
"Do u hav th experience to face th comitee and can u do th job?"
As wel as needing to memrize th 100 word-syns, th readr may face sevrl perceptul dificltis with this script:
2.3 How myt CS afect reading speed?
Wat efect CS myt hav on reading speeds is a complex question. John Kerr gave a sycolojists vew: [5] "Most of the time spent during reading is taken up by the processes involved in understanding the text rather than simply decoding the symbols ... readers of a system like CS may not read faster, for the same reasons." Valerie Yules experimnts [6] at least demnstrated that adult readrs quikly overcom th setbak caused by th initial unfamiliarity of CS. Th presnt riter has no experimentl evidnce, but he needs furthr persuading that no time at al can be saved if fewr y-fixations ar required (th fastr reading of arabic numerals in year-names shos that brevity can help at least somtimes).

Ther is howevr a rathr difrnt reasn wy th gretr brevity of CS may not produce corespondingly fastr reading. Wen word-length is reduced, it autmaticly folos that th variety of word-length is reduced too; but length is in itself one of th distinctiv featurs of words in ther ritn form, so that th words their written form (5, 7, 4 letrs respectiviy) ar in that respect mor obviusly distinct than ar ther ritn form (4 letrs each). Therfor it is posbl that with mor uniform word-length, a givn line-length may hav to be red mor sloly and with gretr concentration, altho, even if 100 lines of text take longer, this dos not mean that 100 words canot stil be red fastr in CS. Only sycolojists can resolv such questions; th experimnts cud be conductd in TO to establish wethr readrs scan texts mor sloly wen word-length is mor uniform.

Th foloing sentnces hylyt by exajration certn efects on th apearance of text that can arise wen word-length is cut.

1CS:Confrnces ar pland anuly in Lestr. (28 letrs)
 TO:Conferences are planned annually in Leicester. (40 letrs)
2CS:He ot to go to th in if lo clouds threin.
 (30 letrs, 9 consecutiv 2-letr words)
 TO:He ought to go to the inn if low clouds threaten.
 (38 letrs, maximm 3 consecutiv 2-letr words)
3TO:The two men had now put the big box in the hut.
 (11/12 words of 3 letrs)

Sentnce 1 is over 40% shortr in CS than in TO, and readrs, wil observ how much fastr th y scans th CS version. Sentnce 2 shos how, by shortning spelings jenrly, CS reduces words to a mor uniform length; in this extreme case th long string of 2-letr words makes them visuly less distinctiv and therfor perhaps requires mor concentrated reading (with th add dificlty here of frequent repetition of <o, t> in a very short space); but th 27% longr TO version may stil take longr to read. TO itself can of corse also contain a succession of words of equal length, as in Sentnce 3; th readr may like to considr wethr it apears hardr to read than mor varid text myt. If experimnts proved that strings of 2-letr words, as in CS sentnce 2, do impede reading, th dificlty cud be reduced by leving th definit articl and som othr comn short words uncut.

It is thus clear that th brevity of CS benefits th riter, but it is not yet clear how far, if at al, such brevity helps th skild readr. But even if th skild readr is scarcely helpd, th lernr wil benefit from th much gretr regularity of CS and its relativ lak of dificlt spelings compared with TO.

2.4 Letrs redundnt in som accents only.
A very difrnt kind of conflict between brevity and readbility in CS arises from discrepncis in pronunciation between accents. One of th advantajs of CS is that it dos not usuly favor a particulr accent by implying one exclusiv pronunciation for a word - most redundnt letrs ar redundnt in al accents. Thus no accent pronounces <b> in debt or <e> in apple, nor dos any accent require dubld consnnts in accommodate. Likewise few problms arise in CS, as they do in many reform proposals, over how th vowls ar pronounced (and hence how they shud be spelt) in sets of words like but, put, truth, suit, hue, or in blood, good, room, food, new.

