Zambezia (2000), XXVII (i).MORPHO(?)PHONO(?)IOG/C4I FUZZY EDGES: THECASE OF {-/I/-}/{-/U/-} SEMANTICO CONTRAST INSHONAK. G. MKANGANWIDepartment of Linguistics, University of ZimbabweAbstractAlthough the Western world of science continues to be dominated by theblack-and-white (or 'binary') laws of logic, the binary faith has always faceddoubt. (Unfortunately, logical positivism has remained the workingphilosophy of linguistic science. This article argues that Linguistics needs toconsider going 'fuzzy' in its logic to handle the fuzzy edges of languages likeShona. The argument of the article (which is an observation in search of a'fuzzy' theory) takes the form of a brief consideration of {-/i/-}/{-/u/-} as anexample of linguistic phenomena which clearly defy the application ofphonological, morphological, or semantic theories as autonomouscomponents.To illustrate a point I have not yet made, let me quote a dialogue betweenRichard Lower (British English) and Marco da Cola (Venetian Italian), inAn Instance of the Fingerpost, an historical mystery story (set in 1663) byIain Pears (London, Jonathan Cape, 1997). The exchanges capture what Ihave always seen as the philosophical difference in scientific methodbetween British (European?) and American (Descriptivist/ Chomskyan)linguists.1] 'You have fallen under the influence of Monsieur Descartes, haveyou not?' he asked eventually.'Why do you say that?''You have constructed a theory, and that leads you to recommenda practice. You have no evidence that it would work. And, if I maysay so, your theory is confused.''Though not disputed even by yourself.''No. That is true.''Do you dispute my theory, though?''No.''And is there any way of finding out whether I am correct except bytesting it against result. That is surely the basis of experimentalphilosophy?''That is Monsieur Descartes's basis', he said, 'if I understand himcorrectly. To frame a hypothesis, and then amass evidence to see ifit is correct. The alternative, proposed by my Lord Bacon, is toamass evidence, and then to frame an explanation which takes intoaccount all that is known.'4748 THE CASE OF {-/l/-}/{-/U/-} SEMANTIC*?) CONTRAST IN SHONAGeorge Fortune (or 'my Lord Bacon') is very definitely NOT Cartesianin his 'scientific' study of Shona. Why do I involve Fortune here? Heindirectly inspires (or is it 'provokes'?) this article. His scientific approachseems to me to be very definitely Baconian in the sense that he sets outto 'amass evidence, and then to frame an explanation which takes intoaccount all that is known'. Unfortunately, he does not seem to have hadthe time to 'take into account all that is known' Š i.e. the Shona language.It is not only Fortune that is the problem. Logic, too, is grappling witha related problem. As long ago as the early stages of Jakobson's parametricapproaches to phonology, which I suspect lies at the root of Chomskyan(or Cartesian) linguistics (philosophically), when he (Jakobson) arguedfor a small group of parameters intrinsically fit to play a linguisticallydistinctive role, saying that despite surface appearances (i.e. phonetics),each of the parameters forms a fixed hierarchy of precedence. Jakobson'shypothesis was that all parameters are binary, supported by themathematical notion that a transmission code is more efficient when ituses only independent binary choices (cornerstone of computertechnology) Š psychologically/accoustically even if it is continuous inarticulatory terms. This was argued for at the time when Descriptivistsemphasized that languages differ unpredictably in the particular phoneticparameters (which are physically continuous) which they utiliseddistinctively. Let us say that the Jakobsonian, as well as mainstreamcomputer technology, position is philosophically governed by what wemay call 'binary logic'.1On the other hand, however, (back to my quote) 'an explanationwhich takes into account all that is known' inevitably runs into what thetitle of this'article calls the 'fuzzy edges' of things. Philosophically, we arepushed to invoke 'fuzzy logic'. (The way forward in linguistics seems tohave been Š and to continue Š to pretend that certain things are notknown).2 Why can't linguistics also adopt the philosophy which is now1 Historically, according to Bart Kosko (1994), this bivalence reaches back in the West to atleast the ancient Greeks, namely Democritus, Plato, and Aristotle whose black-and-whitelaws of logic scientists and mathematicians still use to describe and discuss the grayuniverse. Fortunately, the binary faith has always faced doubt. Philosophically, althoughevery philosopher one asks attacks logical positivism, it remains the working philosophyof modern science (including linguistics), medicine and engineering.2 Peter W. Cullicover (1997), on the methodological foundations of linguistic science,writes:The very richness of linguistic experience requires that we idealize, in thesense that we ignore certain facts at