1989 Markedness theory: the first 150 years

36
Markedness theory - the first 150years Henning Andersen 0. Introduction \larkedness has been hailed by some asoneof themost significant conceptual .tdvances of twentieth-century linguistics. andit has been denounced by others .i\ un entirely vacuous concept. When linguists of recognized standing hol<J such cliametrically opposecl r iews. it should give pause to thought. Onemight envisage thepossibility that thet'have formed their divergent views due to differences in metatheoretical premisses. If thisis lhe case. it is important to expose these deeper differences. There is another possibility as well,namely that theirdiametrically opposed r iews areonly apparently opposed. This occasionally happens whena term is understood differentty by different practitioners, and the notion espoused by trne is in fact different. from the one rejected by the other.Obviously, when a termgives rise to such misunderstandings. it is worthwhile making its different :eilSeS explicit. It seems that the understanding of the notion markedness hasindeed been changing as it was being transplanted fiom one metatheoretical paradigm to .tnother. As a consequence, the term hasbecome polysemous in sucha way that it no longer clearly distinguishes the phenomenon for whichit wascoined liom other phenomena. Thisbeing so. it wouldbeuseful to take a close look at the various phenomena for which the term hasbeen used, to try to establish the relations amongthese, and to determine, if possible, to what extent it is reasonable to continue usingthe term for the diversity of phenomena that currently come within its extension. This is the topic I will broach. Since thecomplex of phenomena to which the term "markedness" hasbeen applied has changedover the years,it would be interesting to trace the historical developmentthrough which the current polysemy has arisen. Although I havechosen a title which may seem to promise such a historical survey, in fact I will limit myselfto sampling someof the influentialworks * The editor of the volume this paper appeared in doctoredthe exposition in numerous places, in some of them. disturbing, distorting or destroying the sense intended. And then I never saw proofs. In the text below, a small number of the editor's oemendations'have been corrected so asto makesense.

Transcript of 1989 Markedness theory: the first 150 years

Markedness theory - the first 150 years

Henning Andersen

0. Introduction

\ larkedness has been hai led by some as one of the most s ign i f icant conceptual. tdvances of twent ie th-century l inguis t ics. and i t has been denounced by others. i \ un ent i re ly vacuous concept .

When l inguists of recognized standing hol<J such cl iametrical ly opposeclr iews. i t should g ive pause to thought . One might envisage the possib i l i ty thatthet ' have formed their divergent views due to dif ferences in metatheoreticalpremisses. I f this is lhe case. i t is important to expose these deeper dif ferences.There is another possib i l i ty as wel l , namely that the i r d iametr ica l ly opposedr iews are only apparently opposed. This occasional ly happens when a term isunderstood dif ferentty by dif ferent practi t ioners, and the notion espoused bytrne is in fact dif ferent. from the one rejected by the other. Obviously, when aterm gives r ise to such misunderstandings. i t is worthwhile making i ts dif ferent:ei lSeS expl ici t .

I t seems that the understanding of the notion markedness has indeed beenchanging as i t was being transplanted f iom one metatheoretical paradigm to.tnother. As a consequence, the term has become polysemous in such a waythat i t no longer clearly dist inguishes the phenomenon for which i t was coinedl iom other phenomena. This being so. i t would be useful to take a close look atthe various phenomena for which the term has been used, to try to establ ishthe relat ions among these, and to determine, i f possible, to what extent i t isreasonable to continue using the term for the diversity of phenomena thatcurrently come within i ts extension.

This is the topic I wi l l broach.Since the complex of phenomena to which the term "markedness" has been

applied has changed over the years, it would be interesting to trace thehistorical development through which the current polysemy has arisen.Although I have chosen a title which may seem to promise such a historicalsurvey, in fact I will l imit myself to sampling some of the influential works

* The editor of the volume this paper appeared in doctored theexposition in numerous places, in some of them. disturbing,distorting or destroying the sense intended. And then I never sawproofs. In the text below, a small number of the editor'soemendations'have been corrected so as to make sense.

12 Henning Andersen

which have shaped this development. I will restrict myself to the early history

of the notion "markedness", the period from its f i rst known del ineation in

1815 to the f irst substantial work on markedness in American l inguist ics'

Greenberg's monograPh from 1966.

1. Marked and unmarked terms

For an init ial clari f icat ion of markedness one might turn to a standard work

such as Lyons lgl l .which contains a section specif ical ly devoted to "Marked

and unmarked terms".Lyons (1977' .305) recognizes "mark ing (or markedness)" as an ext remely

important concept . which "unfor tunate ly , however , . . . covers a number of

disparate and independent phenomena." But his presentation makes i t clear

that this characterization is not apt. In fact, Lyons describes not one but three

dist inct concepts. So, i t would be more appropriate to say that the term

"marking" (or "markedness") is used for a number of disparate and

independent notions. This is, at least in the main, the way Lyons presents

them. His exposit ion does suggest, however, that the three phenomena are not

in fact "disparate and independent". But this is a point to which we will return

only in section 4.7.

1.1 Formal marking

The first of Lyon's concepts. which he terms "formal marking" is exemplified

by such word pairs as host ' . fuostess, count: countess, l ion: l ioness,fr iendly:

un/rienclll ', c'onsistent'. inconsistent, respectJul:,disrespect/ul' The forms of the

second member of each pair evidently contain$an affix which the forms of the

first member lack. "Formal marking", then, consists in a syntagmatic

modification of a morpheme by *.uni of another (bound) morpheme; the

latter is not necessarily to be described as negative (cf'.Lyons 1977 306)'

Interestingly, Lyons Supports this concept of "formal marking" with a

quotation from Trubetzkoy (1958: 67): "One member of the pair of opposites

is characterized by the presence and the other by the absence of a mark

[Merkmal]". Trubetzkoy's statement, however, has nothing to do with

syntagmatic modification. It is taken from a discussion of correlative

l .

Markedne.ss lheort the first 150 vears 13

phonemes, in which, say, the voicing in one obstruent is signif icantly opposedto voicelessness in the other. It has to do, in other words. with paradigmaticopposit ions of dist inct phonological signs, and not with the syntagmaticmodif icat ion of the sense of one sign by the addit ion (aff ixat ion) of another.The quotation from Trubetzkoy would be better placed in a discussion of thethird concept of markedness, to which we turn in section 1.3.

In discussions of "marking (or markedness)", there is another commonusage of the terms "marked" and "unmarked". which Lyons does notmention. I t di f fers from the one exempli f ied above. but might well be termed"formalmark ing" too. I t may be said. for instance. that the p lura l is marked inthe second member of the pairs cat : cots. dng '. dog,s, horse '. horses, ox '. oxen,but unmarked in deer '. deer. sheep : .sheep: or that the genitive plural is markedin Russian byk : bykov 'bul l : (o l ) bu l ls ' , kon ' : konej 'horse : (o l ) horses ' , butunmarked in so ldat : so ldat 'so ld ier : (o0 sold iers ' . In such instances themarking does not consist in the addit ion of a separate modifying morpheme.absent in the unmarked forms. but rMFFcrl€,e_elf a +eal(segmentallyexpressed)alle+norph of one and the same modifying morpheme.

The terms "marked", "unmarked", "marking" seem inescapable for thesedist inct phenomena since they involve the presence of a (morphemic oral lomorphic) mark or marker, but i t would be practical to keep themterminologically distinct from markedness. This is especially so since "mark-ing" as described here is signif icantly related to markedness (cf. sec. 4.7). Therelat ionship between these two dist inct phenomena is dif f icult to clari fy i f onedoes not dispose of dist inct terms for them.

1.2 Distributional marking

The second concept Lyons describes is "distr ibutional marking", which herelates to "formal marking", at the same t ime noting that i t is independent ofi t : "The formally marked member of the opposit ion tends to be morerestr icted in i ts distr ibution ( i . e. in the range of contexts in which i t occurs)than the formally unmarked member. .. . But there are many formally markedlexemes that are not distr ibutional ly marked. And there are manydistr ibutional ly marked lexemes that are not formally marked .. ." (306f.). Atthe same t ime, "distr ibutional marking" "correlates with, and in many casescan be plausibly explained as being determined by" "semantic marking"(307), the third concept of "marking (or markedness)", to be discussed in thenext section.

1: in the occurrence of a real (segmentally expressed) rather than a

zero allomorPh

14 Henning Ander.sen

I f i t is the case that dif ferences in distr ibution within such pairs as happv ' .

unhapp.v or high: low are independent of their "formal marking". there is nopoint in speaking of these dif ferences as a kind of marking, and i t would bemore appropriate to cal l them what they are. "dif ferences in distr ibution". I t isprobably in the nature of such dif ferences that they range from theinsignificant (e. g.. prin<:e '. prince,ss) to the significan t (e . g.. lion : lioness). I f thesignif icant dif ferences are in fact determined or explained by anotherphenomenon, "semantic marking", i t is to this phenomenon they should berelated and not to "formal marking". of which they are conceded to beindependent.

Even though Lyons 'exposi t ion presents the three "d isparate and independ-ent" phenomena of "marking (or markedness)" as coordinate, i t becomesclear that two of them. " formal mark ing" and - 'd is t r ibut ional mark ing" are to

be understood as subordinate in relat ion to the third.

1.3 Semantic marking

Lyons' third concept, "semantic marking" captures the difference between

such pairs of lexemes as lion : lione,s,s, dog : bitch, in which the second,"semantical ly marked" member of each pair " is more specif ic in sense than the

corresponding semantical ly unmarked lexeme" (307). This pecul iari ty has an

interesting consequence for the notion of hyponymy. Normally i t is assumed

that the relat ion of hyponymy is irref lexive: Everl ' rose is a.f lov'er is true. but

Eyer\,.flov,er is a ro,se is false. But since in such pairs as dog '. bitch, "dog is

sometimes in contrast with bit t 'h and sometimes superordinate to i t . . . i t

follows that in certain circumstances r/og can be a hyponym of itself. Is that

dog a clog or a bitch is a meaningful, though perhaps rather odd sentence"(308). One might add that the superordinate term can be used freely to refer to

either of i ts hyponyms, as when we speak of one dog having puppies and of

another dog having sired them.This concept of "semantic marking" has been called markednes.s since the

term was coined, and this is the concept for which the term will be used in this

paper. We wil l return to Lyons' two other concepts in section 4.7.

The discovery of the peculiar semantic relations in which one of the terms

(e.S., r log 'male dog') is opposed to the other (e.g., bitch ' female dog') and at

the same t ime subsumes i t (e. g., dog'dog, regardless of sex') is an interesting

event in the history of linguistics, which deserves a few lines.

r-G.h._._

l\larkedness the'rsry - rhe fir.:t 150 year.s 15

2. From Roth to Hjelmslev

The notion of markedness has a history which goes back much further thanthe term "markedness".