Howevr ther ar a few patrns wher a letr pronounced in one accent is silent in anothr. Shud CS then encuraj som speakrs to cut letrs out wich othr speakrs wud want to keep? In jenrl alternativ spelings must be undesirebl, as they wud undrmine th world-wide unity of ritn english as a comunication standrd; and foren lernrs wud presumably then hav to lern alternativ spelings (as to som extent they do now).

One exampl of a patrn wher perceptions of redundncy vary between accents is found in words like secretary, monastery, raspberry, territory, armoury, jewellery. Many british peple find th speling of th penultimat vowl-grafeme in such words unpredictbl, since they eithr totaly elide th vowl, or at least reduce it to shwa. For these speakrs it wud be very helpful if th letrs concernd wer cut, giving th CS forms secretry, monastry, rasbry, teritry, armry (cf. CS armr), jewlry; a modl for this cut is perhaps seen in wintry, wich has entirely suplantd th oldr alternativ wintery. Howevr americns oftn giv these vowls a clear valu and myt find th cut unreasnbl, tho paradoxicly they alredy rite jewelry.

A reverse anglo-americn exampl is that of th <-ile> words such as fertile, hostile, missile, volatile, hose final sylabl americns tend to reduce to sylabic <l>, so making homofones of hostel:hostile, missal:missile. Th cut forms fertl, hostl, missl, volatl shud therfor be apropriat for americns, if not for th british. It is howevr worth noting that formr speling of fossil as fossile.

Th <wh> words ar similrly contentius. Th distinction between <w> and <wh>, not much made in England, may be insistd upon by americn and scotish teachrs. Shud one therfor rite wat, wen, wich, wy for th sake of those ho do not distinguish th voiced/unvoiced valus of <w, wh>, or shud one keep th <h> in those words to preserv a distinction that for many english is a major speling-trap? (Th authr always hesitates between weather:whether, and much prefers wethr for both.) An argumnt for merjing both spelings as <w> is that al users wud benefit from th econmy and certnty of these forms, wich no mor need to be disfinguishd than do th voiced and unvoiced valus of <th>.

Alredy in TO ther ar ocasionl difrnces of speling between Britn and America wich reflect th absnce of a vowl-foneme in americn english that is presnt in british english:
british aeroplane, aluminium
americn airplane, aluminum.
If worldwide uniformity was not regardd as paramount, such speling distinctions cud provide a modl for difrnt CS forms too: if th british now rite aluminium with one more <i> than th americns, they cud do th same with fertile.

Yet mor dificlt to resolv is th question of redundncy in th word your. Al speakrs agree that TO your shud not apear to rym with our; but ther is no agreemnt as to wethr th form yor or yur best reflects th pronunciation. In jenrl CS trys to cut <ou> wen it dos not represent th vowl in out, as shown by th foloing words:
TO sour, source, scour, course, our, journey
CS sour, sorce, scour, corse, our, jurny
For your CS curently proposes th compromise wordsyn yr, alredy familir as an abreviation.

These exampls concern variations between th domnnt pronunciations of english, RP and jenrl americn. Not surprisingly, discrepncis can also arise between these major accents on th one hand and local accents used by only a few milion peple on th othr; such is th distinction made by som welsh speakrs between th last sylabl of principal and of principle, or th scots pronunciation of plaid as ryming with made rathr than with bad. No global speling systm can atemt to reflect al local variations, and CS here rites principl, plad; but it is not always obvius wher th line shud be drawn. Shud we for instnce, as Robert Craig and Edgar Gregersen hav haf-seriusly sujestd, no longr rite th aspirated <h> because many english peple do not pronounce it (e.g ouse for house)? Such a cut wud doutless be stigmatised by 'educated' speakrs of th major accents, but systemicly it is no difrnt from droping th <h> from th <wh> grafeme. Ultimatly it seems inevitbl that ther shud be a ranje of pronunciations of words that ar aproved as having to be representd by th speling, wile othr pronunciations fal outside orthografic bounds (a point acceptd, from a scotish point of vew, by David Stark).

Howevr, wile speling reforms that start by defining pronunciation constntly fal foul of this problm, CS dos so rathr rarely, th abov patrns being th most widespred.

