245 102 7MB
German Pages 142 [160] Year 1953
Paul
Forchheimer
The category of person in language
THE CATEGORY OF PERSON IN LANGUAGE bY
PAUL FORCHHEIMER
W a l t e r de G r u y t e r & Co. vormals G. J . Goedien'edie Verla gehand lung - J. Guttentag, Verlagsbuchhandlung - Georg Reimer - Karl J . Trübner - Veit & Comp.
B e r l i n 1953
Archivier. 47 54 53 Printed in Germany Alle Rechte d e · Nachdrucke, der photomechaniichen Wiedergabe, der Überfettung, der Herstellung von Pbotokopien und Mikrofilmen, auch auasugsweiee, vorbehalten. Satz und Druck: Buchkunet, Berlin W 35
IN MEMORIAM PATRIS PRAECEPTORISQVE PHILOSOPHIAE DOCTORIS IACOBI FORCHHEIMER NATI IN MONTE NOBILITATIS FRANCONIAE INFERIORIS DIE SEXTO AB CALENDIS IVNII MDCCCLXXVI ET MORTYI IN MONTE REGIO CANADENSE CALENDIS IANVARII MCMXLIV PRIMVM SVORVM LIBRORVM DEDICAVIT GRATVS FILIVS Άνδρό; φιλοϋν-ros σοφ(αν eümpcrfvrrai πατήρ αύτοϋ ΡΓΟΥ.
XXIX, 3
ACKNOWLEDGEMENT It is my pleasant duty, to acknowledge at this place the constant encouragement and help I have received from Professor Guiliano Bonfante, Princeton University, and Professor Karl H. Menges, Columbia University, from the very first conception of this work until its completion. Both have been friends as much as teachers. — Needless to say that all of my teachers have left some traces on my work, though I may not be able to identify consciously the contributions of each and to thank them all by name. Of the almost innumerable authors whose works I used to good advantage I just want to mention Father Wilhelm Schmidt, who probably has influenced my work more than any other individual author. In a work of this nature mistakes are bound to occur. The responsibility for these is my own. I hope that, nevertheless, the audacity of attempting such a far-reaching undertaking will have resulted in some worthwhile findings which justify the risks that are bound to go with this type of work. New York, July 1951 Paul Forchheimer
CONTENTS ρ·β·
Α. General Discussion I. Introduction II. Statement of Method III. Definition of Person IV. Person and Deictic Expression V. Person and Number VI. Distribution of Three-Person System VII. A Fourth Person? VIII. Some Remarks on Personal Pronouns IX. Non-Pronominal Expression of Person X. Objective — Subjective Persons XI. Gender, Class, and Location Distinctions XII. Substitution of One Person for Another
1 2 4 7 11 20 22 23 25 29 33 37
B. Types of Person Patterns I. General Remarks II. Languages with Morphological Plural of Pronouns
39 40
A. Languages without Formal Plural of Nouns or with Optional Plural Only 41 a. Chinese (Pekinese) 41 b. Burmese 42 c. Japanese 43 d. Maidu 44 e. Yuma 45 B. Languages with Formal Plural of Nouns a. b. c. d. e.
Kottish Hurrian Sumerian Shilh West Greenland Eskimo
48 48 49 49 50 50
C. Languages with Different Pluralization for Nouns and Pronouns 51 a. Ostyak 52 b. Mordwinish 53 c. Lapponian 53 d. Primitive Finno-Ugric 53 e. Suomi Finnish 53 f. Osmanli Turkish 54 g. Eastern Suketi 55 h. Chukchee 56 i. Tonkawa 58 k. Tunica 59 1. Khasi 61
III. Languages without Morphological but with Lexical Plural, at least in the First Person 63 1. General 63 2. A. Languages without Formal Plural, Having a Lexical Plural in the First Person Only 65 a. Korean 65 b. Kamanugu 66 c. Massai 67 B. Languages with Lexical Plural for the First Person and Morphological Plural for the Other Persons a. Akkadian b. Arabic c. Hausa d. Hopi e. Aztec (Milpa Alta Dialect) f. Tshimshian
68 68 69 70 71 74 75
C. Languages with Lexical Plurals in the First and Second Persons and Lexical or Morphological Plural in the Third Person a. Latin b. Gothic c. Sanskrit d. Tlingit e. Carrier
75 76 77 77 78 78
D. Languages with Two Lexical Plurals of the First Person 79 a. Malay 79 b. Tagalog 80 c. Nogogu 81 d. Hawaiian 81 IV. Languages with Lexical and Morphological Plural in the First Person 83 1. General 83 2. A. Languages with Lexical Plural Restricted to Inclusive Dual or Limited Inclusive Plural 85 a. Lakota and Related Siouan Dialects 85 b. Southern Paiute 88 c. Shoshone 89 d. Chinook 90 e. Sierra Popoluca 92 B. Languages with Lexical Inclusive Plural and Morphological Exclusive Plural 94 a. Old Nubian 94 b. Ful 95 c. Mikir 95 d. Rotuman 95
