Edward Sapir (1884–1939). Language: An Introduction to the Study of Speech. 1921.
Chapter 4Form in Language: Grammatical Processes
T
Every language possesses one or more formal methods for indicating the relation of a secondary concept to the main concept of the radical element. Some of these grammatical processes, like suffixing, are exceedingly wide-spread; others, like vocalic change, are less common but far from rare; still others, like accent and consonantal change, are somewhat exceptional as functional processes. Not all languages are as irregular as English in the assignment of functions to its stock of grammatical processes. As a rule, such basic concepts as those of plurality and time are rendered by means of one or other method alone, but the rule has so many exceptions that we cannot safely lay it down as a principle. Wherever we go we are impressed by the fact that pattern is one thing, the utilization of pattern quite another. A few further examples of the multiple expression of identical functions in other languages than English may help to make still more vivid this idea of the relative independence of form and function.
In Hebrew, as in other Semitic languages, the verbal idea as such is expressed by three, less often by two or four, characteristic consonants. Thus, the group sh-m-r expresses the idea of “guarding,” the group g-n-b that of “stealing,” n-t-n that of “giving.” Naturally these consonantal sequences are merely abstracted from the actual forms. The consonants are held together in different forms by characteristic vowels that vary according to the idea that it is desired to express. Prefixed and suffixed elements are also frequently used. The method of internal vocalic change is exemplified in shamar “he has guarded,” shomer “guarding,” shamur “being guarded,” shmor “(to) guard.” Analogously, ganab “he has stolen,” goneb “stealing,” ganub “being stolen,” gnob “(to) steal.” But not all infinitives are formed according to the type of shmor and gnob or of other types of internal vowel change. Certain verbs suffix a t-element for the infinitive, e.g., ten-eth “to give,” heyo-th “to be.” Again, the pronominal ideas may be expressed by independent words (e.g., anoki “I”), by prefixed elements (e.g., e-shmor “I shall guard”), or by suffixed elements (e.g., shamar-ti “I have guarded”). In Nass, an Indian language of British Columbia, plurals are formed by four distinct methods. Most nouns (and verbs) are reduplicated in the plural, that is, part of the radical element is repeated, e.g., gyat “person,” gyigyat “people.” A second method is the use of certain characteristic prefixes, e.g., an’on “hand,” ka-an’on “hands”; wai “one paddles,” lu-wai “several paddle.” Still other plurals are formed by means of internal vowel change, e.g., gwula “cloak,” gwila “cloaks.” Finally, a fourth class of plurals is constituted by such nouns as suffix a grammatical element, e.g., waky “brother,” wakykw “brothers.”
From such groups of examples as these—and they might be multiplied ad nauseam— we cannot but conclude that linguistic form may and should be studied as types of patterning, apart from the associated functions. We are the more justified in this procedure as all languages evince a curious instinct for the development of one or more particular grammatical processes at the expense of others, tending always to lose sight of any explicit functional value that the process may have had in the first instance, delighting, it would seem, in the sheer play of its means of expression. It does not matter that in such a case as the English goose—geese, foul—defile, sing—sang—sung we can prove that we are dealing with historically distinct processes, that the vocalic alternation of sing and sang, for instance, is centuries older as a specific type of grammatical process than the outwardly parallel one of goose and geese. It remains true that there is (or was) an inherent tendency in English, at the time such forms as geese came into being, for the utilization of vocalic change as a significant linguistic method. Failing the precedent set by such already existing types of vocalic alternation as sing—sang—sung, it is highly doubtful if the detailed conditions that brought about the evolution of forms like teeth and geese from tooth and goose would have been potent enough to allow the native linguistic feeling to win through to an acceptance of these new types of plural formation as psychologically possible. This feeling for form as such, freely expanding along predetermined lines and greatly inhibited in certain directions by the lack of controlling types of patterning, should be more clearly understood than it seems to be. A general survey of many diverse types of languages is needed to give us the proper perspective on this point. We saw in the preceding chapter that every language has an inner phonetic system of definite pattern. We now learn that it has also a definite feeling for patterning on the level of grammatical formation. Both of these submerged and powerfully controlling impulses to definite form operate as such, regardless of the need for expressing particular concepts or of giving consistent external shape to particular groups of concepts. It goes without saying that these impulses can find realization only in concrete functional expression. We must say something to be able to say it in a certain manner.