2.5 Conclusion: CS brevity no obstacl.
Pending furthr evidnce, wethr from sycolojicl experimnts or from major accents of english, ther wud seem to be no grounds for fearing that CS has been systemicly too drastic jenrly in its treatmnt of TO. One reasn for this optimism is that CS (unlike som forms of speedriting) sets out to respect that fundamentl principl of alfabetic script: that it shud spel out th ful fonemic structur of words, so giving gidance to riters as to speling, and to readrs as to pronunciation.

Readrs may howevr question wethr this principt is observd in a CS form like opration, wher th pronounced <e> is cut out from TO operation. Later sections of this articl and its sequel wil discuss this patrn and othrs wher cuts may indeed at first syt apear exessiv.

3. ACTIV TRANSFER EFICIENCY.

3.1 Ho needs to lern th cuting rules?
An importnt eficiency-criterion for CS, as for any reform that claims to be suitbl for imediat implantation, is th simplicity of its rules for th lernr. We may cal this Activ Transfer Eficiency: how esily th systm can be lernt by adults skild in TO ho wish to use th new systm. Here we must undrstand that th numbr of peple needing to lern th cuting rules wud be very smal. Scoolchildren wud lern CS straitaway as th norm, and nevr need to cut TO: TO for them wud just be a mor complicated systm stil used by adults. Th vast majority of adults wud only need to read th new spelings, and wud nevr be oblijed to rite them. Th only peple ho wud need to mastr th cuting rules as such wud be th relativly few adults ho for professionl reasns had to lern to rite th new systm themselvs; they wud necesrly include teachrs, and in du corse perhaps jurnlists, typ-setrs, secretris, and som othr categris. We myt howevr anticipate that many othr adults wud find th simplicity and brevity of CS an incentiv for lerning it voluntrly.

3.2 Simpl transfr from TO.
For adult lernrs a ke eficiency criterion wud be th simplicity of th rules: th fact that just 3 main rules ar suficient for converting most english words from TO to CS. These rules ar far simplr for instnce than th rules for lerning a ful fonemic orthografy, wich requires 40+ grafemes to be lernt for an agreed set of fonemes, as wel as a standrd pronunciation - for al of wich a major reeducation exrcise wud be necesry. It is esy to se how much closer CS is to TO than a fuly fonemic orthografy, if we compare a short text ritn in th two systms. Th Simplified Spelling Societys New Spelling (NS), th fuly fonemic proposal publishd in 1948 [7], included th foloing sentnce:
NS Agaen let us not forget huu form dhe graet majorrity ov dhoez dhat lurn to reed and riet.
CS Again, let us not forget ho form th gret majority of those that lern to read and rite.
In NS, th speling of 11/18 words has been chanjed, 2 of them shortnd and I lengthnd. In CS, a new speling is needd in only 5 words, and is acheved in evry case merely by omiting a letr from TO. In th fonemic systm adult lernrs hav conciusly to create th speling of each word, wile in CS they only hav to monitr and cut th familir TO form.

3.3 Total mastry unecesry for adults.
Adults lerning to aply th CS rules start by monitring th letrs in words as they rite them, omiting those that ar redundnt. But especialy if first atemts ar chekd and errs corectd, th systm is quikly lernt and confidnce gaind, indeed th relief at dispensing with many uncertntis of TO soon becoms a positiv incentiv to using th systm. Befor long th CS forms becom automatic, indeed one user even abandnd CS because he was afraid he myt be unable to return to TO. No dout adult professionls lik teachrs ho had to mastr CS wud need training, but it wud be less elabrat than th training teachrs receved for i.t.a. For one thing total mastry of CS wud be unecesry - only th words needd in th classroom wud hav to be practiced.