V. Languages with Morphological or Lexical and Composite Plurals of the First Person 98 a. b. c. d. e. f. g.
Melanesian Pidgin English Telugu Tamil Tungus Ordos Mongol Algonquian Somali
A. Languages with Morphological Plural for Inclusive and Composite Plural for Exclusive First Person Plural a. Maya b. Coos c. Siuslawan
99 100 101 102 103 103 104 106 106 107 109
B. Languages with Lexical Plural and Composite Forms 111 a. Iroquoian Ill b. Gäro 112 c. Purik 113 d. Balti 114 e. Chitkuli 114 f. Lower Kanauri 115 VI. Languages with Two Composite Forms for Inclusive and Exclusive Plural of First Person 118 a. Kwakiutl 119 b. Mundari 120 VII. Languages with Variants of One Form Used to Express Inclusive and Exclusive Plural 122 a. Otomi 122 b. Dyirringan 124 c. Kamilaroi 125 d. Worora 126 e. Nyul-Nyul 126 f. Saibalgal 127 g. Kate 128 h. Bongu 128 VIII. Languages with Variants of Second Person Plural for Inclusive First Person Plural 129 a. Yokuts 130 b. Kiowa 131 IX. Languages with Complete Set of Pronouns Plus Composite Forms 132 a. Berber Dialects 132 b. Ewe 132 c. Kele 133 d. Nkosi 134
X. Languages with Less than Three Distinct Plural Forms
136
a. Na-Dene b. Indo-European C. A Note on the Distribution of Person Patterns
136 138 140
D. Speculative Remarks on the Origin of Some First Personell Pronouns
141
PREFACE
Nothing is more comforting than to read works of the type presented here by Dr. Forchheimer. While I could certainly not vouch for the exactitude of all his data or the conclusiveness of all of his results — this could only be done by doing over once more all the enormous work he has done — it is certain that this is the kind of research that should be encouraged the most. While descriptive grammars of every kind — good or bad, useful and useless — are sweeping the learned world, investigations that attempt to go to the root of the spiritual problems of language, as in the creative time of the brothers Schlegel, of Bopp, and of Humboldt, are increasingly rare. It seems almost ironical that, precisely when our linguistic methods have been perfected by G. I. Ascoli, K. J. GilMron, M. Bärtoli, and so many other linguists, when Croce's philosophy has vivified theoretical speculation and saved our science from the blind alley of the neogrammarians, and when the accumulation of material grows every year, the study of general linguistics, so brilliantly initiated in our century by A. Meillet and J. Vendryes, seems to prosper the least. One wonders what the purpose of so many descriptive (not practical or normative) grammars can be, if it is not to furnish material for a deeper penetration into the linguistic phenomenon, into the spiritual activity of man as a speaker, into the origin of language and of man himself. With all their drawbacks or mistakes or illusions, the works of A. Cuny, W. Schmidt, A. Trombetti, A. Sommerfelt, and M. Bärtoli, remain in our century as shining examples of daring thought and loftiness of purpose. We know now — through the efforts mainly of the Italian neolinguists and of the Prague school — that linguistic features of any kind, just as any other form of human activity, can extend from one language or a group of languages to another without any limitation, just as the invention of the fire or the wheel once did, and perhaps even more. The morphological affinities between two contiguous linguistic groups, such as Indo-European and Semitic, can now be observed in a new light that does not need the old genetic applications of Schleicher's stem-theory, as repeatedly (and vainly) presented by Möller, Pedersen, and others. The application of areal linguistics to vast areas of the world, or even to all the languages of the world, as attempted by M. Bärtoli in works that did not by far receive the attention they deserved, is now entirely justified and opens new wonderful vistas upon the greatest problems of languistics. I tried recently, following the lead of an article of Nyklovits', the application of areal linguistics
to the problem of the vigesimal system which appears in the two extreme areas of Eurasia (and parts of America), but not in the central zone, and reached the conclusion that the vigesimal system is much older than the decimal and duodecimal system. The barriers between languages are now falling, although in a different way than older generations of linguists thought, and the road is open to new, daring investigations like the one I am honored to present here. It is with works like this one that linguistics can now fulfil its main task, that is to contribute fully to the understanding of man's history and mind. I am firmly convinced that new, wonderful discoveries can thus be achieved. Prinoeton University, August 1951 G. Bonfante
THE CATEGORY OF PERSON IN LANGUAGE A. GENERAL DISCUSSION I. Introduction If general linguistics is to be developed as an independent science, it will be necessary to study especially such features as are common to all languages, or at least to a great many. While the time may still be premature for a perfect, comprehensive "General Grammar", i. e. a framework into which all specific grammars could be fitted, it is nevertheless possible now to work on some special topics. Research on linguistic developments that are common to several languages was strongly demanded in a meeting of the Paris Linguistic Society on the 26th May, 1945, by Joseph Vendryes. He stated that without such research there could not be any general linguistics. This idea he repeated in several articles. I quote1): "Independamment de toute parente, on observe, entre les langues tres diverses de temps et de lieux, certains traits communs... En reunissant le plus grand liombre possible de ces traits communs, on devrait arriver ä etablir une sorte de grammaire generale,... L'etude des langues revele des besoins communs ä l'ensemble des hommes;... En d'autres termes, il s'agit de rechercher les tendances naturelles de l'esprit humain, . . . " One such general feature is the category of person. I have not been able to find even one language or dialect that did not have it. Next to person we find number, but there are languages that have not developed number into a category. Together, person and number are among the most important and general categories of language and can easily be studied with the historical and comparative material now available. From any fairly reliable grammar, be it conservative or modern, prescriptive or descriptive, scholarly or written by an interested traveller, it is possible to work out the system of person peculiar to a language. It is, therefore, not without interest that Father Schmidt devoted the first two general chapters in his "Sprachenkreise"2* to general problems of person and number, and thatTrombetti wrote the two parts of his "Saggi de Glottologia"3) on personal pronouns and numerals. Jespersen4) also devoted three chapters of his Philosophy of Grammar to number and person. It is all the more strange that Vendryes himself omitted person altogether in his discussion of grammatical categories in "Le Forchheimer,
Language
Language"5). Most texts on general linguistics or linguistic psychology which I have had a chance to consult have little information on person, and I have not seen any thorough treatment of this category. Thus, there is definite need for this research. ') Bulletin de la Societe de Linguistique de Paris, XLIV, 1948, pp. 1—2 ) Schmidt, Pater Wilhelm, Die Sprachfamilien und Sprachenkreise der Erde, Heidelberg 1926, pp. 316—33 3 ) Trombetti, Alfredo, Saggi di Glottologia, 2 vols., Bologna 1922/23 4 ) Jespersen, Otto, The Philosophy οί Grammar, London 1924, pp. 188—225 6 ) Vendryes, Joseph, he Langage, Paris 1921 2
II. Statement of Method In order to evaluate the expression of person as a category in language, it is necessary to work out, for each dialect, the existing systems. For these structural oppositions I shall also use the terms "person pattern" and "person system". By comparing these individual patterns it is possible to establish some basic types and to attempt a classification. While individual patterns must be worked out strictly synchronically, the historical sequence of such patterns in different stages of the "same" language is of great significance for linguistic development and serves to link some of the established basic types. Such connexions have influenced my classification and, to some extent, have guided my selection of criteria for the classification. This broad statement makes it necessary to answer in more detail the following questions: a. What are the sources of the person patterns? How have they been selected? Why is the number of systems examined considered significant? b. Which criteria have been considered most important for the classification, and why? c. How has the material been analysed? The answers are: a. All in all, I have examined about a good five hundred grammars and word lists. Partly, the selection has been arbitrarily guided by previous study and by accessibility. On the systematic side, I have consulted general works and collections on almost all areas. Wherever I have found or suspected unusual or new material I have changed from the extensive to an intensive study of that linguistic stock or area. Thus I have concentrated on Siouan, Uto-Aztec, and Himalayan Tibeto-Burman,while I have treated most Melanesien or Bantu-dialects rather summarily. In this way I hope to have found the main types of person systems. The great number of historically, geographically, and genetically widely scattered 2