Let us now take up a little more systematically, however briefly, the various grammatical processes that linguistic research has established. They may be grouped into six main types: word order; composition; affixation, including the use of prefixes, suffixes, and infixes; internal modification of the radical or grammatical element, whether this affects a vowel or a consonant; reduplication; and accentual differences, whether dynamic (stress) or tonal (pitch). There are also special quantitative processes, like vocalic lengthening or shortening and consonantal doubling, but these may be looked upon as particular sub-types of the process of internal modification. Possibly still other formal types exist, but they are not likely to be of importance in a general survey. It is important to bear in mind that a linguistic phenomenon cannot be looked upon as illustrating a definite “process” unless it has an inherent functional value. The consonantal change in English, for instance, of book-s and bag-s (s in the former, z in the latter) is of no functional significance. It is a purely external, mechanical change induced by the presence of a preceding voiceless consonant, k, in the former case, of a voiced consonant, g, in the latter. This mechanical alternation is objectively the same as that between the noun house and the verb to house. In the latter case, however, it has an important grammatical function, that of transforming a noun into a verb. The two alternations belong, then, to entirely different psychological categories. Only the latter is a true illustration of consonantal modification as a grammatical process.
The simplest, at least the most economical, method of conveying some sort of grammatical notion is to juxtapose two or more words in a definite sequence without making any attempt by inherent modification of these words to establish a connection between them. Let us put down two simple English words at random, say sing praise. This conveys no finished thought in English, nor does it clearly establish a relation between the idea of singing and that of praising. Nevertheless, it is psychologically impossible to hear or see the two words juxtaposed without straining to give them some measure of coherent significance. The attempt is not likely to yield an entirely satisfactory result, but what is significant is that as soon as two or more radical concepts are put before the human mind in immediate sequence it strives to bind them together with connecting values of some sort. In the case of sing praise different individuals are likely to arrive at different provisional results. Some of the latent possibilities of the juxtaposition, expressed in currently satisfying form, are: sing praise (to him)! or singing praise, praise expressed in a song or to sing and praise or one who sings a song of praise (compare such English compounds as killjoy, i.e., one who kills joy) or he sings a song of praise (to him). The theoretical possibilities in the way of rounding out these two concepts into a significant group of concepts or even into a finished thought are indefinitely numerous. None of them will quite work in English, but there are numerous languages where one or other of these amplifying processes is habitual. It depends entirely on the genius of the particular language what function is inherently involved in a given sequence of words.
Some languages, like Latin, express practically all relations by means of modifications within the body of the word itself. In these, sequence is apt to be a rhetorical rather than a strictly grammatical principle. Whether I say in Latin hominem femina videt or femina hominem videt or hominem videt femina or videt femina hominem makes little or no difference beyond, possibly, a rhetorical or stylistic one. The woman sees the man is the identical significance of each of these sentences. In Chinook, an Indian language of the Columbia River, one can be equally free, for the relation between the verb and the two nouns is as inherently fixed as in Latin. The difference between the two languages is that, while Latin allows the nouns to establish their relation to each other and to the verb, Chinook lays the formal burden entirely on the verb, the full content of which is more or less adequately rendered by she-him-sees. Eliminate the Latin case suffixes (-a and -em) and the Chinook pronominal prefixes (she-him-) and we cannot afford to be so indifferent to our word order. We need to husband our resources. In other words, word order takes on a real functional value. Latin and Chinook are at one extreme. Such languages as Chinese, Siamese, and Annamite, in which each and every word, if it is to function properly, falls into its assigned place, are at the other extreme. But the majority of languages fall between these two extremes. In English, for instance, it may make little grammatical difference whether I say yesterday the man saw the dog or the man saw the dog yesterday, but it is not a matter of indifference whether I say yesterday the man saw the dog or yesterday the dog saw the man or whether I say he is here or is he here? In the one case, of the latter group of examples, the vital distinction of subject and object depends entirely on the placing of certain words of the sentence, in the latter a slight difference of sequence makes all the difference between statement and question. It goes without saying that in these cases the English principle of word order is as potent a means of expression as is the Latin use of case suffixes or of an interrogative particle. There is here no question of functional poverty, but of formal economy.
