Dieter Kastovsky


1.1. It is often assumed that the vocabulary of a language, and, correspondingly, the “lexicon” as that subcomponent of the grammar which formally represents the lexical competence of a native speaker, are structured by two unconnected and quite different organizational principles: a semantic and a formal-morphological one. Semantic structures result from the existence of various kinds of sense relations between lexical items, or, rather, the meanings of lexical items, on the basis of which one obtains sets of lexemes sharing a common basic meaning. These sets are usually referred to as lexical fields. Formal-morphological structures derive from the ability of already existing lexical items to combine with other lexical items or with bound morphemes (prefixes, suffixes) forming morphologically complex new lexical items. These processes, i. e. compounding, prefixation, suffixation, etc., characterize the field of word-formation, and they are usually regarded as a means of extending the vocabulary almost without limits in order to adapt it to the ever-changing referential requirements of a speech community. This leads to a formal division of the vocabulary into primary and secondary lexemes, to take up a terminological distinction suggested by Coseriu. Primary lexemes, e. g. big, mountain, give, in,etc., are simple, arbitrary linguistic signs in the sense of Saussure. Secondary lexemes, e. g. spaceship, steamboat, sailing boat, rewrite, atomize, whiten, rider, departure, disestablish-mentarianism,etc., are lexical syntagmas. As such they are characterized by a determinant/determinatum relation; they are relatively motivated with regard to their constituents and parallel formations; and they are based on certain morphological, semantic, and syntactic patterns.

1.2. While the morphological and the semantic aspects of word-formation are quite obvious and uncontested, its syntactic aspect is by no means uncontroversial. This is most obvious in the controversy between the so-called transformational and lexicalist hypotheses in generative-transformational grammar. The former relates word-formation directly to the syntactic rules of the grammar by deriving word-formation syntagmas transformationally from underlying representations which are identical or at least closely related to those also underlying syntactic constructions such as relative or complement clauses, infinitives, gerunds, etc. Accordingly, (l a) and (l b — e), and (2 a) and (3 b, c), respectively, would have similar underlying representations:

(1) a. Peter regretted Harriet’s early departure,

b. Peter regretted Harriet’s departing early.

с. Peter regretted Harriet having departed early.

d. Peter regretted that Harriet had departed early.

e. Peter regretted the fact that Harriet had departed early.

(2) a. I know an alleged discoverer of time-travel.

b. I know someone having allegedly discovered travelling through time.

c. I know someone of whom it is alleged that he has discovered that / how one can travel through time.

Adherents of the lexicalist hypothesis, on the other hand, would relegate such relationships to the lexicon, regarding word-formation as a purelylexical matter.

These two positions do not constitute genuine alternatives, however, but only reflect the inherently ambivalent position of word-formation with regard to syntax and the lexicon. Thus there are numerous phenomena similar to (1) and (2) which require a syntactic explanation and can best be handled transformationally. The close relationship of word-formation to the lexicon, which cannot be denied, of course, can then be regarded as a kind of spin-off effect due to the format of the output of word-formation rules: in contradistinction to regular syntactic transformations, it consists of lexical items (lexemes), which can readily be incorporated into the lexicon as fixed units and can therefore also be recalled as such. This aspect accounts for the view that word-formation is a means of systematically enriching the lexicon. Thus both aspects, the syntactic and the morphological one, are equally important.

1.3. In descriptions of the lexicon, semantic and formal-morphological structures are as a rule kept strictly apart. Thus Coseriu regards both lexical semantics and word-formation as legitimate parts of “lexematics”, as he calls the functional analysis of the vocabulary. Nevertheless, he treats them as two basically unrelated phenomena, viz. as primary paradigmatic lexical structures or lexical fields and secondary paradigmatic lexical structures or word-formation. Both are kept apart from a third type of relation between lexical items, viz. syntagmatic lexical structures or lexical solidarities. These describe the same kind of phenomenon as the selection restrictions in generative grammar, although from a purely semantic rather than a syntactic point of view. Thus no direct relationship between these subdivisions is assumed, although Coseriu does not deny that word-formations may function in lexical fields alongside primary lexemes. But this is regarded as a by-product only of their integration into the lexicon, while the structures themselves are strictly kept apart.