Hjelmslev found i t adumbrated in G. M. Roth's discussion of the semanticrelations of the cases in his Grundriss der allgemeinen Sprachlehre (Frankfurt1815) ( l Hjelmslev 1935: 28, 100f.). Roth had noted that "no case expressespure independence. Independence is only relat ive. that is. i t is independenceand dependence at the same t ime. This complex notion is expressed by thenominative, whereas the other cases express pure dependence" (cf. Hjelmslev1935: 28) . Roth had d iscovered. in other words. that the opposi t ions that arefound within a grammatical category cannot be described simply as con-tradictory opposites. They are not opposit ions of a posit ive and a negativeterm, but - as i t appeared to him - of a simple term and a complex term whichincludes the meaning of the simple one. But Roth's observation was l imited toa single category. the category of case. and he did not attempt to general ize i t .

Jakobson recognized an awareness of semantic relat ions of this kind in theworks of the Russian grammar ians A. Vostokov (1831) , N.P. Nekrasov(1865) , K. Aksakov (1875) , F. For tunatov (1899) , A.A. Saxmatov (1927) .A. M. Pe5kovsku (1914, 1928) , and S. Karcevski j (1921) (1971b: 5) .

Togeby has pointed out that, independently of this Russian tradit ion, thenotion of markedness was defined and applied to a wide range of examples inthe works of the Romanist Theodor Kalepky (1862-1932) from the 1890s on(Togeby 1965: 74tr. \ . Here is Kalepky's characterization of what he saw as aspecial kind of contradictory relat ion typical of language:

( 1) . . . due to the multitude and diversity of the conceptual complexes with whichhuman speech has to cope. there are, besides the many, easily delimitable andeasily categorizable ones, a significant number of cases - one might call themindifferent, intermediate, or neutral - in which it would be diffrcult even forthe most penetrating and trained mind to perform an objectively motivatedcategorization ... Un such cases] the meaning, the sense of one expression isclear, precise. and salient, that of the other full of nuances, shadings,gradings. ... But as long as the meaning of one expression is precise, andtherefore salient, and the other is contradictory to it, that is, negativelydefined, then everything that does not clearly and perceptibly have thefeatures of the former is automatically assigned to the latter, and all difficultyis resolved (1901: 339-40) .

As the passage shows, Kalepky saw grammatical oppositions primarily asrelations between a narrowly or precisely defined term and a contradictory,

16 Henning Andersen

but vaguely defined one. In his application of this understanding to suchcategories as mood (indicative vs. subjunctive), and aspect (passe simple vs.imparfait) he made i t clear that the sense of the vague term could include thatof the more precise term, whereas in other examples he specif ied the condit ionsof use for the marked term, assigning the unmarked term to "elsewhereenvironments" (c/ Togeby 1965: 75).

2.1 Zero meaning

Just prior to the coining of the markedness terminology, Saussure's observ-at ion that " language can be content with opposing something to nothing"(1916: 124) st imulated a number of structural ly oriented l inguists to look forphenomena in language which f i t this descript ion. Jakobson, who hadtouched on the notion of zero meaning in 1921, developed this idea in twopapers in the late 1930s (1971a:217 ff . , 220ff.) , but already in 1928,Peskovski j 's university textbook of Russian syntax integrated the notion ofl inguist ic zeros with the tradit ional Russian understanding of grammaticalopposit ions. Pe5kovski j ends a discussion of the degrees of comparison thiswav:

If we compare the meanings of the positive, comparative and superlativedegrees, we note that a specific meaning of comparison and of degree ofcomparison is found, essentially, only in the comparative and superlativedegrees. When we say beautiful or intell igent ... we are not thinking thatanother object is more beautiful, another person more intell igent, and so on.There is, as it were, no specific meaning present of the degree of the property.But by contrast with the comparative of the superlative, there is sti l l a kind ofzero meaning, that is, by contrast with the other categories. here too weexperience a special, zero category, which is then what we call the category of"the positive" degree. In other words, the very absence of a meaning creates asort of meaning. and this occurs for the very same reason that the absence offormal elements creates .sui generis formal elements in zero forms. ... Ourlanguage is full of such zero categories ...: the indicative mood, for instance,is essentially a "zero" mood; the imperfective aspect, essentially a "zera"aspect: the non-reflexive voice. essentially a "Zero" voice (in the last twoexamples this is apparent in the very terms, which contain negative prefixes,im-, non-) and so forth (1928 28f.).

Here we have in nuce the gist of Jakobson's later papers on "Signe zbro"

and "Das Nullzeichen", including the conceptual and terminological confl-ation of zero allomorphs or absent signs with the unmarked terms of

(2\

.b--

\,lurkedne.ss theort' - the ftrst 150 l,ears 1'l

grammatical oppositions - under the label "zero" - which we noted in section1.1 under the label "marking (or markedness)".

Pe5kovskij's notion of "zero categories" is at first blush diffrcult to reconcilewith Kalepky's understanding that the sense of these unmarked categories iswider than the ones to which they are opposed. In fact Peskovski j 's statementsare difficult to distinguish from the unreasonable claim that the unmarkedcategories have no meaning at al l , or that the posit ive has no meaning ofdegree, the indicative no meaning of mood, the imperfective no meaning ofaspect, and so on, for the distinction between "no specific meaning" and "akind of zero meaning" is not clearly drawn.

PeSkovskij's formulations were severely criticized on precisely thesegrounds by Hjelmslev', w'ho. in an imporfant paper (written in 1933, referredto in Togeby 1965:72, but not publ ished t i l l 1973) analysed the theories ofgrammatical categories of PeSkovskij 1928, Karcevskij 1921, and Jakobson1932.

But i t is not impossible to salvage the intuit ion behind the notion of "zerocategory" and reconcile it with Kalepky's and a more modern understanding.One might relate i t to the tradit ional dist inct ion between intension (semanticdepth) and extension (semantic breadth). In these terms, the perfect ive aspect,for instance, might be said to have a certain,, definite depth, and theimperfective zero depth - in Saussure's terms, something would be opposed tonothing. Since depth and breadth are inversely proport ional, the perfect ivewill have a certain, definite breadth, whereas the breadth of the imperfectivewil l be without l imit. By virtue of their zero intension, in other words, thereference potential of the unmarked 1"zero") categories will include that oftheir marked opposites. However, since both terms of a grammaticalopposition are meaningful, it is necessary to distinguish between meaning andreference potential.

2.2 Meaning and the values of linguistic categories

This conclusion was reached by a different route by Hjelmslev in the papermentioned above.

Hjelmslev pointed out that since similar semantic oppositions may presentdifferent markedness relations in different languages, descriptions mustdistinguish between the meaning (Fr. srgn ification) and what he called thevalues (Fr. valeurs) of linguistic categories. To avoid the misunderstandings towhich the Prague Circle's "markedness" terminology could give rise,

1 8 Hennin.q Andersen

Hjelmslev proposed the terms "intensive" for the marked, and "extensive" lirrthe unmarked members of l inguist ic opposit ions.

Hjelmslev had adopted the notion of inclusive relat ions. under the name of"participative relations". from Levy-Bruhl (tf . 1928:257 ff.) and agreed withthe French anthropologist that these apparently alogical relat ions arefundamental to language to the extent that al l opposit ions in language "aresubject to the law of part icipation: there are no opposit ions between A andnon-.4, but only between A on one hand and A * non-l on the other. There isnothing surprising in this. since the studies of Levy-Bruhl have shown thatlanguage bears the impr int o f a pre- log ica l menta l i ty" (1935: 102) . To re-usethe example given above. i t is because the breadth of dog includes both 'bi tch'

and 'non-bitch' that dog can be used to refer to 'male dog'. 'female dog'. and'dog regardless of sex'.

In a later study Hjelmslev clari f ied the dif ference between the part icipativeopposit ions and the ones that have tradit ional ly been recognized in logic. Theopposit ions of contraries ( l vs. B) and contradictories ( l vs. non-.4) areexclusive relat ions in the sense that the reference potential of each termexcludes that of i ts opposite. Part icipation is an inclusive relat ion inasmuch asthe reference potential of one term includes that of the other (r/ . (3)).

But Hjelmslev went beyond this clari f icat ion and observed that "exclusionis merely a special case of part icipation, in which certain areas (Fr. t 'ases) ofthe extensive term are empty" (1939: 87). We wil l return to this point againbelow.

Exclusioncontrar ies contradictor ies

Inclusion

-F;-l)E

2.3 Types of semantic oppositions

This crucially important insight has not been duly acknowledged in modernstudies of semantics. Lyons, for instance, offers a classification of semanticrelat ions, dist inguishing four main types of opposit ionsi antonymy, com-

( 3 )

Murkedne.s.s theorr the tirst 150 I'ears 19

plementari ty, converseness and direct ional i ty. and cross-cutt ing dist inct ionssuch as equipol lent vs. privative opposit ions (1917:270ff.) . And he observesthat "one can dist inguish a posit ive and a negative member in such contrastingpairs as man : women, parent : child, north : south, lteoven : earth,./ood '. drink, buy . sell,etc." (276). But this observation is not integrated withhis discussion of markedness, which fol lows 30 pages later. His examplesshow, in point of fact, that al l the types of semantic relat ions he recognizes fal lunder the law of part icipation.

The ful l extent to which this is the case can rarely be documentedconvincingly with material from a single language, and the l inguist is ot ienl imited to possibly subjective or circular observations. One manifestat ion ofthis, ci ted by Lyons, is the f ixed order of elements in irreversible binomials(Malkiel 1961), which is apparen+lya*ways "unmarked before marked" (as in Ethe examples immediately above), but where one cannot be sure whether thedist inct ion between the posit ive and the negative member, which is sensed bythe native speaker, is the cause of the f ixed order or results from it .

But there are languages on record which display an extensive organizationof lexical and grammatical categories in terms of markedness and therebyconflrm the claim that in language. inclusive relat ions are the superordinatetype to which the diverse exclusive semantic relat ions are subordinate. Onestr ik ing example is Rot inese, whose ent i re vocabulary is organized in to lex ica lpairs. employed in paral lel l ines in certain tradit ional speech genres, andintegrated with certain r i tuals (cf Fox).

Another example. which I wi l lci te in(4), is the etymological ly mixed lexiconof the Spanish-based creole spoken in Zamboanga in the Phil ippines, where apresumable, earl ier styl ist ic variat ion between Spanish and native lexical andgrammatical morphemes has yielded a language state in which original lySpanish and Phil ippine lexemes have been correlated with respectively theunmarked and the marked terms of opposit ions across a number of semanticcategories so that "where a Phil ippine and a Spanish-derived form part icipatein a marked-unmarked relation in the same contrast set, the Philippine formwil l designate the marked category: i t wi l l signify lesser magnitude, shorterdistance, worse evaluation. female sex, junior generation, or plural i ty" (Frake1971:231). Note that the examples include contraries (e.g., 'sweet' : 'bi t ter ')

and contradictories (e.g., 'straight ' : 'bent ') as well as converses (e.g.,'grandparent ' : 'grandchi ld ' ) .