individual languages examined also guarantees some significance. On the other hand, as this work had necessarily to be limited in scope, it certainly could be extended and also improved intensively. Yet, after a certain stage had been reached, examination of more languages did no longer show new types. While this, by no means, excludes their existence, it nevertheless increases the significance of the material at hand. If the typology is not final, it provides at least a skeleton of sufficient significance to be used in the preparation of a better classification. b. While the distinction of three persons is rather general, differences appear where person is integrated with other categories, such as number, gender or case-inflexion. Number has been given preference as criterium for classification for the following reasons: 1. It is more universal. 2. It affects person more intimately than gender or case. The opposition of masculine and feminine or of agent and object case is purely within the meaning of a definite person and does not modify it. He, she, it, him, her, are all clearly third person, I and me and even mine express the first person. But we includes persons other than the speaker. Number thus can affect person and change the definitions. It is much more essential a modification than gender and case which may be considered rather accidental. 3. From the historical point-of-view number is also closely linked to person. Number offers at least a partial clue for the development of person systems, while cases are less important and gender still less. Thus it may even be argued that number as a criterium leads to a more "natural" system of classification. c. In the analysis of the individual grammars, first of all the given forms have been arranged in uniform tables. Discussion of forms and of syntax have been checked for further information which might affect the interpretation. Singular and plurality forms have been compared. Etymological connexion has only been considered where it was obvious and likely to be felt by the speaker. Scientific etymologies, of course, had to be disregarded for a synchronic scheme, but have been reserved for historical discussions. Pluralizers have been compared with those of nouns and verbs. Often long lists of suffixes, prefixes, and sometimes even nouns and verbs have been examined for possible clues to analysis. Wherever available, works and statements by different scholars have been critically compared. For historical discussions, use has also been made of "internal reconstruction" i. e. diachronic reconstructions based on synchronic patterns, as suggested by Bonfante1'. r
3
The purpose of the classification of person-opposition types has not only been undertaken to satisfy a desire for an orderly arrangement of extant information, but it was also taken as a basis for analysis and understanding to give a background for general linguistic theory and for psychological and anthropological analysis. General conclusions reached, in turn, have been important as background for individual analyses. Thus the present theory linking the Akkadian pronominal plural with the nominal dual has been inspired by conclusions derived from the study of other languages first. For such reasons, the general conclusions derived from the present study have been summarized and presented in the introductory sections as a background for the following classification which then has to support it in details. ) Bonfante, Giuliano, "On Reconstruction and Linguistic Method", Word, vol. 1, nos. 1 & 2, no. 1, par. 8, p. 132, 1945 l
III. Definition of Person The "New English Dictionary" defines person as used in grammar: "Each of three classes of personal pronouns, and corresponding distinctions in verbs, denoting or indicating respectively the person speaking (first person), the person spoken to (second person), and the person or thing spoken of (third person)". These are, as a matter of fact, the traditional classical definitions. In his "Philosophy of Grammar"1) Jespersen objects to these definitions. In I am ill and you must go, he states, the "I" and "you" are spoken of. He redefines, therefore, as follows: first person = ! speaker; second person = spoken to; third person = neither speaker nor spoken to. While Marouzeau's Lexique Linguistique2> still lists the old definition, the American College Dictionary3', following Jespersen, defines grammatical person: "a. (in some languages) a category of verb inflexion and of pronoun classification, distinguishing between the speaker (first person), the one addressed (second person), and anyone or anything else (third person), sometimes with further subdivisions of the third, as I and we (first person), you (second person), and he, she, it, they (third person), b. any of these three (or more) divisions." We can accept these definitions, as last stated, at least for the singular. Another part of defining person is to state the relationships that exist between these three categories. De la Grasserie4) states: "Le point de depart est le moi; du moi on passe au non-moi. Mais le fait du discours introduit un troisieme element et divise le non-moi. On ne parle pas sans interlocuteur; cet interlocuteur se detache du 4
groupe du non-moi et prend une place particuliere". Similarly, Boas5' states: "Logically, our three persons of the pronoun' are based on the two concepts of self and not-self, the second of which is subdivided, according to the needs of speech, into the two concepts of person addressed and person spoken of." Van Ginneken joins these two authors with his statement 6 )":... la difference entre les sentiments et les adhesions est beaucoup plus grande que entre les sortes d'adhesion entre elles, . . . les significations de la 2e et la 3 e personne sont beaucoup plus rapprochees que celles de la l r e et de la 2 e personne." While these analyses by de la Grasserie, Boas, and van Ginneken oppose the speaker to the spoken to and spoken of, Wundt7), Schmidt8^, Bloomfield9' and Bühler10) analyse the persons in a way more in line with Jespersen's classification into speaker, spoken to, and anyone or anything else. Wundt7) states that the third person has a "wesentlich abweichende Stellung gegenüber der ersten und zweiten Person. Nur die erste und zweite Person sind Personen". In a similar way, Bloomfield') calls the first and second person "personal," the third person "definite." Schmidt8), says: Die "Personalpronomina sind ja sicherlich die ersten Sprach-, weil die Deutewurzeln, die eben deshalb, wegen dieses ihres Erstgeburtrechtes, auch später in allem eine beherrschende Stellung sich bewahrt haben, die in der organischen Tatsache gelegen ist, daß die ungeheure Weite dessen, was die 'dritte Person' alles umfassen kann, nichts anderes ist, als was die 'erste Person' sich zum Denk- und Willensinhalt macht und einer gegenüberstehenden ersten, der 'zweiten Person,' als solcher mitteilt." Bühler10), likewise, sees a closer connexion between the first and the second person: "Die Wörter ich und du weisen kurz gesagt auf die Rollenträger der Sprechhandlung hin . . . Die Personalia, ζ. Β. ich und du bezeichnen im Hauptberuf und von allem Anfang an nicht den Sender und Empfänger der Sprachbotschaft wie die Namen Bezeichnungen sind, sondern sie weisen nur hin auf diese Rollenträger in dem Sinne, wie es schon bei Appolonius steht." The first and second persons are personal, I should like to add "subjective." The third person, as Bloomfield states, is not personal but definite, or, as Wundt says, is not a person at all. It is impersonal, objective. Whoever does not act a role in the conversation either as speaker or as addressed remains in the great pool of the impersonal, referred to as "third person." Consequently, between the first andsecond person there is a constant shifting, the actors being alternatingly speakers and addressed, while the "teitius conservationis" always 5