2.1.1. I will begin with some observations from historical linguistics, which already provide a good indication of the kind of relationship that obtains between primary and secondary paradigmatic lexical structures.

2.1.2. These observations relate to the fact that there are transitions between these types of structures in both directions.

2.1.3. One factor involved in a lexeme’s transition from monomorphemic to polymorphemic, i. e. syntagmatic status is folk-etymology. This refers to the reanalysis of a monomorphemic lexical item (moneme) as consisting of more than one morpheme on the basis of phonetic associations with morphemes resembling parts of the reinterpreted item. Semantic analogy may also play a role in this process, which is still little understood. Examples are

(3) OE scamfæst Mod.E shamefaced
asparagus sparrow-grass
E hammock G Hängematte

2.2.1. Similar processes are involved in the reinterpretation of the direction of the derivation with backformations. A case in point is the relationship between peddle and peddler/pedlar. Historically, peddle vb was backderived from the originally monomorphemic noun peddler with the meaning ‘act as peddler’. In the course of time, this relationship came to be reversed by analogy with the usual pattern write: writer. As a result, peddler was reinterpreted as a bimorphemic lexeme meaning ‘someone who peddles’, i. e. as a derivative from peddle, which was originally the derivative but must now be regarded as a moneme acting as the basis for the derivative peddler.

2.2.2. These phenomena, however, concern individual cases only and are not pattern-forming. Closely related, but much more important, are the processes which lead to the adoption of a foreign word-formation pattern and to its becoming productive in the target language.

English prefixes and suffixes of Latin or French origin such as de-, со-, dis-, in-, re-; -able, -ize, -ify, -ive, -ation, etc. were not borrowed into English directly as isolated morphemes. Rather, they were taken over as constituents of foreign word-formation syntagmas which were borrowed into English, such as deplume, decipher, co-author, disallow, recharge, acceptable, harmonize, edification, etc. But these loans could be recognized as word-formations in English only if the corresponding bases had also been borrowed. Thus acceptable would be a moneme in English if the verbal base accept had not been borrowed as well. Only then did a derivative relationship develop between accept and acceptable,and acceptable could be interpreted as a syntagma. And it was only on the basis of such pairs that these affixes could become productive in English itself. There are two aspects to this phenomenon which are of particular interest in this connection.

First of all, often a Latin or French derivative was borrowed before its base was also adopted. Consequently, such loans lost their status as word-formation syntagmas in the process of borrowing and became monemes in English, until their bases were also taken over. Only then did they regain their syntagmatic status. There is thus a constant give and take between primary and secondary lexemes from this point of view, which is only possible, however, if the semantic structures of simple lexical items and word-formation syntagmas closely resemble each other.

Secondly, the base of the original derivative may not have been borrowed, as was the case with laudable, magnify, pensive, receive, discern, inert, inane,which have therefore remained unanalysable monemes in English. But, on the other hand, a word like laudable,due to its origin as the Latin derivative laudabilis from laudare,will have a semantic structure which is completely analogous to the semantic structure of a syntagma such as acceptable. So here again we have a strong indication that the semantic structures of primary and secondary lexemes must be extremely similar.

2.3.1. In contradistinction to the phenomena described so far, the transition from a motivated syntagma to a moneme is a gradual process. Synchronically, we are confronted with a scale or cline of various degrees of motivation; diachronically this means that a word-formation syntagma may move along this scale from complete motivation to complete arbitrariness. This development is triggered by the lexicalization of word-formations, i. e. their incorporation into the general, accepted vocabulary in a fixed, often somewhat idiosyncratic meaning.

2.3.2. The specialization of meaning often accompanying the lexicalization of a word-formation syntagma may be due either to the addition of certain semantic components to the syntagma as a whole, or to some change in the meaning of the constituents, or both. As a result, the overall meaning of the respective word-formation can no longer be deduced completely from the meanings of the constituents and the structural meaning of the word-formation pattern; rather, additional information is required, which constitutes the first step towards complete idiomatization. This can be illustrated by the following examples.

