Sign-oriented vs. form-oriented linguistics and word-formation. (1).

1.1. Simplifying a great deal and abstracting from the details, it can be argued that the history of linguistics in this century until fairly recently was basically dominated by two major currents, which could be characterised by the labels that also figure in the title of this paper, i.e. "sign-oriented" vs. "form-oriented". To these a third trend must now be added, which has emerged (or, maybe more correctly, re-emerged) in the last two decades, viz. "concept-oriented linguistics". This seems to be the root of so-called "cognitive linguistics", a direction that focuses on the relationship between prelinguistic ("cognitive") categorisation and its linguistic representation, with a marked tendency towards identifying the one with the other and thus giving up the distinction between extralinguistic conceptualisation and linguistic meaning. This latter, idealistic stance could be regarded as a position diametrically opposed to the basically materialistic (asemantic) stance of "form-oriented linguistics", wherea s the semiotic position of "sign-oriented linguistics" might be viewed as a compromise between these two, but is rejected by both of them, although for different reasons. Since in the following I will look at the way in which word-formation theory has been influenced by such general trends, and since cognitive approaches to word-formation are still rare (2) and have not yet resulted in a more comprehensive framework, I will concentrate on the dichotomy mentioned in the title, and will leave the idealistic, concept-oriented trend out of consideration here, hoping to return to it elsewhere.

1.2. The first trend obviously goes back to Ferdinand de Saussure's postulate that language is a system of signs, and that signs are characterised by the fact that they have both form and content. Both the formal and the content aspects of a sign are defined negatively on the basis of its differences from other signs. But the sign as a whole, as a combination of form and content, is "un fait positif. ... deux signes comportant chacun un signifie et un signifiant ne sont pas differents, ils sont seulement distincts. Entre eux il n'y a qu'oppositions. Tout le mecanisme du langage ... repose sur des oppositions de ce genre et sur les differences phoniques et conceptuelles qu'elles impliquent" (Saussure [1973]: 166-167).

The second, formal trend is to a certain extent already reflected by the primarily formal approach to historical reconstruction in the wake of the Neo-Grammarians of the 19th century, although it did not become an explicit dogma within this framework. This was different in certain directions of American descriptive linguistics under the influence of behaviourism, beginning with Bloomfield and still clearly visible today. Within this trend, formal aspects had absolute priority, since the semantic side of language is not directly observable, and only directly observable data were considered relevant to "scientific" linguistic investigation. The most extreme formulation of this trend is found in Harris (1951, 1954), where distribution is explicitly claimed to be the only criterion for the identification of linguistic elements on every level. (3) It is certainly no coincidence, therefore, that Chomsky as Harris's pupil continued this tradition, with the consequence that mainstream generative grammar has been pred ominantly form-oriented since its inception. This does not mean, of course, that there were no Saussureans in America and no formalists in Europe, and even apparently strict formalists like Bloomfield were not always consistent. Thus, Winter (1964: 6-7) pointed out that Bloomfield in his identification procedure of morphemes at first relied exclusively on formal criteria (Bloomfield 1935: 161, 163), but later (Bloomfield 1935: 216), when he identified went as consisting of an allomorph of go and a zero allomorph of the past tense morpheme, the criterion used for this analysis could only have been a semantic one, viz, identity of meaning. Also not infrequently definitions, of the basic functional linguistic unit, the morpheme, contained some reference to meaning, cf.:

Any form, whether free or bound, which cannot be divided into smaller meaningful parts is a MORPHEME (Bloch - Trager 1947: 54).

Morphemes are the smallest individually meaningful elements in the utterances of a language (Hockett 1958: 123).

But, as we shall see below in connection with series such as contain, detain, retain, or consist, desist, insist, resist, meaningfulness was usually not really taken seriously in actual morphological analysis. It is therefore justified to claim that on the whole American linguistics was much more form-oriented than European linguistics. This is especially striking in the field of word-formation, to which I will now turn.