remains unchanged. In the same conversation, A refers to himself, A, as "I," and to Β as "you," while Β calls A "you" and himself, B, "I." But C remains C or "he" throughout the conversation, lest he assume one of the two roles of speaker and of addressed. It is perfectly possible to impersonalize speech by treating even the speaker and addressed as third person, as is done frequently in court argot and with babies. The third person, on the other hand, cannot be replaced by a first or second person (unless the speaker "plays its röle" or, in his imagination, addresses it). Thus, in actual speech, not the "self," but the impersonal pool of the "third person" is primary and cannot yet be called a "person" until a first person, a second person, or both, are singled out as persons. By opposition in the system structure, the remainder then, and then only, becomes the third "person." Many languages have no special designation for the verb in the third person, except number, while the first and second person are designated by affix or juxtaposed pronoun. Many languages, likewise, have no pronoun of the third person, and of those that do, many, e. g. Indo-European dialects and Mongolian, lack, or lacked, the nominative. Or, there may be definite pronouns for the first and second persons, but a choice of many demonstratives for the third. It is, therefore, no accident that often closely related languages share the pronouns of the first two persons, but have different pronouns for the third person. Also, the pronouns of the first and second person are often similar in form and flexion, but dissimilar from that of the third person. The third person is much more subject to objective subdivisions such as class, gender and location than are the first two,, but the first person distinguishes number more readily than the second, and the second more readily than the third. Nevertheless, we cannot always oppose the first two persons to the third. To continue! the previously used simile, the first person is first singled out from the pool, and then the second. There must, therefore, be a stage when only the first person is singled out and opposed to all that remains. While such a situation in languages presently known is the exception rather than the rule, it does exist and must be accounted for. The second person is clearly intermediate between the most subjective first and the most objective third person. It is possible to make an objective, impersonal statement about the addressed, using the second person pronoun in the sentence, but talking as if about a third person. This idea, most likely, has caused a peculiar double development of the Tigrigna pronoun of the second person, as studied by Leslau11'. The old Semitic forms have 6
been superseded for the third person, and partly also for the second person, by a compound näfs "person," contracted to ness, with the possessive suffix. The first and second person have retained the old Semitic pronouns. Thus the second person now has a twofold system of expression, depending on how personal the address is meant. This schism even penetrates into the first person plural where the familiar nekhna "we" is occasionally replaced by a less familiar nessatna. But discussion of problems involving plural must be postponed. Similar to Ethiopic Tigrigna, the North American Quileute of the Chimakua group of Mosanic languages has produced a special "vocative" pronoun, i. e. a pronoun of the second person of emphasized subjectivity, used in polite imperative and in general sentences of a subjective nature. Besides that, Quileute, according to Andrade 12 ', has indicative and imperative forms of pronouns. (The vocative pronoun is used in, polite imperatives, more directly addressing than the indicative pronouns,yet less personal than the imperative pronoun.) Some languages also have vocative forms of the first person possessive, e. g. Latin mi Brute (nom. Brutus meus) and Sierra Popoluca13) (Mixe-Lenca) man-cashi "Thou my child" (nom. an-cashi). ') Jespersen, Otto, The Philosophy oi Grammar, London. 1924, pp. 212 f. 2 ) Marouzeau, Jules, Lexique Linguistique, Paris 1943, p. 163 3 ) The American College Dictionary, New York 1947, p. 904 4 ) de la Grasserie, Raoul, De la v0ritable nature du pronom, Etudes de Grammaire Comparee, Louvain 1888, p. 3 5 ) Boas, Franz, Handbook oi American Indian Languages, vol. I., Washington 1911, pp.39—40 e ) van Ginneken, Jacobus, Principes de Linguistique psychologique, Paris 1907, p. 211 7 ) Wundt, Wilhelm, Völkerpsychologie, vol.11, „Die Sprache", Leipzig 1911, pp. 141 ff. 8 ) Schmidt, Pater Wilhelm, „Die Personalpronimina in den australischen Sprachen", Denkschriiten der österreichischen Akademie der Wissenschaften, Philosophisch-historische Klasse, vol. 64, Vienna 1919, p. 203 9 ) Bloomfield, Leonard, Language, New York 1938, pp. 225—26 10 ) Buehler, Karl, Sprachtheorie, Jena 1934, p. 113 ") Leslau, Wolf, Documents Tigrigna, Paris 1941, pp. 45—46 12 ) Andrade, Manuel J., "Quileute", Handbook ot American Indian Languages, vol. III, New York 1933/38, pp. 203—205 ls ) Foster, Mary L., and Foster, George M., Sierra Popoluca Speech, Washington 1948, p. 23
IV. Person and Deictic Expression Connexion between person and deictic function (demonstration) is claimed in different spheres which are best treated separately. These 7
are: A, etymology, Β, 1, supplementation, 2, psychological or logical classification. Etymology again can be subdivided into claims that the personal pronoun is of demonstrative origin and into opposite claims that the demonstratives are derived from personal pronouns. Thus Brockelmann1) states: "Die Pronomina gehören in allen Sprachen einer älteren Entwicklungsschicht an als Nomina und Verba. In ihren Stammformen ist im Semitischen durchweg noch die Herkunft aus Interjektionen, Deutewörtern oder Lautgebärden unverkennbar." Similarly, Barth claims demonstrative origin of Semitic pronouns. On the Indo-European side, Whitney2) states: " . . . there is a little body of so-called pronominal or demonstrative roots which are distinguished from the rest as signifying position or direction with reference to the speaker rather than any more concrete quality. They are few and of the simplest phonetic form: a vowel only or a consonant with following vowel." Later 3 ): "In those demonstrative words, however, which acquired a specific personal character, as denoting the speaker and the spoken to, . . . " Another view is expressed by Capell4': "In attempting to analyse pronouns, it must always be remembered that they form one of the oldest strata in speech. As in the study of Indo-Germanic languages it has been found that no really satisfactory analysis of some pronominal forms can be arrived at, so it will appear in that of Oceanic languages. The forms have existed too long, they have influenced and crossinfluenced each other..." While most older linguists insist on the relative constancy of the forms of pronouns which are often most easily identified in related languages, Hjelmslev5) emphasized rather their being subject to change. We have to bear in mind that the so-called full or free or independent pronouns often consist of a base neutral as to person and of a pronominal affix. This often explains the almost universal phenomenon of a great difference between subject case and oblique case. In Semitic, for instance, we can trace the nominative form to a supposedly demonstrative base plus personal suffix for the first two persons, while the third person is designated by another demonstrative. Oblique cases are indicated by affixes. Affixes, in all languages, are greatly exposed to wear and tear, but have the advantage of being easily identified by their position. Certain differences of position or additional elements distinguish in many languages subject and object forms. Clearly understood in context, these forms can often change so that little, if anything, of the original pronominal element is left, while precision of expression is 8
maintained. Inasmuch as we have often only a vowel, a consonant, or a combination of both left as characteristic of person, comparisons with similar elements in the demonstrative realm are always easily made, but hardly more than speculation, usually prompted by an a-priori hypothesis of language genesis. I could not give this subject a sound scientific treatment within the limited framework of this book and shall, therefore, omit it completely. On the other hand, use of personal elements of whatever origin to differentiate deixis in demonstratives is claimed by van Ginneken®): "L'armenien n'emploie p. e. aucun pronom demonstratif, sans qu'il y fasse sentir le rapport avec une premiere, une deuxieme, ou une troisieme personne. Les trois elements qu'il emploie ä cet effort sont s, d, et η. Joints au nom ou au verbe ils ont la fonction d'un article personnel, comme on dit: ter-s, le monsieur ici; moi, le monsieur ter-d, le monsieur lä; toi, monsieur ter-n, le monsieur lä-bas; lui, le monsieur
xausim-s,
moi ici, je parle
xausis-d,
toi la, tu paries
xausi-n,
lui lä-bas, il parle."
