¹ As an example of within vs. without, take the remark of H. Poutsma, A Grammar of Late Modern English (Groningen: P. Noordhoff, 1928), Part I, First Half, p. 390, that the arrangement We went to school together “could hardly be altered.” We note that to go to school is a stereotype with the sense ‘to attend school’; but, contrary to Poutsma, we can transfer together into the phrase, breaking it up, and destroying the stereotype. We went together to school is less likely in the sense ‘attended school’ and more likely in the sense ‘journeyed to school.‘

² We are caught in the same circumstantial web as when we try to determine the meaning of an intonation contour—the intonation is there, and we know that it means something, but to separate it from the particular locution with which it is used imposes a heavy burden of patience and abstraction.

³ Though my hypothesis has not so far as I know been explicitly announced in the literature of linguistics, there are a few hints of it. One is provided by Sapir, Language (New York: Harcourt Brace, 1921) : “It seems to me that there is a rather important psychological distinction between a language that settles the formal status of a radical element before announcing it … and one which begins with the concrete nucleus of a word and defines the status of this nucleus by successive limitations, each curtailing in some degree the generality of all that precedes. The spirit of the former has something diagrammatic or architectural about it, the latter is a method of pruning afterthoughts. In the more highly wrought prefixing languages the word is apt to affect us as a crystallization of floating elements, the words of the typical suffixing languages … are ‘determinative’ formations, each added element determining the form of the whole anew” (p. 135 n.). He calls such distinctions “elusive, yet important.” Sapir is speaking of the architecture of bound forms, not of the architecture of sentences, but the analogy is apparent, and one of my points (overrun vs. run over, ii, §8) would also be an example of his argument. See n. 9 for another hint, from Samuel Gili Gaya.

⁴ Modification in the most general sense can work backward as well as forward, just as two objects exert a mutual gravitational pull. I do not argue that backward modification is non-existent, but that forward modification is of a peculiar kind by being forward.

⁵ Lg., xxvi (1950), 238.

⁶ Progression is partly crowded out by competition from the está cantando construction, which causes Juan canta to favor the durative present.

⁷ In Lg., xxvi (1950), 240, the Kahanes reported that in Los tecolotes llegaron “the tecolotes are expected,” whereas in Llegaron los tecolotes they are not. This is too narrow an interpretation, for it depends on a uniform stress, but it nevertheless illustrates our point. The authors probably intend a like stress on tecolotes in both sentences, so that any extra “contrastiveness” that the word may possess will come about by its position, not by any other form of highlighting. Despite the stress on tecolotes in both sentences, it is more pointed, more contrastive, when it follows the verb.

⁸ It is just as nearly impossible to say in Spanish ¿Bien se portó o mal se portó?

⁹ Samuel Gilí Gaya, Curso Superior de Sintaxis Española §164 (Mexico: Minerva, 1943), discussing the hermoso example cited above, gives the nearest approximation to my hypothesis that I have encountered, and then, curiously enough, rejects it as central though accepts it as ancillary. “From the logical standpoint,” he says, “the element that follows determines the scope of the one that precedes. In un hermoso edificio the substantive restricts the quality set forth in the adjective by applying it to a particular object. On the other hand, in un edificio hermoso the adjective excludes from the general image of ‘building’ all buildings that are not beautiful.” This reasoning, he says, underlies Bello's view that following adjectives are “specifying.” His motive for considering it secondary is his belief that the “determining” function, which Bello had attributed to the position of the adjective, is largely subsumed by the articles and other determinative adjectives that the grammarians allowed to creep into their examples. He then gives sentences without such determinatives, for instance Valiosos cuadros adornaban el salón versus Cuadros valiosos adornaban el salón, with presumably only a “stylistic” difference, to prove that what is paramount is a “synthetic style” for pre-position and an “analytical style” for post-position. This so-called stylistic difference can be reduced, even in his examples without determinatives, however, to the same pattern of adjective-embracing-whole or adjective-splitting-whole: querying a Spanish colleague as to which phrase—valiosos cuadros or cuadros valiosos—would be used of a poorly-appointed, run-down art-gallery, I received the unhesitating reply cuadros valiosos. My colleague agreed that in valiosos cuadros the valiosos overflows the cuadros and extends to the rest of the gallery. Gili Gaya's impressionistic terms “synthetic” and “analytical” are excellent if we concentrate our attention only on the effect that adjective position has on the noun; we saw in the instance with the English adverb that abruptly back away is synthetic as far as the verb is concerned. These are special cases of linear modification.