2.1. Probably the most widely known sign-oriented approach to word-formation is that of Marchand (1969). (4) His indebtedness to Saussurean linguistics is already visible in his morpheme definition, cf.:

A word, like any other morpheme, is a two facet sign, which means that it must be based on the significant/significate (F signifiant/signifie) relationship posited by Saussure (Marchand 1969: 1).

And word-formation itself is defined as "that branch of the science of language which studies the patterns on which a language forms new lexical units, i.e. words" (Marchand 1969: 2). The new words are thus sign combinations and as such necessarily morphosemantically transparent (i.e. analysable, relatively motivated), cf.:

Word-formation can only be concerned with composites which are analysable both formally and semantically ... The study of the simple word, therefore, in so far as it is an unanalysable, unmotivated sign, has no place in it (Marchand 1969: 2).

It should be added that Marchand (1969: 2) makes a distinction between composites consisting of full linguistic signs (compounds, prefixations, suffixations, zero derivations, back-formations), which he calls "grammatical syntagmas", and composites which are not made up of full linguistic signs (expressive symbolism, blending, clipping, rime and ablaut combinations, word-manufacturing). In contradistinction to Saussure he thus regards onomatopoeia as at least partly motivated. The distinction between these two types of motivation is irrelevant for the following, however, and will therefore not be discussed any further.

2.2. In the light of these sign-based definitions, it is not surprising that Marchand (1969: 6) rejects an analysis of series such as deceive, conceive, perceive, receive; consist, desist, insist, persist, resist; conduce, deduce, induce, produce, reduce as bimorphemic, which was current in American descriptive linguistics. (5) He argues that the resulting segments con-, de-, in-, per-, re-; -ceive, -sist, -duce do not have any meaning and therefore lack sign status, in contradistinction to formations such as delouse, co-author, perchloride, rewrite, etc., where de-, co-, per-, re- do have meaning.

In this connection it is important to note that this kind of analysis was later on also adopted by generative grammarians, and the resulting units were given the same status as genuine, meaningful morphemes. This is a clear indication that this model is indeed also primarily form-oriented, despite the integration of a semantic component. Thus, in Chomsky -- Halle (1968: 94ff.) de=ceive, etc. were treated as bimorphemic in order to account for their phonological behaviour, which otherwise could not be explained in the SPE-framework. This decision at the same time necessitated the introduction of a special boundary type ("=-boundary"), so that the whole argument in fact becomes circular. And in the first genuine generative contribution to the morphological aspects of word formation, (Halle 1973), (6) we find morphological segmentations such as bro+ther, vac+ant, tot+al, serendip+it+y, trans+form+at+ion+al (Halle 1973: 10), which contain "formally recurrent sames", but not signs in the Saussurean sense. Clearly, such purely form-based segmentations are not relevant to word formation and the resulting segments will therefore also not become productive, i.e. enter into new formations. This will only happen if such units are reinterpreted as being meaningful, which was the case, e.g., with elements such as -burger, -ade, -teria, and, most recently, -gate, cf.:

(1) a. hamburger > -burger = 'specific dish' > beefburger, cheeseburger, eggburger, mouseburger (7)

b. lemonade > -ade = 'drink' > orangeade, pineappleade, fruitade

c. cafeteria > -teria = 'shop' > candyteria, washetera

d. Watergate > -gate = 'political scandal involving the President' > Irangate, nannygate, Paulagate

2.3. Halle disregards this semantic aspect completely, since his segmentation rests exclusively on formal and distributional criteria This does not allow him to express the qualitative difference between the meaningless phoneme sequences burger, ade, teria, gate in the originally unanalysable hamburger, lemonade, cafeteria, Watergate (8) and the morphemes -burger, -ode, -teria, -gate after the morphosemantic reanalysis, which produced new signs. But Marchand's absolute rejection of any morphological analysis of such series as deceive, etc., on the other hand, would also seem to create problems, as was, e.g., pointed out by Ljung:

...a stem like -duce in e.g. produce ... exhibits predictable graphonemic/ phonological variation in all the words of which it is a part: the pattern produce -- productive -- production recurs also in e.g. deduce, reduce, introduce, seduce, etc. Similar observations can be made for -ceive, -fend, -ply, and many other Latin or Neo-Latin bases. Obviously our grammar will be simplified if we treat these bases as morphemes instead of regarding each of the words in which they occur as unique (Ljung 1975: 476).