2: if not alu a\ s. at least predominantly

20

(4)

Henning Andersen

a. Zamboanguefro adjectives of polarity, activity, potency, and evaluation.

grande : dyutaylalto '. pandaklihdro : mahinaykorre '. pata?mapv'erso'. maluyo?arde '. ?amantalun?apretao : haluga??agudu'. mapurulliso : makasapsabroso : mata?bandillse : mapa'litmaduru : mihilaulnonso : ma'lilapbunito : ?umalinlimpyo : bulinklaro : lubugdertt'o : tiku?balv|nte : mahuva??umilde : hambukbyeho : bata'/nv,lbo : da?an

'large : small'' tall : short'' fast : slow''fast : slow''strong : weak''bright : dim''t ight : loose''sharp : dul l ''smooth : rough''tasty : tasteless''sweet : bitter''r ipe : raw'' tame : wi ld ''pretty : ugly''clean : dirty''clear : turbid''straight : bent''bold : shy''modest : vain''o ld : young''new : old'

sij' : r

b. Zamboanguefro nouns contrasting in generation, age, or sex:

lolola : ?apu grandparent : grandchildtatalnana : ?anak father/mother : son, daughter'?ohas : talbu.s mature leaf : young leafplores : putut blossom : budsolt1ro : daldga bachelor : unmarried girl

c. Zamboanguefio neutral pronouns, singular and plural:

i": t ' tt t t l rt ,g. ) 'o kani kita

3rd ?dle sila

d. Zamboanguefro 2nd person pronouns, degrees of respect:

p L - polite ?uste, ?usttdesneutral tu. bosotrosfamiliar ?ebos. kamo

;.1

E-

Markednes.s theorv' the first 150 vears 21

3. Tiubetzkoy and Jakobson

The terms marked and unmarked were apparently f irst used in discussionsbetween Trubetzkoy and Jakobson in 1930 (Jakobson l9 l9:90f. , Trubetzkoy1 9 8 5 : 1 6 2 f . ) .

Trubetzkoy had noted that relations between correlative phonemes weremostly conceptually asymmetrical, apparently independently of whether theirreal izat ions in sound were contradictor ies or contrar ies, and he proposed thatthis asymmetry be reflected in the standard terminology that was being de-veloped. I am citing the relevant passage in my own translation, which differsin a few details from Jakobson's. which has been cited by other scholars:

(5) . . . Among fphonetic variants] one can actually consider :L bi.ri€-ene which B "occurs most frequently (or in the position of maximal differentiation). Withrespect to correlative phonemes the situation is different, here statistics arebeside the point . what matters is, so to say, the " ideat ional content" IRuss.ide jnoe soderianie; Jakobson confusingly translates this "intrinsic content";cf .1979:90] of the correlation. Apparently every (or perhaps not every'l)phonological correlation acquires in l inguistic consciousness the form of anopposition between the presence of some feature and its absence (or betweenthe maximum and the minimum of some feature). Thus one of the membersof the correlation invariably is "positive", "active", and the other "nega-tive", "passive". At least if the opposition is binary [Jakobson's translationomits this sentence]. Thus. for instance, the correlative differences in timbreamong the consonants are objectively reduced to the opposition "maximallyhigh vs. maximally low timbre", but subjectively they are always transformedinto either an opposition "raised vs. non-raised timbre" (: "maximally highvs. minimally high") or an opposition "non-lowered vs. lowered timbre"(: "minimally vs. maximally low"): the former type is realized in thecorrelation "softness vs. hardness", the latter, for instance, in the NorthCaucasian correlation "unlabialized vs. labialized consonant" (probablyalso in the Semitic correlation "unemphatic vs. emphatic consonant"). Ineach instance only one of the members of the correlation is conceived of asactively modified, as positively endowed with a certain feature, whereas theother is felt as not cairying this feature, as passively unmedified. [Russ. kak 4 ,"passivno neizmennly; Jakobson translates this "as passively modified" andends his quotation here]. The conception of the two opposite members asequally active, equally deviating in opposite directions from a norm, ispossible only where there is additionally a third member which is absolutelypassive or "neutral" - i.e. only when the correlation is no more binary, butternary. This is the case, for instance, in Abkhaz where in a whole series ofarchiphonemes (the sibilant spirant, the hushing sibilants, the pre-dorsals,the post-dorsals), three kinds of phonemes are distinguished, soft (palata-lized), labialized, and neutral. It is interesting that in Abkhaz, besides thesethree-timbre archiphonemes, there are also two-timbre archiphonemes, but

3 : tha t one bas ic

4: invariant

( 6 )

22 Henning Andersen

in these there is an opposition not of'solt and labialized, but either of soft andnon-soft {v iz.3 i i vs. j c g) or of ' labial ized and non- labial ized (v iz. h vsh . d" t ' t vs . d r l ) .' f

hus I propose that we speak not of "basic" and "secondary" variants ofan archiphoneme, but of active and passive or of positive and negative marks[Russ. priznak 'feature, mark'] of correlations. In any case the terminologymust somehow be revised on this point (162tr.).

Jakobson responded by general iz ing the observed conceptual asymmetr iesto value systems far beyond the l imits of grammar (c/'. 6).

... I am becoming ever more convinced that your idea that a correlation isalways a relation between a marked IRuss. priznakov.t ' l 'bearing or distingu-ished by a feature or mark'] and unmarkcd [Russ. bezpriznakov-1.7'lacking afeature or mark'] series is one of your most remarkable and fruitful ideas. Ith ink i t wi l l turn out to be important not only in l inguist ics, but a lso inanthropology and cultural history, and that such historico-cultural corrcl-ations as l ife and death, l iberty and bondage, sin and virtue, holiday andworkday, and the l ike are always reducible to a vs. non-u relations, and thatit is important to establish fbr each period, group, nation, etc. which is themarked series. For instance. for Majakovskij l i fe was the marked series.realizcd only when motivated, for him not death, but l i fe demanded amotivation. Cornpare the attitudes to l ife and death of Tolstoj 's master andworker. Anothcr example: the Chekists said that everyone is a WhitcGuardsman, and i f not , i t must be proven in each indiv idualcase. Here Sovietallegiance is the marked series. In the current Soviet press you find thefollowing idca expressed. "We used to say that everyone who is not against usis with us, now we say that everyone who is not with us is against us". Thismeans that there has been a shift in the series, i. e. a generclization of thcC--hekist point of view. I am sure that many anthropological phenomena, ofworld view and the l ike. which at f irst glance look identical, often differprecisely by the fact that what in one system is the marked series in the othersystem is evaluated as the absence of a mark (162f .).

When one reads these passages with the distinction between meaning andvalue in mind which Hjelmslev introduced (cf. above, section 2.3). one can seehow close Trubetzkoy's and Jakobson's informal understanding of marked-ness was to Hjelmslev's at this time. Trubetzkoy distinguishes clearly betweenthe substantive sound propert ies in which phonological opposit ions arereal ized and their " ideational content". Jakobson correspondingly dist ingu-ishes between the experiential dimensions, such as life vs. death and allegiancevs. non-al legiance, and the values imposed on them by the individualperiod orsociety.

But two things disturb this general impression. One is Jakobson's formula"a vs. non-a", which expresses a contradictory (exclusive), that is, a logicallysymmetrical relat ion; this is a point Jakobson tr ied to amend in.his later

Mqrkednes's theorr - the first 150 rear.s 23

writ ings on grammatical categories. but without success (see section 3.1). Theother is the terms mark, markecl, unntarkerl, which already betweenTrubetzkoy's f irst usage ("active and passive marks") and Jakobson'sresponse ("marked and unmarked") underwent a subtle change. whichbecame I'ateful for Trubetzkoy's theory of phonological oppositions (c/.sect ion 3.2) .

3.1 Merkmalhattig and merkmallos

Jakobson's f irst publ ished application of the concept of markedness was in his"Zur Struktur des russischen Verbums" ( 1932). Here he began by contrastingthe tradit ionalcontrary and contradictory relat ions with the newly recognizedtype, using the German counterparts nterkmalhaltig and nterkmallo,; of theoriginal Russian terms for "marked" and "unmarked":

( l) As hc examines two opposed morphological categories, the l inguist of ienassumes that these categories are equals, and that each of them lias its ownposit ive meaning: ci t tegory I dcnotes A, category II denotes B. Or, at least. Idenotes A and II denotes the absenc. or negit ion of A. ln real i ty the generalmeanings of correlative categories stand in a different relation:if catigory Iindicates the presence of A. category II does not indicate the presence of A.that is. it does not state whether A is present or not. In comparison with themarked category I. thc general meaning of the unmarkei category II isl imited to the absence of . .A-signal ing. '

i tqZt b: f )In the conclusion to the paper, Jakobson first seems to make it clear that the

essence of the difl 'erence between the two terms of such an opposition residesin their different reference potential (197lb: 14), but then tries to incorporatethis dif ference in value into the dehnit ion of their meaning and does notsucceed in clarifying their relation.

The asymmetry of these relations, he says, consists in this: category I signalsA, whereas category I I either (a) does not signal A or (b) signals non-A(1971b: 15). Whether a token of category II is to be interpreted one way or theother is said to depend on the context. specifically, sense (b) can beunderstood, in modern terms, as an implicature.

But this characterization is deeply unsatisfactory in at least two respects.For one thing, it gives no account of what some of the tokens of categbry II

do signal , vrz. the (a) tokens, which are described merely as not signaling A, anegative characteristic they evidently share with all other unmarked terms,without however being synonymous with these. This is the same difficulty,only in different terms, as the one we noted with peikovskij's ,'rrrocategories" (c/. section 2.2).

II-

24 Hennins Andersen

In later works Jakobson varied his characterization of marked and

unmarked terms - "s tatement of 'A" r , 's . "nt l s tatement of A" (971b: 136) ." indicates" vs. ' is non-committal" (148), "signal izuju5i i j clannyj priznak" vs'

" l i$ennyj podobnoj signal izaci i" (158). "more specif ied" vs. " less specif ied"

(184. 188), but he never arr ived at a satisfactory account of either the meaning

or the inclusive value of unmarked terms. The characterization of Jakobson

1971b:720,"\ack and presence of markedness" r:an be read as an indication

that Jakobson himself foun<J the notion essential ly intractable. The reason for

th is in t ractabi l i ty , however . is no mystery. Jakobson t r ied to reduce marked-

ness re lat ions to contradic tory re lat ions (s ignal ing vs. not s ignal ing and A vs.

non-A) . But an inc lus ive re lat ion is not thus reducib le.More important ly . there is no way in which Jakobson's character izat ion of

markedness can be applied tt ' r any other types of semantic opposit ion than the

ones exempli f ted in his analyses of grammaticalcategories, which he viewed as

a srr i generi,s variety of contradictories (cf. Jakobson 1 911b 136) or expl ici t ly

ca l led contradic tor ies (213) (antonyms. in Lyons ' terminology) , but which in

lact are inc lus ive.The tense opposit ion "present" : "past" is indeed paral lel to the erample

, lug ' , b i t t 'h . ment ioned above (sect ions 1.3,2.3) . But complementary oppo-

sit ions such as mule: lentule, converse relat ions such as p(lrent: t 'hi ld. and

directional relat ions such as up : clon'n cannot be construed thiswaYlfrom a

senrantic point of view: malc is never used indif ferently of males and f 'emales

the way' r/og is, purent cannot refer to parents and chi ldren al ike, nor is the

unmarke d up ever non-committal with respect to direct ion. These types of

opposit ion do document the general i ty of markedness, but at the same t ime

they show the necessity of dist inguishing between semantic relat ions and value

relat ions. They urge the importance of Hjelmslev's insight that the value

relat ions obtain independently of the semantic relat ions, and that the value

relations hold even where. for semantic reasons, a part of the reference

potential associated with the extensive (unmarked) term of an opposit ion

cannot be real ized.