Despite the example of the Spanish descriptive adjective to suggest it, no Spanish grammarian to my knowledge has hit upon the fact that the Spanish descriptive adverb exhibits the same traits as the English descriptive adverb. Gili Gaya seems to be alone in extending the analogy of the adjective a little way, in the direction of the gerundio (see below, n. 20). It is curious that the phenomenon has been so completely overlooked by grammarians of English, or perhaps not so curious if we reflect that the most comprehensive grammars of English have been written by non-natives, and the distinction is not a highly obvious one. Non-natives are handicapped by having to work with affirmative data only; but a pattern cannot be fixed unless in addition to what can be said we determine what cannot be said. Poutsma's principle, to which he keeps recurring in the body of his chapter on word order, is (387) : “The best way of throwing any element of the sentence into particular relief is to give it end-position… . The first words of the sentence, like the cautionary words of a command, put the listener on the alert. As the discourse proceeds, he is kept in suspense, so that his mind is prepared to receive that part of the communication on which his attention should chiefly be centered.” Poutsma also recognizes front-position for emphasis, but does not reason it further. “Emphasis,” “suspense,” and “relief” are vague terms; we need to know more, for instance, about the nature of the relief that is provided by end-position.

Curme, in Syntax §16 2 (New York: Heath, 1931), says that “an adverb can freely stand in almost any position except between a verb and its direct object… . This usage rests upon the principle that an adverbial element is usually more important than a direct object and, like important elements in general, gravitates toward the end.” He cites, among other examples, Yesterday I met your father and I met your father yesterday. In what sense yesterday is more “important” in one example than in the other is hard to see—in one it is more important by being more inclusive, in the other it is more important by being more selective. “Importance,” “stylistic difference,” and the like are traps. The adverb again may be both more “important” and more “emphatic” in Again he told me than in He told me again, but it precedes, nevertheless, for a reason that neither importance nor emphasis can explain.

¹⁰ Two different intonations to answer the two different questions.

¹¹ Or, with loud stress on halted as well as on advancing, we intending to go on ourselves to add something like And the retréating soldiers were thrown into a rôut—in any case, contrastive stress on a pre-posed adjective seems to refer to some other verbally expressed element.

¹² The adverbs that I have included mainly give a mental picture of an action. Though I leave out subordinating adverbial conjunctions of the when-where-as-because type, since they are fixed in a single position, I include non-subordinating conjunctive adverbs of the afterward-then-therefore type. The latter behave much like descriptive adverbs—which is just as well, since drawing a sharp line between them is practically impossible.

¹³ Time, 23 Jan. 1950, p. 46. The writer not perfectly versed in English betrays himself in the following from J. Gonda, Lingua, viii (1950), 316: “Notwithstanding his shortcomings the disinterested and idealistic Swiss scholar has essentially contributed to the advancement of Indonesian studies.” As it stands, this is probably a feebler compliment than the author intended.

¹⁴ Commands are typically short. We are not likely to combine two orders in one. Thus in Do it carefully, the do it part is almost certainly resumptive, carefully being the only new datum. If the do it is also a new datum, the dual command will probably take some such shape as Do it, and be careful about it.

¹⁵ Cortésmente me abrió la puerta implies ‘cortés to be expected.‘ Furiosamente se abalanzó sobre mi produces the same effect as furioso león.