In jenrl, an importnt practicl advantaj of CS over a comprehensiv or fonemic reform is that even if not al redundnt letrs ar omitd, words ar stil imediatly recognisebl. Thus if we compare TO accommodate, CS acomodate with th two posbl intrmediat forms accomodate, acommodate, we se that al four forms ar equaly readbl. Here th moto "if in dout, dont leve out" is a useful safegard, in that it ensures that th speling used wil lie somwher on th continuum between TO and CS, and wil not be randmly mangld.

3.4 Ar ther any othr redundnt letrs?
Ar al redundnt letrs covrd by th 3 rules? Brodly speaking they ar, but a few patrns of redundncy may not be entirely self-evidnt and so may require special lerning - or even be too controversial to be, acceptbl:
Th abov forms ar inevitbly among th most controversial proposed by CS. Esentialy th justification for forms like u, brek, brod, grup is that th TO digrafs <ou, ea, oa> ar seriusly misleading here, and altho th CS vowl-letrs may not represent th sound unambiguusly or precisely, they ar closer to it and so at least constitute an improvemnt over TO. It wud howevr be esy for CS not to make these cuts, if ther wer a consensus against them.

3.5 Eficiency for beginrs: consnnt strings.
It must also be askd wethr any particulr lerning dificlis can be forseen for childrn or forenrs in CS, wich ar not alredy presnt in TO. Th advantajs of CS over TO (econmy, regularity) for th lernr ar evidnt, but som teachrs fear problms with consnnt-strings. Because CS cuts out mor vowl- than consnnt-letrs, consnnt strings tend to be longr and mor frequent than in TO, and since children find consnnt-strings dificlt in TO, teachrs wondr wethr th problm myt be agravated in CS. TO contains som complex 5-letr consnnt-strings, as in eighths, strengths, but they ar fairly rare. In CS, on th othr hand, strings ocur quite regulrly with up to 7 consnnt-letrs, as in govrnmnts, circmstnce, aftrwrds, complmnts. Ther ar howevr sevrl reasns for beleving that, watevr trubl consnnt-strings in jenrl may cause, in CS they make th speling esir rathr than hardr to handl:
3.6 Conclusion: inherent simplicity.
This section has tryd to sho that th CS rules ar inherently simpl to lern and to oprate. Howevr, ther ar cases wher this criterion of Activ Transfer Eficiency conflicts with othr criteria, and wher rathr sutlr discriminations hav to be made than th 3 basic rules themselvs cater for.

4. PASSIV YRANSFER EFICIENCY.

4.1 Compatbility
Next to be considrd is th criterion of compatbility between old and new orthografis. CS is based on th premiss that a Staje 1 reform that wud radicly chanje th apearance of ritn english is politicly unrealistic and sycolojicly unwise. Th old and new orthografis must be compatbl with each othr in both directions: adults must be able to read th new systm esily (forwrds compatbility), and children must be able to read th old systm esily (bakwrds compatbility), without extensiv re-education. This two-way compatbility between new and old, wich we may cal Passiv Transfer Eficiency, means that words must remain esily recognisebl. CS acheves this by its tecniqe of mainly just omiting sycolojicly and fonograficly redundnt letrg, wheras a reform that chanjes many letrs, especialy stressd vowls, is visuly or disturbing and hence less compatbl, as wil now be shown.

4.2 Forwrds compatbility.
Th sentnce "To the learner interested in the history of the language the old spelling would be easily accessible" is now givn in 3 reformd orthografis, 1 as quoted from th 1948 New Spelling, 2 in Simplified American Spelling, [8] and 3 in CS, togethr with statistics indicating th degree of chanje from TO:

1.To dhe lurner interested in dhe history ov dhe langgwej dhe oeld speling wood be eezily aksesibl.
    15/80 chanjed letrs, length = TO -5%
2.To th lurner interested in th history of th langgwej th oeld speling wuud be eezily acsesibl.
    10/76 chanjed letrs, length = TO -10%
3.To th lernr intrestd in th histry of th languaj th old speling wud be esily accessbl.
    1 chanjed letr out of 68, length = TO -20%