We have already seen something of the process of composition, the uniting into a single word of two or more radical elements. Psychologically this process is closely allied to that of word order in so far as the relation between the elements is implied, not explicitly stated. It differs from the mere juxtaposition of words in the sentence in that the compounded elements are felt as constituting but parts of a single word-organism. Such languages as Chinese and English, in which the principle of rigid sequence is well developed, tend not infrequently also to the development of compound words. It is but a step from such a Chinese word sequence as jin tak “man virtue,” i.e., “the virtue of men,” to such more conventionalized and psychologically unified juxtapositions as t’ien tsz “heaven son,” i.e., “emperor,” or shui fu “water man,” i.e., “water carrier.” In the latter case we may as well frankly write shui-fu as a single word, the meaning of the compound as a whole being as divergent from the precise etymological values of its component elements as is that of our English word typewriter from the merely combined values of type and writer. In English the unity of the word typewriter is further safeguarded by a predominant accent on the first syllable and by the possibility of adding such a suffixed element as the plural -s to the whole word. Chinese also unifies its compounds by means of stress. However, then, in its ultimate origins the process of composition may go back to typical sequences of words in the sentence, it is now, for the most part, a specialized method of expressing relations. French has as rigid a word order as English but does not possess anything like its power of compounding words into more complex units. On the other hand, classical Greek, in spite of its relative freedom in the placing of words, has a very considerable bent for the formation of compound terms.
It is curious to observe how greatly languages differ in their ability to make use of the process of composition. One would have thought on general principles that so simple a device as gives us our typewriter and blackbird and hosts of other words would be an all but universal grammatical process. Such is not the case. There are a great many languages, like Eskimo and Nootka and, aside from paltry exceptions, the Semitic languages, that cannot compound radical elements. What is even stranger is the fact that many of these languages are not in the least averse to complex word-formations, but may on the contrary effect a synthesis that far surpasses the utmost that Greek and Sanskrit are capable of. Such a Nootka word, for instance, as “when, as they say, he had been absent for four days” might be expected to embody at least three radical elements corresponding to the concepts of “absent,” “four,” and “day.” As a matter of fact the Nootka word is utterly incapable of composition in our sense. It is invariably built up out of a single radical element and a greater or less number of suffixed elements, some of which may have as concrete a significance as the radical element itself. In the particular case we have cited the radical element conveys the idea of “four,” the notions of “day” and “absent” being expressed by suffixes that are as inseparable from the radical nucleus of the word as is an English element like -er from the sing or hunt of such words as singer and hunter. The tendency to word synthesis is, then, by no means the same thing as the tendency to compounding radical elements, though the latter is not infrequently a ready means for the synthetic tendency to work with.
There is a bewildering variety of types of composition. These types vary according to function, the nature of the compounded elements, and order. In a great many languages composition is confined to what we may call the delimiting function, that is, of the two or more compounded elements one is given a more precisely qualified significance by the others, which contribute nothing to the formal build of the sentence. In English, for instance, such compounded elements as red in redcoat or over in overlook merely modify the significance of the dominant coat or look without in any way sharing, as such, in the predication that is expressed by the sentence. Some languages, however, such as Iroquois and Nahuatl, employ the method of composition for much heavier work than this. In Iroquois, for instance, the composition of a noun, in its radical form, with a following verb is a typical method of expressing case relations, particularly of the subject or object. I-meat-eat, for instance, is the regular Iroquois method of expressing the sentence I am eating meat. In other languages similar forms may express local or instrumental or still other relations. Such English forms as killjoy and marplot also illustrate the compounding of a verb and a noun, but the resulting word has a strictly nominal, not a verbal, function. We cannot say he marplots. Some languages allow the composition of all or nearly all types of elements. Paiute, for instance, may compound noun with noun, adjective with noun, verb with noun to make a noun, noun with verb to make a verb, adverb with verb, verb with verb. Yana, an Indian language of California, can freely compound noun with noun and verb with noun, but not verb with verb.
On the other hand, Iroquois can compound only noun with verb, never noun and noun as in English or verb and verb as in so many other languages. Finally, each language has its characteristic types of order of composition. In English the qualifying element regularly precedes; in certain other languages it follows. Sometimes both types are used in the same language, as in Yana, where “beef” is “bitter-venison” but “deer-liver” is expressed by “liver-deer.” The compounded object of a verb precedes the verbal element in Paiute, Nahuatl, and Iroquois, follows it in Yana, Tsimshian, and the Algonkin languages.
