Chapter XIX · RISE OF WORD-FORMATION AND INFLECTION.

We have in former chapters dealt with, and frequently alluded to, the fact that much which is new in derivation and inflection is due to analogy. Much is due to this, but not all; and we must now ask whence originated these processes of derivation and flection, which cannot be explained as due to analogy, i.e. those which, instead of being moulded on a given pattern, have, on the contrary, served as the model for others. It is clear that as soon as language arose, even in its most primitive state, words must have been combined syntactically, in however simple a manner. Groups of etymologically connected words, words derived the one from the other by suffixes (as long, length; king, kingdom) or by flection (as book, books; go, goes),—such groups need not have existed at once, nay, must have arisen only gradually, and in course of time. How did they arise? Theoretically, three ways only seem possible.

Words formed independently for cognate ideas, might accidentally resemble each other so closely as to group themselves also phonetically, i.e. to be sounded more or less alike; or—what is essentially the same, though not quite so improbable—words originally different and expressing different ideas, might, in course of time, so develop in meaning and sound as to become members of a group. A case somewhat of this nature we studied in our word bound (cf. page 194), which, originally different in sound and form from the then existing past participle of to bind, has come to resemble it so much in form, and was used in such a sense as to cause all but students of language to group these forms together.

A second way is a differentiation in sound, i.e. two forms may arise, under the influence of accent or other causes, from the same word, which two forms then come to be differentiated in meaning. We have in this way, for instance, the two forms of the past tense of the verb werden (to become) in German, ward and wurde. These arose absolutely independently of any difference in meaning; once having arisen, a custom sprang up of using the one (ward) as aorist and the other (wurde) by preference as imperfect tense.

That in the above examples, the form which later on became bound is not itself an original creation, or that, in German, the two forms of the past tense were due largely to analogy, does not affect their value as illustrative of our point. We readily understand that both these ways were and are possible, but, at the same time, that in only very few cases they have been followed.

Only one way of explaining the origin of flection remains—‘composition.’

In order to explain how derivation and flection can have been derived from composition, we will go somewhat deeply into the nature and application of the latter. We shall then see how impossible it is to draw a sharp line between syntactical co-ordination, composition, derivation, and flection anywhere, and then—and only then—we shall acquire an insight into the true nature of the subject of this chapter.

If we study the composition of words in the various Indo-European languages, we soon learn to distinguish two different kinds. In one we find the so-called crude forms (that is to say, those forms of the words which, WITH THE CASE-ENDINGS, make up what we now consider the complete word) combined with other crude forms, the last of which alone assumes these case-endings. To illustrate this we must of course go back to ancient languages, in which this crude form is clearly distinct from the nominative or any other case. We have plenty of such compounds even now in English and other modern languages; but, in consequence of the wearing off of terminations, the most undoubted examples would illustrate (i.e. throw light upon) nothing. In Sanscrit, for instance, there are three plants which in the nominative singular would be called çaças (or çaçaḥ), kuças (ḥ) or kuçam (masc. or neut.), and palâçam. It is the crude forms of these nouns (without their nominative—s and m) which are used in the compound çaça-kuça-palâçam, which indicates a collection of the three. Again râjâ (with long â) is the nominative form of a stem râjan (‘king’) or râja (with short a). In the compound râja-purushas (h) we again find the crude form, this time the shorter form of the base: purushas means ‘man’ and the whole (= ‘king-man’) stands for king’s man. We might illustrate this kind by such words as our tragi-comic, melodramatic (melos = ‘song’).

In the other kind of compounds we find two or more fully inflected forms combined in one group. This is the method of composition which survives in our present linguistic consciousness, which sees compounds of the second kind even in those which are historically connected with the Indo-European type, illustrated in the former paragraph by râja-purushas. The wearing off of well-nigh all case-endings has in the present language almost completely obliterated the difference between crude forms and nominatives of nouns and adjectives or the infinitives of verbs. Hence, at present, the ordinary speaker realises no difference between, e.g., noon in noon-tide and the word noon in It is noon. Yet the compound noon-tide belongs historically to the former class, and noon is there a ‘crude form,’ if we may still so call it. In our following study of composition as at present employed in the English language, we neglect the scientific origin, but base our classification on appearance; in the present case, on present linguistic consciousness. One of the fullest and best-known lists of compounds in the English language is perhaps that given by Morris (Histor. Outlines, p. 222). We shall largely draw upon it in the following study, though we have, in our enumeration, rather considered the character of the component parts than, as Mr. Morris does, that of the function of the compound.