¹⁶ Bull. Amer. Assn. of Univ. Professors, xxxvi (1950), 684.

¹⁷ There is a further side-effect in the fact that agreeable generally admits of two types of selective contrast: ‘agreeable in all ways instead of just some ways’ and ‘agreeable in an over-all sense but maybe specifically disagreeable.‘

¹⁸ An example where the adverb has been half captured is Well he knows that it is not true, He well knows that it is not true, He knows (perfectly) well that it is not true, He knows thai it is not true (perfectly) well. Here there is partial interference from intonation. Many adverbs of time may be captured by one or the other of two or more verbs: Often I hope he will suffer, I often hope he will suffer, I hope he will suffer often.

¹⁹ The term “conjunctive adverb” does not satisfy me, but as it is not material to the thesis I copy it from Poutsma (e.g., p. 444).

²⁰ He says (§146) the gerundio “is generally placed after the verb; but it can go before, and in this case the adverbial modification takes on … a relief similar to that of the adjective put before its noun.”

²¹ There may be a partial side-effect here. My feeling is that when we say It's fun playing golf the it has to some extent retained the status of a true subject, and is not merely a vicarious subject. We might, for example, ask, Is it fun? (it a true subject=‘what you are doing‘), and receive the reply, Yes, it's fun playing golf. In this way it is possible for the effect that we observed in the participle-as-adverb to carry over to the participle-as-noun.

²² It might be argued that contiguity with the verb rather than the relationship of before-after is responsible for this. If so, the examples lose their value, though the theory is not contradicted.

²³ The contrary effect may be noted in the manner in which a terminal adverb weakens a contrast on some other element even when that element bears contrastive stress. Thus Put your hát on is less effective as regards singling out hat than is Put on your hát.

²⁴ Only can be made captive in a variety of ways. In Only, I saw my friend yesterday it is ‘except for the fact that.’ In Only Í saw my friend yesterday it is ‘I alone.’ In I only saw my friend yesterday there are various major implications according to stress, plus a minor one due to position. In I saw only my friend yesterday there are two major implications according to stress (mý or friénd) plus a minor one due to position, and the latter implication competes with I saw my friend ónly [optional pause] yesterday. In I saw my only friend yesterday, only is captured as an adjective. In I saw my friend only yésterday we get either ‘no other time but yesterday,’ competing in its order implications with some of the foregoing, or the stereotyped ‘no longer ago than yesterday.’ In I saw my friend yesterday only we get ‘no other time but,’ with competing order implications.

²⁵ The first is an announcement of mere result; we might expand to She got into the habit of insisting so, that as usually happens in such cases she got her way. The second is more: an interpolation pointing to mere result is less suitable.

²⁶ Part iii Syntax 2nd vol. 10.3₅.

²⁷ Another relevant effect is noted by Poutsma (p. 472) : that end-position of the preposition is sometimes avoided “when the verb or nominal is felt to be of particular weight.” Thus in “On some of the most important issues of modern life there is no free discussion within the groups on whose decisions everything depends.” It is desired to focus on depends at the narrow end of the sentence, and a following on would dilute this, even though depends is still stressed.

²⁸ The entire paper, and this section of it particularly, has benefited from the detailed and painstaking criticism of Professor Anna Granville Hatcher of Johns Hopkins University.

²⁹ See n. 11, above, and accompanying text for the effect of contrastive stress on pre-posed advancing, analogous to the effect of contrastive stress on pre-posed everlasting here. The two examples just given correspond respectively to a theologian's and an astronomer's view of the world. The latter speaks of a world whose existence is the very measure of time—whence everlasting is expletive.