Ljung in this connection also mentions Chomsky -- Halle's bimorphemic analysis of permit, repel, etc. on phonological grounds and concludes that the definition of the morpheme as the smallest meaningful unit of the language should be replaced by "any minimal morphological unit required by the rules set up for the description of a language" (Ljung 1975: 477). This then would result in "a hierarchy of morphemes with the meaningful ones at the top and those required by the stress rules at the bottom" (Ljung 1975: 477). And it cannot be denied that such a morphological analysis allows the statement of certain subregularities in the grammar, which without it cannot be stated at all -- but at the cost of multiplying the types of minimal, functionally relevant units.

There is also a diachronic argument supporting such a dine, viz, the phenomenon of lexicalisation/idiomatisation, by which originally meaningful constituents may lose their semantic identity, cf. the idiomatised vs. non-idiomatised examples (in brackets) in (2):

(2) a. G aufgeben, aufhoren (vs. aufsteigen), zerstoren, zersetzen (vs. zerbei[beta]en), vertragen, vergleichen(vs. verspielen)

b. E make out, find out (vs. drag out), make up, take up (vs. push up), butterfly (vs. buttermilk), blackboard (vs. blackbird)

From a general morphological point it would indeed be problematic to treat such examples as totally unanalysable, since their morphological behaviour is predictable on the basis of the properties of the simple constituents geben, horen, storen, setzen, tragen, gleichen or make, find, take, fly, board. Thus, if we recognise them as constituents, we simplify the morphological description. In fact, Marchand (1969: 1) himself noticed the problem in connection with the analysis of phrasal verbs such as make out, where the constituents behave like words, but cannot be regarded as signs proper. He therefore introduced, a distinction between genuine free morphemes with sign character (= words) and so-called "free forms". (9) These latter distributionally behave like words, but just form part of a discontinuous sign, e.g. make...out, i.e. do not have sign status themselves. This distinction is not, however, extended by him to the domain of prefixes and suffixes, where a similar solution might have been conceivable. Th e problem of meaningless but morphologically/syntactically/phonologically relevant elements is also discussed in Kastovsky (1982: 70-71) and Bauer (1983:16), where a binary distinction between morphemes proper (with sign status) and formatives (without sign status) is proposed. It has to be admitted, however, that such a binary solution is a gross simplification, and that Ljung's concept of a scale or hierarchy of "signhood" is closer to the truth, although much more difficult to apply. Nevertheless, Marchand's assumption that only signs, i.e. meaningful elements, will become productive in word-formation patterns, is not contradicted by this problem of the demarcation between signs (morphemes) and formatives.

3.1. The influence of sign-oriented linguistics in Marchand's theory of word-formation is also obvious in other respects, e.g. by basing it on the notion of the syntagma, and with it on the notion of opposition, cf.:

Word-formation can only be concerned with composites which are analysable both formally and semantically ... a composite is a syntagma consisting of a determinant and a determinatum ... Both parts are morphemes, i.e. signs based on a significate/significant relation. In the event of full compounds, the syntagma is opposable to either morphemic element (head-ache to head and ache). Prefixal and suffixal composites must be opposable to their unprefixed and unsuffixed bases (un-do to do, father-hood to fat father), and to other composites containing the same dependent morpheme (un-do to un-fasten, un-roll, etc., father-hood to mother-hood, boy-hood, etc.) (Marchand 1969: 2-3).