Of all grammatical processes affixing is incomparably the most frequently employed. There are languages, like Chinese and Siamese, that make no grammatical use of elements that do not at the same time possess an independent value as radical elements, but such languages are uncommon. Of the three types of affixing—the use of prefixes, suffixes, and infixes—suffixing is much the commonest. Indeed, it is a fair guess that suffixes do more of the formative work of language than all other methods combined. It is worth noting that there are not a few affixing languages that make absolutely no use of prefixed elements but possess a complex apparatus of suffixes. Such are Turkish, Hottentot, Eskimo, Nootka, and Yana. Some of these, like the three last mentioned, have hundreds of suffixed elements, many of them of a concreteness of significance that would demand expression in the vast majority of languages by means of radical elements. The reverse case, the use of prefixed elements to the complete exclusion of suffixes, is far less common. A good example is Khmer (or Cambodgian), spoken in French Cochin-China, though even here there are obscure traces of old suffixes that have ceased to function as such and are now felt to form part of the radical element.
A considerable majority of known languages are prefixing and suffixing at one and the same time, but the relative importance of the two groups of affixed elements naturally varies enormously. In some languages, such as Latin and Russian, the suffixes alone relate the word to the rest of the sentence, the prefixes being confined to the expression of such ideas as delimit the concrete significance of the radical element without influencing its bearing in the proposition. A Latin form like remittebantur “they were being sent back” may serve as an illustration of this type of distribution of elements. The prefixed element re- “back” merely qualifies to a certain extent the inherent significance of the radical element mitt- “send,” while the suffixes -eba-, -nt-, and -ur convey the less concrete, more strictly formal, notions of time, person, plurality, and passivity.
On the other hand, there are languages, like the Bantu group of Africa or the Athabaskan languages of North America, in which the grammatically significant elements precede, those that follow the radical element forming a relatively dispensable class. The Hupa word te-s-e-ya-te “I will go,” for example, consists of a radical element -ya- “to go,” three essential prefixes and a formally subsidiary suffix. The element te- indicates that the act takes place here and there in space or continuously over space; practically, it has no clear-cut significance apart from such verb stems as it is customary to connect it with. The second prefixed element, -s-, is even less easy to define. All we can say is that it is used in verb forms of “definite” time and that it marks action as in progress rather than as beginning or coming to an end. The third prefix, -e-, is a pronominal element, “I,” which can be used only in “definite” tenses. It is highly important to understand that the use of -e- is conditional on that of -s- or of certain alternative prefixes and that te- also is in practice linked with -s-. The group te-s-e-ya is a firmly knit grammatical unit. The suffix -te, which indicates the future, is no more necessary to its formal balance than is the prefixed re- of the Latin word; it is not an element that is capable of standing alone but its function is materially delimiting rather than strictly formal.
It is not always, however, that we can clearly set off the suffixes of a language as a group against its prefixes. In probably the majority of languages that use both types of affixes each group has both delimiting and formal or relational functions. The most that we can say is that a language tends to express similar functions in either the one or the other manner. If a certain verb expresses a certain tense by suffixing, the probability is strong that it expresses its other tenses in an analogous fashion and that, indeed, all verbs have suffixed tense elements. Similarly, we normally expect to find the pronominal elements, so far as they are included in the verb at all, either consistently prefixed or suffixed. But these rules are far from absolute. We have already seen that Hebrew prefixes its pronominal elements in certain cases, suffixes them in others. In Chimariko, an Indian language of California, the position of the pronominal affixes depends on the verb; they are prefixed for certain verbs, suffixed for others.
It will not be necessary to give many further examples of prefixing and suffixing. One of each category will suffice to illustrate their formative possibilities. The idea expressed in English by the sentence I came to give it to her is rendered in Chinook by i-n-i-a-l-u-d-am. This word—and it is a thoroughly unified word with a clear-cut accent on the first a—consists of a radical element, -d- “to give,” six functionally distinct, if phonetically frail, prefixed elements, and a suffix. Of the prefixes, i- indicates recently past time; n-, the pronominal subject “I”; -i-, the pronominal object “it”; -a-, the second pronominal object “her”; -l-, a prepositional element indicating that the preceding pronominal prefix is to be understood as an indirect object (-her-to-, i.e., “to her”); and -u-, an element that it is not easy to define satisfactorily but which, on the whole, indicates movement away from the speaker. The suffixed -am modifies the verbal content in a local sense; it adds to the notion conveyed by the radical element that of “arriving” or “going (or coming) for that particular purpose.” It is obvious that in Chinook, as in Hupa, the greater part of the grammatical machinery resides in the prefixes rather than in the suffixes.