³⁰ One must guard against distracting elements such as occur in Poutsma's example If the nation will rise to the occasion and make the united effort necessary, it can overcome this German machination. We “feel” that the position gives the same effect here as in the example in the body of the text (the money necessary), but discover that selective contrast seems to be ruled out by the fact that in both the united effort necessary and the necessary united effort only one particular effort is under consideration. The reason is the presence of the added modifier united, which of itself particularizes the noun and which must, by the mechanics of the English sentence, go in pre-position. In order not to be misled we must then either strip away the surplus adjective, or bear in mind that both united and necessary particularize, but that united is fixed in position.

³¹ At the risk of laboring the point, I again insert the caution that He is the pure musician, with contrastive stress on pre-posed pure and with musician de-stressed, is iterative.

³² Jespersen, Syntax, 1st vol. 15.54 (Heidelberg, 1914).

³³ A troublesome case of grouped adjectives is provided by comparisons of inequality. In We use a fairly bright light in indoor photography, like this one, because a dimmer light than this would slow the photograph too much, I feel that dimmer in pre-position is slightly inappropriate because of the faint suggestion that the alternative light would be actually dim, whereas the context says that it probably would not be. A better contrast seems to be got by saying a light dimmer than this, dim in post-position being more contrastive. Similarly in For this job we need an older man than John, which suggests ‘really old,’ as against For this job we need a man older than John, which leans more to the purely contrastive, relative sense of old. It is hard to manipulate comparisons like these for three reasons. First, all comparatives imply a contrast regardless of their position. Second, ungrouped comparatives are required to precede most nouns (a dimmer light, not a light dimmer), and grouped ones assimilate to them. Third, there is a side-effect from the use of such expressions to pay compliments. In There is no prettier girl here than Mary, pretty overspreads and makes Mary a pretty girl; but There is no girl here prettier than Mary can be either the matter-of-fact ‘and she is not necessarily pretty’ or the hyperbolic ‘none conceivably prettier than she.‘ An identical effect is observable in the Spanish deliciosas comidas, which makes delicious non-differentiating and so implies a compliment, against comidas deliciosas which may be either an insult or a hyperbolic compliment.

³⁴ The latter paraphrases Poutsma 498.

³⁵ The expression Es un infame hombre does not appear in the textbooks, but is good colloquial Spanish, such as might be used by a woman complaining to a judge about the treatment she received from her husband—is more emotional, less matter-of-fact than Es un hombre infame, where infante has less spread. The same would be true of the grouped Es un perverso e infame hombre. In Es un estudiante pervertido we get the figure

‘a student who happens to be perverted’; Es un pervertido estudiante gives

‘a perverted one who happens to be a student‘—perverted in all senses, not merely qua student. It might be argued that since Spanish can convert almost any descriptive adjective to a noun, in the second construction we have the equivalent of Es un pervertido, with another noun in apposition, which is therefore not comparable with Es un estudiante pervertido, where pervertido has to be an adjective. The argument can be countered in part, however, by casting about for an adjective that would be unusual as a noun; thus Es un inicuo, though possible, is distinctly unusual, while even the person unfamiliar with the meaning of the word would accept Es un inicuo hombre. This seems to suggest that in Es un pervertido hombre the pervertido is an adjective and so is comparable with hombre pervertido. In any case our main thesis is unaffected, for even with noun-noun combinations the effect of linear modification may be noted: mi viuda madre refers to only one mother, a widow, while mi madre viuda would give a ludicrous selective contrast (Gili Gaya §164).

³⁶ Compare money ready, a momentary state, and ready money, a prearranged class.

³⁷ To put it subjectively, the “characterizing” -able words give a mental picture of the thing; the others do not.

³⁸ If Citizens replaces All persons, the result is somewhat un-English. See the next section for the reason, person being a more inclusive noun.

³⁹ Non-connectedness is made explicit by certain interpolations: Mary being beautiful, Having stick countries friendly, All persons found loyal. It is commonest of all in the quasi-predicatives I saw John unhappy, We found the report unsuitable, etc. The pattern probably evolved from these types.