First reactions to th thre difrnt spelings wil be impressionistic, but almost certnly th readr wil hav found th first version hardst to read, th secnd version esir, and th CS version esiest th <j> in languaj being th only unfamilir letr. Th implication is clearly that th mor chanjed letrs an orthografy contains, th hardr it is to read unprepared. CS indeed positivly lends itself to imediat fluent reading: th esentials of th TO gestalt of most words ar preservd, and th fastr one reads, th less one notices that letrs ar missing. Th eficiency observd here, then, is a matr of how fluently th uninstructd readr scans text in th reformd orthografy. But altho this forwrds compatbility is a gret strength of CS, it may somtimes conflict with th first eficiency criterion, that of Activ Transfer Eficiency for adults, in othr words with th regularity of th 3 cuting rules.

4.3 Degrees of forwrds compatbility in CS.
Ocasionly th regulr aplication of th 3 CS rules results in considrbl disturbnce to th familir apearance of words in TO. Th foloing grups of words sho a progressivly increasing degree, of visul disturbnce, from th very slyt to th seriusly disruptiv. In th first grup, th cut is not very conspicuus:
unconstitutionl
receit (cf deceit - also etymolojicl eficiency)
leve, sleve, receve, beleve (cf eve, but receit, belief)
In th next grup of words th cut is visuly mor disturbing because th initial letr (i.e. th most promnnt letr) is dropd from th TO form:
nat, neel, nemonic, syche, rong
As wel as undrgoing a 50% cut and losing both its first and last letrs, th foloing speling introduces an aditionl elemnt of disturbnce by merjing th homofones know, no:
know→CS no
Perhaps th most severely cut words of al (if th norml CS rules ar mecanicly aplyd) ar th foloing, th first indeed losing 66% of its letrs:
eye→CS y, eyesight→CS ysyt
honour→CS onr, honourable→CS onrbl.

Th absnce of a letr from th midl of a word of medium length or longr may not even be noticed in fluent reading (any mor than we notice many misprints), because most of th time we read wat we expect to read. Th conflict of criteria we observ in th mor disturbing of th abov exampls is between compatbility with TO, wich is rathr lo, and regularity both of th cuting-rules and of sound-symbl corespondnce, wich is hy; and we hav to ask wich criterion shud hav priority. Shud we say that for instnce th word y shud keep its first fonograficly redundnt <e>, and onr keep its redundnt initial <h> (ey, honrbl, in ordr to remain esily recognisebl, or shud eye be spelt regulrly, as my without th <m>, and honour like on with a sylabografic <r> add? Speling reformrs may prefer regularity in these circmstnces, but th public, wich must be persuaded to accept th forms, is likely to atach hyr priority to familiarity, in othr words to forwrds compatbility.

4.4 Repeatd consonnts.
At first syt disturbing in CS ar th repeatd consonnts with repeatd pronunciation, as in probbl, needd, maximm, linn, terr. This fonografic device dos not ocur at al in TO, and is therfor a complete novlty for th readr encountring it for th first time. Such repeatd consnnts must be clearly distinguishd from th dubld consnnt letrs that ar such a comn but iregulr and trublsm featur of TO; but they do hav som afinity with th repeatd <c> with difrnt pronunciations in words like accent, success.

Th visul disturbnce of repeatd consnnts in CS is a direct conseqence of th regularity of th systm. Th speling of th last sylabl of words like hooligan, beaten, cotton, important, different is regulrised by reduction to sylabografic <n> (hoolign, beatn, cotn, importnt, difrnt). Regularity then requires th same reduction even if, as in linen, cannon etc, th preceding letr is also <n>: linn, cann. Th dilema we face is wethr to complicate th cuting rules and introduce systemic iregularitis by making exeptions in these cases for th short term benefit of readrs transfering from TO, or wethr th visul disturbnce for these readrs is a price worth paying for th long term regularity of th systm. Readrs do aftr al soon becom acustmd to new forms.