certain times in the interest ofarticulating the theory ... One strategy for idealization is to ignore certainfacts that do not naturally fit. Another, more refined, approach is todecompose phenomena into components such that it is (potentially) morefeasible to come up with simple theories of the parts than to cume up witha theory of the undifferentiated whole (p. 10).The classical components are phonology, morphology: syntax and semantics.K. G. MKANGANWI 49operating in the technology of those robots that are meant to work theway the Shona language (or mind) works Š fuzzy logic?That was the conclusion to my article. Now the reasons for theconclusion. Consider first the standard practice in linguistics (see Footnote2 above). In its approach to the 'fuzzy' nature of language (generallyknown as the immense complexity of language) linguistics has over theyears been generally binary, in the sense of dealing with particular datain the context of phonology or morphology or semantics. But when Shonacontributes something as fuzzy as the data we shall examine shortly,then the case for a fuzzy approach (or logic) strengthens.Now back to Fortune, and a slice of 'all that is (not?) known'. I repeatthat Fortune is certainly not Cartesian. His explanations tend to try andtake into account all that was known (by him). After throwing in all thebits and pieces he goes back and attempts to identify and label3 thepieces as being from the tail, foot or ear. Unfortunately, from this type ofeffort, we are not able to see exactly what type of animal it is Š cat, dogor human. His logic seems quite fuzzy in the non-technical sense. Theidentity of the animal is obscured by the other fact that his approach isalso 'binary' Š i.e. morphology or phonology. For example, in hisGrammatical Constructions4 Fortune would make the following observation,among numerous others, that Shona verb radicals can be extended in avariety of ways, one of which he would call the 'reversive'5 extension, asthe following examples illustrate:2] (a)(b)(c)Fortune's analysis here is quite obviously morphological. Thesegmentation into radical and extension is very clear, though, of course,/-ur-/, /-unur-/, and /-onor-/ are all treated as the same morpheme which-sim--sim-ur--sung--sung-unur-pomb--pomb-onor-(plant)(pull out)(tie)- (untie)(wind round)(unwind)3 The Procrustean 'curse'! In his review of Kosko (1994) in the Sunday Times, Gilbert Adairwrites:The best defense of fuzzy logic is that it works ... Its fundamentalbreakthrough is to have shown how it is possible to emancipate logicalthought from the Procrustean bed of the mathematical or scientific model4 I assume that the reader is familiar with Fortune's work.5 Fortune appears to have gone silent on the 'reversive' extension. I wonder why. I, however,vividly recall what he used to say when he taught us Shona grammatical constructionsmany years ago.50 THE CASE OF {-/l/-}/{-/U/-} SEMANTICS) CONTRAST IN SHONA'has the meaning' 'reversive'. The other complication we shall also not gointo is illustrated by the following example:3] -ching- (receive, meet, e.g. baby, word)-ching-ur- (receive, meet, e.g. visitor, someone arriving).There is a sense in which we may say there is no difference betweenthe 'meaning' of the simple radical and the extended one. This, however,may not be seen as being so 'fuzzy'. But in this context of the 'reversive'morphological analysis, Fortune makes one of the most curiousobservations ever seen in linguistics. He includes what seems to be aclear case of a semantic (sense) relationship between pairs of Shonawords under his clearly morphological analysis of the reversive extension.Consider the following pairs of examples:4] (a)00(c)There seems to be a clear 'semantic' sense relationship between theabove pairs of words, which is one of 'oppositeness' of meaning, or whatFortune prefers to call 'reversive'. What is curious about Fortune'sobservation is that he sees the (reversive) contrast or opposition betweenthe morphemes /-ur-/ and /-ir-/. Unfortunately, examples in 4](a) and4](b) cannot be morphologically segmented, and while 4](c) examplescan be segmented, /-ir-/ does not 'have meaning'. Without /-ir-/ in 4](c),the example is exactly the same type as those under 2] and means exactlythe same as it does with /-ir-/. If we exclude 4](c) from Fortune'sobservation, it seems clear that what we have is a 'semantic' issue ratherthan a morphological one, a sense relationship of oppositeness of meaningbetween pairs of lexical items.The matter is, however, not quite as simple as my comments so farwould suggest. The pairs concerned display a 'minimal' phonologicaldifference Š that between /i/ and /u/. With a little imagination, moreexamples can be lined up to reinforce the semantico-phonological featureof pairs of Shona words, e.g.5] (a) -dur- (confess, pour out)-dir- (pour in)(b) -nunur- (separate)-ninir- (beckon)-zarir--zarur--warir--warur--nam-(ir-)-nam-ur-(close)(open)(spread, e.g. sleeping