C H A P T E R I
The ways in which new words are formed, and the factors which govern their acceptance into the language, are generally taken very much for granted by the average speaker. To understand a word, it is not necessary to know how it is constructed, whether it is simple or complex, that is, whether or not it can be broken down into two or more constituents. We are able to use a word which is new to us when we find out what object or notion it denotes. Some words, of course, are more ‘transparent’ than others. For example, in the words unfathomable and indescribable we recognize the familiar pattern of negative prefix + transitive word + adjective-forming suffix on which many words of similar form are constructed. Knowing the pattern, we can easily guess their meanings – ‘cannot be fathomed’ and ‘cannot be described’ – although we are not surprised to find other similar-looking words, for instance unfashionable and unfavourable for which this analysis will not work. We recognize as ‘transparent’ the adjectives unassuming and unheard-of , which taking for granted the fact that we cannot use assuming and heard-of . We accept as quite natural the fact that although we can use the verbs to pipe , to drum and to trumpet , we cannot use the verbs to piano and to violin .
But when we meet new coinages, like tape-code , freak-out , shutup-ness and beautician , we may not readily be able to explain our reactions to them. Innovations in vocabulary are capable of arousing quite strong feelings in people who may otherwise not be in the habit of thinking very much about language. Quirk quotes some letter to the press of a familiar kind, written to protest about ‘horrible jargon’, such as breakdown , ‘vile’ words like transportation , and the ‘atrocity’ lay-by .
Many linguists agree over the fact that the subject of word-formation has not until recently received very much attention from descriptive grammarians of English, or from scholars working in the field of general linguistics. As a collection of different processes (compounding, affixation, conversion, backformation, etc.) about which, as a group, it is difficult to make general statements, word-formation usually makes a brief appearance in one or two chapters of a grammar. Valerie Adams emphasizes two main reasons why the subject has not been attractive to linguists: its connections with the non-linguistic world of things and ideas, for which words provide the names, and its equivocal position as between descriptive and historical studies. A few brief remarks, which necessarily present a much over-simplified picture, on the course which linguistics has taken in the last hundred years will make this easier.
The nineteenth century, the period of great advances in historical and comparative language study, saw the first claims of linguistics to be a science, comparable in its methods with the natural sciences which were also enjoying a period of exciting discovery. These claims rested on the detailed study, by comparative linguists, of formal correspondences in the Indo-European languages, and their realization that such study depended on the assumption of certain natural ‘laws’ of sound change. As Robins observes in his discussion of the linguistics of the latter part of the nineteenth century:
The history of a language is traced through recorded variations in the forms and meanings of its words, and languages are proved to be related by reason of their possession of worlds bearing formal and semantic correspondences to each other such as cannot be attributed to mere chance or to recent borrowing. If sound change were not regular, if word-forms were subject to random, inexplicable, and unmotivated variation in the course of time, such arguments would lose their validity and linguistic relations could only be established historically by extralinguistic evidence such as is provided in the Romance field of languages descended from Latin.
The rise and development in the twentieth century of synchronic descriptive linguistics meant a shift of emphasis from historical studies, but not from the idea of linguistics as a science based on detailed observation and the rigorous exclusion of all explanations depended on extralinguistic factors. As early as 1876, Henry Sweet had written:
Before history must come a knowledge of what exists. We must learn to observe things as they are, without regard to their origin, just as a zoologist must learn to describe accurately a horse or any other animal. Nor would the mere statements that the modern horse is a descendant of a three-toed marsh quadruped be accepted as an exhausted description... Such however is the course being pursued by most antiquarian philologists.
The most influential scholar concerned with the new linguistics was Ferdinand de Saussure, who emphasized the distinction between external linguistics – the study of the effects on a language of the history and culture of its speakers, and internal linguistics – the study of its system and rules. Language, studied synchronically, as a system of elements definable in relation to one another, must be seen as a fixed state of affairs at a particular point of time. It was internal linguistics, stimulated by de Saussure’s works, that was to be the main concern of the twentieth-century scholars, and within it there could be no place for the study of the formation of words, with its close connection with the external world and its implications of constant change. Any discussion of new formations as such means the abandonment of the strict distinction between history and the present moment. As Harris expressed in his influential Structural Linguistics : ‘The methods of descriptive linguistics cannot treat of the productivity of elements since that is a measure of the difference between our corpus and some future corpus of the language.’ Leonard Bloomfield, whose book Language was the next work of major influence after that of de Saussure, re-emphasized the necessity of a scientific approach, and the consequent difficulties in the way of studying ‘meaning’, and until the middle of the nineteen-fifties, interest was centered on the isolating of minimal segments of speech, the description of their distribution relative to one another, and their organization into larger units. The fundamental unit of grammar was not the word but a smaller unit, the morpheme.
The next major change of emphasis in linguistics was marked by the publication in 1957 of Noam Chomsky’s Syntactic Structures . As Chomsky stated it, the aim of linguistics was now seen to be ‘to make grammatical explanations parallel in achievement to the behavior of the speaker who, on the basis of a finite and accidental experience with language can produce and understand an indefinite number of new sentences’ . The idea of productivity, or creativity, previously excluded from linguistics, or discussed in terms of probabilities in the effort to maintain the view of language as existing in a static state, was seen to be of central importance. But still word-formation remained a topic neglected by linguists, and for several good reasons. Chomsky made explicit the distinction, fundamental to linguistics today (and comparable to that made by de Saussure between langue , the system of a language, and parole , the set of utterances of the language), between linguistic competence, ‘the speaker-hearer’s knowledge of his language’ and performance, ‘the actual use of language in concrete situations’ . Linked with this distinction are the notions of ‘grammaticalness’ and ‘acceptability’; in Chomsky’s words, ‘Acceptability is a concept that belongs to the study of competence’ . A ‘grammatical’ utterance is one which may be generated and interpreted by the rules of the grammar; an ‘acceptable’ utterance is one which is ‘perfectly natural and immediately comprehensible... and in no way bizarre or outlandish’ . It is easy to show, as Chomsky does, that a grammatical sentence may not be acceptable. For instance, this is the cheese the rat the cat caught stole appears ‘bizarre’ and unacceptable because we have difficulty in working it out, not because it breaks any grammatical rules. Generally, however, it is to be expected that grammaticalness and acceptability will go hand in hand where sentences are concerned.
The ability to make and understand new words is obviously as much a part of our linguistic competence as the ability to make and understand new sentences, and so, as Pennanen points out, ‘it is an obvious gap in transformational grammars not to have made provision for treating word-formation.’ But, as we have already noticed, we may readily thing of words, like to piano and to violin , against which we can invoke no rule, but which are definitely ‘unacceptable’ for no obvious reason. The incongruence of grammaticality and acceptability that is, is far greater where words are concerned than where sentences are concerned. It is so great, in fact, that the exercise of setting out the ‘rules’ for forming words has so far seemed to many linguists to be out of questionable usefulness. The occasions on which we would have to describe the output of such rules as ‘grammatical but non-occurring’ are just too numerous. And there are further difficulties in treating new words like new sentences. A novel word (like handbook or partial ) may attract unwelcome attention to itself and appear to be the result of the breaking of rules rather than of their application. And besides, the more accustomed to the word we become, the more likely we are to find it acceptable, whether it is ‘grammatical’ or not – or perhaps we should say, whether or not is was ‘grammatical’ at the time it was first formed , since a new word once formed, often becomes merely a member of an inventory; its formation is a historical event, and the ‘rule’ behind it may then appear irrelevant.