For the correct interpretation of rattlesnake we need nothing more than the simple syntactic paraphrase ‘snake which can rattle’. But for callboy or callgirl a definition as ‘boy who calls’ and ‘girl who is called’, respectively, is by no means sufficient. The restriction to these two paraphrases is already an indication of their lexicalization and the beginnings of idiomatization, because these two nouns could theoretically just as well mean ‘boy who is called’ and ‘girl who calls’. In order to understand callboy correctly, one has to know that this formation refers to someone who summons the actors onto the stage; and a callgirl is by no means just any girl one talks to on the phone, but rather some female one calls up for some very specific, well-known purpose. It would not do to call one’s girlfriend a callgirl just because one happens to call her up once in a while and invites her home. Idiomatization has progressed even further in the case of blackboard due to a change in the object referred to by this compound: blackboards today are usually green and not black. And the compound butterfly is completely idiomatic, since its meaning no longer has anything to do with the meanings of its constituents. From a purely formal point of view, butterfly is still a syntagma; from the point of view of the relation between meaning and form it behaves like a simple lexical item. This borderline case thus has the same status as syntactic idioms like kick the bucket, spill the beans, pull someone’s leg,etc.

Idiomatization can be accompanied by formal demotivation, which results in a moneme on the morphological level, too, e. g.

(4) OE hlāfweard ‘loaf-guard’ lord

OE hlæfdīge ‘loaf-kneader’ lady

OE hūswīf ‘housewife’ hussy

OE freond ‘friend’ freon ‘love’ friend

OE feond ‘enemy’ feon ‘hate’ fiend.

Friend and fiend have lost their syntagmatic status because their verb­al bases no longer exist and, moreover, the derivative pattern has also become extinct.

It is not likely, however, that this process will have fundamentally altered the underlying semantic structures, although some modifications have of course taken place: all the words in (4) continue to refer to persons, butterfly still denotes some kind of insect, etc. Thus, from a diachronic point of view there is considerable evidence for the assumption that on the semantic level there exist close parallels between primary and secondary lexical structures.

3.1. Turning now to synchronic considerations, the same conclusion results from a comparison of lexical solidarities and word-formation syntagmas. Lexical solidarities are defined by Coseriu as syntagmatic implications holding between various types of lexical structures, which are due to the fact that the meaning of some lexeme, or of a whole lexical field (i. e. of an archilexeme) is contained in the meaning of some other lexeme where it functions as a semantic component. In this sense, the verbs bark, neigh, miaow imply the nouns dog, horse, cat as agents; fell implies tree as object; see, look and hear, listen imply eye and ear,res­pectively, as instruments; similarly, kiss implies lips, sweep implies broom,etc. Bally talks in this connection of implicit syntagmas or of “motivation par cumul des signifies”; Gruber uses the term “incorporation”, and Lyons the term “encapsulation” for basically the same phenomena.

Now Porzig, who was probably the first to call attention to this phenomenon, referring to it as “wesenhafte Bedeutungsbeziehungen”, pointed out that pairs like hämmern: Hammer (hammer vb: hammer sb) etc. exhibit the same type of relation. Coseriu, however, rejects this partial identification of word-formation processes and lexical solidarities. He regards word-formation as a primarily paradigmatic process having a syntagmatic basis, e. g. mit dem Hammer + verbalization → hämmern, in contradistinction to lexical solidarities, which are basically a syntagmatic phenomenon caused by paradigmatic oppositions, e. g. schneiden + Zahn = beßen, schneiden + Baum = fällen, schneiden + Gras/Getreide = mähen. From a purely analytic point of view focussing on morphological distinctions, this may be justified. From a genuinely semantic point of view, however, this strict separation strikes me as unsatisfactory and incoherent. Rather, it seems to me that these two phenomena are much more closely related than Coseriu is inclined to admit. In fact, pairs like hammer vb: hammer sb, shovel vb: shovel sb, telephone vb: telephone sb, knife vb: knife sb, etc. differ from lexic­al solidarities only in that the semantic implication is accompanied by a formal implication. In word-formation syntagmas, one lexeme is contained in the other both semantically and formally, and it is this kind of formal-semantic implication which is the essential characteristic of word-formation. Conversely, we might speculate that the sense-relations characterizing the structure of the primary vocabulary are taken up in word-formation and are made explicit by corresponding formal relations, which then accounts for the relatively motivated character of the resulting lexical items.