Iroquoian7)has two t prefixes, onemeaning"here,"andoneexpressing inclusion of the addressed, "I and thou." According to deictic theory, there would be a clear connexion. There are, however, other facts known about the t, too. We find t in many American Indian dialects as a pluralizer. In some, under the influence of a later, more general pluralizer (see next chapter), it became restricted to inclusive or dual or inclusive dual. In others it generally expresses the plural of the first person or even is an alternative form for the η or whatever element indicates "I." Besides, t is a frequent formative in TibetoBurman and Munda, and has almost any meaning in the various languages and dialects. Thus, though we do not know the exact historic origin and all the contacts and absorptions in the history of the Iroquoian dialects, we can very well conceive of an explanation which makes the identity of the demonstrative t and that of the inclusive dual (and plural) t a mere coincidence which may not be interpreted psychologically any more than many other homonyms. A more convincing situation, however, is found in Kwakiutl and Bella-Bella, Mosanic Wakash dialects studied by Boas. Number as a clear cut grammatical category does not exist in these languages. There is a pronoun of the first person1 and one of the second, both existing as affixes, and with a base as free pronouns. The pronouns of the third 9
person have a sevenfold classification. They distinguish near first person, near second person, and near third person, and for each one of the three they oppose visible and invisible. Besides, there is a seventh form of the demonstrative, denoting absence. There is no obvious connexion between: the first two persons and these demonstratives. Yet, by adding the suffixes distinguishing visible and invisible, the pronoun of the first person is pluralized. "I plus visible" becomes the inclusive, "I plus invisible" the exclusive plural. Here the second and third persons joined to the first person are clearly indicated by deictic elements. These last mentioned illustrations fall already under "supplementation of personal pronouns by deictic elements." In general, we meet the same difficulties as in etymology, and most analysts find what they set out to find. No doubt, valuable research could be done in this direction, but again, for the purpose of this study the problem is best left open. While we thus leave the etymological connexions between personal and demonstrative pronouns in abeyance, it is certainly worthwhile to discuss the logical and psychological connexions between these elements. For Bühler, who distinguishes pointers ("Zeigwörter") and descriptive names, pronouns are definitely deictic elements, as they are not proper names. He states 8 ': "Wenn ein Sprecher auf den Sender des aktuellen Wortes 'verweisen' will, dann sagt er ich, und wenn er auf den Empfänger verweisen will, dann sagt er du. Auch 'ich' und 'du' sind Zeigewörter und primär nichts anderes." Thus, what w e initially singled out as personal from the impersonal pool of the third person, the words for "I" and "you" which shift from one person to another, depending on whether it is the speaker or the addressed, are analysed as deictic or demonstrative. Büihler distinguishes Zeigwörter or demonstratives which take their meaning from the situation, and Nennwörter or names which have an independent meaning. In this sense, personal and demonstrative elements certainly can be classed together. This is also expressed by Gray9*: "Semantically, pronouns differ from nouns in that they are essentially deictic. They do not designate persons, things, concepts, or qualities in general (as do some nouns man, stone, thought, goodness), but, without limitation to any single category of ideas, they denote a specific individual or specific individuals of any category." The deictic (pointing) nature of the personal pronoun is more pronounced in the singular than in the plural. To quote Bühler 10 ' 10
again: "Wie das 'ich,' so setzt natürlich auch das 'wir' zu seiner Erfüllung eine Zeighilfe voraus; aber es scheint von vornherein schon um einen Schritt weiter als das 'ich' vom Grenzwert des reinen Zeigwortes entfernt zu sein, denn es fordert doch irgendwie zur Bildung einer Klasse von Menschen auf; das inklusive 'wir' ζ. B. verlangt eine andere Gruppenbildung als das exklusive. Und Klassenbildung ist nun einmal das Vorrecht der nennenden Wörter, der sprachlichen Begriffszeichen... Gewiß kann jedes Zeigzeichen eine Nennfunktion übernehmen; denn sonst gäbe es ja keine Pronomina." These statements concerning the deictic nature of the personal pronouns are certainly valid. It was the purpose of our initial discussion to work out their peculiar nature which singles them out from the wide field of demonstratives, but equally, a fortiori, from any nominal category. As far as the following chapters are concerned, this distinction is of little consequence. It is the purpose of this book to describe the person systems as found in language, and, as far as possible, to work out their likely origin and cause, based purely on linguistic considerations. The results, no doubt, need interpretation by psychologists and anthropologists. It cannot be the purpose to force linguistic data into pre-conceived schemes, but rather to try to reveal manifestations of the human mind and connexions on which psychological and anthropological theories can safely be based. ') Brockelmann, Carl, Grundriß der vergleichenden Grammatik der semitischen Sprachen, 2 vols., Berlin 1908, par. 103, p. 296 2 ) Whitney, William Dwight, Life and Growth of Language, N e w York 1897, p. 201 3 ) op. cit., p. 207 4 ) Capell, Α., The Linguistic Position oi SE Papua, Sydney NSW 1943, p. 203 6 ) Hjelmslev, Louis, Principes de grammaire ginirale, Copenhagen 1928, pp. 325—26 β ) van Ginneken, Jacobus, Principes de iinguistique psychologique, Paris 1907, pp. 209—10. (These examples, originally quoted from Meillet, are also discussed by Brugmann and Bühler.) 7 ) Barbeau, C. Marius, Classification of lroquoian Pronominal Prefixes, Ottawa 1915, pp. 17—23 e ) Bühler, op. cit., p. 79 ·) Gray, Louis H., Foundations of Language, N e w York 1939, p. 173 10 ) Bühler, op. cit., p. 143
V. Person and Number Person and number are not only most easily recognised and compared, hence the logical starting point for a general grammar, but 11
these categories seem also intimately connected, if not in all, then at least in a good many languages. On this subject, Father Schmidt1) states: "Die Numerusbildung ist überall zuerst vom Personalpronomen ausgegangen und hat sich dort am reichsten entwickelt. Es gehört deshalb auch zu den Ausnahmen — die semitischen Sprachen stellen eine solche dar —, daß eine reiche Entwicklung des Numerus beim Nomen vorhanden wäre, aber beim Personalpronomen fehlte. Die Regel ist vielmehr, daß die Entwicklung vom Personalpronomen ausgeht und auf das Nomen übergreift... Denn gerade bei der Erfassung der Person, besonders bei der ersten Person, tritt der Begriff der individuellen Einheit am stärksten hervor. Gerade die scharfe Erfassung der Einheit ist aber auch die notwendige psychologische Vorbedingung für die Herausarbeitung der Mehrheitsformen." My research has borne out Schmidt's statement for the plural without exception. As for dual and trial, however, Semitic is not the only exception, and I shall offer a full explanation in my theory on the origin of these numbers after discussing plural. Assuming, as shall be explained later, that plural develops before dual, I discuss it first. Although quite a number of languages lack formal number, while the plural in others has all signs of being a recent development, idioms without an expression for "we" are extremely rare. Though I have found several languages where the word for "I" can also serve to express "we," they all possess, besides