I. Nouns are compounded with Nouns—

1. Both in the same case; i.e. in apposition, the one explanatory of, or defining the other (in which case one of the nouns has a function almost, if not quite, identical with that of an adjective). Instances are spear-plant, noon-tide, church-yard, headman, oak-tree, master-tailor, merchant-tailor, prince-regent, water-course, watershed, head-waiter, plough-boy, bishopdom (found in Milton, dom = ‘jurisdiction’), bishopric (ric = A.S. rîce, ‘power,’ ‘domain’), bandog (= band + dog), barn (bere, i.e. barley + ern, i.e. ‘storehouse’), bridegroom (bride + groom = goom = A.S. guma, ‘man’191), bridal (bride + ale = ‘bride-feast’), cowslip (cow + slip, A.S. cu-slyppe = ‘cow dung’), hussy (= ‘house-wife’—Skeat, Prin. Eng. Etymol., p. 422), Lord-lieutenant, earlmarshal, wer-wolf (‘man-wolf,’ A.S. wer = ‘a man’), world (weoruld, wer = ‘man’ + ældu = ‘age,’ ‘old age,’ ‘age of man’), yeoman (= ‘village-man’—see Skeat), orchard (A.S. orceard, ortgeard, metathesis = wort-yard = ‘vegetable-garden’), Lammas (= hláf-maesse = ‘loaf-mass,’ ‘day of offering,’ ‘first-fruits’), handi-work (hand + geweorc = ‘hand-work’), mildew (= ‘honey dew,’ mil = ‘honey,’ A.S. mele), penny-worth.

2. Genitive + Nominative. Doomsday, Thursday, Tuesday (day of Tiw, the godhead), kinsman, trades-union, calf’s-foot (calf’s-foot jelly), lady day (lady as a feminine had no s in the genitive), daisy (‘day’s eye,’ A.S. dæges 4 éage), Wednesday (‘Wodan’s day’), shilling’s-worth.

3. Noun + Verbal Noun (the former having the function of object to the verb cognate with the latter). Man-killer, blood-shedding, auger (i.e. ‘nauger,’ a nauger having been divided as if = an auger; A.S. nafu-gár, ‘nave (of a wheel)’ ‘-borer,’ ‘-piercer’), groundsel (A.S. grunde + swelge = ‘ground-swallower’ = ‘abundant weed;’ already in the Saxon corrupted from gunde-swilge = ‘poison-swallower,’ with reference to healing effects),192 lady (hláf-dige, ‘loaf-kneader’), soothsayer (= ‘truth-speaker’).

4. Two Nouns in other relations: nightingale (A.S. nihte-gale = ‘night-singer’), nightmare (mara, ‘an incubus,’ by night).

II. Nouns are compounded with Adjectives.

1. Adjective and Substantive.

a. Nouns. Nobleman, upperhand, good-day, sometime, meanwhile, freeman, blackbird, long-measure, sweet-william, lucky-bag, midday, alderman (ealdor-man = ‘elder-man’), Gospel (god-spell = ‘good-spell’ = ‘good tiding’), holiday (= ‘holy day’), halibut (= ‘holy but’ = ‘holy plaice for eating on holy days’), hoar-frost, hoar-hound (the hoar or greyish húna, i.e. the plant now called horehound), hind-leg, neighbour (= ‘near-dweller’), midriff (mid + hrif = belly), titmouse (small sparrow; mouse here = A.S. máse, small bird, not the A.S. mûs from which the common word mouse).

b. Adjectives. Barefoot.

2. Substantive and Adjective.

a. Nouns. Furlong (= ‘furrow long’ = ‘the length of a furrow’).

b. Adjectives. In many of these the noun has very much the functions of an adverb. Blood-red, snow-white, fire-proof, shameful, beautiful, manly (i.e. ‘man-like’), scot-free (free from paying scot, i.e. a contribution).

3. Substantive and Participle.

a. Earth-shaking, heart-rending, life-giving, blood-curdling.

b. Airfed, earthborn, moth-eaten.193

4. Numeral + Substantive.

Sennight (= ‘seven night’), fortnight (‘fourteen night’), twi-light (= ‘double light’ = ‘doubtful light’).

III. Pronoun and Substantive. Self-will, self-esteem.

IV. 1. Substantive and Verb (or Verbal Stem).

Verbs. Back-bite, blood-let, brow-beat, hoodwink, caterwaul (= ‘to wail like cats’).

2. Verb and Substantive.

Nouns. Grindstone, bakehouse, wash-tub, pickpocket, brimstone (i.e. brenstone = ‘burning stone’), rearmouse (hrére-mús, hreran, ‘to flutter’), wormwood (A.S. wermód = weremód, werian, ‘to defend,’ mód = ‘mood’ = ‘mind;’ ‘that which preserves the mind’), breakfast, spend-thrift (cf. wast-thrift—Middleton, A Trick to Catche the Old One, II. i.).

V. Adjective + Adjective (or Adverb + Adjective; it is not always possible to decide which).

1. Old-English, Low-German, deaf-mute, thrice-miserable.

2. Adjective (or Adverb) + Participle.

a. Deep-musing, fresh-looking, ill-looking.

b. Dear-bought, full-fed, high-born, dead-beat.

(In well-bred, well-disposed, etc., there is, of course, no doubt that the first element is an adverb.)