Two aditionl peculiaritis shud be mentiond in this context. Th first arises if, as apears necesry, final <ss> is not simplifyd in CS, as in words like class, miss. In that case, forming inflexions by th adition of just <s>, as is th norml CS patrn, rathr than with <es> as in TO (classes, misses), results in endings with 3 consecutiv <s>s: classs, misss. Like th othr repeatd consnnts, this patrn is not in itself a problm, and th readr soon becoms acustmd to it; but at first syt it undoutdly apears stranje. Mor awkwrd on transfer from TO is th past tense inflexion of th verb to ad, wich by th regulr CS rule becoms add (cf.needd). If this word is taken out of context, ambiguity dos apear to constitute a real problm of both forwrds and bakwrds compatbility between TO and CS. Howevr, th context usuly makes th meaning clear, as in th sentnce: to form th past tense, in CS, th letr <d> is simply add to th root, but a sentnce like we add <d> to th root myt at first be misundrstood as th presnt rathr than th past tense.

4.5 Bakwrds compatbility.
Now let us considr bakwrds cornpatbility. How esy wud it be for children ho had lernt CS to read TO? They wud aftr al need to be able to do so for many years, since ther parents wud mostly stil use it, as wud al erlir printd material. We can esily juj forwrds compatbility just by considring how hard we ourselvs find it to read text in th new speling; but asessing bakwrds compatbility is mor dificit, as we hav to imajn ourseivs having lern to read and rite in a mor fonografic orthografy than TO, and then looking at TO with difrnt ys from our own.

John Downing pointd to th kind of problm that can arise, wen he described [9] how on transfer from i.t.a. children tend to misread TO shoe as show, since show is spelt <shoe> in i.t.a. TO one can also be such a trap if its speling is reformd to represent its pronunciation: if th lernr is familir with 'majic' <e> aftr a consonnt as a way of indicating a preceding long vowl (as in bone), th form <one> must apear to be pronounced as own (just as in TO beginrs ofn pronounce the word once as tho it wer spelt onki). Anothr problm wud arise with miniml pairs if childrn wer taut <s> for th unvoiced siblnt and <z> for th <s> inflexion in TO: a child ho lerns hence with th speling hens is bound to be confused on encountring TO <hens>. Similrly th TO forms come, comb, comma, coma contain th seeds of multipl confusion if a reformd orthografy atemtd to spel them fonemicly. Even CS wud merj coma, comma if Rule 2 for simplifying dubld consnnts wer aplyd rijidly (it is here asumed that in such cases th dubld consonnt has to be kept, and in jenrl that CS shud not cut letrs if hetrofones wud result). Th abov exampls sho th dificltis of bakwrds compatbility that wud arise particulrly from a speling reform that actuly chanjed th letrs in words.

4.6 Bakwrds compatbility of CS.
By not changing many letrs, CS larjly avoids this problm. One way to visulise how TO myt apear to those ho had been taut CS is to look at elizabethan speling, wich difrs from TO much as TO difrs from CS: mainly by extra letrs. Th foloing sentnce has been selectd from th 1588 Bishops Bible for its particulrly markd deviation from TO, with th TO and CS versions aftr it for comparisn:

BB Beholde the fowles of the ayre: for they sowe not, neyther doo they reape, nor carrie into barnes. (77 letrs),
TO Behold the fowls of the air: for they sow not, neither do they reap, nor carry into barns. (69 letrs)
CS Behold th fowls of th air: for they so not, neithr do they reap, nor cary into barns. (64 letrs)

Th 1588 version is not dificlt to read today, and it contains a similr proportion (about 11%) of letrs that ar cut in TO as TO jenrly contains letrs that ar cut in CS. From this comparisn we can convincingly imajn how TO wud apear to a readr educated in CS: not hard to undrstand, but arcaicly grotesq in th irationality of its forms.