3.2 Distinctive properties and binary opposition

In phonology, Trubetzkoy originally distinguished neatly between "mark"

(Russ. priznatk,Gm. Merkntal) and "dist inct ive property" (Russ. ot l i iatel 'noe

svo.jstvo, Gm. distinkrive Eigenscha.ft), as can be seen in the quotation above

(sect ion 3.1) .Jakobson has claimed that Trubetzkoy consistently maintained this

dist inct ion (1971a:736f.), but this is not quite correct. In 1931, Trubetzkoy

lvlurkcdne:'.s theor.v , the ftrst 150 t,ear,s 25

clearly used Merkrnal in the sense of 'dist inct ive property' (p.9i f . ; note alsothe use of marque in 1933). True. in Grunclzi)gehe pointed out that from aphonetic point of view, al l sound dif ferences are equipol lent and gradual. andalmost consistently l imited the use of Merkmctl (and i ts derivatives) to thedescript ion of phonological ly privative opposit ions. As a consequence.however, the terms Merkmal and (cli:;tinktive) Eigensc,hafr tend to occur incomplementary distr ibution, for instance in his account of the classif icat ion ofoppositions (Trubetzkoy 1958: 66ff.). And occasionally in Grundzi)ge he saysMerkmalwhere one would expect Eigen.sclta./t (e.g. ". . . wo ein aufhebbarerGegensatz tatsdchl ich aufgehoben ist. verl ieren die spezif ischen Merkmaleeines Opposit ionsgl iedes ihre phonologische Geltung", Trubetzkoy 1958: 70:Cant ineau t ranslates "marques speci f iques" . 1957: 81: Bal taxe has "speci f icmarks". 1969:78). I t is interesting to note that Trubetzkoy's conceptual andterminological innovations have been translated into Russian with theneologism Qrc)markirovann)'.i for Gm. merkmalhaft. merkmallo.r(Trubetzkoy's original priznakovy.fi and with priznak for Gm. Eigenscha./'t(Trubetzkoy's sl 'o7s tvo) (cf ' . Jakobson l9j l a: l-34tr.).

In any case, there is no doubt that the general, received understanding ofTrubetzkoy's concept of privative opposit ion is the one ref lected by Lyons asthe theoretical basis for the notion of marking, "one member of the pair ofopposites is characterized by the presence and the other by the absence of amark" (Trubetzkoy 1958: 67; cf. section 1.1), with ntark understood asposit ive (phonetic) property. Thus the original notion that marked andunmarked values are imposed on dist inct ive propert ies gave way to thesimpler idea that privative opposit ions consist in the presence vs. the absenceof some sound property - despite Trubetzkoy's expl ici t acknowledgement ofthe essentially gradual character of all differences in sound properties(1958: 67. the very same^quoted by Lyons) . :

ln the further development of phonological theory, after Jakobson hadshown in 1938 that al l phonological dist inct ions can be understood as binaryopposit ions (19l la:272ff.) , this oversimpli f icat ion was left behind, for therewas no way in which i t could be extended to the descript ion of dist inct ivefeature oppositions in general. To be applicable to both phonologicallyungradable ( logical ly contradictory) dimensions, such as voicing, and phono-logical ly gradable ( logicalty contrary) dimensions, such as compactness,markedness relations had to be understood as the distinction between a morenarrowly defined and a less narrowly defined term: "the marked term isopposed to the unmarked one by i ts closer concentrat ion on a certain eitherpositive or negative perceptual property polar to that of the unmarked term"(Jakobson 1979:91f . ) .

26 Henning Ander.rcn

Short ly after he had establ ished the binary character of al l phonologicaldist inct ions. Jakobson wrote his pioneering studies of the acquisit ion ofphonology. which entai led further developments relevant to the understand-ing of markedness.

The apparently universal sequence of acquisit ion of phonological oppo-sit ions - counter-paral leled by the loss of phonological dist inct ions in aphasia,and paral leled by the incidence of phonological opposit ions in the languagesof the world - pointed to the substantive basis of markedness relat ions inphonology: markedness values could be expl icated direct ly in terms ofperceptual (and ar t icu latory) opt imalness (unmarked) and at tenuat ion(marked) (c' f , 1971a 373ff. , 491 ff .) . Hereby phonological markedness wasdistanced doubly from the understanding of markedness Jakobson had givenexpression to in his correspondence with Trubetzkoy. For one thing,phonological markedness values could apparently not be freely imposed on agiven perceptual dimension the way they can on such experiential dimensionsas l i fe and death ( r ' / .sect ion 3.0) ; they are universal . For another , the i rimposit ion on phonological opposit ions could now appear to be not aconceptual matter. but to be determined by the human perceptual apparatus.

This is i l lust rated by Jakobson's d iscussion of markedness in 1979,where hetr ied to clari fv the dif ference between markedness relat ions in grammar and inphono logy ( r / , ( t t ) )

ln the search for general meanings of paired grammatical categories it wasfound that one category signals a certain grammatical concept which theother one leaves unsignaled. In French, with its grammatical distinction ofthe feminine and masculinc genders. the marked feminine l ionne ' l ioness'

specifies the female; the unmarked, called "masculine", l ion'l ion' mayincludc both the male and the female. Thus the general meaning of l ion, incontradistinction to that of l ionne , implies no sexual specification, and onlythe basic meaning (Grundhedeutung) of lion prompted by informativecontexts suggests a sexual specification, e. g. l ions et l ionnes.

There is an intrinsic communality between markedness on the level ofgrammatical categories and markedness on the level of distinctive features.This communality is, however, cornbined with the considerable differencebetween these tv;o tt,pes o./'opposition lMy emphasis; HA], one based on thesemantic level of signatum, the other on the sound level of the signans. Theconstraining, focusing character of the marked term of any grammaticalopposition is directed toward a more narrowly specified and delimitedconceptual item. In the dyads of distinctive features the marked term isopposed to the unmarked one by its closer concentration on a certain eitherpositive or negative perceptual sound property polar to that of the unmarkedterm (1979: 91f.) .

( 8 )

---

. t '

i .

le1e. r I( t l

cdI S

' ter

: ed

tr'lurkctlnc,t.: theorr thc lirst 150 t,eur.r 27

Jakobson was obviously r ight in h is in tu i t ion that there is a palpabledifference between markedness relat ions in grammar and in phonology. Butthere is no reason to bel ieve that perceptual sound propert ies are lessconceptual than dist inct ions of gender. They are in fact processed by the samehemisphere of the human brain (c/. Jakobson 1979:1751. and passim). Morel ikely, the observed dif ference has nothing to do with the dif ferences betweensemantics and phonology. but rather - as he himself suggested in the quotedpassage - with the types of opposit ions involved.

The grammatical opposit ion ' femrnine' vs. 'masculine' is ful ly integratedwith the inclusive relat ion of markedness values: hence l ion,s can be used aboutmale l ions, female l ions. and l ions irrespective of sex. Phonological oppo-sit ions, by contrast. form exclusive relat ions - not surprisingly. for theirdist inct ive, or diacri t ic, function depends on this. By their exclusive characterthey are completely analogous to such semantic opposit ions as"male : female". "married : unmarried", " 'straight : crooked". But l ikethese, and despite their f unctional exclusivity, phonological opposit ions fal lunder the " law of part icipation" and are conceptual ly framed in terms ofmarkedness values.

4. Greenberg

Undoubtedly the most important publ icat ion on markedness since the notiongained currency in the 1 930s is Greenberg's monogra ph Language Universals( 1966). which presents a survey of phenomena to which the notion had beenapplied t i l l then and extends the application to the lexical f ield{ of kinshipterminology.

The monograph is remarkable by not departing from a definition orexpl icat ion of the concept of markedness. Instead, the concept is assumed tobe familiar to the reader from, for instance, Trubetzkoy's Grundzilge,Jakobson's publ icat ions from the 1930s. and Hjelmslev (1935), but to lack aprecise definit ion. The exposit ion consequently concentrates on i l lustrat ingdiverse manifestations of markedness in phonology, grammar, and lexis withthe aim, principal ly, of determining whether there is "some one characterist icwhich might serve as definitional for this notion which tends to take onProtean shapes" (p. 11) .

This approach marks a significant theoretical departure from the earlierwritings on markedness. The concept of markedness had emerged, from Roth

28 Henning Andersen

(1815) through the European structural ist period. as a (hypothetical) formal

principle which (deductively) explained a number ot 'observed characterist ics

of language. Greenberg, by contrast, undertook to find in the diverse

manifestat ions of markedness a unifying characterist ic. a single general

cri ter ion which would serve as a heurist ic. As one might expect ' this inductive

approach would tend to lead in the wrong cl irect ion. Instead of r ising from the

otservable phenomena to the explanatory principle, the investigation re-

peatedly descends from the phenomena of grammar to the epiphenomenon of

text frequency.Given that Greenberg did not succeed in determining any single character-

ist ic of the observed phenomena (p.70), the monograph did not serve to

cement the fundamental importance of markedness as a formal principle' I t

offered nothing that would resolve the conflict between the two most

widespread, equally nominal ist att i tudes to markedness, one, that i t is a handy

label for a number of disparate observable phenomena, the other, that i t is a

hazy concept which contributes nothing to the characterization of the diverse

observable phenomena to which i t has tradit ional ly been attached.