that, a word for "we." The only exception is Chinese Pidgin English. Yet the source on which I base my description of that idiom is probably not comprehensive and thorough enough to justify from it alone a statement that langugages without a plural of the first person exist. The same goes for Tasmanian. While Chinese, Chitkuli (vide infra) and other languages suggest such a previous stage, I have not been able to put my finger with certainty on any definite pattern, to say that it exists there. While such a state may have existed, I prefer to leave the question open. On the other hand, it must be said that many languages without a plural of nouns and of pronouns in the third person do distinguish singular and plural in the first person, often also in the second. I found no record of a language distinguishing "thou" and "you," but not "I" and "we," whereis the opposite is frequent. There is no doubt that plural starts from the first person, spreads to the second, and then to the third person and nouns designating person, then animate nouns, and last to the names of objects. If we now remember Bühler's distinctions of pointers and names, we see a perfect parallelism. The first two per12
sons are more deictic (i. e. pointing) than the third. Names of relatives (father, mother, uncle, child), and next classes of persons (man, woman, boy, girl, chief, etc.) are more deictic than class names of animate beings, and last of objects. This analysis of the words to be pluralized from the point of view of deictic significance (i. e. as pointing or defining) is very fruitful. Names exist in many languages in two degrees of deictic determinations: indeterminate and determinate. Whether we have two grammatical forms, as in Aramaic, or absence or presence of the definite article, or two articles as in the Western languages, is irrelevant. The determinate form refers usually to a definite object or being. (I am only concerned with the basic use, not with stylistic possibilities after a long development.) The indefinite form does not so much refer to a specific individual object as to the class. It can be singular as well as plural in many languages, or, better, it is without number. While Western languages cannot express any word without number, they must use either the definite or the indefinite article, the absence of article, esp. in plural, being interpreted as indefinite article. A look at Turkish, however, will bring out the indefinite or indeterminate state better. Turkish really has three states. Case endings or possessives form the determinate state. The numeral bir "one," which has almost become an indefinite article in Southern Turkish, according to Grönbech, designates the individual, but not the determined object. The word without determinant then means just the class. Sommerfelt2) states even that in Old Turkish the noun meant not an object, but a quality, or, better still, a force. This fundamentally important function has been studied in great detail by Grönbech"), and it will help our appreciation of number first to gain some insight into the workings of a language which is half-way between languages without number and languages where number is such a fixed category that nouns cannot exist independently of it. Grönbech writes (par. 195): "Die unflektierte Form eines Nomens, die in allen Fällen, wo nicht von Pronomina die Rede ist, mit dem Wortstamm zusammenfällt, gibt an sich keinerlei bestimmte grammatische Beziehungen an." He then continues saying that the word stem of a noun stands for an indefinite, general form of the ideas expressed by that noun. It expresses only the notion, not any definite individual specimen. A Turkish noun without suffix is not "protected" in the sentence. It can be absorbed by any following noun, losing its individuality completely. (Par. 196) These two words then fuse to yield a new, unified idea. 13
Grönbech goes on (par. 216) explaining the importance of the nominal attribute of a verb fort the designation of the logical object. In such a context, the indefinite case (i. e. the stem without affixes) indicates that the object of the action is not a definite individual, but that the centre of interest is the action as such. There is, therefore, no distinction made between one or several individuals as object. I may illustrate this point of Grönbech's with a few examples, adam means "man," ev means "house." adorn evi "a man's house," i. e. not an animal's, but adamm evi "the house of the man," the house of a definite person. Similarly: ev adami "a man of a house," i. e. a man resigned to home life, but evin adami "the man of the house," i. e. the one who lives in it or who owns it. adam ev seviyor "man house loving," i. e. the (or a) man loves the institution of houses. House-loving is one idea. But: adam evi seviyor "the (or a) man loves the house," i. e. a definite house. In par. 246 Grönbech states that the grammatical distinction between definite and indefinite is based on the same shift in linguistic structure which also accounts for the origin of plural. The development of plural also contributed to the separation of the indefinite case from the general idea. A plural cannot have a general meaning in any Turkish language. Thus, if a plural is used as object in the indefinite case (i. e. without suffix) it must refer to a more or less definite number of individual objects. Summarizing Grönbech's analysis, we may say that the development of a plural depends on the previous or simultaneous crystallization of a singular. Structuralistically speaking, we either do not distinguish number at all, or we have to oppose at least two numbers, singular and plural, to establish such a category. But the crystallization of a singular, which automatically creates a plural, is correlated to the deictic nature of the word. Words that are pointers rather than names, such as the pronouns, call for number much more than names of objects. Terms for relationships or stations in life, in which some deictic qualitiy is inherent, would come next. The clearer the pointer, the less specification is necessary. The speaker of the word is always identified automatically by the first person. Thus a choice of forms for the first person, except for reasons of courtesy, is the exception. The first person is usually neutral as to class or gender; and classification of location, as is often found within the third person, is superfluous. The speaker determines what is here and there, all is relative to him. While such distinctions, especially gender, occur already more frequently 14
in the second person, they are still, all in all, rare. But in the third person, which is neither speaker nor addressed, pointers have to be more specific. Where we have one standard pronoun of the third person, it is a grammatical device, pointing to a noun previously used or implied. For general pointing, we still have a choice of demonstratives. But, the more a pointer is qualified, the closer it comes to a name. "I" means whoever is speaking. "The invisible yonder" comes closer to a class name, and, therefore, does not automatically imply number. Aside from the degree of deictic function, we also have to consider another distinction, closely connected with it, that of subjective and objective relationship. In the first person, the speaker speaks about himself. In the second person, which he addresses, he still is more subjective than when talking about a third person or object, not actively involved in the conversation. A survey of many languages seems to indicate that the distinctions made of subjective persons are different from those made for third persons. Inflexional distinctions such as number and case are most elaborate for the first person. The first person is most interested in a specification of its relations. The third person is more or less "the object of the conversation'," and its