What exactly is a word? From Lewis Carroll onwards, this apparently simple question has bedeviled countless word buffs, whether they are participating in a game of Scrabble or writing an article for the Word Ways linguistic magazine. To help the reader decide what constitutes a word, A. Ross Eckler suggests a ranking of words in decreasing order of admissibility. A logical way to rank a word is by the number of English-speaking people who can recognize it in speech or writing, but this is obviously impossible to ascertain. Alternatively, one can rank a word by its number of occurrences in a selected sample of printed material. H. Kucera and W.N. Francis's Computational Analysis of Present-day English is based on one million words from sources in print in 1961. Unfortunately, the majority of the words in Webster's Unabridged do not appear even once in this compilation – and the words which do not appear are the ones for which a philosophy of ranking is most urgently needed. Furthermore, the written ranking will differ from the recognition ranking; vulgarities and obscenities will rank much higher in the latter than in the former.
A detailed, word-by-word ranking is an impossible dream, but a ranking based on classes of words may be within our grasp. Ross Eckler proposes the following classes: (1) words appearing in one more standard English-language dictionaries, (2) non-dictionary words appearing in print in several different contexts, (3) words invented to fill a specific need and appearing but once in print.
Most people are willing to admit as words all uncapitalized, unlabeled entries in, say, Webster's New International Dictionary, Third Edition (1961). Intuitively, one recognizes that words become less admissible as they move in any or all of three directions: as they become more frequently capitalized, as they become the jargon of smaller groups (dialect, technical, scientific), and as they become archaic or obsolete. These classes have no definite boundaries – is a word last used in 1499 significantly more obsolete than a word last used in 1501? Is a word known to 100,000 chemists more admissible than a word known to 90,000 Mexican-Americans? Each linguist will set his own boundaries.
The second class consists of non-dictionary words appearing in print in a number of sources. There are many non-dictionary words in common use; some logologists would like to draw a wider circle to include these. Such words can be broadly classified into: (1) neologisms and common words overlooked by dictionary-makers, (2) geographical place names, (3) given names and surnames.
Dmitri Borgmann points out that the well-known words uncashed , ex-wife and duty-bound appear in no dictionaries (since 1965, the first of these has appeared in the Random House Unabridged). Few people would exclude these words. Neologisms present a more awkward problem since some may be so ephemeral that they never appear in a dictionary. Perhaps one should read Pope's dictum "Be not the first by whom the new are tried, nor yet the last to lay the old aside."
Large treasure-troves of geographic place names can be found in The Times Atlas of the World (200,000 names), and the Rand McNally Commercial Atlas and Marketing Guide (100,000 names). These are not all different, and some place names are already dictionary words. All these can be easily verified by other readers; however, some will feel uneasy about admitting as a word the name, say, of a small Albanian town which possibly has never appeared in any English-language text outside of atlases.
Given names appear in the appendix of many dictionaries. Common given names such as Edward or Cornelia ought to be admitted as readily as common geographical place names such as Guatemala, but this set does not add much to the logological stockpile.
Family surnames at first blush appear to be on the same footing as geographical place names. However, one must be careful about sources. Biographical dictionaries and Who's Who are adequate references, but one should be cautious citing surnames appearing only in telephone directories. Once a telephone directory is supplanted by a later edition, it is difficult to locate copies for verifying surname claims. Further, telephone directories are not immune to nonce names coined by subscribers for personal reasons. A good index of the relative admissibility of surnames is the number of people in the United States bearing that surname. An estimate of this could be obtained from computer tapes of the Social Security Administration; in 1957 they issued a pamphlet giving the number of Social Security accounts associated with each of the 1500 most common family names.
The third and final class of words consists of nonce words, those invented to fill a specific need, and appearing only once (or perhaps only in the work of the author favoring the word). Few philologists feel comfortable about admitting these. Nonce words range from coinages by James Joyce and Edgar Allan Poe (X-ing a Paragraph) to interjections in comic strips (Agggh! Yowie!). Ross Eckler and Daria Abrossimova suggest that misspellings in print should be included here also.
In the book “Beyond Language”, Dmitri Borgmann proposes that the philologist be prepared to admit words that may never have appeared in print. For example, Webster's Second lists eudaemony as well as the entry "Eudaimonia, eudaimonism, eudaimonist, etc." From this he concludes that EUDAIMONY must exist and should be admitted as a word. Similarly, he can conceive of sentences containing the word GRACIOUSLY'S ("There are ten graciously's in Anna Karenina") and SAN DIEGOS ("Consider the luster that the San Diegos of our nation have brought to the US"). In short, he argues that these words might plausibly be used in an English-language sentence, but does not assert any actual usage. His criterion for the acceptance of a word seems to be its philological uniqueness (EUDAIMONY is a short word containing all five vowels and Y).
The available linguistic literature on the subject cites various types and ways of forming words. Earlier books, articles and monographs on word-formation and vocabulary growth in general used to mention morphological, syntactic and lexico-semantic types of word-formation. At present the classifications of the types of word-formation do not, as a rule, include lexico-semantic word-building. Of interest is the classification of word-formation means based on the number of motivating bases which many scholars follow. A distinction is made between two large classes of word-building means: to Class I belong the means of building words having one motivating base (e.g. the noun doer is composed of the base do- and the suffix -er ), which Class II includes the means of building words containing more than one motivating base. They are all based on compounding (e.g. compounds letter-opener, e-mail, looking-glass ).
Most linguists in special chapters and manuals devoted to English word-formation consider as the chief processes of English word-formation affixation, conversion and compounding.
Apart from these, there is a number of minor ways of forming words such as back-formation, sound interchange, distinctive stress, onomatopoeia, blending, clipping, acronymy.
Some of the ways of forming words in present-day English can be restored to for the creation of new words whenever the occasion demands – these are called productive ways of forming words , other ways of forming words cannot now produce new words, and these are commonly termed non -productive or unproductive . R. S. Ginzburg gives the example of affixation having been a productive way of forming new words ever since the Old English period; on the other hand, sound-interchange must have been at one time a word-building means but in Modern English (as we have mentioned above) its function is actually only to distinguish between different classes and forms of words.
It follows that productivity of word-building ways, individual derivational patterns and derivational affixes is understood as their ability of making new words which all who speak English find no difficulty in understanding, in particular their ability to create what are called occasional words or nonce-words (e.g. lungful (of smoke), Dickensish (office), collarless (appearance)). The term suggests that a speaker coins such words when he needs them; if on another occasion the same word is needed again, he coins it afresh. Nonce-words are built from familiar language material after familiar patterns. Dictionaries, as a rule, do not list occasional words.
