!
S
Istanbul 11/2/55
!
Dear Miss Fischer-Jorgensen, 1 hope you received the brief replies to your questions by the 15th# small matter if you did not, since as I said,there were no important misinter- pretations, and none at all for which you could be to blame; I wonder if you could let me have a copy of the critical part of your presentation (even if it has melted bitter on it)? A few more details. "Intrinsic similarity" and "composition" are of course quite distinct when it comes to different mediums, since then intrinsic similarity has no applica- tion at all, whereas one may still distinguish between isomorphism of composl- tion and isomorphism of distribution. In a conventional phonemic transcrip- tion, the letters may be sa4d to have the same formal distribution as the phon- ernes they represent, but not the same formal composition. **here^ei£ the different features were represented each by invariable letter-parts, thej/could said to have the same formal composition as the phonemes. Of course, to say that two wholes (e.go the wholes constituted b£ / phonemes divisible into features) have the same formal composition, is to say that the distribution of the parts (up to the level of the whole, but not beyond) is the same (e.g. that if nasality combines with ocslihsiveness/ but not with vocal- ity, then e.g. there is a dot which combines with a stroke but not with a circle and so on, in the compositionally equivalent graphic system). So statements of formally equivalent composition may be turned into statements of formally equi- valent distribution. But the reverse is equally true. Statements on the dls- tribution of phonemes can be turned into statements on the composition of wordso ■°ut this points to an important terminological error in my paper, to which am grateful that you have indirectly drawn attention, when opposed to lntrlns c , distribution does not have the same meaning as when it is opposed to compositiono In the former case it means only distribution among substantially defined units. Within a given medium, this distinction is naturally superfluous9 since however they are defined, the units will be of like substance. But of course within the fgiven medium, the distinction between "composition" and "intrinsic similar- ity" is also (with the relatively trivial exception made in my letter) also sup- erfluous• My terminological solution Is to separate the opposition intrinsic/extrinsic from the opposition compositional/distributional, -they are parallel oppositions, but the one should be general, and the other apply specially to a given medium. rhere is hence another confusion that I ought to have dealt with. My fault, according to you, Is that I made a false distinction. make a false identification. Anyway * was wrong. I am lookimg forward to the studies you announce, especially to that on the overlapping manifestations of phonemes. ‘lhis Is closely connected with the question of the arbitrary", on which I am grateful to have your agreement in principle. I wish you could have dealt with this in ypur paper for -“eta. Por instance the Japanese distribution Ha Hi fu is not to be set on a par with the conceivable distribution fa fl hu» which would not be easy to motivate -- henc- in the latter cage one would be more Inclined (other things equal) to assume ^accidental gaps in the distribution. ‘lhis is the only criterion which (In Linguistic lorm" ) Iadded to those which you gave (in the second paragraph, Si«8 2fv0n?^d a haif H?*3' on P*1® --did you recognise that this was but A\for llnSul!tl° in general, nobody approves of tualS von ^ seem to approve of some part. Yoy are nearly the first (ac- your Extension i apJ\ro7® °f graphical parallel. I accept your extension of the analogy, which can be developed*
But
also be
ihe fault was rather to
Yours sincerely
2. t,
p.S. Ishould say more about the form/substance opposition, since here lies your main charge against my treatmenht. But after all the notion of isomorphism is clear enough. Now one may say that one has given a formal description when all the terms used may be indifferently substituted for any other terms, pro- viding onljp that the same logical relations hold between them. All other terms are substantial• 'ihe mere fact that one can use after of temporal as well as spacial rela- ^fter is not a formal term when
tlons does not turn it into a formal term. used in such a way that one could not ihdifferently substitute before for after and vice-versa, since the two terms obey the same logical rules. Nor is it a formal term if one could not substitute above and below for before and after. Within a system of discrete successions, as phonemics is, one could also substit- ute the relation "father of", since all the logical relations are equivalent -- at least if 6ne takes after in the sense "immediately after", as is usual in phonemics. If of course it means "anywhere after" then this meaning would for- mally include "ancestor of" (or for that matter "descendant of"). It is not necessary that a formal term should have any application whatsoever outside a given domain, “hat is necessary id that it should be applicable, whenever the logical relations are the same. Ihe relation "simultaneous with", if taken In the substantial sense, has no application in phonemics (as opposed to feature-analysis); but formally it has an application, for "simultaneous with obeys the same logical rules as "being in the same word with", and hence, if formally used, must include this. (If Å is 'aimultanous with b, and B with G , A is simultaneous with C -- and this holds if one substitutes "is in the same word with" for "simultaneous", etc«). In other words a term is formally used if all substitutions are synonyms when they do not (in the context) affect the calculus of relations« And here, I think, one may lay(|one*s finger on another reason why distri- bution is regarded as more "formal than composition. In a logical calculus, all the rules are either rules of distribution (known as "rules of formation"«? or else rules presupposing such rules (e.g. rules of transformation -- cf. r>ar- Hillel's last article In Language). Composition ddes not ehter into the picture - It has no need to, since the terms are defined exclusively in respect to their relations to other terms In the calculus; and a calculus tells one how terms behave, not "how they are made up" — it would be meaningless to ask how they are made up, since they have no make-up. (It is a notatlonal fact, and not an arithmetical fact, that the sign "plus" Is made up of two strokes; yet if one were (absurdly) to regard arithmetic as a "language", the composition of the sign would be just as interesting as Its behaviour.) Yet since matthem- atics serves as a model of the "formal", is It not natural to take distribution as eminently formal".”. I believe that this that lied behind a lot of stuff in Harris and such peopleo xhey have the algebraic model In front of them. It is a ridiculous model to take for a science like linguistics which begins with complexes and seeks to break them upo For the purpose of my paper, differences in substance were identified with differences in medium, cut other differences could equally well be treated in the same way. Bor instance "paradigmatics and "syntagmatics may be treated as two substances having the same form, ^very relation in the one has a logi- cally equivalent relation in the other. It is for this reason that a contrast is profitable, although there is no immediate logical opposition, i am here using "synfea]gmatic in its commonest sense (adopted in my booklet). On the other hand the genuine logical opposition between rules of substitution and rules of combination aannot be treated in this way, for they are not isomor- phous. it Is a tautology to say that a unit may be substituted for Itself, but not a tautology to say that it can combine with itself (as in congruence)« Whereas in syntagmatics (the relations of units in combination-- not the rules for their combination) there Is the corresponding tautology; namely that the unit occupies the same segment as itself.
at all
c4/h