exact position as agent or receiver of the action, and especially further modifications as to how the action is received, are of lesser interest. On the other hand, objective qualities, such as number, location, class or gender, are of greater interest and no longer implied in pointing. These consequences of the subjective-objective opposition could have been, so far, just as well explained by reference to pointing and naming only. The two oppositions seem to go parallel, almost inviting the mechanist to scrap the objective-subjective distinction and to replace it by naming-pointing. The real merit of the objective-subjective opposition appears when we find that the object of the first person has moved on the subjective-objective line away from the subjective position slightly toward the objective position. It would be hard to say that "me" is less pointing and more naming than "I." As far as plural is concerned, both analyses coincide. Bühler remarked already4' that "we" is less pointing and more naming than "I." As "we" includes also outsiders, second or third persons, it is also more objective than' "I." The perhaps strange theory that the object form of the first person is less subjective than the agent form is borne out by the observation that the oblique forms show fewer distinctions than 15
the agent forms. Some Athapaskan languages, to quote an extreme example, have separate agent forms for the plurals of the first and second persons, but only a common object form. Although similar, though less extreme data can be collected from unconnected languages, I have yet to find a case where object forms show more distinctions of the subjective type than subject forms. It is, therefore, not quite so surprising if w e find in Semitic (e. g. Akkadian) and in Siouan (e.g. Ponca, Winnebago) an identity of the singular object and plurality (dual in Siouan) subject forms, which seem also to have existed in Proto-Tibetan. Once established in the first person, number distinction is easily carried over to the second person. Plural of the first person is a difficult concept. Wundt claims that "Meist wird die Mehrzahl als ein inhaltlich anderer Begriff aufgefaßt: ich-wir, du-ihr." It is true, many languages have different words for "I" and "we." Yet the old grammarians of Indo-European and Semitic languages, where this state exists, considered these forms singular and plural of the first person, i. e. two numbers of the same name. On the other hand, languages from Eastern Asia across most of Asia to Finland, and on across the North of Scandinavia to Greenland and down through many parts of the Americas really form a plural from "I." We have now to consider how language can express plurality. Of the first two persons there are only simultaneous or real plurals possible, not distributive plurals. For the second person it might have collective meaning, but the distributive forms which occur for objects can hardly be found with persons. Languages without a formal category of number nevertheless can express the idea of more than one. Only, grammatically it is expressed as a quality parallel with colour, shape, size, and other physical qualities. This can be understood if we remember Grönbech's and Sommerfelt's discussions of the Turkish noun. The very fact that Sommerfelt makes his remarks while analysing the Central Australian Aranda (also spelled Aranta) shows that it is not a peculiar feature of Turkic, though perhaps best studied in this family. If, as in Turkish and Chinese, a plural of "I" is formed by adding a suffix, this can spread throughout the pronouns and thence to other words in the sequence of decreasing deictic and increasing descriptive qualities. This seems to happen even now in Chinese. In Turkish the spread was stopped and another suffix was used for nouns and verbs. Many languages have different pluralizers for 16
animate and inanimate, or pluralize only animate nouns. Again, prefixes or suffixes may be restricted to nouns of relationship. As the plural spreads from pronouns to nouns, it may spread semantically only, or carry over its morphemes. On the other hand, distributive and collective forms may develop, by analogy, into real plurals. Different pluralizations also exist in better known languages, although obscured by conventional grammars, e. g. in classical Greek and Arabic, which, in a way, still carry over animate-inanimate distinction. Not only does the IE neuter not distinguish subject and direct object form, its plural takes the verb in the singular, as do also the broken plurals, originally collective forms, in Arabic. It is clear that the different pluralization for the various classes of words appeared at different times. It can happen that later pluralizations, comprising a larger territory by taking in groups of words not hitherto pluralized, instead of being confined to these new plurals, spread out and displace older plural formatives. But it may also happen that the two coexist for a while, and that a word takes both endings or either ending. Then a peculiar semantic development takes place. The later, more general pluralizer indicates a general plural. The older plural, originally more restricted in its distribution, now assumes a more restricted meaning. It can become a "paucal" plural as in Haussa or Hopi, or the most limited pluralitiy, dual, as in Dakota or Tonkawa. Semantically, the development may be compared with the specialization undergone by brothers and brethren once two plurals were available, or German Worte — Wörter. This development of a dual may start from pronouns and spread to nouns, or start from nouns and spread to pronouns. It may also be restricted to the form where it originated. If it starts from the pronoun of the first person, originally the only plural in the language, but not a morphological plural, and therefore easily subject to addition of the new suffix, another specialization may result. The older form may be restricted to an inclusive or exclusive form. On the other hand, if the plural of pronouns is well developed, but essentially different from that of nouns, the new form may never spread to pronouns and dual may exist with nouns only. This is illustrated by Akkadian and American Zuni. In any such case the plural of pronouns should exhibit the older forms, often the dual of nouns. I shall treat this further for Akkadian. Dual, aside from this development, may also occur as a loantranslation (caique semantique) from a substratum. This then FoTChheimer,
Language
explains dual forms composed with the numeral for two. In some Melanesian languages, and in Northern Australia and parts of Papua, where this feature occurs, the analogy with the dual formed with "two" has led to a similar trial, and partly even quatrial. Where these languages have a plural, it is essentially different from the numeral-numbers. Often the singular form (indefinite?) is used again for plural. This explanation of the origin of dual is all the more satisfactory, as dual is no logically or psychologically necessary form. It also occupies no special step on the scale of development. Some languages never had it. Others had it and give it up after a while. Singular and plural express basic concepts, and in the first person practically no language does without this opposition. The long trend is universally toward number. (Even if French pronunciation often does not distinguish singular and plural, the distinction is clear in the article.) But dual is not a logical desideratum, as it can easily be replaced by the use of a numeral. Now we can understand why Schmidt mentions Semitic as an exception. While number, i. e. the distinction of singular and plural, originates from the personal pronoun of the first person, the development of dual may also originate from nouns and need not spread to pronouns, just as dual may also exist with pronouns only. It