The real value of the monograph. therefore, is in the survey i t provides of

these observables. Greenberg identi f ies half a dozen dist inct phenomena in

phonology (13 -24) and a few more in grammar and lexis (25-55), which can

be taken as indications of or cr i ter ia for markedness. In a synthesizing chapter

(56-71) he then argues that f ive of the major phenomena in phonology are

closely isomorphic to phenomena in grammar and lexis (cf ' (9)) '

(9) a. neutralization/participationb. text frequencYc. greater allophonic/allomorphic variabilityd. unPaired Phonemes/sYncretisme. basic allophone/agreement a potiori

and discusses a few grammatical and lexical phenomena which do not seem to

have direct .orr.rpondents in phonology, the most important of these being,'zero expression'i and "facultative expression". In the following sections I

will review these phenomena and try to point out their interrelations and

ultimate dependence on the single principle Greenberg despaired of finding' It

will become apparent that the main obstacle to Greenberg's search, besides his

inductive appioach, was insufficient attention to the radically different

dimensions to which the phenomena he discussed belong'

As a preliminary to this survey, it is necessary, first of all, to distinguish

between gru*-u,i 1o. code) and text (or messages), for the frequency

phenomena ((9) (i)l t.tong to the latter, whereas all the other manifestations

of markedness relations belong to the former. Secondly, it is useful to make

Murkedncss theort' rlrc first 150 years 29

explici t the assumption, apparently underlying Greenberg's exposit ion, thatgrammar can be viewed as a series of mapping relations between successivelevels of representation through which referential matter is transformed intophonetic representations, minimally as in (10). Thirdly. these mappingrelat ions should be described consistently with the same direct ional i ty. say. rnterms of the encoding process (from A to D in (10)).

(10) A. referential mattersB. semantic representationsC. phonological representationsD. phonetic representations

4.1 Frequency

Unmarked phonemes have a greater text frequency than their markedopposites. Greenberg documents this lact with results of phoneme couirts in anumber of languages ( 14-21).

Unmarked grammatical categories similarly have a greater text frequencythan their marked counterparts. Greenberg presents a fair amount of datai l lustrat ing this general izat ion with the categories of number, case, g€nder,degrees of comparison. gradation, ordinals, person, voice, mood. tense andaspect, negation, parts of speech, and in lexis, polari ty adjectives (31-53).

In his discussion of the common characterist ics of markedness phenomenaon different levels of language (63-70). he tries to relate all the differentphenomena which will be reviewed in the following sections to differences intext frequency and tends towards the conclusion that they can all be reducedto'such dif ferences. Actual ly, he is ambivalent in this reductionist effort. Onone hand, he notes the advantages of interpreting markedness relationsconsistently as observable differences in text frequency. On the other hand, herepeatedly points out that relative text frequency is merely a historicalresultant of markedness relat ions (65), and synchronical ly is a symptom, thatis, an epiphenomenon, which i tself is in need of explanation.

I will not review Greenberg's discussion in detail. But it may be recalled thatTrubetzkoy initially conceived of markedness as in principle independent offrequency (c.f.the quotation in section 3.0). In Grundzi)ge he conceded therelevance of markedness to frequency and explained it as in part the result ofneutralization, which produces a different distribution of marked andunmarked phonemes. But at the same t ime he pointed to data in whichmarkedness values and relative frequency do not go together (1958: 235 f.). In

30 Ilenning Ander,sen

French, plk-obstruents, taken together. are more frequent than bdS-

obstruents, just as in Russian; but in French the plfr-series is marked (the

opposit ion is based on tenseness). whereas in Russian i t is unmarked (the

opposit ion is based on voicing).This example suggests an important source of divergences between

markedness and frequency. In the historical development of languages,changes may occur through which, for instance, the dist inct ive propert ies onwhich an opposit ion is based may be reinterpreted. Such "rephonologiz-at ions" may entai l inverse markedness values, and when this occurs. relat ivetext frequency cannot be expected to correlate with markedness values.Greenberg presents a convincing picture of the interrelat ions betweenmarkedness values and universal diachronic tendencies (63ff.) . But i t is clearthat in some of these, markedness values motivate changes in frequency, andnot vice versa. A paradigm case is changes involving styl ist ic variat ion. Areinterpretat ion of the markedness values of styl ist ic variants can explainchanges in their relative frequency. But without such (covert) changes invaluation. the observable changes in f iequency are inexpl icable.

One may conclude that text frequency may be a useful, but not a universallyrel iable indicator of markedness values. Where i t does not conform withmarkedncss values, the independence and the primacy of the values is evident.Where text trequency does correlate directly with markedness values, it seems,there are two possibilities. The markedness values may be language specific,and the relative frequency is the result of a historical development which hasadjusted usage to the norms or the system of the language. Alternatively,themarkedness values may be universal, as is presumably the case with the basicpolarity adjectives (52f.). ln such instances the question of whether marked-ness values or relative frequency is primary appears to be moot. But hererelative frequency is a consequence of a difference in distribution which is theprimary manifestation of the different values of the two opposites.

l t is notable that Lyons, in his exposit ion of markedness, does not speak of,

relative text frequency, but of differences in distribution (c/. section 1.2).

Differences in distribution are a matter of grammar, which can be correlated

with markedness values. But text frequencies reflect a variety of different

factors, linguistic as well as extralinguistic. If one defines markedness in terms

of text frequency * as Greenberg explicitly wished to do (but in the end shied

away from doing), and many linguists have been prone to do since - one

merely empties the concept of all meaning other than "relative text

frequency", and the term might as well be dispensed with (y' Lass 1975).

- -

Marketlne,s.s thertry the tirst 150 r'eor.s 31

4.2 Neutralization and participation

Greenberg compares neutral izat ion in phonology with what he cal ls contex-tual neutral izat ion in grammar and lexis (58).

ln envi ronments where a phonologica l opposi t ion is neutra l ized. usual ly .the unmarked term occurs (13f.).

In grammar and lexis, the opposit ion between trvo or more categories maybe suppressed, and i t is then the unmarked member which appears. Greenbergidenti f ies this phenomenon with Hjelmslev's part icipation. but prefers to cal li t "contextual neutra l izat ion" (28f . ) . His only grammat ica l example of th is isthe use of the singular form of nouns in phrases with cardinal numerals inHungarian, Turkish, and other languages (which wil l be mentioned again insection 4.5), but in the discussion of lexis he gives a variety of other examples(52 f . ; .

Greenberg's exposit ion of phonological neutral izat ion is couched in termsof the simpli f ied, received understanding, which was mentioned in section 3.3and ref lects the pre-1938 conception of segments as the ult imate phonologicalcomponents. Viewed in these terms, which do not disturb the picture here. thedistr ibution of phonetic segments that results in cases of neutral izat ion isparal lel to the reference values of antonymic pairs: we have, say. [b] : [p]where a voicing opposit ion is dist inct ive (corresponding to bitch ' female

dog' : dog'male dog', when used contrastively) and phonetic [p] where theopposit ion is suspended (corresponding to dog'dog, irrespective of sex'). Onecan even elaborate the paral lel ism with further dist inct ions, in posit ions ofneutralization, between [p] as the outcome of irresolvable neutralization (cf."generic dog"), tp] representing underlying lbl ( ' female dog'), and tp]representing lpl ( 'male dog') (cl . the examples in section 1.3).

It appears that the different reference potential associated with markednessvalues in grammatical and lexical antonyms is entirely isomorphic with thedifferent " realization potential" of (neutralized) phonological opposites.

There is an apparent discrepancy, which needs to be specified andexplicated. The grammatical and lexical antonyms are, so to say, born withthese differences in reference potential; they form genuine inclusive relations.ln phonology, where oppositions are functionally exclusive, the similardistributional configuration depends on the existence of a neutralization rule(which deletes phonological specifications) and an additional rule by whichthe resulting "archiphonemes" are consistently assigned the phonetic pro-perties of the unmarked term of the opposition.

This apparent lack of parallelism disappears, however, the moment"reference potential" and "realization potential" are considered from the

32 Henning Ander.sen

encoding point o f v iew. From th is point o f v iew. contextual neutra l izat ion ingrammar or lexis means the omission of semantic specif icat ions for whichthere is no communicative need and, concomitantly. the oblrgatory selectionof the unmarked term of the relevant grammatical or lexical opposit ion (e.9..when we talk of a dog having puppies). Similarly, neutral iz.at ion in phonologymeans omission of a phonological dist inct ion and the obl igatory selection ofthe phonetic propert ies associated with the unmarked term.

The omission of information is in both cases sanctioned or st ipulated by thenorms of the language. One may have the impression that in grammar andlexis the omission of semantic specif icat ions is optional, whereas in phonologyneutral izat ions are obl igatory. In fact, however. this is not a dif ference inprinciple, but one of degree.

The two paral lel neutral izat ion phenomena can be understood as arising inthe mapping between levels A and B and between levels C and D in (10).

4.3 Allophonic and allomorphic variability

Greenberg cites C. F. Hockett for the observation that phonemes which areunmarked for a certain feature tend to show greater al lophonic variat ion thantheir marked counterparts (21).

In morphology, as a rule. there is less morphological irregulari ty in markedthan in unmarked categories (29).Greenberg's examples here involve stemalternations and alternations in aff lxes. This general izat ion may only be atendency; yet, it is clearly related to the established fact that marked (derived)

categories are more subject to analogical leveling than unmarked ones (69).Although the similari ty between these two, phonological and morpholog-

ical, phenomena is clear enough, they are also clearly different from thephenomena reviewed in section 4.2. Neither al lophonic variat ion noral lomorphy have to do with the omission of information. On the contrary.The difference between the single, invariant morpheme shape as bearer of a

certain meaning and the contextual ly condit ioned alternation of two or more

al lomorphs is that the latter carry more information. They have their propergrammatical or lexical meaning, but in addit ion indicate features (be theylexical, grammatical, or phonological) of the environments in which they

occur. or, in other words, besides their symbolic content they have specificindexical content. Similarly in phonology. When in Nootka the unglottal izedstops have both aspirated and unaspirated al lophones (21), the dif ferent

degrees of aspiration are in effect subsidiary indexical signs which serve to

specify the phonological surroundings.

L-

Murketlnc.ts thertrt' the first 150 t'eur,s 33

Interpreted in these terms, the two phenomena at hand can be restated asthe observation that unmarked terms of phonological or morphologicalopposit ions are more compatible with subsidiary. indexical signs than aretheir marked opposites. We wil l return to this general izat ion in section 4.4.

The two phenomena considered here concern the mapping relat ionsbetween levels B and C. respectively levels C and D in (10)

4.4 Unpaired phonemes and syncretism

The number of phonemes in a language with a certain marked f 'eature isalways less than or equal to the number with the corresponding unmarkedfeature but not greater. Greenberg's key example is the universal numericalrelat ion between nasal and non-nasal vowels (21).

A morphological and lexical counterpart to this, in Greenberg's view, is thegreater susceptibi l i ty to syncretism of marked than of unmarked categories, aswhen masculine and feminine gender is dist inguished in the singular, but notin the plural. or when cases that are morphological ly dist inct in the singularare syncretized in the plural (21).

This comparison is not fel ici tous. The phonological side of i t has to do withthe combination of dist inct ive features into simultaneous syntagms (or"bundles") and amounts to an abservation that a simultaneous syntagm ofdist inct ive features with a given marked term wil l not be expanded withsubordinate feature opposit ions unless the corresponding syntagm with i tsunmarked opposite is. This observation goes back ( in essence) to Brondal(1943: 105 f.), who spoke of it as "a principle of compensation" (cf. Jakobson1971b:214,487 and pas.sim, Andersen 1974, Gvozdanovic 1985: 37f.). I t isnotably similar to the general izat ion about subsidiary indexical signs insect ion 4,3.