VI. Adjective and Verb. White-wash.

VII. Adverb and Verb. Cross-question, doff (do-off), don (do-on).

Further compounds we meet are made up of—

VIII. Pronouns with Pronouns. Somewhat.

IX. Adverbs with Adverbs. Each (= á (aye) + gelic = like, A.S. aelc).

X. Adverbs with Pronouns. None (= ne + one), naught (= ne + aught).

XI. Adverbs with Prepositions. Therefrom.

XII. Adverbs with Adverbs. Henceforth, forthwith.

XIII. Prepositions with their Case. Downstairs, uphill, instead.

XIV. Adverbs with Verbs. Foretell, gainsay, withstand, etc.

We also find more than two members formed into one; such as man-o’-war, will-o’-the-wisp, brother-in-law, nevertheless, whatsoever, etc. Sentences and phrases coalesce; as in good-bye (= ‘God be with you’), the provincial beleddy (= ‘By our lady,’ i.e. the Virgin Mary), may-be (provincially in America written mebbe), and, aided by metaphorical usage, forget-me-not, kiss-me-quick, etc.

The student should carefully go over these examples, and, in each of them, attentively study the full force of the compound, and see what is really expressed by the component part, and what implied by the mere fact that they are thus joined.194 If he is acquainted with any foreign languages, he should also study all the various habits of these languages as regards composition. He will then gain a clear insight into the nature of the process, and see how impossible it is to fix a line of demarcation between compounds and syntactical combinations. This is further illustrated by the fact that much, which in one language is looked upon as a compound, in another is kept asunder; nay, in the same language one calls a compound what the other would count as two distinct words. Thus a German writes derselbe (= ‘the self,’ i.e. ‘the same’) as one word, whereas an Englishman writes the same; an Englishman writes himself where the German has, in two words, sich selbst. Cf. the Eng. long-measure with the Ger. langenmass; the Fr. malheureux (from malum augurium, ‘evil omen’) with the Eng. ill-starred, etc. It is this uncertainty, this vacillation, to which we owe the compromise of writing such combinations with a hyphen; e.g., a good-for-nothing. Though even this usage is not fixed and invariable; for one author will write, e.g., head-dress, another headdress, etc.

If there is no line of logical demarcation between compound and syntactical groups, no more is there a phonetic one. Misled by the fact that the words of a syntactical group are written asunder, and a compound written as one word, we might think that the members of such a compound were pronounced as though more intimately connected than those of a syntactical group. But combinations like those of article and noun, preposition and noun, are really pronounced as one continuous whole as much as any compound. Nor is there an essential difference in the accent, either in place or in force. Compare, for instance, with him and withstand or withdraw; the degree of strength (or perhaps rather the absence) of emphasis on the first word in Lord Randolph, Lord Salisbury, with that on the last ‘syllable’ in landlord; or, again, the quantity of stress we give to the preposition in the expression in my opinion with that on the first syllable of insertion. If the example of Lord Randolph v. landlord seemed to show that the PLACE of the accent has some significance, we have but to read the sentences Not Lord Randolph but Lady R. Churchill, or Not the landlord but the landlady spoke to the lodger, to find the accents in exactly the opposite relations and places. No special place of accent, then, is characteristic of a compound. A very instructive example we have in the compound Newfoundland. This is actually pronounced by various speakers in three different ways: one says Néwfoundland, another Newfóundland, and, again, another Newfoundlánd. What, then, makes every one feel this word, in all three pronunciations, to be compound? Nothing physiological, but simply and solely the psychological fact that the meaning of the group new-found-land has become specialised, and no longer corresponds to what once would have been a perfectly equivalent group, land-newly-discovered. Semasiological development and isolation is the criterion of a compound. What degree of such isolation is required cannot be stated in any hard and fast rule.

Such isolation can be effected in four different ways. (1) In the first place, the whole group, as such, can develop its meaning in a manner, or to a degree, not shared by the compound members. An example of this we saw just now in Newfoundland. (2) Or, again, the component parts, as separate words, may develop and change their meaning, without being followed in that development by the same words as part of the group. Thus, e.g., with originally meant against. This meaning it still has in withstand, whilst as a separate word it is not now used in that meaning. (3) Thirdly, the compound parts may become obsolete as separate words; as, for instance, ric in ‘bishopric’ (cf. supra, p. 317). (4) And lastly, the peculiar construction according to which the parts are connected or combined may become obsolete, surviving only in the formula, which thus becomes isolated. Thus, e.g., the genitive singular of feminine nouns can no longer be formed without s; hence Lady-day is now felt as a compound word, whilst ladies’-cloak or ladies’-house would not be so felt.

Though such isolation is necessary and may suffice to stamp a group as compound, we must not conclude that every group, where such isolation in one way or another has commenced, is ipso facto looked upon as a compound. Many considerations are here of importance, some of which will be brought out in a further study of some examples in which we can observe the commencement of the fusion.

The first step which a syntactical group takes on the road towards complete isolation and consequent fusion into a compound, is commonly the one we described under No. 1. in the former section. We must here distinguish two cases, which, though perhaps not easily distinguished in words, are yet clearly different.