⁴⁰ We have noted the comparatives as one set of grouped adjectives. In the contrast between A lovelier face I have never seen and A face more lovely I have never seen the general principle is clear, as is also the suggestion of ‘striking unexpectedness’ (when coupled with the never phrase) offered by the post-position. In ordinary constructions, such as You need a more intelligent manager, the comparative is more likely to precede, though, in the nature of the case, comparatives are more detachable than non-comparatives and more readily follow. Note that the -er comparatives, except in the type discussed in n. 33, regularly precede; they are the most-used adjectives, those representing qualities most likely to be embalmed in standard types.

⁴¹ Spanish post-position is not limited to “detachability,” the latter being relegated to a totally different pattern (ser-estar), whence “standard type” is not excluded from postposition. The chief use of pre-position is a special kind of “norm”: that of known or assumed-to-be-known characteristic (cf. n. 15 and corresponding text). Just as English postposition is an extreme, so Spanish pre-position is an extreme: that of “standard” which given a setting is so standard that just mentioning the noun is supposed to suggest the adjective (as in world everlasting versus everlasting worl d). Chopping up the spectrum so as to fit the two languages, we get the following:

⁴² Charles Hockett in Lg., xxvii (1951), 441.

⁴³ Poutsma (506) attributes the pre-position in At last one of the advancing bulls stood still to there being “no time association” in the participle when it precedes the noun. In many instances the time association of the post-posed participle is obvious, as in working man and man working; but the advancing bulls example is a poor one because post-position of the participle has as its primary effect a selective contrast; there is potential time association in both arrangements. This is best shown by a ludicrous rearrangement of another of Poutsma's examples, She was now quite alone with the dying Henry, where unless there are two Henrys no other arrangement is possible—time association is secondary.

⁴⁴ However one may define “compound,” the compounds that admit of simple change of position without a change of meaning too radical to apply the theory, are few indeed. Heavier-than-air craft is ‘craft heavier than air,’ reversible because grouped modifiers take both positions more freely—it is ‘craft characteristically heavier than air’ just as under-the-counter-sale is ‘a sale characteristically under the counter.‘ But the stress-defined compound drummer boy, though it refers to a boy who is a drummer, is not the same as boy drummer, nor may other similarly stress-defined noun-noun compounds like beggarwoman, teacher lady, washerwoman, soldier boy, etc., be matched with their reversals. The meaning-defined black death, horned owl, heavy water, etc., cannot be reversed. Such varied extra-linguistic considerations seem to influence the order of the “dual capacity” meaning-defined compounds described by Hatcher (Modern English Word-Formation and Neo-Latin, Baltimore, 1951) that it is difficult if not impossible to draw any conclusions: the sander-polisher, harvester-thresher type gives the natural sequence of the two combined operations, and so cannot be reversed; the theater-auditorium type cannot be reversed without suggesting a proper name; lend-lease and lease-lend were interchangeable, though the latter was less frequent.

As might be expected, since the type underseas craft is stereotyped as to position, in order to get a non-iterative stress on underseas we must re-word just as with vacant house. If a person ignorant of the fact that a submarine is a vessel asks what a submarine is, he will not be told It is an underséas craft, for that would presuppose craft, but will be told It is a ship that runs under the water—the function of non-iterative selective contrast is taken over by a clause.

In all likelihood the participle that enters into the formation of a compound is contaminated with the sense of gerund plus noun, if, indeed, we can tell which is which in many cases. Thus sewing machine is not only ‘a machine that sews,’ but also ‘a machine for sewing,’ just as reading lamp is unambiguously ‘a lamp for reading.’ Though the missing link is clearly a participle, and a losing battle probably is one, it would be hard to say how to class a working arrangement.

⁴⁵ Although combined with other modifiers there may be a selective contrast, e.g., The jewels stolen were the ones she prized most of her whole collection, where the gives stolen the power to distinguish jewels from jewels, there must always be an implied self-contrast as well. The same “jewel from jewel” distinction is present in The stolen jewels were the ones she prized most of her whole collection, but the example changes because of the missing “previous state versus present state” contrast.