The morphological and lexical side of the comparison cannot be viewed, asGreenberg did, as a reduction in the number of "semantemes" in markedcategories (59). Nouns retain their lexical gender in the plural even when theirmorphological markings do not reflect gender distinctions; and a noun phraseis in whatever syntactic case it has been assigned regardless of whether itsmorphological case affrx is distinct from other case affrxes; and similarly withall the other examples. Syncretism is not a reduction in meaning or content,but a reduction in the number of distinct expressions relative to the number ofdistinct combinations of grammatical or semantic features.

The phonological side of the comparison, then, is a matter of the segment

34 Henninc Ander,sen

internal phonotactics of dist inct ive f 'eatures: i t concerns the inventory ofreadymade dist inct ive feature syntagms avai lable for phonological represent-at ions ( level C in (10)) . Syncret ism, on the other hand. is a pecul iar i ty of themapping relat ion between levels B and C,

4.5 Basic allophone and agreement a potiori

When a phoneme has several al lophones. i t may be possible to dist inguish oneor more non-basic variants which occur in specif ic environments. whosefeatures they may share. and a basic variant which is independent of i tsenvironments and is unmarked with respect to the features carr ied by the non-bas ic va r ian ts Q2f . . 59 ) .

Greenberg sees a paral lel to this in the default agreement rules, common inlanguages with gender dist inct ions. which apply" for instance, to adjectivesmodifying cc'rnjoined nouns of dif ferent ger-rder (59f.).

The phonological side of this comparison clearly has to do with themapping relat ion between levels C and D and should be connected with thestatement about neutra l izat ion in sect ion 4.2. \n h is d iscussion of phonolog-ical neutral izat ion. Greenberg mentions only the type in which an "archi-phoneme" is consistently represented by the unmarked term of the opposit ion(13i.). More commonly. however. the phonetic propert ies of both oppositesare distr ibuted in complementary environments. In such cases, apparently, thegeneral izat ion at the beginning of this section holds, ?s, for instance, inRussian, where in posit ions where the voicing opposit ion is suspended,obstruents are voiced belore a tbllowing voiced obstruent (regardless ofintervening boundaries), but voiceless everywhere else, that is, before voicelessobstruents (regardless of intervening boundaries), at encl i t ic and wordboundaries before an init ial vowel or sonorant, and before pause.

The morphosyntactic side of the comparison, by contrast, is internal to levelB. I t concerns the selection of the grammatical category to be expressed in asyntagmatic environment where the grammatical opposit ion in question isneutral ized. This is in principle no dif ferent from the neutral izat ion of thecategory of number in numeral phrases, mentioned in section 4.2. Note that innumeral phrases the number in the noun is redundant. I f , for morphologicalreasons, number cannot be left unspecified in the noun (say, by the use of abare noun stem or of a special numerative form of the noun), there are threelogical possibilities: the noun may be in the unmarked number (paradigmatic-ally motivated neutralization); it may be in the marked number (syntagmatic-ally motivated neutralization); or the different numbers may be complemen-

---t--

^ - -

Mrtrketlness theort the ftr,st 150 t'cur,s 35

tar i ly distr ibuted depending on the magnitude of the numeral or on theadjacent consti tuent of the numeral (as in Russian. which uses the singularwi th numerals ending in 2,3, or 4 and the p lura l wi th h igher numerals : orcontrast Danish and Englrshenogtt ' t ,e nrci ler '21 nights' . but tu,sind og en not'1001 night.s') . Mutatis mutandis, default agreement rules for gender show asimi lar d ivers i ty .

But in al l such cases, the grammatical neutral izat ion is operated on the levelof semantic representations ( level B in (10)). prior to the selection ol- theappropriate elements of expression. The apparent paral lel ism betweenphonology and morphology here vei ls a basic dif ference.

4,6 Zero expression and facultative expression

Greenberg defines two morphological phenomena involving markedness.which seem to lack phonological counterparts.

The f irst of these is the high incidence in languages of zero morphemes asexpressions for unmarked categories, which involves the mapping betweensemantic and phonological representations (62\, levels B and C in (10). This isLyons ' f i rs t type of "mark ing (or markedness)" (sect ion 1.1) , which was notedby Pe5kovski j (r ' f , section 2.1) and was thematicized by Jakobson in "Signe26,10" (cf. section 2.2) and exploited in his many papers on morphologicalpatterns of the Slavic languages.

By "facultat ive expression" Greenberg means the omission of semanticspecification which is not communicatively necessary, as in the use of uuthorfor a writer regardless of sex, or the Korean optional use of the singular formof nouns also where more than one is ref 'erred to. [n essence, this is only weaklydif ferentiated from his "contextual neutral izat ion". which was discussed insect ion 4.2.

4.7 Summing up

We can now summarize the different phenomena considered by Greenberg interms of the mapping relat ionships mentioned in section 4.0.

A -' B. Contextual neutralization (section 4.2) and facultative expression(section 4.6) concern obligatory or optional omissions of semantic specific-ations. They arise in the mapping of referential representations into semanticrepresentations and can be understood in terms of the inclusive paradigmatic

36 Henning Andersen

relat ions described by Hjelmslev (c' / . section 2.3). Since the reference potentialof the unmarked mernber of an opposit ion encompasses that of the markedone, it can be selected whenever there is no need for the specific reference of themarked member. Such omissions may be more or less normative. they may begrammatical ized or ref lect genuine pragmatic options. the latter beingmot ivated by the maxlm of quant i ty . I f one v iews them as neutra l izat ions, theycan be said to be paradigmatical ly motivated.B. Agreement u potiori (4' . section 4.5) arises in the concatenation ofgrammatical categories and can be seen as a syntagmatical ly motivated kindof neutral izat ion. But when default agreement rules st ipulate the selection ofthe unmarked member of a gender opposit ion. they are evidently paradigmat-ical ly motivated as well : here. too. the undelimited reference potential of theunmarked member is essent ia l .B -+ C. The assignment of zero expression to unmarked categories (c/.section 4.6) is a simple example of iconicity (c./ . Jakobson 1965): thequanti tat ive opposit ion between zero depth (unmarked) and definite depth(marked) (c ' l . sect ion 2.2) is d iagrammed by a s imi lar ly quant i ta t ive d is t inc-t ion between zero and real expression. But i t must be noted that this iconicrelat ion obtains also in the numerous cases where the value relat ion forsemantic reasons cannot be real ized as a dif ference in reference potential , viz.when semant ica l ly non- inc lus ive re lat ions are involved (e.9. ,rnarried : unmarried (r'f . section 2.4).

The greater amount of al lomorphy in unmarked categories was seen to be aspecial manifestat ion of Brsndal 's principle of compensation: unmarkedmembers of opposit ions being more hospitable to the addit ion of indexicalcontent (c/. section 4.3). Syncretism may be understood in similar, thoughinverse, terms. While al lomorphy provides symbolic signs which carrysubsidiary, indexical information about their context, syncretism provides lessspecific symbolic signs, which for their specific interpretation are dependenton the context (r.1. section 4.4). Both al lomorphy. slanted in favor of theunmarked categories, and syncretism, biased towards the marked ones,produce skewed mappings between content and expression and therebydiagram markedness values.C, The f act that languages never have more, but often have fewer phonemeswith a marked feature than with a corresponding unmarked one wasinterpreted as a consequence of Brondal's principle of compensation, whichacts as a constraint on the expansion of dist inct ive feature syntagms withsubordinate distinctive feature oppositions (c/. section 4.4).C -) D. The same principle constrains the expansion of dist inct ive featuresyntagm s; with (subordinate) al loph onic di f ferences: unma rked phonemes are

l -

,M urkadnas,t' thetrt thc first 150 vcars 37

more hospitable to the addit ion of subsidiary indexical phonic signs than theirmarked opposites (r/ . section 4.3).

Phonological neutral izat ion was viewed as the normative (but not neces-sari ly involuntary) omission of phonological specif icat ions and the sub-sequent assignment of phonetic propert ies to the thus reduced simultaneoussyntagms (c' / . section 4.2). Greenberg discussed only the cases where thesegments in question are consistently assigned the propert ies of the unmarkedmember of the phonological opposit ion. However, as soon as one considerswhat Trubetzkoy termed "external ly condit ioned" real izat ions. the rules thatspecify these are clearly shown to be of the same kind as the ones that ref lectGreenbergs general izat ion about non-basic and basic variants (r. / . section4'5): they assign marked propert ies to specif ied (marked) environments, andunmarked proper t ies e lsewhere (22f . .6 j f . ) .

This is a convenient place to return to the three types of "marking (ormarkedness)" described by l-yons 1977.

Lyons ' " formal mark ing" ( t : / .sect ion 1.1) corresponds to Greenberg 's"zero express ion" (B - ) , h the e x tent that the1, are re lated to markedness.dist inct ions between zero anrJ real expressions are an iconic ref lect ion of 'nrarkedness values.

Lyons" 'semant ic mark ing" ( t f ' . sect ion 1.3)covers the markedness valueson the semantic side ( level B) of the phenomena summarized above underA --+ Ef , whereas his "distr ibutional marking" (r ' / . section 1.2) concerns thedifferences in extension ( level A) engendered by these asymmetrical values.

In v iew o l ' the f act that Lyons ' top ic is semant ics. i t is natura l that he wouldl imi t h is exposi t ion to phenonlena that in one way or other mani fest the valuere lat ions at level ts . But h is fa i lure to in t roduce h ierarchica l re lat ions amonghis three concepts of "mark ing (or markedness)" is a ser ious shor tcoming.L,vidently. markedness values ( '"semantic marking") are primary. But theycould hardly be exhibited i f they were not. in conjunction with some semanticopposit ions, manifested in dif ferences in distr ibutions. Without doubt, someof these distributional differences are universal. By contrast, differencesbetween zero and real expressions may, but do not necessarily reflectmarkedness values. Their possible correlat ion with markedness values; is inevery case a language part icular matter, which can be ascertaincd onlyprovided that the markedness values have been establ ished irrespective oftheir zero or real expressions.

It appears f iom the preceding review of the markedness phenomenadiscussed by Greenberg that they can all be explicated in terms of one and thesame formal principle. Al l paradigmatic relat ions in language, both semantic(grammatical and lexical) and phonological dist inct ions - and, in phonology,

38 Henning Anderscn

both phonemic dist inct ions and al lophonic dif ferences - are establ ished as

inclusive opposit ions. As a consequence, they al l incorporate asymmetrical

value relat ions even though, from a functional (semantic or phonological)

point of view, many of these paradigmatic relat ions are non-inclusive and

hence symmetrical.The asyrnmetry of these value relat ions is ref lected in skewed mappings

between referential and semantic representations. between semantic content

ancl phonological expression, and between phonological and phonetic

repreScntatlonS'T]he$"a$€+{ewRe'g'inde|:ruIagrc@refe+egeets t*re+r,r++nruL,cd tern1 i*certair] sy$taltmatic environments. irr{he*l+flb+enee

of' read+ne*<*' m a r kecl an<l u n m ar k ecl+enr+s kr beco m bi n ed w i t h s u bo rdi*a+e

distine+io*r*- ryherlrer. -phoner*ie epposi tiot*- er @rallophonicv+r+ation.