An example will best serve to explain it. We have already more than once stated that in Lady-day the grammatical isolation of the genitive lady, as against the present genitive lady’s, serves to emphasise the fusion of the two parts into one compound. But we must not forget that this form of the genitive in this combination would not have been preserved if, at the time when the word lady by itself began to assume the genitive s—or, rather, began to follow analogically other genitives in s,—if, we say, the compound had not then already been isolated to a sufficient degree to protect the first component part against the influence which affected it when standing in other combinations. The absence of the s is therefore NOT the CAUSE of the isolation of the group, or the fusion of its parts. We must seek for that cause most likely in the fact that the genitive was, in this combination, used in a sense which always was or had become unusual. Lady-day, even when the form lady was still felt as genitive, would but mean ‘the day consecrated to the service of our Lady,’ or ‘the day sacred to our Lady.’ Now this use of the genitive must always have been an exceptional one. Never, for instance, could a man’s book or a lady’s cloak have had a similar meaning. It was therefore at first not so much the meaning of the component parts, as the MEANING EXPRESSED BY THEIR SYNTACTICAL CO-ORDINATION, which stood apart and became isolated. We see something of the same influence if we compare St. John’s wood and St. John’s Church. In the second group, the latter of the component parts has a meaning which suggests and helps to keep alive the correct meaning of the genitive-relation expressed by the flection of the former part. In St. John’s wood this is not so. This compound is therefore felt to be more intimately fused together than the other, and, while every one who uses the expression St. John’s Church thinks of the Saint who bore the name of John, but few speakers will do so in speaking of St. John’s wood. There is a very clear instance of this at hand in the German Hungersnot, lit. = hungersneed, i.e. ‘famine’ (need, suffering caused by hunger). Here the genitive with the word need has a very special sense, which, e.g., could not be expressed by the otherwise equivalent construction with of. ‘The need of hunger,’ if ever used in German, would be a very forced and uncommon way of expressing the idea ‘famine,’ a way which only a poet could adopt (die Not des Hungers). Here, then, again, it is not the sense of the words, but the sense of their syntactical relation which stands isolated.

On the other hand, if we consider forms like upstairs, always, altogether, we shall find that it is not this relation, but the whole meaning of the group as such, which has become isolated by development or specialisation of meaning. Upstairs has become equivalent to ‘on a floor of the building higher than we are now;’ always has been extended so as to include the relation of time, etc. This development has then generally given rise to what grammarians term ‘indeclinabilia,’ which sometimes, by secondary development have become capable of flection. Thus the German preposition zu (to, at), and the dative case frieden (peace), in a sentence like Ich bin zufrieden, gave rise to the compound zufrieden (lit. = ‘at peace’), ‘contented.’ When once the prepositional phrase at peace had developed into the adjective content, the compound was declined like other adjectives: ein zufriedener mann = ‘a contented man;’ etc.

Again, when the groups round-about and go-between had become nouns, they could be treated as such, and we find the plurals round-abouts and go-betweens.

The more highly a language is inflected, the less liable will the parts of a syntactical group be to fuse into one. It is much easier for a combination like Greenland or Newfoundland to pass into a real compound than for one like the German (das) rote Meer, ‘(the) Red Sea,’ though the amount of isolation of meaning is the same in both. Whether the group Green + land is nominative or dative or genitive, no change in the form of green occurs; in German, das rote Meer is nominative, des roten Meeres is genitive, dem roten Meer is dative. Every time one of the two latter cases is used, the addition of the flection n reminds us of the independence of the two words rot and Meer.

Just as by means of suffixes, etc., we derive new words from others, whether the latter are simple or compound forms (love, love-able; for-get, forget-able; etc.), so we sometimes find whole syntactical groups, which are not yet considered as having been fused into one compound, used with similar suffixes. Instances are: good-for-nothingness, a stand-off-ishness, a devil-may-carish face; That fellow is such a go-a-header; He is not get-at-able, etc., which no doubt scarcely belong to the literary language, but which show that the linguistic feeling of the speaker must have already apprehended these groups as unities; in other words, that the first step on the road towards welding them into a compound has been taken. A well-established instance appears in our ordinal numerals, such as one-and-twentieth, five-and-fortieth, etc.

A similar commencement of fusion we can observe in copulative combinations like wind and weather or town and country, as soon as the whole may be conceived as a single conception. In wind and weather this is the case, the two terms being in this combination SYNONYMOUS, describing the same object from different points of view. Other instances of this we have in bag and baggage, kith and kin, moil and toil, safe and sound, first and foremost, house and home, far and wide.195 In town and country, on the other hand, we have two elements which, whilst CONTRASTING, supplement one another. Such groups are old and young, heaven and hell, gown and town, big and small, rich and poor, hither and thither, to and fro, up and down, in and out. In a few, the same member is repeated; as, out and out, through and through, again and again, little by little. A careful consideration of the real meaning of such groups will show that, strictly speaking, these form a subdivision of our second class.

Inflected languages like German afford a criterion not applicable to English, as to the fusion of such combinations. We find there, for instance, a group—Habe und Gut (Etymol. = have, as a noun, for ‘property,’ and good = ‘chattels’), for ‘all a man’s possessions.’ The first of these nouns is feminine, and consequently ‘with all (his) belongings’ would be ‘mit aller Habe;’ Gut, on the other hand, is neuter, and requires the form (dative after mit) ‘mit allem Gut.’ Goethe has treated the group Hab’ und Gut as a neuter noun, and written ‘mit allem mobilen Hab’ und Gut’ (‘with all movable possessions’).

