13/9- 50
Dear professor hazel1,
I am sorry that 1 have not answered your let, ter a till new. ihe reason that I nave not been well for a very long
is , as I wrote to you briefly, time. I have only been abiéuto give my lectures,ana it has been necessary to postpone many other things , among these in* letters' which aemanaea a relatively high amount of thinking. - And you Know perhaps that your ar- tides make rather difficult reading, even for us in Copenhagen who are accustomed to Hjelmsrev, - ana a suppose that they are still more fiifficut for people who have not had this hard training. But at the same time i find them highly stimulating.
But before putting questions about your articles 1 will answer the questions in your letters. - Mou asked ( in connection with my articl in "Recherches" , how one could know that the alternation public- pumiici- ty is irrelevant before knowing that k ana s are two different phonemes. - In a certain sense you cannot, but you nave to investigate the whole ma- teriai a as a whole. - And then you will flndh that u and v in Danish axe never coiamutable, whereas k and s are commutable in bngiish. And the theoretical possibility of saying that k and s in te ^iveri example are variants of one phoneme, whereas other s'es or k's are variants of other phonemes ±&xHfitkxkftxy win give unnecessary complications for the a©scrip- tion of the'manifestation. - In your first letter you asked what glossemauicians are thin- king of Roman Jakobson's distinctive features. - inis can ue answered very briefly so, that Hjeimalev finds that the breaking down of the ponemes into distinctive f eatures is a purely substantial matter, and that xxars. this analysis constitutes a jump from xsmxM&ih&x. a formal point of view into a substantial one. - Stjamaxxmm*. Roman Jakobsen was here in the month of May ana „ave kikke various lectures , a very brilliant one on Slavic Mythology, an interesting lecture on linguistics in Sovjet- Russia, and finally a lecture on Sound and-s*teanin_, which treated parti- cuiariy tne problem of the distinctive features, in a meeting in the lin- guistic Circle I tried to give a sketch -of the iundamental ksus;* differen- ces between Jakobson ana Hjeimsiev, and we had a very lively diskussion, where Jakobson protested violently against Rjelmslev's sharp criticism. - I for my part am rather hesitating . 1 find that much can oe said in favour of Jakobson's view. - As you wifi nave seen from my article in Recherches" I do not. think, that na purely formal analysis is possible at any stage, i.e. that it can. lead to an unambiguous result. - hjeimsiev has sometimes in .his lectures useaths example that you may , describe the minister of education, the mini at ei of IV- -ign affairs!purely on account of tneir functions without any- regard oi trie persons, ana so in linguistics you m&j ..describe trie elements on trie oasis of their function alone, - but in the given example we have the case that•each element has a relation of its own, which the others have not, and Ju&*yxluiJi.*xa«fc in this case you may determine La© inventory ana 6ive tne definitions witnout any regard oi the suogthnee. nut. in language the elements nave arways some relations inaa©iifii©nr©iatiii§®yDebwgaag^hethetwowetvehevåsaåi nesibiim« in common, the consonants at any rate many, ana inen the question of identification arises, and furthermore language is a system of signs with two levels, and
2portance, tne generally adopted method is this that elements having aiffe- rent relations to the otuer level are considered as formally different, whaxaaaxafffltaiLi; ( and these elements must have one or more relations in the chain in common) whereas elements with totally different relations in tne chain are reduced ( as far as possible; to variants of tne same .element , ana fox these variants and for those with partly the same reia- tions there are problems of identification which cannot be resolved xoUsxgi unamuiguously from the standpoint of form ( I taxe form in Hjelms lev's ,sense-- relational form# ). - The example-with the ministers is in fact very bad# lor in language you would try to reduce them to one having dif- feient £bu&sJsx relations, and in connection with these different relations aiso variation,in tne.personai substance . The endeavour to reduce the in- vent cry ie involves that you -cannot have phonemes with totally 'different relations in the chain. - .But once the inventory established you employ the remaining relational differences to define different categories.- Hut the jb& further the reduction, tne more elements wifi get the same definition. - 1'his has always troubled me somehow, bafxix as a somewhat contradictory method, but 1 cannot see how to avoid it. - nnyway you cannot maxe the luentiiicatxons without any regard of the substance if yop, want a result which is apt quite useless. - - The inventory £xa£ is at any stage of the ahaiydis' ( accentual u ties, .syllables phonemes — ana correspondingly in the' content) made on the oasis of commutation and identification of variants ( where tne sub- stance always comes in). - Ana now the same can be said about the aistinc tive fee tuxes, i'he inventory is made on- the oasis of commutation and idee tificetion, Furthermore' just as the phoenemes , once the inventory estab- iisned, can be defined pn purely relational terms, so tne distinctive fee tures may oe defined in this way ; nasality always combines with stop <j presupposes the feature. "stop" J and so on. - Ana the terminology "stop" "nasality” is no more substantial than the names f, p, A-etc. - ©a tne othef hand there, is a difference between the .