The delimitation between productive and non-productive ways and means of word-formation as stated above is not, however, accepted by all linguists without reserve. Some linguists consider it necessary to define the term productivity of a word-building means more accurately. They hold the view that productive ways and means of word-formation are only those that can be used for the formation of an unlimited number of new words in the modern language, i.e. such means that “know no bounds” and easily form occasional words. This divergence of opinion is responsible for the difference in the lists of derivational affixes considered productive in various books on English lexicology.
Nevertheless, recent investigations seem to prove that productivity of derivational means is relative in many respects. Moreover there are no absolutely productive means; derivational patterns and derivational affixes possess different degrees of productivity. Therefore it is important that conditions favouring productivity and the degree if productivity of a particular pattern or affix should be established. All derivational patterns experience both structural and semantic constraints. The fewer are the constraints, the higher is the degree of productivity, the greater is the number of new words built on it. The two general constraints imposed on all derivational patterns are: the part of speech in which the pattern functions and the meaning attached to it which conveys the regular semantic correlation between the two classes of words. It follows that each part of speech is characterized by a set of productive derivational patterns peculiar to it. Three degrees of productivity are distinguished for derivational patterns and individual derivational affixes: (1) highly productive , (2) productive or semi-productive and (3) non-productive .
R. S. Ginzburg says that productivity of derivational patterns and affixes should not be identified with the frequency of occurrence in speech, although there may be some interrelation between then. Frequency of occurrence is characterized by the fact that a great number of words containing a given derivational affix are often used in speech, in particular in various texts. Productivity is characterized by the ability of a given suffix to make new words.
In linguistic literature there is another interpretation of derivational productivity based on a quantitative approach. A derivational pattern or a derivational affix are qualified as productive provided there are in the word-stock dozens and hundreds of derived words built on the pattern or with the help of the suffix in question. Thus interpreted, derivational productivity is distinguished from word-formation activity by which is meant the ability of an affix to produce new words, in particular occasional words or nonce-words. For instance, the agent suffix –er is to be qualified both as a productive and as an active suffix: on the one hand, the English word-stock possesses hundreds of nouns containing this suffix (e.g. writer, reaper, lover, runner , etc.), on the other hand, the suffix –er in the pattern v + –er - N is freely used to coin an unlimited number of nonce-words denoting active agents (e.g. interrupter, respecter, laugher, breakfaster , etc.).
The adjective suffix –ful is described as a productive but not as an active one, for there are hundreds of adjectives with this suffix (e.g. beautiful, hopeful, useful , etc.), but no new words seem to be built with its help.
For obvious reasons, the noun-suffix –th in terms of this approach is to be regarded both as a non-productive and a non-active one.
Now let us consider the basic ways of forming words in the English language.
Affixation is generally defined as the formation of words by adding derivational affixes to different types of bases. Derived words formed by affixation may be the result of one or several applications of word-formation rule and thus the stems of words making up a word-cluster enter into derivational relations of different degrees. The zero degree of derivation is ascribed to simple words, i.e. words whose stem is homonymous with a word-form and often with a root-morpheme (e.g. atom, haste, devote, anxious, horror , etc.). Derived words whose bases are built on simple stems and thus are formed by the application of one derivational affix are described as having the first degree of derivation (e.g. atomic, hasty, devotion , etc.). Derived words formed by two consecutive stages of coining possess the second degree of derivation (e.g. atomical, hastily, devotional , etc.), and so forth.
In conformity with the division of derivational affixes into suffixes and prefixes affixation is subdivided into suffixation and prefixation . Distinction is naturally made between prefixal and suffixal derivatives according to the last stage of derivation, which determines the nature of the immediate constituents of the pattern that signals the relationship of the derived word with its motivating source unit, e.g. unjust (un – + just ), justify (just + –ify ), arrangement (arrange + –ment), non-smoker (non– + smoker ). Words like reappearance, unreasonable, denationalize , are often qualified as prefixal-suffixal derivatives. R. S. Ginzburg insists that this classification is relevant only in terms of the constituent morphemes such words are made up of, i.e. from the angle of morphemic analysis. From the point of view of derivational analysis, such words are mostly either suffixal or prefixal derivatives, e.g. sub-atomic = sub – + (atom + –ic ), unreasonable = un– + (reason + –able), denationalize = de– + (national + –ize), discouragement = (dis– + courage) + –ment .
A careful study of a great many suffixal and prefixal derivatives has revealed an essential difference between them. In Modern English, suffixation is mostly characteristic of noun and adjective formation, while prefixation is mostly typical of verb formation. The distinction also rests on the role different types of meaning play in the semantic structure of the suffix and the prefix. The part-of-speech meaning has a much greater significance in suffixes as compared to prefixes which possess it in a lesser degree. Due to it, a prefix may be confined to one part of speech as, for example, enslave, encage, unbutton , or may function in more that one part of speech as over – in overkind, overfeed, overestimation . Unlike prefixes, suffixes as a rule function in any one part of speech often forming a derived stem of a different part of speech as compared with that of the base, e.g. careless – care; suitable – suit , etc. Furthermore, it is necessary to point out that a suffix closely knit together with a base forms a fusion retaining less of its independence that a prefix which is as a general rule more independent semantically, e.g. reading – ‘the act of one who reads’; ‘ability to read’; and to re-read – ‘to read again’ .
Prefixation is the formation of words with the help of prefixes. The interpretation of the terms prefix and prefixation now firmly rooted in linguistic literature has undergone a certain evolution. For instance, some time ago there were linguists who treated prefixation as part of word-composition (or compounding). The greater semantic independence of prefixes as compared with suffixes led the linguists to identify prefixes with the first component part of a compound word.
At present the majority of scholars treat prefixation as an integral part of word-derivation regarding prefixes as derivational affixes which differ essentially both from root-morphemes and non-derivational prepositive morphemes. Opinion sometimes differs concerning the interpretation of the functional status of certain individual groups of morphemes which commonly occur as first component parts of words. H. Marchand , for instance, analyses words like to overdo, to underestimate as compound verbs, the first component of which are locative particles, not prefixes. In a similar way he interprets words like income, onlooker, outhouse qualifying them as compounds with locative particles as first elements.
R. S. Ginzburg states there are about 51 prefixes in the system of Modern English word-formation.
Unlike suffixation, which is usually more closely bound up with the paradigm of a certain part of speech, prefixation is considered to be more neutral in this respect. It is significant that in linguistic literature derivational suffixes are always divided into noun-forming, adjective-forming and so on; prefixes, however, are treated differently. They are described either in alphabetical order or sub-divided into several classes in accordance with their origin,. Meaning or function and never according to the part of speech.
Prefixes may be classified on different principles. Diachronically distinction is made between prefixes of native and foreign origin. Synchronically prefixes may be classified:
(1) According to the class of words they preferably form. Recent investigations allow one to classify prefixes according to this principle. It must be noted that most of the 51 prefixes of Modern English function in more than one part of speech forming different structural and structural-semantic patterns. A small group of 5 prefixes may be referred to exclusively verb-forming (en–, be–, un –, etc.).
(2) As to the type of lexical-grammatical character of the base they are added to into: (a) deverbal, e.g. rewrite, outstay, overdo , etc.; (b) denominal, e.g. unbutton, detrain, ex-president , etc. and (c) deadjectival, e.g. uneasy, biannual , etc. It is interesting that the most productive prefixal pattern for adjectives is the one made up of the prefix un – and the base built either on adjectival stems or present and past participle, e.g. unknown, unsmiling, untold , etc.