From these basic assumptions several interesting consequences result. First of all, syntagmas are in principle binary structures; non-binary structures are the exception. Secondly, syntagmas are headed constructions, i.e. they consist of a head (determinatum) and a modifier (determinant). This was, incidentally, also recognised by American structuralist Immediate Constituent analysis in connection with endocentric constructions (cf., e.g., Hockett 1958: 184ff.), which where treated as headed, as against exocentric ones, which were not. This insight could at first not be incorporated into generative grammar, since the phrase structure rules of the Standard Theory generated syntactic base structures without specifying a head. And since originally word-formations were transformationally derived from such syntactic deep structures (cf. Lees 1960 and the ensuing "transformational hypothesis", cf. Kastovsky 1982: ch. 6), they could also not be described as headed in this framework. The notion of headedness was even tually introduced into the generative description of word-structures by Williams (1981), but without any reference to its pre-existence outside X'-syntax.

3.2. The centrality of the syntagma concept and of the sign-constitutive role of oppositions also has consequences for the handling of conversions (derivations without an overt derivational marker), i.e. pairs such as [cheat.sub.V]: [cheat.sub.N], [father.sub.N] : [father.sub.V], [clean.sub.Adj]: [clean.sub.V]. These derivatives have to be regarded as syntagmas, too, and since they stand in a formal-semantic opposition to parallel explicit syntagmas with overt suffixes, they have to be analysed not just as conversions or functional shifts from one word-class to another, but as derivatives containing a zero morpheme instead of an overt derivational suffix (10) cf.

(3) cheae-[[empty set].sub.N] (: wrft-[er.sub.N]), father-'[[empty set].sub.V] (: deput-[ise.sub.V]), clean-[[empty set].sub.V] (: dark-[en.sub.V])

This also holds for the so-called pseudo-compounds or exocentric compounds of the type

(4) paleface-[[empty set].sub.N] pickpocket-'[[empty set].sub.N] (: backbench-[er.sub.N]),

which for the same reason are treated as zero-derivatives in Marchand (cf. Marchand 1969: 380ff).

The notion of opposition, which establishes parallels between related structures not only on a formal, but also on a semantic basis, (11) and the syntagma concept as a frame of reference for oppositions, were not part of the American linguistic tradition. It is not surprising, therefore, that the formations in (3) and (4) were often treated as totally different from regular word-formation patterns. Thus examples such as (4) were interpreted as metaphorical extensions of regular compounds, i.e. as just involving a shift of their referential domain. And the examples in (3) were occasionally regarded as in principle non-derivational and interpreted as related by non-directional redundancy rules in the lexicon, cf. Lieber (1981) and others. The problem here is not so much whether one accepts the device of zero derivation or not - this is basically a metatheoretical-formal question; rather, the issue involved is whether conversion (or whatever one might want to call this process) is derivational or not. In a synta gma-based word-formation theory, formations such as those in (3) are necessarily binary on account of the parallels to explicit derivatives, i.e. they contain some kind of implicit derivational element. Note, incidentally, that with most of these zero patterns at some historically earlier stage an overt antecedent of zero can be found, which was subsequently lost by phonological attrition, cf. Kastovsky (1980; 1996). In a basically formally oriented approach, on the other hand, one would indeed be inclined to treat such formations as totally different from suffixal derivatives, despite the semantic parallels between the two types of derivation, which would be lost in such an analysis.

3.3. In this connection, another peculiar consequence of treating the examples in (4) as related by non-directional redundancy rules in the lexicon, and not as derivationally connected, should be mentioned, viz, the analysis of prefixations such as

(5) en-noble, de-louse, un-saddle, dis-bar

Lieber (1981: 47ff., 1983: 252ff.) postulated a set of percolation. rules, which transfer the properties of the final nodes filled with lexical entries to the dominating node, thus ultimately accounting for the overall properties of the complex lexical item in question. She furthermore assumes that all affixes with the exception of those belonging to the so-called "null-category class" (e.g. counter-in counter-attack), which do not change the word class, act as heads of the respective combination and percolate their features to the next highest branch. Since in (5) the bases are adjectives or nouns, but the resulting combinations are verbs, and since there are no other affixes present except the prefixes, it follows automatically that the prefixes must be considered the heads of the combinations effecting the category change from adjective/noun to verb. There are several rather serious problems with this purely formal, mechanistic approach, which is ultimately due to the fact that conversion is not treated as a derivational process.