Al l these observable phenomena, i t must be noted. presuppose. but are not

identical to the asymmetrical value relat ions they ref lect. The asymmetrv of

the "singular" : "plural" opposit ion may in a given language be ref lectedf in

case syncretism in the plural paradigrns or in greater al lomorphic vartat lon In

the singular ene. But it obtains independently of these specific manifestations.

Similarly with lexical opposit ions. These may be functional ly incompatible

with the inclusive relation of markedness values (as we saw in section 2.4) and

may not necessarily, in any given language, be reflected by any of the types of

phenomena reviewed above, not even by what Greenberg called contextual

neutral izat ion (section 4.2) or facultat ive expression (section 4.6). That they

do indeed obtain can often be inferred only from more subtle manil'estations,

such as their preferred sequence in binomials (c/. section 1.3), or through word

association tests, as Greenberg suggests (53 f.). Only occasionally do we find

such spectacular demonstrat ions ot ' the pervasiveness of markedness relat ions

as the regular lexical distribution of Spanish and Philippine etyma in

Zamboanguefro, mentioned in sectton 2.4.But by whatever means markedness relations are manifested, their existence

is logically prior to their manifestation, and they may consequently be

presumed to obtain even when their existence has not yet been ascertained.

5. Understanding markedness

5.1 Markedness as a formal principle

In the preceding sections I have distinguished between markedness as a formal

principle which determines the form of paradigmatic relations in language

5: It is reflected. for instance in default agreement ruies. br thepreference for the unmarked terrn in cenain s) ntagmatlcenvironments. It is reflected, as w'ell, in the unequal faciliry' w'ithwhich marked and unmarked terms are combined with subordinatedistinctions, whether morphological categories or phonemicoppositions, or allomorphic or allophonic variation.

| ,

5

Murkedness theorr' - the ftrst 150 vear.r 39

and the numerous - as yet far from ful ly charted - manifestat ions thisprinciple has jn languages. In this section I wi l l br ief ly touch on the nature ofthis principle.

In his 1966 monograph, Greenberg discussed some of the asymmetries hedescribed, in terms of implicational relat ions, noting that whenever one of theobservable phenomena can be stated as a universal implication. " i t is theunmarked member which is the implied or basic term and the marked which isimplying or secondary". This led him to the observation that the dist inct ionbetween the marked term and the implied, fundamental character of theunmarked term appears very similar to the famil iar dist inct ion in Gestaltpsychology between figure and ground (60).

There is no doubt that this is a fruit ful comparison, part icularly i f thedist inct ion between perception and what seems to be involved in markednessrelations can be specified. Note that in visual perception, the ground in factincludes the figure, but figure and ground are experienced as contradictories.ln the framing of l inguist ic opposit ive concepts ( lexical, grammatical, orphonological). i t appears as i f there is an init ial division of an experientialdimension into a sal ient, del imited area and a less sal ient, unbounded onewhich uncludes i t . Only subsequentlyf i t corne{ to an identi f icat ion of thefunctionalcharacter of their relat ion. *1, ich nray'be inclusive or exclusive, andif exclusive, contrary, contradictory, converse, etc.

With this interpretat ion we are not very far from Levy-Bruhl 's notion ofpart icipative relat ions, part icularly as this notion was harnessed by Hjelmslevto account for the paradoxical conjunction in grammatical opposit ions ofIogically diverse relations in meaning with inclusive relations in value (y'section 2.3). But where Levy-Bruhl spoke of language as bearing the imprintof a pre-logical mental i ty, i t would seem more to the point to recognize in theubiquitous markedness values the effect of a cognitive strategy which takesprecedence, ontogenetically, over the functional (and logical) analysis of theexperiential dimensions encoded in language and culture.

The inclusive relat ions which this cognit ive strategy imposes on al lexperiential dimensions are apparently not superseded, in the life of theindividual, by the results of later cognit ive activi ty. I t is for this reason that thel inguist can conclude that al l opposit ions in language "fal l under the law ofparticipation" (Hjelmslev 1935: 102), and the logician who ventures beyondthe confines of traditional logic must recognize that "exclusion is merely aspecial case of participation, in which certain areas (Fr. cases) of the extensiveterm are empty" (1939: 87).

The cognitive interpretation of markedness suggested here forms a goodbasis for understanding Brsndal's principle of compensation as well, for it can

40 Henning Ander,sen

be expanded into an account of both the conceptual elaboration of individual,

homogeneous experiential dimensions and the formation of complexes of

heterogeneous concepts.In terms of visual perception, it stands to reason that once a figure-ground

distinction has been established, it is easier to perceive secondary differences in

saliency in the (primary) ground than within the (primary) figure' Corre-

spondingly, in the conceptual differentiation of any single experiential

dimension, the less sal ient, undelimited area wil l be a more fert i le ground for

secondary divisions than the more narrowly del imited one.The difference carries over into the formation of conceptual complexes

(semantic or phonological), where the undelimited area of an unmarked term

of an opposition will form a better background for secondary, subordinate

dist inct ions than the narrowly del imited one of i ts marked counterpart.

Considering the fundamental importance of the figure-ground distinction

for perception. i t is not surprising i f the most basic paradigmatic relat ion, the

inclusive opposit ion, is founded on a homologous cognit ive operation.

5.2 Definitions

Against the background of this understanding, one can evaluate the different

attempts that have been made in the history of the markedness concept to

capture i ts essence.To Roth (1815), the unmarked vs. marked (U/M) relat ion was between a

complex (U) and a simple (M) term (dependence and independence vs.

dependen ce; cf'. section 2.1), a view which adequately describes the different

reference potential of the two terms.Kalepky (1901) emphasized the vagueness and indeterminacy of the

unmarked and the saliency of the marked term in words that very neatly

circumscribe the ground vs. figure relation. When he defined the difference

between two opposites in terms of their privileges of occurrence. assigning one

to specific conditions and the other to elsewhere environments, he in effect

performed a similar analysis on the total range of relevant environments (c'/.

sect ion 2.1 ) .Pe5kovskij (1928) spoke of zero (U) vs. definite (M) meaning and drew the

parallel to zero vs. real (segmental) morphemes. As I have tried to show

(section 2.2), this quantitative conception of inclusive semantic relations

corresponds perfectly to the opposition in semantic depth such relations

emborJy, and it accounts well for their widespread iconic representation in

quantitative distinctions between zero and real morphemes.

--

-L-_

Markedne,s,s theor.\, - the.first 150 year,s 4l

Hjemslev, in his early writings on the subject. spoke of markedness as arelation between a vague term (u) and a precise term (M) (thus in1933:102ff.) , a conception very similar, in essence, to Kalepky's. His ult imateunderstanding of part icipation as a value relat ion, dist inct but inseparablefrom the semantic (or phonological) relations with which it is conjoined, hasproven superior, I think, to al l the other views that have been examined here.

Both Roth and Kalepky explicitly viewed grammatical oppositions as.r,/r'generis contradictory relat ions. Jakobson init ial ly contrasted the logicians'exclusive relations with those of language, but in the end was unable to freehimself from the strait jacket of tradit ional logic (r/ . section 3,1). In attempringto dissolve inclusive opposit ions into conjunctions of exclusives he grappledunsuccessfully with the impossible. Hjelmslev saw correctly that exclusiverelat ions can be reduced to inclusive ones, but not vice versa (1939:8i).

5.3 Applications

With the understanding of markedness sketched in section 5.0, one can seethat the range of phenomena to which the concept is applicable cannot beexhausted with the manifestat ions surveyed in Greenberg's monograph. I f i t istrue' as Hjelmslev maintained, that al l paradigmatic opposit ions in languageare framed in terms of inclusive relat ions, then markedness relat ions obtain inall cases where a language presents its speakers with a choice.

5.3.1 Synchrony

Phonemic distinctions have traditionally been recognized as paradigmaticrelat ions. But also al lophonic dist inct ions form paradigmatic sets. And therange of environments in which any phonological sign may occur can beconstrued as a paradigm (c/. section 4.3).

In morphology and lexis, not only do the content categories formparadigmatic relations, but allomorphs form paradigmatic sets, and theirselection is dependent on a construal of their ranges of occurrence, such thatone or more specified (M) allomorphs are assigned to specified (M)environments, and a single (U) allomorph is assigned to (U) elsewhereenvironments.

Sentence syntax offers paradigmatic choices among grammatical construc-tions (active vs. passive, nominative vs. ergative) and element orders (direct vs.inverted) and text syntax among pragmatically conditioned focusing, topi-

42 Henning Andersen

cal izat ion. and cohesion devices (element order, i terat ion, substi tut ion,

anaphora, el l ipsis).lndividual speech act types present paradigmatic choices between indirect-

ness and directness. Speaking involves paradigmatic choices among speech

genres, and these among different registers and styles.

In al l such cases one can reasonably presume that asymmetrical value

relat ions obtain, and i t makes good sense for the l inguist to posit them and, to

the greatest possible extent, determine their manifestat ions - individual ly as

well as in their interaction in complexes governed by the principle of

compensationAmong the manifestations of markedness in texts one can recognlze

juxtaposit ions of opposite values such as in the irreversible binomials, f i rst

described by Malkiel 11967: cf. section 1.3), whose sequence relat ions

(unmarked before marked) are paral leled in phonology by thc juxtaposit ion

of opposite phonemic values in diphthongs (c/. Andersen 1912).

On the other hand. when in discourse the paradigmatic relat ions ( in

grammar) are transposed into syntagmatic relat ions ( in speech)l th{ nrocess is

in part governed by a principle by which members ol 'di f ferent paradigms

which have identical markedness value tend to be distr ibuted in contiguity

relat ions. This principle was f irst formulated by Jakobson ( 1960). Some of i ts

manifesrations are unwitt ingly mentioned by Greenberg (c' / . sections 4.3. 4.5).

Irs general i ty and possible cognit ive basis is discussed in Andersen 198?.

5.3.2 Diachronv

The relevance of markedness relations for linguistic change in semantic and

phonological oppositions was succinctly sketched by Greenberg (1966: 69 tr.).

In an entirely different dimension. speakers of a language evidently perform

choices between what is innovative and what is not, between what is archaic

and what is not, and they may have a more or less clear conception of

differences in productivity, just as they

structural features of their language aremay have an awareness that some

more. while others are less congenial

to i ts general plan.As it is well known, Sapir's conception of drift rested on the assumption

that such long term developments in languages result from generations of

speakers' subconscious, but similarly weighted choices between alternative

means of expression, among which they favor innovations tending to change

the system of the language consistently in the same direction, into greater

conformity with a definite general plan or type.