(3) Semantically prefixes fall into mono– and polysemantic.
(4) As to the generic denotational meaning there are different groups that are distinguished in linguistic literature: (a) negative prefixes such as un–, non–, in–, dis–, a–, im–/in–/ir– (e.g. employment - unemployment, politician - non-politician, correct - incorrect, advantage - disadvantage, moral - amoral, legal - illegal , etc.); (b) reversative of privative prefixes, such as un–, de–, dis–, dis – (e.g. tie - untie, centralize - decentralize, connect - disconnect , etc.); (c) pejorative prefixes, such as mis–, mal –, pseudo – (e.g. calculate - miscalculate, function - malfunction, scientific - pseudo-scientific , etc.); (d) prefixes of time and order, such as fore–, pre–, post–, ex – (e.g. see - foresee, war - pre-war, Soviet - post-Soviet, wife - ex-wife , etc.); (e) prefix of repetition re – (e.g. do - redo, type - retype , etc.); (f) locative prefixes such as super –, sub–, inter–, trans – (e.g. market - supermarket, culture - subculture, national - international, Atlantic - trans-Atlantic , etc.).
(5) When viewed from the angle of their stylistic reference, English prefixes fall into those characterized by neutral stylistic reference and those possessing quite a definite stylistic value . As no exhaustive lexico-stylistic classification of English prefixes has yet been suggested, a few examples can only be adduced here. There is no doubt, for instance, that prefixes like un–, out–, over–, re–, under– and some others can be qualified as neutral (e. g. unnatural, unlace, outgrow, override, redo, underestimate , etc.). On the other hand, one can hardly fail to perceive the literary-bookish character of such prefixes as pseudo–, super–, ultra–, uni–, bi – and some others (e. g. pseudo-classical, superstructure, ultra-violence, unilateral, bifocal , etc.).
Sometimes one comes across pairs of prefixes one of which is neutral, the other is stylistically coloured. One example will suffice here: the prefix over– occurs in all functional styles, the prefix super– is peculiar to the style of scientific prose.
(6) Prefixes may be also classified as to the degree of productivity into highly-productive, productive and non-productive .
Suffixation is the formation of words with the help of suffixes. Suffixes usually modify the lexical meaning of the base and transfer words to a different part of speech. There are suffixes however, which do not shift words from one part of speech into another; a suffix of this kind usually transfers a word into a different semantic group, e. g. a concrete noun becomes an abstract one, as is the case with child—childhood, friend—friendship , etc.
Chains of suffixes occurring in derived words having two and more suffixal morphemes are sometimes referred to in lexicography as compound suffixes: –ably = –able + –ly (e. g. profitably, unreasonably ) –ical–ly = –ic + –al + –ly (e. g. musically, critically ); –ation = –ate + –ion (e. g. fascination, isolation ) and some others. Compound suffixes do not always present a mere succession of two or more suffixes arising out of several consecutive stages of derivation. Some of them acquire a new quality operating as a whole unit. Let us examine from this point of view the suffix –ation in words like fascination, translation, adaptation and the like. Adaptation looks at first sight like a parallel to fascination, translation . The latter however are first-degree derivatives built with the suffix –ion on the bases fascinate–, translate–. But there is no base adaptate –, only the shorter base adapt – . Likewise damnation, condemnation, formation , information and many others are not matched by shorter bases ending in –ate , but only by still shorter ones damn–, condemn–, form–, inform–. Thus, the suffix –ation is a specific suffix of a composite nature. It consists of two suffixes –ate and –ion , but in many cases functions as a single unit in first-degree derivatives. It is referred to in linguistic literature as a coalescent suffix or a group suffix. Adaptation is then a derivative of the first degree of derivation built with the coalescent suffix on the base adapt–.
Of interest is also the group-suffix –manship consisting of the suffixes –man and –ship . It denotes a superior quality, ability of doing something to perfection, e. g. authormanship, quotemanship, lipmanship, etc.
It also seems appropriate to make several remarks about the morphological changes that sometimes accompany the process of combining derivational morphemes with bases. Although this problem has been so far insufficiently investigated, some observations have been made and some data collected. For instance, the noun-forming suffix –ess for names of female beings brings about a certain change in the phonetic shape of the correlative male noun provided the latter ends in –er, –or, e.g. actress (actor), sculptress (sculptor), tigress (tiger) , etc. It may be easily observed that in such cases the sound [∂] is contracted in the feminine nouns.
Further, there are suffixes due to which the primary stress is shifted to the syllable immediately preceding them, e.g. courageous (courage), stability (stable), investigation (investigate ), peculiarity (peculiar) , etc. When added to a base having the suffix –able/–ible as its component, the suffix –ity brings about a change in its phonetic shape, namely the vowel [i] is inserted between [b] and [l], e. g. possible - possibility, changeable - changeability, etc. Some suffixes attract the primary stress on to themselves, there is a secondary stress on the first syllable in words with such suffixes, e. g. 'employ'ee (em'ploy), govern'mental (govern), 'pictu'resque (picture ).
There are different classifications of suffixes in linguistic literature, as suffixes may be divided into several groups according to different principles:
(1) The first principle of classification that, one might say, suggests itself is the part of speech formed. Within the scope of the part-of-speech classification suffixes naturally fall into several groups such as:
a) noun-suffixes, i.e. those forming or occurring in nouns, e. g. –er, –dom, –ness, –ation, etc. (teacher, Londoner, freedom, brightness, justification, etc.);
b) adjective-suffixes, i.e. those forming or occurring in adjectives, e. g. –able, –less, –ful, –ic, –ous, etc. (agreeable, careless, doubtful, poetic, courageous, etc.);
c) verb-suffixes, i.e. those forming or occurring in verbs, e. g. –en, –fy, –ize (darken, satisfy, harmonize , etc.);
d) adverb-suffixes, i.e. those forming or occurring in adverbs, e. g. –ly, –ward (quickly, eastward , etc.).
(2) Suffixes may also be classified into various groups according to the lexico-grammatical character of the base the affix is usually added to. Proceeding from this principle one may divide suffixes into:
a) deverbal suffixes (those added to the verbal base), e. g. –er, –ing, –ment, –able, etc. (speaker, reading, agreement, suitable , etc.);
b) denominal suffixes (those added to the noun base), e. g. –less, –ish, –ful, –ist, –some, etc. (handless, childish, mouthful, violinist, troublesome , etc.);
c) de-adjectival suffixes (those affixed to the adjective base), e. g. –en, –ly, –ish, –ness, etc. (blacken, slowly, reddish, brightness, etc.).
(3) A classification of suffixes may also be based on the criterion of sense expressed by a set of suffixes. Proceeding from this principle suffixes are classified into various groups within the bounds of a certain part of speech. For instance, noun-suffixes fall into those denoting:
a) the agent of an action, e. g. –er, –ant (baker, dancer, defendant , etc.);
b) appurtenance, e. g. –an, –ian, –ese , etc. ( Arabian, Elizabethan, Russian, Chinese, Japanese , etc.);
c) collectivity, e. g. –age, –dom, –ery (–ry), etc. ( freightage, officialdom, peasantry , etc.);
d) diminutiveness, e. g. –ie, –let, –ling , etc. ( birdie, girlie, cloudlet, squirreling, wolfing , etc.).