We have seen that groups like one and twenty, five and forty, etc., were really far advanced on the way of fusion, as was shown by the formation of the corresponding ordinals. In the case of those which begin with one, we have a further proof of this in the use of the plural noun, e.g. ‘one and twenty men.’

It will be readily felt that in expressions like a black and white dog, the group black and white really is in a similar state of fusion. We have but to separate the parts into two really independent words by the insertion of a second indefinite article, to see at once that ‘black and white’ is the description of one quality of one object, a compound word to express one (though not psychologically simple) conception.

So, again, the group one and all is sufficiently welded into one to resist, e.g., the insertion of the preposition of before its second part. Thus we should say It was for the good of one and all (i.e. for the entire community) and not of one and of all.

We may assume that complete fusion between the parts of such copulative groups would be more common if it were not checked by the connecting particle and. In some of the most common of these the accent of and has become so much depressed that the word becomes almost inaudible: cf. hare and hounds, half and half, etc. In combinations where the connecting particle has become unrecognisable in consequence of such phonetic sinking, it no longer resists the fusion. Thus, Jackanapes has become to all intents and purposes one word. It stands196 with the common preposition on, instead of of (cf. the very frequent use of this ‘on’ in Shakespeare and contemporaries), for Jack-of-apes, i.e., originally, ‘the man of the (or with the) [performing] apes,’ just as Jack-a-lantern stands for ‘Jack of the (or with the) lantern,’ etc. Combinations without any such connecting link pass, of course, all the more easily into compounds: cf. Alsace-Lorraine, as against such combinations as Naples and Sicily.

In the period of the Indo-European languages before inflections had taken their rise, or when they were not yet indispensable, the fusion into a ‘copulative compound’ (dvand-va) must have been simple and easy.

When a substantive has been specialised in meaning by being combined with an attributive, as blackbird, the combination may pass through all the changes of signification described in Chapter IV. without the uncombined substantive as such being affected. The result is commonly to make the combination richer in contents than the simple combination of the parts. Thus, by ‘a blackbird’ we understand the familiar songster to which we give the name, and no longer understand such birds as rooks, crows, etc., which might have been classed under the name ‘blackbird.’197 Further modifications may set in, which may cause the epithet, strictly interpreted, to become wholly inapplicable. Thus, ‘a butterfly’198 is applied to a whole class of insects quite irrespective of their colours. When we talk of the Middle Ages, we mean a strictly defined period of time, though no such definition is involved in the word middle. Privy Councillor denotes a definite rank; and the idea of privacy hardly enters into our heads as we pronounce the word: cf. also such expressions as the Holy Scriptures; the fine Arts; cold blood; Black Monday; Passion Week; the High School; the wise men from the East. It must be observed that the substantival determinants are only able to fuse with the word defined if they are employed in an abstract sense. This restriction does not, however, apply in the case of proper names.

A subdivision of this great class of words, thus specialised, is formed by common place-names which have become proper nouns by the aid of some determinant, itself possibly also unspecific. Such are the Red Sea, the Black Forest, Broadway, the Sublime Porte, the Watergate, the Blue Mountains, High Town, Beechwood, Broadmeadows, Coldstream, Troutbeck, Dog-island. It is similar, too, when an epithet attached as a distinguishing mark to a proper name comes to be apprehended as an integral portion of the proper name—in fact, as attaching to the individual; as, Richard the Humpback, Charles the Bald, William the Conqueror, Alexandra Land, the Mackenzie River, Weston-super-mare.

Compare also such compounds as Oldham, Littleton, Hightown, Lower-Austria, Great Britain.

The metaphorical application of a word is generally rendered intelligible by the context; especially and chiefly by the addition of a determinant: cf. ‘the head of the conspirators;’ ‘the heart of the enterprise;’ ‘the life of the undertaking;’ ‘the sting of death.’ Similarly, a determinant forming an element in a compound helps to render the metaphorical application intelligible; indeed, we are able by the aid of such a determinant to give to compounds a metaphorical sense, which we could hardly venture upon for the undetermined word alone: so, for instance, we give the name of German-silver to a material which we should not call merely silver; the name of sea-horse to what we would not call a horse: cf. further, sea-cow, elder-wine, ginger-beer, etc.

There are some cases, again, in which the compound has a proper, as well as a metaphorical meaning, and only as a compound acquires its metaphorical use: such are swallow-tail, negro-head, mothers’ joy, cuckoo-spittle, woolly bear, etc.