A reverse case, one in which the grammatically significant elements cluster, as in Latin, at the end of the word is yielded by Fox, one of the better known Algonkin languages of the Mississippi Valley. We may take the form eh-kiwi-n-a-m-oht-ati-wa-ch(i) “then they together kept (him) in flight from them.” The radical element here is kiwi-, a verb stem indicating the general notion of “indefinite movement round about, here and there.” The prefixed element eh- is hardly more than an adverbial particle indicating temporal subordination; it may be conveniently rendered as “then.” Of the seven suffixes included in this highly-wrought word, -n- seems to be merely a phonetic element serving to connect the verb stem with the following -a-; -a- is a “secondary stem” denoting the idea of “flight, to flee”; -m- denotes causality with reference to an animate object; -o(ht)- indicates activity done for the subject (the so-called “middle” or “medio-passive” voice of Greek); –(a)ti- is a reciprocal element, “one another”; -wa-ch(i) is the third person animate plural (-wa-, plural; -chi, more properly personal) of so-called “conjunctive” forms. The word may be translated more literally (and yet only approximately as to grammatical feeling) as “then they (animate) caused some animate being to wander about in flight from one another of themselves.” Eskimo, Nootka, Yana, and other languages have similarly complex arrays of suffixed elements, though the functions performed by them and their principles of combination differ widely.
We have reserved the very curious type of affixation known as “infixing” for separate illustration. It is utterly unknown in English, unless we consider the -n- of stand (contrast stood) as an infixed element. The earlier Indo-European languages, such as Latin, Greek and Sanskrit, made a fairly considerable use of infixed nasals to differentiate the present tense of a certain class of verbs from other forms (contrast Latin vinc-o “I conquer” with vic-i “I conquered”; Greek lamb-an-o “I take” with e-lab-on “I took”). There are, however, more striking examples of the process, examples in which it has assumed a more clearly defined function than in these Latin and Greek cases. It is particularly prevalent in many languages of southeastern Asia and of the Malay archipelago. Good examples from Khmer (Cambodgian) are tmeu “one who walks” and daneu “walking” (verbal noun), both derived from deu “to walk.” Further examples may be quoted from Bontoc Igorot, a Filipino language. Thus, an infixed -in- conveys the idea of the product of an accomplished action, e.g., kayu “wood,” kinayu “gathered wood.” Infixes are also freely used in the Bontoc Igorot verb. Thus, an infixed -um- is characteristic of many intransitive verbs with personal pronominal suffixes, e.g., sad- “to wait,” sumid-ak “I wait”; kineg “silent,” kuminek-ak “I am silent.” In other verbs it indicates futurity, e.g., tengao- “to celebrate a holiday,” tumengao-ak “I shall have a holiday.” The past tense is frequently indicated by an infixed -in-; if there is already an infixed -um-, the two elements combine to -in-m-, e.g., kinminek-ak “I am silent.” Obviously the infixing process has in this (and related) languages the same vitality that is possessed by the commoner prefixes and suffixes of other languages. The process is also found in a number of aboriginal American languages. The Yana plural is sometimes formed by an infixed element, e.g., k’uruwi “medicine-men,” k’uwi “medicine-man”; in Chinook an infixed -l- is used in certain verbs to indicate repeated activity, e.g., ksik’ludelk “she keeps looking at him,” iksik’lutk “she looked at him” (radical element -tk). A peculiarly interesting type of infixation is found in the Siouan languages, in which certain verbs insert the pronominal elements into the very body of the radical element, e.g., Sioux cheti “to build a fire,” chewati “I build a fire”; shuta “to miss,” shuunta-pi “we miss.”