(4) Still another classification of suffixes may be worked out if one examines them from the angle of stylistic reference. Just like prefixes, suffixes are also characterized by quite a definite stylistic reference falling into two basic classes:
a) those characterized by neutral stylistic reference such as –able, –er, –ing , etc.;
b) those having a certain stylistic value such as –old, –i/form, –aceous, –tron, etc.
Suffixes with neutral stylistic reference may occur in words of different lexico-stylistic layers. As for suffixes of the second class they are restricted in use to quite definite lexico-stylistic layers of words, in particular to terms, e.g. rhomboid, asteroid, cruciform, cyclotron, synchrophasotron , etc.
(5) Suffixes are also classified as to the degree of their productivity.
Distinction is usually made between dead and living affixes . Dead affixes are described as those which are no longer felt in Modern English as component parts of words; they have so fused with the base of the word as to lose their independence completely. It is only by special etymological analysis that they may be singled out, e. g. –d in dead, seed, –le, –l, –el in bundle, sail, hovel; –ock in hillock; –lock in wedlock; –t in flight, gift, height . It is quite clear that dead suffixes are irrelevant to present-day English word-formation, they belong in its diachronic study.
Living affixes may be easily singled out from a word, e. g. the noun-forming suffixes –ness, –dom, –hood, –age, –ance , as in darkness, freedom, childhood, marriage, assistance , etc. or the adjective-forming suffixes –en, –ous, –ive, –ful, –y as in wooden, poisonous, active, hopeful, stony , etc.
However, not all living derivational affixes of Modern English possess the ability to coin new words. Some of them may be employed to coin new words on the spur of the moment, others cannot, so that they are different from the point of view of their productivity. Accordingly they fall into two basic classes — productive and non-productive word-building affixes.
It has been pointed out that linguists disagree as to what is meant by the productivity of derivational affixes.
Following the first approach all living affixes should be considered productive in varying degrees from highly-productive (e. g. –er, –ish, –less, re –, etc.) to non-productive (e. g. –ard, –cy, –ive , etc.).
Consequently it becomes important to describe the constraints imposed on and the factors favouring the productivity of affixational patterns and individual affixes. The degree of productivity of affixational patterns very much depends on the structural, lexico-grammatical and semantic nature of bases and the meaning of the affix. For instance, the analysis of the bases from which the suffix –ize can derive verbs reveals that it is most productive with noun-stems, adjective-stems also favour ifs productivity, whereas verb-stems and adverb-stems do not, e. g. criticize (critic), organize (organ), itemize (item), mobilize (mobile), localize (local) , etc. Comparison of the semantic structure of a verb in –ize with that of the base it is built on shows that the number of meanings of the stem usually exceeds that of the verb and that its basic meaning favours the productivity of the suffix –ize to a greater degree than its marginal meanings, e. g. to characterize — character, to moralize — moral, to dramatize — drama, etc.
The treatment of certain affixes as non-productive naturally also depends on the concept of productivity. The current definition of non-productive derivational affixes as those which cannot hg used in Modern English for the coining of new words is rather vague and maybe interpreted in different ways. Following the definition the term non-productive refers only to the affixes unlikely to be used for the formation of new words, e. g. –ous, –th, fore – and some others (famous, depth, foresee ).
If one accepts the other concept of productivity mentioned above, then non-productive affixes must be defined as those that cannot be used for the formation of occasional words and, consequently, such affixes as –dom, –ship, –ful, –en, –ify, –ate and many others are to be regarded as non-productive.
The theory of relative productivity of derivational affixes is also corroborated by some other observations made on English word-formation. For instance, different productive affixes are found in different periods of the history of the language. It is extremely significant, for example, that out of the seven verb-forming suffixes of the Old English period only one has survived up to the present time with a very low degree of productivity, namely the suffix –en (e. g. to soften, to darken, to whiten ).
A derivational affix may become productive in just one meaning because that meaning is specially needed by the community at a particular phase in its history. This may be well illustrated by the prefix de – in the sense of ‘undo what has been done, reverse an action or process’, e. g. deacidify (paint spray), decasualize (dock labour), decentralize (government or management), deration (eggs and butter), de-reserve (medical students), desegregate (coloured children), and so on.
Furthermore, there are cases when a derivational affix being nonproductive in the non-specialized section of the vocabulary is used to coin scientific or technical terms. This is the case, for instance, with the suffix –ance which has been used to form some terms in Electrical Engineering, e. g. capacitance, impedance, reactance . The same is true of the suffix –ity which has been used to form terms in physics, and chemistry such as alkalinity, luminosity, emissivity and some others.
Conversion , one of the principal ways of forming words in Modern English is highly productive in replenishing the English word-stock with new words. The term conversion, which some linguists find inadequate, refers to the numerous cases of phonetic identity of word-forms, primarily the so-called initial forms, of two words belonging to different parts of speech. This may be illustrated by the following cases: work — to work; love — to love; paper — to paper; brief — to brief , etc. As a rule we deal with simple words, although there are a few exceptions, e.g. wireless — to wireless.
It will be recalled that, although inflectional categories have been greatly reduced in English in the last eight or nine centuries, there is a certain difference on the morphological level between various parts of speech, primarily between nouns and verbs. For instance, there is a clear-cut difference in Modern English between the noun doctor and the verb to doctor — each exists in the language as a unity of its word-forms and variants, not as one form doctor . It is true that some of the forms are identical in sound, i.e. homonymous, but there is a great distinction between them, as they are both grammatically and semantically different.
If we regard such word-pairs as doctor — to doctor, water — to water, brief — to brief from the angle of their morphemic structure, we see that they are all root-words. On the derivational level, however, one of them should be referred to derived words, as it belongs to a different part of speech and is understood through semantic and structural relations with the other, i.e. is motivated by it. Consequently, the question arises: what serves as a word-building means in these cases? It would appear that the noun is formed from the verb (or vice versa) without any morphological change, but if we probe deeper into the matter, we inevitably come to the conclusion that the two words differ in the paradigm. Thus it is the paradigm that is used as a word-building means. Hence, we may define conversion as the formation of a new word through changes in its paradigm.
It is necessary to call attention to the fact that the paradigm plays a significant role in the process of word-formation in general and not only in the case of conversion. Thus, the noun cooker (in gas-cooker ) is formed from the word to cook not only by the addition of the suffix –er , but also by the change in its paradigm. However, in this case, the role played by the paradigm as a word-building means is less obvious, as the word-building suffix –er comes to the fore. Therefore, conversion is characterized not simply by the use of the paradigm as a word-building means, but by the formation of a new word solely by means of changing its paradigm. Hence, the change of paradigm is the only word-building means of conversion. As a paradigm is a morphological category conversion can be described as a morphological way of forming words.
Compounding or word-composition is one of the productive types of word-formation in Modern English. Composition like all other ways of deriving words has its own peculiarities as to the means used, the nature of bases and their distribution, as to the range of application, the scope of semantic classes and the factors conducive to productivity.