mat)(fold up, remove, e.g. sleeping mat)(stick on, seal)(unseal)K. G. MKANGANWI 51The following, which, some analysts may argue, are segmentablemorphologically,6 still display the same semantico-phonological featuresillustrated by 5] above:6] (a)(b)(c)(d)(e)-pakur--pakir--turur--turir--tungur--tungir--dzimur--dzimir--kungur-(share out, dish out)(put into, stuff into container)(take down)(place on top, hand over)(make a ritual offering to bring out)(stick in, thread)(extinguish, control, moderate)(lose sense of direction)(regret)-kungir- (prepare headpad Š to carry something away)Linguistic meaning is a relation between words and expressions, withany degree of phonological difference. For example:(a) Oppositeness of meaning:7] (a)(b)(c)-vat--muk--sek--chem-babamai'b) Sameness of8] (a)(b)(c)-w--punzik--dhan--tsveruk-godopfupa(sleep)(wake up)(laugh)(cry)(father)(mother)meaning:(fall)(fall(be shy)(be shy)(bone)(bone)Finally, there seems to be more than a simple phonological contrastbetween /i/ and /u/ as Shona vowels. When the contrast is simplyphonological the difference in meaning is arbitrary and/or unpredictable,7for example:9] (a)(b)(c)-sik--suk--pir--pur--kiy--kuy-(create, twirl)(wash, clean)(make a ritual offering)(thresh)Gock)(grind)6 Such segmentations seem to be motivated by (non?) proportional analogy.7 Looking at phonology as an autonomous component (Cullicover, 1997, 11), this is whatsimple phonological contrast is all about.52 THE CASE OF {-/l/-}/{-/U/-} SEMANTICS) CONTRAST IN SHONAThe phonemes /i/ and /u/, though meaningful, do not themselves'have meaning'. In other words, they are not morphemes. When, however,they differ in a predictable way, as illustrated by the examples givenearlier, then we have a problem. A further complication lies in the factthat although, with a little imagination, dozens of such minimal pairsdiffering in a non-arbitrary/predictable way can be cited, they are still avery limited, small set of verbs. To further complicate matters, although/-ur-/ and /-ir-/ can be shown to be morphological extensions (but not inall cases), extensions do not normally have meaning relationships witheach other as /-ur-/ and /-ir-/ seem to do as distinct morphemes as well asin cases where they are not segmentable as separate extentions. (When/-ir-/ and /-ur-/ are segmentable as separate extensions, the minimaldifference can only be seen between the vowels /i/ and /u/, as in examples4](a) and 5]). Besides, derivational meanings of extensions are in relationto verb radicals and not to each other. It seems clear, therefore, thatFortune applied a 'binary' solution to a 'fuzzy' situation. He is, however,correct in his observation that there is a 'reversive' relation. The questionis whether this so-called reversive relation is phonological, ormorphological, or semantic.I could go on and on making further observations. The point seemsclear, that some things defy 'binary' logic, because they are fuzzy. 'MyLord Bacon' (i.e. George Fortune), after amassing a lot of evidence,attempted to frame an explanation (or is it a mere description?) whichdid not take into account all that is known (i.e. the fuzzy edges), while'Monsieur Descartes' (i.e. Noam Chomsky, etc.) seems unable to constructa theory which can handle the fuzzy edges. I cannot frame an hypothesisto which the examples cited in this article can be related. Can linguisticsgo sufficiently 'fuzzy' in its logic to handle the fuzzy edges of languages?Does it need to do that? How, then, should one approach this problem? Italso seems clear that as long as the world of linguistic science continuesto sheepishly follow the traditional practice of idealisation8 (whateverthe methodological justification and convenience), it remains hard put toaccount for the real world illustrated by the fuzzy Shona mind and/orlanguage. Corpus linguistics and optimality look better placed to breakwith the scientific/theoretical tradition that continues to haunt alllinguistics paradigms.8 The question being asked here is: Can the issue of the fuzzy edges have morphological, orphonological, or semantic, solutions? No! Whichever way one tries to go, something getsin the way each time. Perhaps it is time we tried fuzzy logic, fumbling along the best wecan, since we are much better than Deep Blue.K. G. MKANGANWI 53ReferencesCULLICOVER, PETER W. (1997) Principles and Parameters: An Introductionto Syntactic Theory (Oxford University Press, New York).FORTUNE, GEORGE (1984) Shona Grammatical Constructions (MercuryPress, Harare).HARRIS, RANDY ALLEN (1993) Linguistics Wars (Oxford University Press,New York).HURFORD, JAMES R. and HEASLEY, BRENDON (1983) Semantics: ACoursework (Cambridge University Press, Cambridge.)KOSKO, BART (1994) Fuzzy Thinking (Flamingo, London).PEARS, IAIN (1997) An Instance of the Fingerpost (Jonathan Cape, London).SAMPSON, GEOFFREY (1980) Schools of Linguistics: Competition andEvolution (Hutchinson, London).