First, Germanic morphology is generally right-headed, even in those instances where a semantic analysis would seem to suggest the inverse sequence of head/modifier, as, e.g., with diminutives or sex-denoting suffixes. Thus, a formation such as Taschchen 'baglet' would seem to be most naturally interpretable semantically as 'little bag', and stewardess as 'female steward', i.e. the suffix looks like the modifier. Nevertheless, from a morphological point of view it acts as head and determines the overall morphological and syntactic properties of the derivative (gender, inflectional class, pronominal reference, etc.), which are different from those of the nominal base, cf. the following examples, where the apparent semantic analysis is contrasted with the morphological one:

(6) a. Tasch / chen                       duke / let
       modifier / head/       morphology  modifier / head/
       head / modifier        semantics   head / modifier
       'bag' 'little'                     'duke' 'petty'
       (= 'little bag')                   (= 'petty duke')
    b. Wirt / in                          steward / ess
       modifier / head/       morphology  modifier / head/
       head / modifier        semantics   head / modifier
       'landlord' 'female'                'steward' 'female'
       (= 'female landlord')              (= 'female steward')

Secondly, Lieber's analysis is internally contradictory. Thus there are formations containing both a prefix and a suffix, and with these it would not be clear which of the two should be regarded as the head, cf.

(7) em-bold-en, de-sulphur-ate, de-gas-ify

Thirdly, the formations in (5) and (7) are also rivalled by simple zero derivations, cf.

(8) em-bold-en: en-noble : clean; de-sulphur-ate: de-louse: gut; en-capsul-ate : en-cage : bottle

which poses the question as to the morphological-semantic relationship between these three types of formation.

All these problems can of course be solved quite easily if on the one hand we treat simple zero derivation (conversion) as on a par with other derivational processes, and if we assume that zero derivation is also involved in those instances where we only have prefixes as overt derivational elements: in these formations, we have a cooperation of prefixation with zero derivation, just as we have a cooperation of prefixation and suffixation in (7). This, however, entails that the syntagmatic approach to word-formation, and with it the sign-based framework as a whole, is preferable to a purely formal approach, which tends to disregard the indispensible presence of a signifie.

(1.) This paper is based on some ideas that I had the privilege of discussing with Roger Lass and his team in Cape Town in October 1991 during a staff seminar. Since Roger at the time expressed interest in some of my ideas, I offer this little contribution as a thank-you note -- not only for the wonderful time in Cape Town -- but also for the many years of friendship and fruitful co-operation. Unfortunately, some more mundane (liquid and solid) aspects of all this friendship can't be included in the publication but will be remembered nevertheless (and hopefully repeated). And so says BKK.

(2.) But cf., e.g., Gorska (1994), Lipka (1996), Mettinger (1996), Szymanek (1988), Zbierska-Sawala (1993) and some others.

(3.) Cf.: "Since there is no independently known structure of meanings which exactly parallels linguistic structure, we cannot mix distributional investigations with occasional assists from meaning whenever the going is hard" (Harris 1954: 152).

(4.) Another sign-oriented approach, popular especially in Romance linguistics, is the one developped by Coseriu (1977, 1982) and his pupils, e.g. Dietrich (1994), Laca (1986, 1994), Ludtke (1978). This framework, however, does not regard the formal-morphological and the semantic aspects as equally relevant, but accords the semantic aspect absolute priority, which to me seems as problematic as the negligence of the semantic aspect in predominantly form-oriented approaches, cf. Kastovsky (1995, forthcoming).