Compounds, as has been mentioned elsewhere, are made up of two ICs which are both derivational bases. Compound words are inseparable vocabulary units. They are formally and semantically dependent on the constituent bases and the semantic relations between them which mirror the relations between the motivating units. The ICs of compound words represent bases of all three structural types. The bases built on stems may be of different degree of complexity as, for example, week-end, office-management, postage-stamp, aircraft-carrier, fancy-dress-maker, etc. However, this complexity of structure of bases is not typical of the bulk of Modern English compounds.
In this connection care should be taken not to confuse compound words with polymorphic words of secondary derivation, i.e. derivatives built according to an affixal pattern but on a compound stem for its base such as, e. g. school-mastership ([n + n] + suf), ex-housewife (prf + [n + n]), to weekend, to spotlight ([n + n] + conversion).
Structurally compound words are characterized by the specific order and arrangement in which bases follow one another. The order in which the two bases are placed within a compound is rigidly fixed in Modern English and it is the second IC that makes the head-member of the word, i.e. its structural and semantic centre. The head-member is of basic importance as it preconditions both the lexico-grammatical and semantic features of the first component. It is of interest to note that the difference between stems (that serve as bases in compound words) and word-forms they coincide with is most obvious in some compounds, especially in compound adjectives. Adjectives like long, wide, rich are characterized by grammatical forms of degrees of comparison longer, wider, richer . The corresponding stems functioning as bases in compound words lack grammatical independence and forms proper to the words and retain only the part-of-speech meaning; thus compound adjectives with adjectival stems for their second components, e. g. age-long, oil-rich, inch-wide , do not form degrees of comparison as the compound adjective oil-rich does not form them the way the word rich does, but conforms to the general rule of polysyllabic adjectives and has analytical forms of degrees of comparison. The same difference between words and stems is not so noticeable in compound nouns with the noun-stem for the second component.
Phonetically compounds are also marked by a specific structure of their own. No phonemic changes of bases occur in composition but the compound word acquires a new stress pattern, different from the stress in the motivating words, for example words key and hole or hot and house each possess their own stress but when the stems of these words are brought together to make up a new compound word, 'keyhole — ‘a hole in a lock into which a key fits’, or 'hothouse — ‘a heated building for growing delicate plants’, the latter is given a different stress pattern — a unity stress on the first component in our case. Compound words have three stress patterns:
a) a high or unity stress on the first component as in 'honeymoon, 'doorway , etc.
b) a double stress, with a primary stress on the first component and a weaker, secondary stress on the second component, e. g. 'blood-ֻvessel, 'mad-ֻdoctor , 'washing-ֻmachine , etc.
c) It is not infrequent, however, for both ICs to have level stress as in, for instance, 'arm-'chair, 'icy-'cold, 'grass-'green , etc.
Graphically most compounds have two types of spelling — they are spelt either solidly or with a hyphen. Both types of spelling when accompanied by structural and phonetic peculiarities serve as a sufficient indication of inseparability of compound words in contradistinction to phrases. It is true that hyphenated spelling by itself may be sometimes misleading, as it may be used in word-groups to emphasize their phraseological character as in e. g. daughter-in-law, man-of-war, brother-in-arms or in longer combinations of words to indicate the semantic unity of a string of words used attributively as, e.g., I-know-what-you're-going-to-say expression, we-are-in-the-know jargon, the young-must-be-right attitude. The two types of spelling typical of compounds, however, are not rigidly observed and there are numerous fluctuations between solid or hyphenated spelling on the one hand and spelling with a break between the components on the other, especially in nominal compounds of the n+n type. The spelling of these compounds varies from author to author and from dictionary to dictionary. For example, the words war-path, war-time, money-lender are spelt both with a hyphen and solidly; blood-poisoning, money-order, wave-length, war-ship — with a hyphen and with a break; underfoot, insofar, underhand —solidly and with a break . It is noteworthy that new compounds of this type tend to solid or hyphenated spelling. This inconsistency of spelling in compounds, often accompanied by a level stress pattern (equally typical of word-groups) makes the problem of distinguishing between compound words (of the n + n type in particular) and word-groups especially difficult.
In this connection it should be stressed that Modern English nouns (in the Common Case, Sg.) as has been universally recognized possess an attributive function in which they are regularly used to form numerous nominal phrases as, e. g. peace years, stone steps, government office , etc. Such variable nominal phrases are semantically fully derivable from the meanings of the two nouns and are based on the homogeneous attributive semantic relations unlike compound words. This system of nominal phrases exists side by side with the specific and numerous class of nominal compounds which as a rule carry an additional semantic component not found in phrases.
It is also important to stress that these two classes of vocabulary units — compound words and free phrases — are not only opposed but also stand in close correlative relations to each other.
Semantically compound words are generally motivated units. The meaning of the compound is first of all derived from the combined lexical meanings of its components. The semantic peculiarity of the derivational bases and the semantic difference between the base and the stem on which the latter is built is most obvious in compound words. Compound words with a common second or first component can serve as illustrations. The stem of the word board is polysemantic and its multiple meanings serve as different derivational bases, each with its own selective range for the semantic features of the other component, each forming a separate set of compound words, based on specific derivative relations. Thus the base board meaning ‘a flat piece of wood square or oblong’ makes a set of compounds chess-board, notice-board, key-board, diving-board, foot-board, sign-board; compounds paste-board, cardboard are built on the base meaning ‘thick, stiff paper’; the base board– meaning ‘an authorized body of men’, forms compounds school-board, board-room . The same can be observed in words built on the polysemantic stem of the word foot . For example, the base foot– in foot-print, foot-pump, foothold, foot-bath, foot-wear has the meaning of ‘the terminal part of the leg’, in foot-note, foot-lights, foot-stone the base foot – has the meaning of ‘the lower part’, and in foot-high, foot-wide, footrule — ‘measure of length’. It is obvious from the above-given examples that the meanings of the bases of compound words are interdependent and that the choice of each is delimited as in variable word-groups by the nature of the other IC of the word. It thus may well be said that the combination of bases serves as a kind of minimal inner context distinguishing the particular individual lexical meaning of each component. In this connection we should also remember the significance of the differential meaning found in both components which becomes especially obvious in a set of compounds containing identical bases.
Compound words can be described from different points of view and consequently may be classified according to different principles. They may be viewed from the point of view:
(1) of general relationship and degree of semantic independence of components;
(2) of the parts of speech compound words represent;
(3) of the means of composition used to link the two ICs together;
(4) of the type of ICs that are brought together to form a compound;
(5) of the correlative relations with the system of free word-groups.
From the point of view of degree of semantic independence there are two types of relationship between the ICs of compound words that are generally recognized in linguistic literature: the relations of coordination and subordination, and accordingly compound words fall into two classes: coordinative compounds (often termed copulative or additive) and subordinative (often termed determinative).