-breakin up of phonemes into distinctive features ana tne division of the syilabija into phonemes. - nut perna,s it is only this tnat phonemes generally succeed - another in tne chain , distinctive feau^res are simultaneus t but this difference is not absolute, cp. /that ng may be manifested as 9. , and one phoneme as ph). And this .aimuitaneousness max-s tne.substantial identifi- cation more difficult* - In.one of your articles you say that the units of content can onif be identified if tney qan hava .different expressions but the units of ’expression can be identified without having different separate contents. Perhaps this can be aon.e because the perception of ■successive sounds v of a raider restricted number; list is relatively easy, i .e. people agree dr. this point ( the content is much more comple but'wnen we come to .srmuic&nous souna-elaments we are hot so sure. - Ana whereas acoustical,, physiological and p'ercepti. nal anaiyaés uo not give very different results for -successive' phenomena, it is not so simp! for the. simultanous elements ( cp. : the vowels i,yyu -e',0,0 make physi logically a three-dimensional system ( front-oacit- rounded- unrounded, ■high -iop. • physically a two-oimeaeione 1 'system ( according to upper and ib/.er fr.,maiit ) ana perceptionaiiy ( as far as i have observed through * some experiments with studenta one.démånsionai system. 1'his gives ■quit' different distinctive features. - Ana 1 find Roman Jakobsen's mixi up o... physiiiogipai, physical ana perceptional terms lather confusing. - i’hu8, it is more difficult to identify tne aistmctive features,, this does not involve, that tne method-or the principle has changed. - Hjefmsiev said that it. is at any Bate-a new- step in the analysis, for m, nasality ana. stop cannot enter into tne same paradigm. 1'hat is true, out the question is ii the new step means a change of method. If it doei not, it must be taken, for it gives a further ieduction of the inventor,1
but
3and it carries the particular analysis one step further, wheieas Bgeims lev’s breax,ng-down of the taxemes into glossernes is a'universar anaiysi which would be identical for two languages having the same number of tax a es in a category. - As his analysis is always made so that the smallest number of flossernes is arrived at ( 6 taxemes ar« resoivea into d times 5 glossernes, 9 into 3 times 3 etc.) there will be solidarity between the dimensions, and he can say that the anaiysis according to selection stops. But if the features are established according to commutation ana substantial identification, the number may - be- greater, out there may ais •be selection.
jSjs X am thus inclined to find that <jaxobson*s t and one the whoi the phenologists’; attempt to set up distinctive features is very in- teres ting. JBut j. do not find the give;., examp lest French and neroian) very good, and perhaps it is so difficult that we must give it up. -—
I have now read your article^ again, with grafeat interest. But i am not sure that i have understood everythin^,. - In jour article about "le principe de compensation" i find fch jutiuaxipixxxxxyxxxxxxa'aLjcåHgx tire new formulation of cn© principle very interesting, and you are certainly right -in asserting that generally the so-called indefinite ana definite articles are, not members of the same paradigm, As for singular ana plural 1 am not so sure. Can you have a plural without a singular ? - (if the singular is unmarned it ijiay iaci e p-iuiaiidy in some cases). - And why is the singular considered as unmarfc in any language? It may be the piural that is unmanned. - ■everything in I tninx that i nave understood your article on the morpheme ( and i agree with you on almost' aii points) except the distinction between neutralisation,and syncretism at the end; - - 1 find this new conception ol the "morpheme " very useiui 'in i.e languages ( but pexhapsyou might have given it another name, ""morpheme" has already too many senses. - XM-t Ana cue formulation of the definition p . 11 > £ s somewhat va0ue. X find. cno definition which you give Acta Bing. V,2 better, if "association" is not banen psychoio&i caliy . -
i was glad that I had so many articles of yours together, because various problems were-treated in more tnan one of these articles, e.g. the fundamental syntactic relations. - I find -your three syntactic relations; determination, subordination and cohesion very useful,parti- vurariy the distinction between determination ana subordination, which h not bees uacle in this way before. - 1 agree with' you in your criticism of' lijeiihsiev ’ s "determination". It must te presupposition at a given point in the chain, his formulation is too vague. - But I do not iquite see that this involves that there; is no determination between vowel ana consonant ( A ataamlxhx consonant presupposes a vowel in the same syl laoie, tnis must oe jamaEm enough as for the point in the chain). - Ana I do not.quite see either that your definition of determination beioggs- to wiiat you have called pattern relations in your article in''"nechefches", whereas•Hjeimslév* s belongs to the syntagmatie relation at any rate not if you maintain cue definition which you _>ive in the same article p; /j>, — X have'read it once more, perhaps you - are right, out you nave nad a different idea of it your sell in your article on the ruhdåméhtal relations, where you say (p io) that inthe case of 'subordination i as'opposed to determination;•there is not any presuppo sitioii of one morpheme by another , out 'merely the presupposition or one place in the syntactic chain by another, ^oea not this mean that subordination - is a, pattern relation ana determination is a syrta^atxc
4ft . ..