⁴⁶ Professor Hatcher notes that these have an element in common: they are gnomic predications with generic subject.

⁴⁷ Pre-position assimilates participles to ordinary adjectives; post-position assimilates ordinary adjectives to participles. Note the “adjective” sense of C and H is the best refined sugar against C and H is the best sugar refined (fallen woman—woman fallen; a given number—a number given; a studied remark—a remark studied). In He is an easily-satisfied person the participle is not only adjectivized but broadened to suggest ‘easy-going,’ ‘friendly,’ ‘comfortable,’ etc. He is a person easily satisfied is verbal and restricted.

⁴⁸ In the absence of iterative stress on appointed.

⁴⁹ Poutsma (508-509) is unsure of any distinction here.

⁵⁰ Poutsma's remark (523) that “Late Modern English hardly tolerates an attributive participle after a noun that is modified by a possessive pronoun or genitive” is pertinent as an incidental result of post-position for selective contrast (bear in mind that “attributive” here means ‘adjunct‘). His example is He heard his dear and doted-on Mary Ann say, where doted-on cannot follow the noun. Since a possessive usually pins a thing down so as to make it unique (and moreover in this instance the proper noun clinches it), to split it by means of post-position is not common. Rather than make this a mechanical rule applying to possessives, however, we must realize that it is due to the nature of linear modification. In Our two articles accepted were both on natural science we have post-position despite the presence of a possessive, because thing-to-thing contrast is possible.

⁵¹ Poutsma denies this (519), citing He was a liar born to disprove it; he reasons that one cannot be bom a liar, hence the literal-figurative dichotomy won't work. But he forgets that it is always possible to employ a literal form rather than a figurative one for the purpose of a special figure: hyperbole. If I can say He was literally a liar from the day he was born I can also use the equivalent and equally manifest exaggeration He was a liar born.

⁵² The potential difference is not always actualized; thus with arms folded and with folded arms do not show the striking difference of head erect and erect head. Its absence is an accident. On the one hand, there is no creature characterized by folded arms as there are animals characterized by erect heads, so that a radical difference in meaning is impossible. On the other hand, there is a standard pose “folded arms,” which therefore admits the “standard” position of the adjective.

⁵³ There is a side-effect with myself taken as a word-modifier of I alone: ‘even I seldom do it that way.’ This depends on intonation. To highlight the difference, change the locution to I myself don't feel that I have the right to do it that way. With I and my at fairly high pitch and with falling pitch and loud stress on -self, or with I and my at mid-pitch and -self at high pitch and stressed, the implication is ‘even I.’ With the latter conditions but with -self falling slightly in pitch and then rising slightly (or with some surrogate of rising-falling-rising pitch, such as the fall alone accompanied by a slowing down of -self), the implication is as in the text. ‘Even I’ is again implied when myself is terminal and there is a steady rise in pitch and no stresses until -self, which is stressed on a rising-falling pitch. In both cases there is contrast with some other “self,” which is sharpened by end-position; in one case the meaning is ‘why should anyone else feel that he has the right to do it that way?’ and in the other it is ‘I concede another's right to do it that way.’

⁵⁴ The same effect may be observed in Spanish with Juan canta, putting contrastive stress on Juan so that canta is resumptive. In Canta Juan both elements are new data, unless canta is given contrastive stress, in which case Juan is resumptive.

⁵⁵ Or, as with sentence adverbs in general, put Tuesdays at the end following a pause.

⁵⁶ No more than we would find “How did you do it?”—“Carefully I did it.”

⁵⁷ I suspect, though the two do not automatically condition each other, that linear modification is one reason for a primary feature of the English sentence, viz., the tendency toward loud stress at the end. If the “point” of the sentence is there, so will the stress be, more often than not. The same is true of Spanish.