In coordinative compounds the two ICs are semantically equally important as in fighter-bomber, oak-tree, girl-friend, Anglo-American . The constituent bases belong to the same class and той often to the same semantic group. Coordinative compounds make up a comparatively small group of words. Coordinative compounds fall into three groups:
a) Reduplicative compounds which are made up by the repetition of the same base as in goody-goody, fifty-fifty, hush-hush, pooh-pooh . They are all only partially motivated.
b) Compounds formed by joining the phonically variated rhythmic twin forms which either alliterate with the same initial consonant but vary the vowels as in chit-chat, zigzag, sing-song, or rhyme by varying the initial consonants as in clap-trap, a walky-talky, helter-skelter . This subgroup stands very much apart. It is very often referred to pseudo-compounds and considered by some linguists irrelevant to productive word-formation owing to the doubtful morphemic status of their components. The constituent members of compound words of this subgroup are in most cases unique, carry very vague or no lexical meaning of their own, are not found as stems of independently functioning words. They are motivated mainly through the rhythmic doubling of fanciful sound-clusters.
Coordinative compounds of both subgroups (a, b) are mostly restricted to the colloquial layer, are marked by a heavy emotive charge and possess a very small degree of productivity.
c) The bases of additive compounds such as a queen-bee, an actor-manager , unlike the compound words of the first two subgroups, are built on stems of the independently functioning words of the same part of speech. These bases often semantically stand in the genus-species relations. They denote a person or an object that is two things at the same time. A secretary-stenographer is thus a person who is both a stenographer and a secretary, a bed-sitting-room (a bed-sitter ) is both a bed-room and a sitting-room at the same time. Among additive compounds there is a specific subgroup of compound adjectives one of ICs of which is a bound root-morpheme. This group is limited to the names of nationalities such as Sino-Japanese, Anglo-Saxon, Afro-Asian , etc.
Additive compounds of this group are mostly fully motivated but have a very limited degree of productivity.
However it must be stressed that though the distinction between coordinative and subordinative compounds is generally made, it is open to doubt and there is no hard and fast border-line between them. On the contrary, the border-line is rather vague. It often happens that one and the same compound may with equal right be interpreted either way — as a coordinative or a subordinative compound, e. g. a woman-doctor may be understood as ‘a woman who is at the same time a doctor’ or there can be traced a difference of importance between the components and it may be primarily felt to be ‘a doctor who happens to be a woman’ (also a mother-goose, a clock-tower ).
In subordinative compounds the components are neither structurally nor semantically equal in importance but are based on the domination of the head-member which is, as a rule, the second IC. The second IC thus is the semantically and grammatically dominant part of the word, which preconditions the part-of-speech meaning of the whole compound as in stone-deaf, age-long which are obviously adjectives, a wrist-watch, road-building, a baby-sitter which are nouns.
Functionally compounds are viewed as words of different parts of speech. It is the head-member of the compound, i.e. its second IC that is indicative of the grammatical and lexical category the compound word belongs to.
Compound words are found in all parts of speech, but the bulk of compounds are nouns and adjectives. Each part of speech is characterized by its set of derivational patterns and their semantic variants. Compound adverbs, pronouns and connectives are represented by an insignificant number of words, e. g. somewhere, somebody, inside, upright, otherwise moreover, elsewhere, by means of , etc. No new compounds are coined on this pattern. Compound pronouns and adverbs built on the repeating first and second IC like body, ever, thing make closed sets of words
On the whole composition is not productive either for adverbs, pronouns or for connectives.
Verbs are of special interest. There is a small group of compound verbs made up of the combination of verbal and adverbial stems that language retains from earlier stages, e. g. to bypass, to inlay, to offset. This type according to some authors, is no longer productive and is rarely found in new compounds.
There are many polymorphic verbs that are represented by morphemic sequences of two root-morphemes, like to weekend, to gooseflesh, to spring-clean , but derivationally they are all words of secondary derivation in which the existing compound nouns only serve as bases for derivation. They are often termed pseudo-compound verbs. Such polymorphic verbs are presented by two groups:
(1) verbs formed by means of conversion from the stems of compound nouns as in to spotlight from a spotlight, to sidetrack from a side-track, to handcuff from handcuffs, to blacklist from a blacklist, to pinpoint from a pin-point;
(2) verbs formed by back-derivation from the stems of compound nouns, e. g. to baby-sit from a baby-sitter, to playact from play-acting, to housekeep from house-keeping, to spring-clean from spring-cleaning.
From the point of view of the means by which the components are joined together, compound words may be classified into:
(1) Words formed by merely placing one constituent after another in a definite order which thus is indicative of both the semantic value and the morphological unity of the compound, e. g. rain-driven, house-dog, pot-pie ( as opposed to dog-house, pie-pot). This means of linking the components is typical of the majority of Modern English compounds in all parts of speech.
As to the order of components, subordinative compounds are often classified as:
a) asyntactic compounds in which the order of bases runs counter to the order in which the motivating words can be brought together under the rules of syntax of the language. For example, in variable phrases adjectives cannot be modified by preceding adjectives and noun modifiers are not placed before participles or adjectives, yet this kind of asyntactic arrangement is typical of compounds, e. g. red-hot, bluish-black, pale-blue, rain-driven, oil-rich. The asyntactic order is typical of the majority of Modern English compound words;
b) syntactic compounds whose components are placed in the order that resembles the order of words in free phrases arranged according to the rules of syntax of Modern English. The order of the components in compounds like blue-bell, mad-doctor, blacklist ( a + n ) reminds one of the order and arrangement of the corresponding words in phrases a blue bell, a mad doctor, a black list ( A + N ), the order of compounds of the type door-handle, day-time, spring-lock ( n + n ) resembles the order of words in nominal phrases with attributive function of the first noun ( N + N ), e. g. spring time, stone steps, peace movement .
(2) Compound words whose ICs are joined together with a special linking-element — the linking vowels [ou] and occasionally [i] and the linking consonant [s/z] — which is indicative of composition as in, for example, speedometer, tragicomic, statesman. Compounds of this type can be both nouns and adjectives, subordinative and additive but are rather few in number since they are considerably restricted by the nature of their components. The additive compound adjectives linked with the help of the vowel [ou] are limited to the names of nationalities and represent a specific group with a bound root for the first component, e. g. Sino-Japanese, Afro-Asian, Anglo-Saxon .
In subordinative adjectives and nouns the productive linking element is also [ou] and compound words of the type are most productive for scientific terms. The main peculiarity of compounds of the type is that their constituents are nonassimilated bound roots borrowed mainly from classical languages, e. g. electro-dynamic, filmography, technophobia, videophone, sociolinguistics, videodisc .
A small group of compound nouns may also be joined with the help of linking consonant [s/z], as in sportsman, landsman, saleswoman, bridesmaid . This small group of words is restricted by the second component which is, as a rule, one of the three bases man–, woman–, people –. The commonest of them is man –.
Compounds may be also classified according to the nature of the bases and the interconnection with other ways of word-formation into the so-called compounds proper and derivational compounds.
Compounds proper are formed by joining together bases built on the stems or on the word-forms of independently functioning words with or without the help of special linking element such as doorstep, age-long, baby-sitter, looking-glass, street-fighting, handiwork, sportsman. Compounds proper constitute the bulk of English compounds in all parts of speech, they include both subordinative and coordinative classes, productive and non-productive patterns.