A subsidiary but by no means unimportant grammatical process is that of internal vocalic or consonantal change. In some languages, as in English (sing, sang, sung, song; goose, geese), the former of these has become one of the major methods of indicating fundamental changes of grammatical function. At any rate, the process is alive enough to lead our children into untrodden ways. We all know of the growing youngster who speaks of having brung something, on the analogy of such forms as sung and flung. In Hebrew, as we have seen, vocalic change is of even greater significance than in English. What is true of Hebrew is of course true of all other Semitic languages. A few examples of so-called “broken” plurals from Arabic will supplement the Hebrew verb forms that I have given in another connection. The noun balad “place” has the plural form bilad; gild “hide” forms the plural gulud; ragil “man,” the plural rigal; shibbak “window,” the plural shababik. Very similar phenomena are illustrated by the Hamitic languages of Northern Africa, e.g., Shilh izbil “hair,” plural izbel; a-slem “fish,” plural i-slim-en; sn “to know,” sen “to be knowing”; rmi “to become tired,” rumni “to be tired”; ttss “to fall asleep,” ttoss “to sleep.” Strikingly similar to English and Greek alternations of the type sing—sang and leip-o “I leave,” leloip-a “I have left,” are such Somali cases as al “I am,” il “I was”; i-dah-a “I say,” i-di “I said,” deh “say!”
Vocalic change is of great significance also in a number of American Indian languages. In the Athabaskan group many verbs change the quality or quantity of the vowel of the radical element as it changes its tense or mode. The Navaho verb for “I put (grain) into a receptacle” is bi-hi-sh-ja, in which -ja is the radical element; the past tense, bi-hi-ja’, has a long a-vowels, followed by the “glottal stop”; the future is bi-h-de-sh-ji with complete change of vowel. In other types of Navaho verbs the vocalic changes follow different lines, e.g., yah-a-ni-ye “you carry (a pack) into (a stable)”; past, yah-i-ni-yin (with long i in -yin; -n is here used to indicate nasalization); future, yah-a-di-yehl (with long e). In another Indian language, Yokuts, vocalic modifications affect both noun and verb forms. Thus, buchong “son” forms the plural bochang-i (contrast the objective buchong-a); enash “grandfather,” the plural inash-a; the verb engtyim “to sleep” forms the continuative ingetym-ad “to be sleeping” and the past ingetymash.
Consonantal change as a functional process is probably far less common than vocalic modifications, but it is not exactly rare. There is an interesting group of cases in English, certain nouns and corresponding verbs differing solely in that the final consonant is voiceless or voiced. Examples are wreath (with th as in think), but to wreathe (with th as in then); house, but to house (with s pronounced like z). That we have a distinct feeling for the interchange as a means of distinguishing the noun from the verb is indicated by the extension of the principle by many Americans to such a noun as rise (e.g., the rise of democracy)—pronounced like rice—in contrast to the verb to rise (s like z).
In the Celtic languages the initial consonants undergo several types of change according to the grammatical relation that subsists between the word itself and the preceding word. Thus, in modern Irish, a word like bo “ox” may under the appropriate circumstances, take the forms bho (pronounce wo) or mo (e.g., am bo “the ox,” as a subject, but tir na mo “land of the oxen,” as a possessive plural). In the verb the principle has as one of its most striking consequences the “aspiration” of initial consonants in the past tense. If a verb begins with t, say, it changes the t to th (now pronounced h) in forms of the past; if it begins with g, the consonant changes, in analogous forms, to gh (pronounced like a voiced spirant g or like y, according to the nature of the following vowel). In modern Irish the principle of consonantal change, which began in the oldest period of the language as a secondary consequence of certain phonetic conditions, has become one of the primary grammatical processes of the language.
Perhaps as remarkable as these Irish phenomena are the consonantal interchanges of Ful, an African language of the Soudan. Here we find that all nouns belonging to the personal class form the plural by changing their initial g, j, d, b, k, ch, and p to y (or w), y, r, w, h, s and f respectively; e.g., jim-o “companion,” yim-’be “companions”; pio-o “beater,” fio-’be “beaters.” Curiously enough, nouns that belong to the class of things form their singular and plural in exactly reverse fashion, e.g., yola-re “grass-grown place,” jola-je “grass-grown places”; fitan-du “soul,” pital-i “souls.” In Nootka, to refer to but one other language in which the process is found, the t or tl of many verbal suffixes becomes hl in forms denoting repetition, e.g., hita-’ato “to fall out,” hita-’ahl “to keep falling out”; mat-achisht-utl “to fly on to the water,” mat-achisht-ohl “to keep flying on to the water.” Further, the hl of certain elements changes to a peculiar h-sound in plural forms, e.g., yak-ohl “sore-faced,” yak-oh “sore-faced (people).”