A slyt dificlty myt be th gretr variety of letrs CS cuts from TO, including special hazrds like <gh>; but it is probbly no mor serius than our momentry puzlmnt wen confrontd with th Elizabethan habit of using <i, j> and <u, v> intrchanjebly in forms like ivdge, Iesvs for judge, Jesus. Th foloing sentnce has been composed to exajrate th dificlty that cud arise if a TO text containd an exeptionl density of unprodictbl extra letrs:

CS: Tho thot tuf, english speling ot to be taut ryt enuf. (41 letrs = TO -30%)
TO: Though thought tough, English spelling ought to be taught right enough. (59 letrs = CS + 44%)

This exampl incidently demnstrates poor compatbility both bakwrds and forwrds. Th econmy of th CS version is striking, but it is particulrly th many non-fonolojicly motivated extra letrs in th TO version wich reduce bakwrds compatbility (i.e. make reading hardr for CS-educated readrs). Obviusly, howevr, such an absurdly artificial exampl dos not imply that CS-educated readrs wud normly hav dificlty in decoding TO in a real reading situation; and presumably in th erly years of reform they wud be warnd of th <gh> anomly in TO, altho they wud not hav to lern it.

4.7 Conclusion.
We hav here examnd conflicts between Activ Transfer Eficiency (mecanicly aplying th 3 cuting rules) and Passiv Transfer Eficiency (bakwrds and forwrds compatbility, making CS as esy as posbl for adults, and TO as esy as posbl for children). We hav found that ther is a dilema: if we try to minmise difrnces in apearance between TO and CS, we need exeptions to th main cuting rules of CS; but if we want to make CS as simpl, regulr, fonografic and predictbl as posbl, then we shud giv priority to th 3 main CS rules, howevr stranje th resulting speling may look. We wud then hav a betr speling-systm for futur jenrations - but probbly at th expense of imediat public acceptbility.

ENDING PART 1, INTRODUCING PART 2.

Part 1 of Conflicting Eficiency Criteria in CS ends by stating the dilema that now arises for th furthr developmnt and promotion of CS. In fact it is a dilema wich faces al speling reform scemes: wethr to giv priority to a systm that is linguisticly and sycolojicly sound in itself, or to make concessions at th outset to expectd public dislike of th weirdr-looking forms proposed. This paper has atemtd to catalog som of th detaild choices that wil hav to be made, along with th considrations that need to be born in mind in making those choices. Th secnd part of th study, to apear in isu J10 1989/1 of th Jurnl, wil then deal with furthr importnt choices that th CS systm presents; th most importnt hav to do with th distinction between short and long vowls and with th hierarchy of ambiguitis in TO and CS (homofones, homografs, etc).

Meanwile, readrs ar urjd to considr th points alredy made, and send in ther observations.

REFERENCES FROM PART I.

JSSS Journal of the Simplified Spelling Society,
SSSN Simplified Spelling Society Newsletter [J1-J3].

[1] Upward, C. 'Cut Speling - a Linguistic Universl?', in JSSS, J5 1987/2, pp.17-25.

[2] -, 'Can Cut Speling Cut Mispeling?', in JSSS, J6 1987/3, pp.21-24.

[3] Citron, A. 'Spelling for the Computer Age', SSSN, J1 Autumn 1985, pp.10-14.

[4] Knowles, F. 'Information Theory and its Implications for Spelling Reform' in SSSN, J2 Spring 1986, pp.5-13, §3.1.

[5] Kerr, J.M 'The Implications of Spelling Reform for the Skilled Reader', in JSSS, J8 1988/2, p.20.

[6] Yule, V. & Greentree. 'Readers' Adaptation to Spelling Change', in Human Learning, 1986/5, pp.229-41.

[7] Ripman, W & Archer, W. New Spelling, London: on behalf of the Simplified Spelling Society by Sir Isaac Pitman & Sons Ltd, revised by Jones D & Orton H, 1948.

[8] Rondthaler, E & Lias E J Dictionary of Simplified American Spelling, New York: American Language Academy, 1986.

[9] Downing, J 'The Transfer of Skill in Language Functions', in JSSS, J5 1987/2, pp.5-12, §4.8. reprinted in JSSS J28 2000/2, pp3-11.


Back to the top.
See Part 2 in J10.