, e. g. long-legged, three-cornered, a break-down, a pickpocket
differ from compounds proper in the nature of bases and their second IC. The two ICs of the compound long-legged — ‘having long legs’ — are the suffix –ed
meaning ‘having’ and the base built on a free word-group long legs
whose member words lose their grammatical independence, and are reduced to a single component of the word, a derivational base. Any other segmentation of such words, say into long–
is impossible because firstly, adjectives like *legged
do not exist in Modern English and secondly, because it would contradict the lexical meaning of these words. The derivational adjectival suffix –ed converts this newly formed base into a word. It can be graphically represented as long legs
- [ (long–leg) + –ed
he suffix –ed becomes the grammatically and semantically dominant component of the word, its head-member. It imparts its part-of-speech meaning and its lexical meaning thus making an adjective that may be semantically interpreted as ‘with (or having) what is denoted by the motivating word-group’. Comparison of the pattern of compounds proper like baby-sitter, pen-holder
[ n + ( v + –er ) ] with the pattern of derivational compounds like long-legged [ (a + n) + –ed ] reveals the difference: derivational compounds are formed by a derivational means, a suffix in case if words of the long-legged type, which is applied to a base that each time is formed anew on a free word-group and is not recurrent in any other type if words. It follows that strictly speaking words of this type should be treated as pseudo-compounds or as a special group of derivatives. They are habitually referred to derivational compounds because of the peculiarity of their derivational bases which are felt as built by composition, i.e. by bringing together the stems of the member-words of a phrase which lose their independence in the process. The word itself, e. g. long-legged, i s built by the application of the suffix, i.e. by derivation and thus may be described as a suffixal derivative.
Derivational compounds or pseudo-compounds are all subordinative and fall into two groups according to the type of variable phrases that serve as their bases and the derivational means used:
a) derivational compound adjectives formed with the help of the highly-productive adjectival suffix –ed applied to bases built on attributive phrases of the A + N, Num + N, N + N type, e. g. long legs, three corners, doll face. Accordingly the derivational adjectives under discussion are built after the patterns [ (a + n ) + –ed], e. g. long-legged, flat-chested, broad-minded ; [ ( пит + n) + –ed], e. g. two-sided, three-cornered ; [ (n + n ) + –ed], e. g. doll-faced, heart-shaped .
b) derivational compound nouns formed mainly by conversion applied to bases built on three types of variable phrases — verb-adverb phrase, verbal-nominal and attributive phrases.
The commonest type of phrases that serves as derivational bases for this group of derivational compounds is the V + Adv type of word-groups as in, for instance, a breakdown, a breakthrough, a castaway, a layout . Semantically derivational compound nouns form lexical groups typical of conversion, such as an act or instance of the action, e. g. a holdup — ‘a delay in traffic’' from to hold up — ‘delay, stop by use of force’; a result of the action, e. g. a breakdown — ‘a failure in machinery that causes work to stop’ from to break down — ‘become disabled’; an active agent or recipient of the action, e. g. cast-offs — ‘clothes that he owner will not wear again’ from to cast off — ‘throw away as unwanted’; a show-off — ‘a person who shows off’ from to show off — ‘make a display of one's abilities in order to impress people’. Derivational compounds of this group are spelt generally solidly or with a hyphen and often retain a level stress. Semantically they are motivated by transparent derivative relations with the motivating base built on the so-called phrasal verb and are typical of the colloquial layer of vocabulary. This type of derivational compound nouns is highly productive due to the productivity of conversion.
The semantic subgroup of derivational compound nouns denoting agents calls for special mention. There is a group of such substantives built on an attributive and verbal-nominal type of phrases. These nouns are semantically only partially motivated and are marked by a heavy emotive charge or lack of motivation and often belong to terms as, for example, a kill-joy, a wet-blanket — ‘one who kills enjoyment’; a turnkey — ‘keeper of the keys in prison’; a sweet-tooth — ‘a person who likes sweet food’; a red-breast — ‘a bird called the robin’. The analysis of these nouns easily proves that they can only be understood as the result of conversion for their second ICs cannot be understood as their structural or semantic centres, these compounds belong to a grammatical and lexical groups different from those their components do. These compounds are all animate nouns whereas their second ICs belong to inanimate objects. The meaning of the active agent is not found in either of the components but is imparted as a result of conversion applied to the word-group which is thus turned into a derivational base.
These compound nouns are often referred to in linguistic literature as "bahuvrihi " compounds or exocentric compounds, i.e. words whose semantic head is outside the combination. It seems more correct to refer them to the same group of derivational or pseudo-compounds as the above cited groups.
This small group of derivational nouns is of a restricted productivity, its heavy constraint lies in its idiomaticity and hence its stylistic and emotive colouring.
The linguistic analysis of extensive language data proves that there exists a regular correlation between the system of free phrases and all types of subordinative (and additive) compounds . Correlation embraces both the structure and the meaning of compound words, it underlies the entire system of productive present-day English composition conditioning the derivational patterns and lexical types of compounds.
 Randolph Quirk, Ian Svortik. Investigating Linguistic Acceptability . Walter de Gruyter. Inc., 1966. P. 127-128.
 Robins, R. H. A short history of linguistics . London: Longmans, 1967. P. 183.
 Henry Sweet, History of Language . Folcroft Library Editions,1876. P. 471.
 Zellig S. Harris, Structural Linguistics . University of Chicago Press, 1951. P. 255.
 Leonard Bloomfield, Language . New York, 1933
 Noam Avram Chomsky, Syntactic Structures . Berlin, 1957.
 Ibidem, p. 15.
 Ibidem, p. 4.
 Ibidem, p. 11.
 Ibidem, p. 10.
 Jukka Pennanen, Aspects of Finnish Grammar . Pohjoinen, 1972. P. 293.
 K. Zimmer, Levels of Linguistic Description . Chicago, 1964. P. 18.
 A. Ross Eckler’s letters to Daria Abrossimova, 2001.
 Kucera, H. & Francis, W. N. Computational analysis of present-day American English . University Press of New England, 1967.
 Webster's Encyclopedic Unabridged Dictionary of the English Language . Random House Value Pub. 1996.
 A. Ross Eckler’s letters to Daria Abrossimova, 2001.
 Dmitri Borgmann. Beyond Language . Charles Scribner’s Sons. 1965.
 The Times Atlas of the World . Times Books. 1994.
 Rand McNally Commercial Atlas and Marketing Guide . Rand McNally & Co. 2000.
 Prof. Smirnitsky calls them “potential words” in his book on English Lexicology (p. 18).
 Ginzburg R. A Course in Modern English Lexicology . Moscow, 1979. P. 113.
 Ibidem. P. 114-115.
 Marchand H. Studies in Syntax and Word-Formation. Munich, 1974.
 Ginzburg R. A Course in Modern English Lexicology . Moscow, 1979. P. 115.
 The spelling is given according to Webster’s New Collegiate Dictionary , 1956 and H.C. Wyld. The Universal English Dictionary , 1952.
 Prof. A. I. Smirnitsky as far back as the late forties pointed out the rigid parallelism existing between free word-groups and derivational compound adjectives which he termed “grammatical compounds”.