We have now to consider how syntactical and formal isolation contributes to further the fusion of the determinant with the determinate. If we compare two combinations such as kinsman with man-of-war, or man of deeds, we shall find that whilst the one has become an undoubted composition, the others are still groups of more or less independent parts. This is of course due to the fact that even now the word man is inflected, and that consequently the plurals, men of war and men of deeds, remind us of the fact that the first member of the group is an independent word. Formerly, when the flection was far more elaborate, this was, naturally, much more the case, and this alone would have sufficed to establish the feeling that, in compounds, the genitive which remained the same in all ‘cases’ of the compound had to precede. Of course, as long as flection sufficiently indicated the cases, both orders could be used in any group, but as then only such groups in which the genitive did precede became ‘compounds,’ those compounds became models, and the practice arose gradually and gradually became a rule. Another force then came to exert its influence in the same direction. In such genitival combinations it is, as a rule, the genitive which has the accent. When, then, this genitive was placed first, the whole group thereby resembled in accent the existing composites of the oldest formation, and so was more easily considered in the same light as these. The main cause must, however, be sought in a syntactical isolation, i.e., in our examples, an isolation in the construction of the article. As long as flectional terminations existed in their entirety, the Teutonic languages could dispense with the article before declined cases of nouns; in fact we may say the article did not exist, the demonstrative pronoun not yet having been degraded into what it became later on—a mere sign of case. Hence it was in old Teutonic languages quite possible, and a frequent practice, to use the genitive case of a noun alone without an article at all. We may be sure that this has also been true for the other cases. Phonetic decay, however, levelled the terminations of the other cases of a noun long before the genitive; and accusative and dative had long been alike (or very nearly so) at a time when in the masculine and neuter singular the genitive s was still preserved: in fact, as we know, in English it is all that has remained to us of the old flectional endings, with the exception of those s’s, in the plural which are original and not due to analogy. In that older stage of the language it was common to express an idea like the son of man by constructions just as in Ancient Greek, where the genitive stood between the article and the noun, which were both, of course, in the same case. Thus we find in Old High German, ther (NOM. SING. masc.) mannes sun (= ‘the man’s son’199). In Anglo-Saxon, Heofona rice ys gelíc ðám hiredes ealdre (‘of heaven’s (the) Kingdom is like the (DAT. sing.) household’s prince’). Gradually, however, the use of a noun without the article, largely, no doubt, owing to the levelling of all other cases, became more and more rare even in the genitive. Such rare standing expressions as remained without article, naturally assumed the appearance of compounds, and, especially in the case where the article belonging to the second noun preceded the genitive, the fusion was complete: the + kin’s + man became the + kinsman.200

We have already pointed out how the adjective and the noun entered into composition, and seen how, even in many combinations which we are not yet accustomed to look upon as fused into one, derivatives show that this fusion has at least partly been accomplished. Such are the many forms in ed, like black-eyed, etc., which are derived from the groups black eye, etc., and cannot be looked upon as compounds of black + eyed. We do not speak of an eyed person, for one who has eyes: cf. left-handed, self-willed, one-handed, etc.

In English, especially in Scottish dialects, many adverbs which commonly follow the verb, are occasionally made to precede it; as, to uplift, to backslide, etc. We may gather that in such forms no composition strictly so called has as yet set in, from the fact that the order is frequently transposed, as in sliding back, to lift up, etc. On the other hand, the fact that the words are joined in writing shows that the whole has begun to be apprehended as a unity.

In the case of most of these combinations we can trace the commencement of an isolation, which proves that the linguistic sense is ceasing to apprehend the elements as distinct. For instance, in English the old prepositional adverbs cannot be used independently and freely to form new combinations at will, but are confined to a definite group of combinations. Thus we can say, enfold and entwine: but not enthrow, for throw in. We can talk of onset, and onslaught, but not of on-run: of overflow, but not of over-pour. In many cases this isolation has led to a special development of meaning, and the word becomes still more definitely a compound; cf. such words as inroad, after-birth, offset, over-coat. From the union of the verb with the adverb, there arise nominal derivatives in which the sense is yet more specialised, such as offset, output, offal, under-writer.

An adverb derived from an adjective sometimes fuses with the nominal forms of the verb. The first impulse to this fusion is often given by the metaphorical application of one part of the compound: cf. deep-feeling, far-reaching, high-flying. The combination becomes even closer when the first part retains a meaning which has become unusual to it in general. For instance, in such a combination as ill-favoured, ill retains a trace of the time when it could be used as synonymous with bad.

In German, the comparative and superlative forms are actually used, showing the completeness of the fusion; as, der tieffühlendste Geist (Goethe), (lit. = ‘deep-feelingest ghost,’ i.e. ‘spirit’).

There are a few combinations of verbal-forms with an object accusative, which similarly occupy an intermediate position between the compound and the syntactic group; such as laughter-provoking, wrath-stirring, fire-spitting. No sharp line can be drawn between these instances of spontaneous and natural fusion, and the analogical formations coined by the poets; as sea-encompassed, storm-tossed, etc.

Again, and even in English, where the application of the inflected comparative and superlative is of so very limited application, it is the use of the comparative or superlative which affords a test as to the degree of fusion. It is, of course, possible to analyse most laughter-provoking, as provoking much laughter. But few would adopt such an explanation in a sentence like This is the most fire-spitting speech I ever heard.