Nothing is more natural than the prevalence of reduplication, in other words, the repetition of all or part of the radical element. The process is generally employed, with self-evident symbolism, to indicate such concepts as distribution, plurality, repetition, customary activity, increase of size, added intensity, continuance. Even in English it is not unknown, though it is not generally accounted one of the typical formative devices of our language. Such words as goody-goody and to pooh-pooh have become accepted as part of our normal vocabulary, but the method of duplication may on occasion be used more freely than is indicated by such stereotyped examples. Such locutions as a big big man or Let it cool till it’s thick thick are far more common, especially in the speech of women and children, than our linguistic textbooks would lead one to suppose. In a class by themselves are the really enormous number of words, many of them sound-imitative or contemptuous in psychological tone, that consist of duplications with either change of the vowel or change of the initial consonant—words of the type sing-song, riff-raff, wishy-washy, harum-skarum, roly-poly. Words of this type are all but universal. Such examples as the Russian Chudo-Yudo (a dragon), the Chinese ping-pang “rattling of rain on the roof,” the Tibetan kyang-kyong “lazy,” and the Manchu porpon parpan “blear-eyed” are curiously reminiscent, both in form and in psychology, of words nearer home. But it can hardly be said that the duplicative process is of a distinctively grammatical significance in English. We must turn to other languages for illustration. Such cases as Hottentot go-go “to look at carefully” (from go “to see”), Somali fen-fen “to gnaw at on all sides” (from fen “to gnaw at”), Chinook iwi-iwi “to look about carefully, to examine” (from iwi “to appear”), or Tsimshian am’ am “several (are) good” (from am “good”) do not depart from the natural and fundamental range of significance of the process. A more abstract function is illustrated in Ewe, in which both infinitives and verbal adjectives are formed from verbs by duplication; e.g., yi “to go” yiyi “to go, act of going”; wo “to do,” wowo “done”; mawomawo “not to do” (with both duplicated verb stem and duplicated negative particle). Causative duplications are characteristic of Hottentot, e.g., gam-gam “to cause to tell” (from gam “to tell”). Or the process may be used to derive verbs from nouns, as in Hottentot khoe-khoe “to talk Hottentot” (from khoe-b “man, Hottentot”), or as in Kwakiutl metmat “to eat clams” (radical element met- “clam”).
The most characteristic examples of reduplication are such as repeat only part of the radical element. It would be possible to demonstrate the existence of a vast number of formal types of such partial duplication, according to whether the process makes use of one or more of the radical consonants, preserves or weakens or alters the radical vowel, or affects the beginning, the middle, or the end of the radical element. The functions are even more exuberantly developed than with simple duplication, though the basic notion, at least in origin, is nearly always one of repetition or continuance. Examples illustrating this fundamental function can be quoted from all parts of the globe. Initially reduplicating are, for instance, Shilh ggen “to be sleeping” (from gen “to sleep”); Ful pepeu-’do “liar” (i.e., “one who always lies”), plural fefeu-’be (from fewa “to lie”); Bontoc Igorot anak “child,” ananak “children”; kamu-ek “I hasten,” kakamu-ek “I hasten more”; Tsimshian gyad “person,” gyigyad “people”; Nass gyibayuk “to fly,” gyigyibayuk “one who is flying.” Psychologically comparable, but with the reduplication at the end, are Somali ur “body,” plural urar; Hausa suna “name,” plural sunana-ki; Washo gusu “buffalo,” gususu “buffaloes”; Takelma himi-d- “to talk to,” himim-d- “to be accustomed to talk to.” Even more commonly than simple duplication, this partial duplication of the radical element has taken on in many languages functions that seem in no way related to the idea of increase. The best known examples are probably the initial reduplication of our older Indo-European languages, which helps to form the perfect tense of many verbs (e.g., Sanskrit dadarsha “I have seen,” Greek leloipa “I have left,” Latin tetigi “I have touched,” Gothic letot “I have let”). In Nootka reduplication of the radical element is often employed in association with certain suffixes; e.g., hluch- “woman” forms hluhluch-’ituhl “to dream of a woman,” hluhluch-k’ok “resembling a woman.” Psychologically similar to the Greek and Latin examples are many Takelma cases of verbs that exhibit two forms of the stem, one employed in the present or past, the other in the future and in certain modes and verbal derivatives. The former has final reduplication, which is absent in the latter; e.g., al-yebeb-i’n “I show (or showed) to him,” al-yeb-in “I shall show him.”