Or what is really meant by syntagmatic relations?
relation? I find it practical to distinguish as you dé in "Recnerches" between relaions betweens members of pattern ( between subject, 'object, predicate etc.) ana tne faculty of. enter.. ng relation's which may characterize certain morphemesC• and which must be the basis for establishing the parts of speech). — It is not clear to, me what is meant by "reference"."Is it deter- mination between terms which are not immediate constituents?
I was very glad to read your criticism of the aqjuucxx&k: cone ep- tion that syncretism of cases etc. are syncretisms of content. - bider- richeaen anai nave maintained tne same view as you ( that those cases 'st&z constitute syncretisms of format!ves - i.e. of sign-expreasions) in a discussion with hjelmslev. - xxxxKxxitKxy:« I aiso find your examples from finalsh, nreek and Chinese ( final aj Very goon and illustrating a theoretical difference, out I should life« to have your definition of the worn "role". - Your contribution to tne Hjelms lev -vo lume was: the most dificuit to understand. - Bn p. ?o you distinguish between phonetic order, phone- mic order aadz, order in the system ana structural order. How are thexat fmaxx relations oetween these four types of order ana the four types of relations given at the start. - Structural order is a relation in the pattern ( but 1 do not quite understand . Kk&i what is meant. Could you give more examples from the expression?)*- Order in th system must - correspond to number n (relations in the system^system usd for Hjeimsle^s fomrn))- Are both phonemic ana phonetic order relations in praesentia tzpzxz number one) and has number 3 no correspondence within orders? — Åre overt relations tne some as relations in tne system?
In some cases ( so when you speak about a neutralisation of or>- der where the oraer is fixed and about neutralisation of relations within the word) I think that it would oe useful to distinguish between neutra- lisation and defective distribution, when there is no criterion to tell which terms are neutralised, it is better to say that one is found ana the other cannot be found in this place. hate sos XM£.-usaxKgx sm, si etc, but no sp, or sr. - I should not speak of neutralisation here. - Ana in the same way: you have in Danish une order sp, sm etc.'initially, but not os- ms to consider XXXXSxxxxxxxxphxxxma:xxiisxche syiiaole as a pattern with diiferenc places ( central, marginal, initial, filial, first otx second in initial or final group etc.; whether tne permutation oetween phonemes is relevant or not, ana to define tne dixfi'erent phonemes on the basis of the places they may occupy ip,this pattern (if only relevant order may pe used, it canxax oi eoux5etoe done). X'his is also a good basis fox comparing languages, whereas the specific relations between specific phonemes are mor« complex and uo not yield a gbod basis for' comparison. But I nav« already written too much. - And I could say a good deal more on this question. I have just $ given a lecture in Finland about .the relational definition of phonemes, which will p maps be pubiisned m Acta Linguistica, and we are also going to dxscuss tne questions of order in the Linguistic Circle next week*- Uidall is giving a lecture about xmxm linguistic relations. - — lout article on tne asymmetry oi tne iihg.sign has been published in AL V,3, which appeared in July. But i have just realised that tne editors have not sett any off-prints yet. i nave now given then your address in England, ana you will get tnem very soon.
So in Banish you
i think it is useful
etc.
!
I hope you will forgive this iate*.outin return very long, answer to your letters.
tours sincerely