L-.

Marketlness theort' - the ftr,rt 150 y'ears 43

This conception, certainly, presumes the existence of a kind of valuerelat ions in internal ized grammars of speakers. But whether such relat ions canproperly be subsumed under markedness must depend on how Sapir 's idea ofdrift can be made explicit as a theory of long term change in language.

6. Conclusion

In the sketch of the early history of the concept of markedness offered here. Ihave tr ied to show both the progressive clari f icat ion of the concept (summedup in section 5.1) and some ol the unclari t ies that developed as aconsequence of the terminology establ ished in the Prague school. and in theprocess of the transfer of the concept from one metatheoretical paradigm toanother .

I t was Togeby who hai led the discovery of inclusive relat ions as "unev6nement comparable d la decouverte de I 'Amerique par Christophe Colornben 1492 - ' ou peut-etre plutot l 'oeuf de Christophe Colomb" (1965:71).Effusive comparisons aside. Hjelmslev's definit ion of part icipation - andpart icularly his insight that the exclusive types of opposit ion are special casesof (are included by) the inclusive type - these were signif icant achievements.One can only regret that they did not become widely known at the t ime.

Hjelmslev found the Prague school terminology apt to confuse, and he wasright. Trubetzkoy's ini t ial , tentat ive dist inct ion between "an active mark" (M)and a "passive mark" (U) might have been improved on and might have led toa clari f icat ion of the concept involved. But the "marked vs. unmarked"terminology that carr ied the day in the 1930s at f i rst led to a conceptualconflation of value relations with the distinction between real and zeroetlressions and in the longer run made it practically impossible to liberate theconcept of markedness from the exclusive, contradictory relation coded in theterms "marked" and "unmarked".

Some of the most vehement denouncements of markedness have been putforward by scholars who have equated markedness with text frequency andthen discover a vicious circle, the "statistical fallacy", as Lass and Andersonhave called it (197 5: 290).ln my critique of Greenberg's monograph (section4.0) I suggested that his concern with text frequency as a heuristic criterion formarkedness reflected an inductivist approach, radically different from thehypothetico-deductive one of the European structuralists. Since markedness

/ I

44 Henning Andersen

is a formal principle and not an immediately given, quantifiable observable,

one should not expect to be able to define it by boiling data down to statistics.But only the inductivist need fear the "statistical fallacy". If one distinguishesproperly between the formal principle, the phenomena it determines, and the

epiphenomena these phenomena entail, there is no circularity. On the

contrary, the appropriate dist inct ions wil l make i t possible to undertakeserious investigations of how, in individual language states, the diverseskewed mapping relations between referential and phonetic representationsmay conspire to produce differences in text frequency which correlate with themarkedness values of l inguist ic enti t ies and how, in diachrony, markednessvalues and the frequencies of semantic and phonologicalenti t ies can mutual ly

influence each other. A confusion of markedness with text frequency

obviously stands in the way of such investigations.The sketch of the first 1 50 years of markedness theory which I have offered

here has deliberately been centered, as much as possible, around the notions,the terminology, and the phenomena that were discussed within that period.

In the decades since the publication of Greenberg's monograph, markedness

theory has flourished in part on terms independent of the earlier history of the

concept. An account of this development, which belongs to the contemporaryhistory of linguistics. will obviously require a very different approach. Butperhaps a retrospective like the one presented here can offer a usefulperspective also for that future account.

ReJbrences

Aksakov , K .1875 Soiineni jaf i lologi ieskie[Philologicalstudies],/ .

Andersen, Henning1968 " lE * . ra f t e r l . u . r . k i nBa l t i candS lav i c " ,Ac tuL ingu i s t i caHu /n iens ia l l : l 7 l - 190 .1972 "Diphthongization". Languuge 48: 1 l - 50.1g'74 "Markectness in vowel systems", Prot'eeding.s o.f the Eleventh International

Congres.r\ / Linguisr.s,edited by L. Heilmann (Bologna: I l Mulino): 1136-1141.1975 "Variance and invariance in phonological typology", Phonologica 1972, edited by

r 980

' etl

Wof fgang U. Dresslcr & Franti5ek V. Mare5 (Mtinchen: Fink): 67-78'"Morphological change: towards a typology", Hi.storical Morpholog1', editcd by

Jacek F is iak (The Hague: Mouton) : 1-50."On the projection of equivalence relations into syntagms", New'vislas in grammur:invariunce and voriation, edited by Stephen Rudy and Linda R. Waugh, (Amster-

dam: John Benjamins).Brsndal, Viggo

1943 Essars de linguistique gtntrale (Kobenhavn: Munksgaard).Fortunatov. F.

1899 "C) russkix zalogax" [On Russian voice], Izvestiia Otdeleni.ia ru.sskogo iazylls I

slovesnosti Akademii Nauk.

1I

Markedne.t.s theory the fir.st 150 t,ears 45

Fox, James197 | "semantic parallelism in Rotinese ritual langu age" . Biidrag(.n tot de Taal , Land- en

Volkenkunde. l2i : 21 5_ 255."Our ancestors spoke in pairs: Rotinese views of language. dialect and code".Explorations in the ethnographr- o.f .speaking. edited by R

-Bou.on & J. Sherzer

(Cambridge: University press): 65-g5."On binary categories and primary symbols" . Interpretotirtn of's.r'mholism, editedby R. Wi l l i s (London: Malaby press) : 99-132.

F'rake. Charles O.1971 "Lexicalorigin and semantic structure in Phi l ippine Creole Spanish". pit lginizarisn

and t'reolization. edited by Delt Hymes (Cambridge: Univeisiry press): 223-242.Greenberg. Joseph H.

1914

197 5

1966 Language uni,er.ral.s, tt ' i th spe t,iul relerenrc to f eature hierarchies (: JunuaLinguarurn ser. min. . 59), (The Hague: Mouton,) .

Gvozdanovic. Jadranka1985 Language s)'stem und irs t'hangc. On theor)' ancl ra.stabilil.y'(: Tiend.s in Lingui.srit,s.

studie's and Monogruph.s, i0) (Berrin: Mouton de Gruvter).I { je lmslev. Lou is

1928 Prin<,ipc:; de grumrnaire ginerale (: Der Kgl. Dan.ske Vitlen,rkuherne.s Selskub,Historiskfilologi'ske Meddelel.ser, 16.1 ) (Kobenhavn: Host & Son).

1935 La <.ut,!gorie tle.s t,us. Etude de grammaire gtnerale (: Actu Jutlantlica, 7,1)(Aarhus: Universitetsforlaget).

1939 "Notes sur les oppositions supprimables", Travaux du Cercle linguistique dePrague. tl: 5l 57. Reprinted in his Essars linguistique.s (: Travaux du Cerclcln{utrttrtut da e;pr;thagra. tti iild ;d; l[;rffirilili ij;;jt,k i;;;*_ ;;Kulturfor lag).

19'73 "srrucrurc g6nerare des c'orrerations ringuisriques" (rg3g), rrxt: .t.ygil.lllgusl, //t: Tratau.r dt cert/e ,/t'n:4ui.rti'que dbZ:op"ntogur,. ra),(Ksbenhavn: NordiskSprog_ og Kulrurforlag).

Jakobson, Romanl92l Nove.i'w-ia rus.sku-iu prt\zija [Recent Russian poetry] (praha).1932 "Zur Siruk,", Jit irrrir.r,.n v.rt,u^:i' lotouron I97rb:3_r5.l9- l9a . .Sisne

7elo, , , . l . f . "Ur"" ' t97t b: 2f i 219.releb . .DIs Nulrz. i ; ; ; ; ; , i ;kobron teTtb:220 _222.le60 "Linguistic' ;; l ' ; ;;r*. stt ' te i tnirrui,ru, edited bv Thomas A, sebeok(Cambridge MA: na.r.r. pr.ssji iso-- i1.i1",1971a Selectert ir it inss,i. ' i io,uotnsiiol, irdirlr. itt. urgue: Mouron).te, I b Scre.recr n,ritin|:5','tt.- i"rd iia t"rir"i; iil. Haguc: Mouron).Jakobson, Roman & Linda W,ugf,

-J'il?r.,n.#jr't" und shape-'|1=io'g,ugc ( Bloomington: Indiana Univcrsiry press).

;,3i:Tj;:1;;:"""adiktorische Gegensdt ze". Zeit.schri"ft fi)r romanisc.he phitoto-

Karccvskij. Serge

- 1927 Systime du verhe rasse (praha).

Lo.tlalo*.r & John M. Anderson

arfJrf ,"n" otrt EngIish pnn'"iost' (Cambri<JgeiUniversity press).

n, jl;:,, yukf;r"ntrn r it s, r -// (Cambridge: Cambridge University press).

1e67 ;?i:ffi,ll1l:":::1"

binomiars", in his r.,.rc-r.r on tinguistit themes(oxrord: Basir

46 Henning Andersen

Nekrasov. N. P.1865 O znaleni i . form ru,sskogo glagola [On the meaning of Russian verb torms] (Sankt

Peterburg).Pe5kovski j , Aleksej M.

1956 Russki j ,r intaksis v nuutnom osveitcni i IRussian syntax in scienti f ic l ight] 7th edn.(Moskva: Udpedgiz).

Roth, G. -M.1815 Grundriss der allgemeinen Sprachlehre (Frankfurt).

Sapir, Edward1921 Language; an introduction to the.stud.v o/ .rpet'ch (Ncw York: Harcourt. Brace and

World).Saussure. Ferdinand de

1916 Cours de l ingui.st ique gtnirule (Paris: Librair ie C. Kl incksieck).Saxmatov, A. M.

1927 Sintaksis russkogo .jaz.lka [Russian syntax] 1/.Togeby. Knud

1965 "Theodor Kalepky et les oppositions participatives". Acta Linguistica lla.fnicnsiu9 : 7 1 - 7 6 .

Trubetzkoy, Nikolaj Sergecvitsch1931 "Die phonolog ischen Systeme". TCLP 4:96 115,l9-13 "La phonologie actuelle". Journul de p'st'chologie 30:219-246.1951 Print ipe.s t le phonologlc. trad. par J. Cantineau (Paris: Librair ie C. Kl incksieck).l95tt Grund:i)ge der Phonologir, (Gottingen: Vandenhoeck & Ruprechi).1969 Principles o/ phonolog.l ' . translated by C'hrist iane A.M. Baltaxe (Berkeley & Los

Angcles: Universrty of Cali fornia Press). i1985 l\'t.5. Truhet:ko.t"s lctters and notes, preparcdfrtr publication by Roman Jukobson

(Ber l in : Mouton de Gruyter ) .Vostokov. A.

1831 Ru.sskuja grummatika IRussian grammar].