Besides this, there are many verbal combinations which must be apprehended as compounds, from the fact that they represent a single notion only; such as with regard to, as soon as possible, forasmuch as, seeing that, none the less,—which must be considered to stand on the same footing as notwithstanding, nevertheless. This fusion is sometimes accompanied by a displacement of the psychological conception as to the parts of the sentence, whereby the natural mode of construction is altered, and the combination performs a new function, and becomes practically a different part of speech. For instance, we commonly hear I as good as promised it to them, where ‘as good as’ is nearly equivalent to ‘almost,’ and is construed like that adverb. We even meet with sentences like unclassified and prize-cattle, where a member of a compound is placed on the same footing as an independent word. Moreover, the first, or determinant member of the compound may be followed by determinants, as if it were itself independent; thus Milton can write hopeless to circumvent us; fearless to be overmatched: as if it had been ‘without hope to circumvent us;’ ‘having no fear to be overmatched.’ All this shows over and over again how completely impossible it is to draw the line between syntactical groups and compounds.

In this manner, then, syntactical isolation favours the fusion of a group into a compound. In our discussion of the form Jackanapes, we had already an instance how phonetic changes may have the same effect. This we shall now investigate and illustrate rather more in detail.

Though it would be impossible to prove the fact historically, it seems involved in the nature of the case that, for the most part, such phonetic changes at first arose in EVERY case of such closer and more intimate syntactical union; that they were re-adjusted and re-equalised later on, and were only preserved in groupings which, as a consequence of development of meaning, had become so far fused into one whole as to be capable of resisting the re-adjusting tendencies.

The simplest of such general effects of syntactical grouping is that the final consonant of a syllable is transferred in pronunciation to the next syllable. Thus, for instance, an apple is pronounced a-napple, without any pause; here + on is pronounced he + ron, etc. If, then, as in French, this final consonant disappears from pronunciation, save when thus made an initial, i.e. save before a word beginning with a vowel, we may expect its presence to have an isolating effect, and consequently to be sufficient to stamp the group as a compound. This, however, is only the case if such a preservation is not sufficiently frequent to be realised as a rule of pronunciation for all similar cases. In French, il peut = ‘he can,’ is pronounced without the t; in peut-être = ‘may be,’ ‘perhaps,’ the t is heard. Yet this has not isolated the form peut with t from the usual third person singular present indicative without t, because this t is preserved not in peut-être alone, or in a few such groups, but in all cases where the following word begins with a vowel; e.g., il peut avoir = ‘he can (may) have,’ pronounced with the t likewise. If we suppose the French language to discard at some time this liaison, as it is called, and always to pronounce peut without t even before vowels, then, and not till then, would the pronunciation peut-être with t stamp the combination as a compound.

So, again, the well-known process of avoiding hiatus by contraction or elision, in the case of a word ending in a vowel preceding one that begins with a vowel, has been sufficient to fuse two elements into one compound in many cases (e.g., about = a + be + ut (an); Lat. magnopere = magno + opere; Gothic sah, ‘this’ = sa + uh), but has no such effect in the case of the French article, or of the French preposition de, because the elision of the unaccented e and a is there an almost invariable and still ‘living’ rule.

A third general effect of close syntactical combination is the assimilation of a final and initial consonant. This, in present European languages, is scarcely, if at all, noticed or expressed in writing. It is, however, an exceedingly common occurrence in the spoken language, a fact of which every one can and ought to convince himself by a little attention to his own and other’s NATURAL pronunciation. It is only in cases where further reasons, in addition to this assimilation, such as, e.g., isolation by development of meaning or other phonetic development, have welded the group into a compound, or at least have advanced it a considerable distance on the road towards complete fusion, that the written language sometimes takes cognisance of the change, and, by the very spelling, indicates the compound nature of the group. We say ‘sometimes’ takes cognisance; for while spelling in no living language follows all the variations in pronunciation, no European tongue is further from accurately representing the spoken—that is, the real—language in its writing than English. Hence the instances even of acknowledged compounds, in which the assimilation in sound is indicated by the spelling, are comparatively rare. Such are gossib, for god + sib = ‘sib, or related, in God;’ leoman, for leof + man = ‘dear man;’ quagmire = quakemire, i.e. ‘quaking mire.’ Instances where the assimilation exists in pronunciation, but is not represented in writing, are plentiful: cupboard, pronounced cub-board (or cubberd); blackguard, pronounced blagguard, etc. In all these we must, on the one hand, admit with respect to the recognition of the group as compound, that, even if it has not promoted assimilation, it has at least checked the tendency to restore the theoretically correct pronunciation of the final consonant of the former member in each group. On the other hand, however, it is as certain that the very facility thus afforded to the working of the assimilating tendency has aided the phonetic isolation of the group and promoted the fusion.

The most effective cause of phonetic isolation, however, lies of course in the influence of accent. This has been sufficiently illustrated in the course of the foregoing discussions.

In all these discussions we have mainly regarded the transition of a syntactical group into a compound. Several of our examples, however, well illustrate the fact that, just as the fusion between the two members of some group may be insufficient to stamp the combination as a compound, so, also, such a compound loses its character as such for the consciousness of all but the student of language, when the fusion proceeds too far. The compound then becomes, to all intents and purposes, a simple word; it serves no more as model for analogical compounds with the same members, and at the very most gives the impression of having been ‘derived’ from its first member by a suffix. To instance this, we need only recall a few of our examples to the reader’s mind—bandog, auger, furlong, etc., or (with the suffixes) bishopric, kingdom, etc.