(5.) Cf., e.g., Bloomfield (1935: 109, 242), Harris (1951: 161; 1954: 152), Nida (1949: 162, 191) and others.

(6.) Lees (1960), although also dealing with compounds and derivatives, i.e. word-formation, is primarily a contribution to the syntactic, and not the morphological aspects of this field.

(7.) In actual fact, this is an Austrian newspaper coinage prompted by the discovery of a mouse in the bun that forms part of this delicious dish, but it shows the potential of the pattern.

(8.) I assume that Watergate as the name of a hotel in Washington D. C. is idiomatised and has the same status as butterfly, i.e. the apparent constituents are pseudo-signs, cf. below.

(9.) Other terms used by him are "pseudo-morpheme" and "pseudo-sign", cf. Marchand (1969: 122, 429, 437).

(10.) Cf. Marchand (1969: 359); for an extensive discussion of the relevance of zero in world-formation and morphology, cf. Kastovsky (1968: 31-53; 1980; 1996).

(11.) Note that the notion of contrast, which is used in American structuralism instead of "opposition", is again geared primarily to distribution and formal differences rather than semantic ones.

REFERENCES

Bauer, Laurie

1993 English word-formation. Cambridge: CUP.

Bloch, Bernard -- George L. Trager 

1947 Outline of linguistic analysis. Baltimore, Md.: Linguistic Society of America.

Bloomfield, Leonard 

1935 Language. London: Allen and Unwin.

Brekle, Herbert E. -- Dieter Kastovsky (eds.)

1977 Perspektiven der Wortbildungsforschung. Beitrage zum Wuppertaler Wortbildungskolloquium vom 9.-10. Juli 197S. Anla[beta] des 70. Geburtstages von Hans Marchand am 1. Oktober 1977. (Wuppertaler Schriftenreihe Linguistik 1.)

Britton, Derek (ed.)

1996 English historical linguistics 1994. (Current Issues in Linguistic Theory 135.) Amsterdam: Benjamins.

Chomsky, Noam -- Morris Halle

1968 The sound pattern of English. New York: Harper and Row.

Coseriu, Eugenio

1977 "Inhaltliche Wortbildungslehre (am Beispiel des Typs "coupe-papier")", in: Herbert E. Brekle -- Dieter Kastovsky (eds.), 48-61.

1982 "Les procedes semantiques dans la formation des mots", Cahiers Ferdinand de Saussure 35: 3-16.

Dietrich, Wolf

1994 "Grundfragen einer funktionellen Wortbildungslehre, dargestellt am Beispiel derata und -aria-Bildungen im Franzosischen und Spanischen", in: Bruno Staib (ed.), 93-111.

Fisiak, Jacek (ed.)

1980 Historical morphology. (Trends in Linguistics. Studies and Monographs 17.) The Hague: Mouton de Gruyter.

Gorska, Elzbieta

1994 "Moonless nights and smoke-free cities, or: What can be without what? A cognitive study of privative adjectives in English", Folia Linguistica 18: 413-435.

Halle, Morris

1973 "Prolegomena to a theory of word-formation", Linguistic Inquiry 4: 3-16.

Harris, Zellig S.

1951 Structural linguistics. Chicago: Chicago University Press.

1954 "Distributional structure", Word 10: 147-162.

Hockett, Charles

1958 A course in modern linguistics. New York: Macmillan.

Hoinkes, Ulrich (ed.)

1995 Panorama der lexikalischen Semantik. Thematische Festschrift aus Anla[beta] des 60. Geburtstags von Horst Geckeler. Tubingen Narr.

Hoinkes, Ulrich -- Dietrich Wolf (eds.)

forthcoming Kaleidoskop der lexikalischen Semantik. Tubingen: Narr.

Kastovsky, Dieter

1968 Old English deverbal substantives derived by means of a zero morpheme. Esslingen/N.: Langer.

1980 "Zero in morphology. A means of making up for phonological losses?", in: Jacek Fisiak (ed.), 213-250.