We come now to the subtlest of all grammatical processes, variations in accent, whether of stress or pitch. The chief difficulty in isolating accent as a functional process is that it is so often combined with alternations in vocalic quantity or quality or complicated by the presence of affixed elements that its grammatical value appears as a secondary rather than as a primary feature. In Greek, for instance, it is characteristic of true verbal forms that they throw the accent back as far as the general accentual rules will permit, while nouns may be more freely accented. There is thus a striking accentual difference between a verbal form like eluthemen “we were released,” accented on the second syllable of the word, and its participial derivative lutheis “released,” accented on the last. The presence of the characteristic verbal elements e– and -men in the first case and of the nominal -s in the second tends to obscure the inherent value of the accentual alternation. This value comes out very neatly in such English doublets as to refund and a refund, to extract and an extract, to come down and a come down, to lack luster and lack-luster eyes, in which the difference between the verb and the noun is entirely a matter of changing stress. In the Athabaskan languages there are not infrequently significant alternations of accent, as in Navaho ta-di-gis “you wash yourself” (accented on the second syllable), ta-di-gis “he washes himself” (accented on the first)
Pitch accent may be as functional as stress and is perhaps more often so. The mere fact, however, that pitch variations are phonetically essential to the language, as in Chinese (e.g., feng “wind” with a level tone, feng “to serve” with a falling tone) or as in classical Greek (e.g., lab-on “having taken” with a simple or high tone on the suffixed participial -on, gunaik-on “of women” with a compound or falling tone on the case suffix -on) does not necessarily constitute a functional, or perhaps we had better say grammatical, use of pitch. In such cases the pitch is merely inherent in the radical element or affix, as any vowel or consonant might be. It is different with such Chinese alternations as chung (level) “middle” and chung (falling) “to hit the middle”; mai (rising) “to buy” and mai (falling) “to sell”; pei (falling) “back” and pei (level) “to carry on the back.” Examples of this type are not exactly common in Chinese and the language cannot be said to possess at present a definite feeling for tonal differences as symbolic of the distinction between noun and verb.
There are languages, however, in which such differences are of the most fundamental grammatical importance. They are particularly common in the Soudan. In Ewe, for instance, there are formed from subo “to serve” two reduplicated forms, an infinitive subosubo “to serve,” with a low tone on the first two syllables and a high one on the last two, and an abjectival subo-subo “serving,” in which all the syllables have a high tone. Even more striking are cases furnished by Shilluk, one of the languages of the headwaters of the Nile. The plural of the noun often differs in tone from the singular, e.g., yit (high) “ear” but yit (low) “ears.” In the pronoun three forms may be distinguished by tone alone; e “he” has a high tone and is subjective, -e “him” (e.g., a chwol-e “he called him”) has a low tone and is objective, -e “his” (e.g., wod-e “his house”) has a middle tone and is possessive. From the verbal element gwed- “to write” are formed gwed-o “(he) writes” with a low tone, the passive gwet “(it was) written” with a falling tone, the imperative gwet “write!” with a rising tone, and the verbal noun gwet “writing” with a middle tone. In aboriginal America also pitch accent is known to occur as a grammatical process. A good example of such a pitch language is Tlingit, spoken by the Indians of the southern coast of Alaska. In this language many verbs vary the tone of the radical element according to tense; hun “to sell,” sin “to hide,” tin “to see,” and numerous other radical elements, if low-toned, refer to past time, if high-toned, to the future. Another type of function is illustrated by the Takelma forms hel “song,” with falling pitch, but hel “sing!” with a rising inflection; parallel to these forms are sel (falling) “black paint,” sel (rising) “paint it!” All in all it is clear that pitch accent, like stress and vocalic or consonantal modifications, is far less infrequently employed as a grammatical process than our own habits of speech would prepare us to believe probable.