A careful study of these and similar examples will show that in the first-class of compounds, no longer recognised as such, sometimes both members have become obsolete, and in both classes almost always one.

We have now reached a point whence we can observe the conditions necessary to give birth to a suffix, or, if the phrase be preferred, necessary to degrade an independent word into a suffix.

We have seen a suffix originate in a noun which either (as in a case of ‘-ric’) became obsolete as an independent word, or whose connection with the etymologically identical independent form ceased to be felt in the linguistic consciousness of the community.

But such a fate may and does often befall a word without converting it into an acknowledged suffix. It has befallen the noun ðyrl (‘a hole’), in nostril (= nose-thirl), or the word búr (‘a dweller’) in neighbour (‘a near-dweller’), and yet neither -tril nor -bour have become recognised as suffixes in the English language.

What more, then, is required?

First of all, the first element must be etymologically perfectly clear; cf. kingdom, bishopric as against nos-tril, gos-sip.

Secondly, the second element must not occur in one or two combinations only, but in a sufficiently large group of words, in all of which it modifies the meaning of the first member in the same way; cf. nos-tril, gos-sip, as against ‘kingdom,’ ‘widowhood.’

This second condition can scarcely be fulfilled except in cases where—

Thirdly, the second element has originally, or in its combination with the others, some such abstract and general meaning as state, condition, quality, action, etc.

A few words on one of the best-known suffixes in English will make this clear. Though the phrase would hardly stand in written or literary language, we might indicate a dealer in pianos as the piano-man, i.e. ‘the man who has pianos.’ In the oldest stages of language, not only could a single noun be thus used with an almost adjectival force, but even a compound (or what was then still a syntactical co-ordination) of two or more nouns, or of adjective and noun, could be thus employed. Thus, e.g., in Sanscrit, a much-rice-king, would mean ‘a king who possesses much rice,’ i.e. ‘is rich;’ and the group man-shape (or its equivalent) might have been used for man-shape-having. Such compounds abound in Sanscrit, and could be formed at will. They were called Bahuvrîhi compounds. Now, without of course wishing to assert that the very combination man-ly is an original one, it is to such a combination of a noun with the noun which afterwards became lic in Anglo-Saxon that we owe the suffix ly. The phonetic differentiation and the development of meaning from shape-having to appearance or quality-having, isolated the member from its corresponding independent form (which in German and Dutch still exists as Leiche and Lyk = body or corpse), and gave us lic (later ly) as a suffix.

From all that we have said it must be clear that this process has gone on neither in prehistoric nor in historic times only, but is one which is repeated again and again, and consequently—seeing that prehistoric times are of unknown, but certainly enormous length—we must be on our guard against assuming that all these prototypes of Indo-Germanic suffixes must necessarily have existed at one time as independent words in the language, before the process which transformed them into suffixes began to operate. We may, nay, we are almost compelled to assume that there, too, they arose in succession, and that then as now, whenever phonetic decay or other causes had affected a suffix to such an extent as to take away the appearance of a derivative from what was once a compound, the suffix was no longer felt as such; it ceased to serve for new combinations, and another more weighty suffix took its function and supplanted it in all but a few remaining cases.

The most superficial knowledge of any modern language, or of Latin etymology, is sufficient to show that it is as impossible to draw a line between suffix and flectional termination, as between syntactical group and compound. Even a Frenchman, unless he has had the true historical explanation pointed out to him, feels in a future tense like j’aimerai, a verb-stem aim, and a termination -erai indicative of futurity, though, nowadays, there are but few students of French grammar who ignore the fact that aimerai is a compound of the infinitive aimer and the first person singular, present, indicative, ai = (I) have. Similarly, we may safely assume that few Romans felt in a pluperfect amaveram a perfect stem amav and eram the imperfect of sum, much less in amabo a present stem ama and a suffix derived from the same root as their perfect fu-i. It is certainly useless to illustrate this further.

We may now conclude with three observations, the truth of which will be apparent from what has gone before.

First. Even when an inflected form, by means of comparative study of all its oldest forms and equivalents in cognate languages, has been brought back to its prototype, and analysed into what are commonly considered to be its component parts, we must remember that these parts cannot have been fused into the integer which we now find made up of them, and yet have retained their original form and original meaning. Just as kingdoms has certainly not arisen from king + dom + s, a Greek optative pherois is not a compound of pher + o + i + s, though, undoubtedly, each of these elements have their regular representatives in other words of the same function, and most probably had their prototypes in fuller forms, in a more independent state. We have no means of knowing what these forms were, or what their original function was when still independent.

Second. Many words which we now consider as “simple” may have been compound or derivative. Our inability to further analyse does not prove primitive unity.

Third. In the history of Indo-European flection we do wrong if we assume the separate existence of a period of construction and one of decay.