1982 Wortbildung und Semantik. (Studienreihe Englisch 14.) Tubingen/Dusseldorf: Francke/Bagel.

1995 "Wortbildungssemantik, ein historischer Lagebericht", in: Ulrich Hoinkes (ed.), 385-398.

1996 "Verbal derivation in English: A historical survey. Or: Much ado about nothing", in: Derek Britton (ed.), 93-117.

forthcoming "Einige Uberlegungen zur Wortbildungstheorie", in: Ulrich Hoinkes -- Dietrich Wolf (eds.).

Laca, Brenda

1986 Die Wortbildung als Grammatik des Wortschatzes. Untersuchungen zur spanischen Subjektnominalisierung. (Tubinger Beitrage zur Linguistik 286.) Tubingen: Narr.

1994 "Un probleme de semantique suffixale: Les derives espagnoles et francais en-ero/-ier", in: Bruno Staib (ed.), 63-91.

Lees, Robert B.

1960 The grammar of English nominalisations. Bloomington, Ind.: Indiana University Press.

Lieber, Rochelle

1981 On the organisation of the lexicon. Bloomington, Ind.: Indiana University Press.

1983 "Argument linking and compounds in English", Linguistic Inquiry 14: 251-286.

1996 "Words, metaphors and cognition: A bridge between domains", in: Jan Svartvik (ed.), 49-69.

Lipka, Leonhard

Ljung, Magnus

1975 Review of Marchand 1974. Foundations of Language 13: 475-480.

Ludtke, Jens

1978 Prudikative Nominalisierungen mit Suffixen im Franzosischen, Katalanischen und Spanischen. Tubingen: Niemeyer.

Marchand, Hans

1969 The categories and types of present-day English word-formation. (2nd edition.) Munchen: Beck.

1974 Studies in syntax and word-formation. Selected articles. On the occasion of his 65th birthday on October 1st, 1972. (edited by Dieter Kastovsky.) (Internationale Bibliothek fur allgemeine Linguistik 18.) Munchen: Fink.

Mettinger, Arthur

1996 "Cognitive linguistics and lexicography: Suggestions for a diagrammatic representation of un-Adj-formations in a monolingual dictionary", in: Arne Zettersten - Viggo Hjomager Pedersen (eds.), 199-211.

Nida, Eugene A.

1949 Morphology the descriptive analysis of words. (2nd edition.) Ann Arbor, Mi.: University of Michigan Press.

Saussure, Ferdinand de 

1973 Cours de linguistique generale. Paris: Payot.

[1916]

Staib, Bruno (ed.)

1994 Wortbildungslehre. (Munstersches Logbuch zur Linguistik 5.) Munster. LIT.

Svartvik, Jan (ed.)

1996 WORDS. Proceedings of an international symposium. Lund, 25-26 August 1995. (Konferenser 36.) Stockholm: Kungl. Vitterhets Historie och Antikvitets Akademien.

Szymanek, Bogdan

1988 Categories and categorization in morphology Lublin: Redakeja Wydawnictw Katolickiego Uniwersytetu Lubelskiego.

Williams, Edwin

1981 "On the notions 'lexically related' and 'head of a work'", Linguistic Inquiry 12: 245-274.

Winter, Werner

1964 "Form and meaning in morphological analysis", Linguistics 3: 5-18.

Zbierska-Sawala, Anna

1993 Early Middle English word formation. Semantic aspects of derivational affixation in the AB language. (Bamberger Beitrage zur Englischen Sprachwissenschaft 32.) Frankfurt: Lang.

Zettersten, Arne - Viggo Hjomager Pedersen (eds.)

1996 Symposium on lexicography VII. Proceedings of the Seventh Symposium on Lexicography May 5-6, 1994 at the University of Copenhagen. (Lexicographica. Series Maior 76.) Tubingen: Niemeyer.