is interesting now to examine in details the distribution of dual in Semitic. Brockelmann e) states: "Außer dem Plural besaß schon das Ursemitische einen Dual, der aber ursprünglich nur zur Bezeichnung der durch Natur oder Kunst zusammengehörigen Paare diente. Nur im Arabischen wird der Dual auf jede beliebige Zweizahl übertragen." Ungnad 7 ', however, gives the following information for Akkadian: "Von paarweise vorhandenen Dingen wird ein D u a l . . . gebildet... Im ältesten Akkadisch ist der Dual noch weiter verbreitet; vgl. saweran annian 'diese beiden Spangen'." In classical Arabic, where the dual is general and the correct way of indicating "two", it has also penetrated into the pronominal paradigm, with the exception of the first person. Schmidt8' mentions the same for Ethiopic, but I could only verify it for South-Semitic, a group of dialects close to Ethiopic. There, in Soqotri"), the dual has even been extended to the first person. Pronominal duals do not distinguish gender, as do the plurals of the second and third person. It is not surprising to find the greatest spread by analogy in that family. A related dialect, that of Hadramaut, has, according to Brockelmann10), the unique feature of extending gender distinction to the first person singular. — When discussing Akkadian, I shall present my arguments as to why I believe the plural endings of 18
the Semitic pronouns of the second and third person to be identical, historically, with the nominal dual, i. e. an older plural. In the majority of Semitic languages, then, the dual, originated by the development of a more general plural, was soon restricted to parts of the body and to objects occurring in pairs, where it was more justified semantically than as a synonym for two. As such, it was lost, except in Arabic and South-Semitic where it became an integral part of the language and spread to pronouns. Zuni, an isolated American language east of El Paso, Texas, shows dual in nouns, but only the first person of pronouns. Aleut has a general dual in nouns, but only in the second and third person of pronouns. Turkish has different endings for pronominal and nominal plural, the latter ones of rather recent origin and not common to the various Altaic languages. The pluralizer -iz of the personal pronouns seems to be common Altaic and seems to be present also in the -z found with some words that express things existing in pairs, e. g. göz "eye." Bang and Menges consider this -z an old dual, an excellent confirmation of the present theory. The third person pronoun (demonstrative) has the nominal plural suffix. This latter, however, is optional, while the -iz of "we" and "you" is essential. Based on this theory I consider it proper, when arranging the pronominal paradigms of the individual languages under discussion, to talk about dual and plural, even though the word for "we" may be the only such form, and category of number may be absent in the language. On the other hand, such a language with a dual category as Aleut will be shown to have only one plural form for the first person, though Veniaminov11' fills in the dual-position by repeating the plural form. The person pattern may know a dual that is not known for nouns, and vice versa. *) Schmidt, Pater Wilhelm, Die Sprachlamilien und Spiachenkreise der Erde, Heidelberg 1926, p. 316 2 ) Sommerfeit, Alf, La langue et la societS: caracteres sociaux d'un type archaique, Oslo 1938, p. 198 3 ) Grönbech, Κ., Der türkische Sprachbau, Copenhagen 1930, par. 195 ff. 4 ) Spraditheorie, p. 143 6 ) Grundriß I. par. 244, pp. 455 ff. 7 ) Ungnad, Arthur, Grammatik des Akkadischen, 3rd. ed., Munich 1949, par. 22, pp. 32—33 8 ) Sprachfamilien, p. 323 ") Leslau, Wolf, Lexique Soqotri, Paris 1938, p. 9 10 ) Grundriß I, par. 104, p. 297 ") Veniaminov, Ivan, The Aleut Language, from the Russian Opyi Grammatik! Aleutsko-Lis'evskago Yazika translated by Richard Henry Geoghegan, ed. Fredricka L. Martin, Washington 1944, p. 32 19
VI. Distribution of Three-Person System Do we always find the three persons as defined? Schlauch1* states on this subject: "There is reason to believe, indeed, that the first person pronoun (singular) was a comparatively late development in some languages. This is vivid grammatical testimony to the relative unimportance of the individual as opposed to the tribe! It seems to imply that in such tribes men could conceive of themselves only as parts of a larger social whole. In such cases it was not possible to say Ί do this,' but something like 'people do so-and-so by means of me, John'." This passage is also quoted in full by Balken2'. The basic thoughts of Schlauch's speculative statement have been thoroughly disproved by Schmidt3' both linguistically and anthropologically. There are, however, languages with only enclitic pronouns. Without free pronouns, we cannot express an absolute, abstract "I" or "thou" without saying anything about it. But whenever we use a word, we use it in some context, in relation with other words. We must say something about it. Our use of a pronoun by itself, unless it be a lexical translation exercise, is usually only an eliptic answer. A language without free pronouns then would have to repeat part of the sentence. They would have to say "it is I", or "my body," "I go," etc., and there is no possessive without something possessed. This is merely a matter of formal grammar, not of presence or absence of a basic concept. Often form words, such as "body", "person", or "head" with a personal suffix develop into subject pronouns. By analysing them into their constituent parts, we certainly cannot claim that the language really had no word for "I". Even if the Japanese choose an expression of modesty and submission to refer to the speaker, and one of respect to refer to the addressed, these words become, in due course, pronouns, designating the shift between speaker and addressed. The question whether or not such languages distinguish three persons becomes a mere play with words. There is no language we know of that would not single out speaker and addressed as special categories. According to Porteus4', the Central Australian tribes are the most primitive peoples known. Yet they have a common system of personal pronouns, while numbers above two, if they exist, are local innovations. Similarly, the Ural-Altaic languages have common pronouns, but different numerals. In this connexion it may also be of interest to study the first appearance of conscious use of personal pronouns in children. To quote again Bühler5', this time from his book on child psychology: 20
"Eine besondere Aufmerksamkeit hat man früher der Entwicklung der Pronomina der ersten Person gewidmet, weil man meinte, in dem Auftreten des ich, mir, mich dokumentierte sich unmittelbar das Erwachen des Selbstbewußtseins. Genaueres Zusehen zeigte indes, daß auf der einen Seite diese Formen vielfach nachgesprochen werden ohne eigentliches Verständnis, und daß auf der anderen Seite im Bewußtsein des Kindes die Absonderung der eigenen Person von der Welt und die Geltendmachung seiner Ansprüche und Wünsche lange schon stattgefunden haben kann, wenn es immer noch sich selbst mit seinem Eigennamen benennt. Ob das Kind früher oder später zu dem Gebrauch des ich, mir, mich kommt, das hängt ganz von seiner Umgebung ab." Children often use their proper name as a pronoun. Thus, Leopold") reports of his daughter Karla at two years and three months: "Karla eat my meat." The essence of this quotation is that the formal linguistic expression of the deictic content of the personal pronoun by the child depends entirely on the surrounding, but the concept to be expressed develops within the child itself. Sometimes, we find statements that certain persons coincide in one language or another, e. g. Paleosibirian languages or Papuan Kiwai. But this refers only to the verb, and on checking the statements made by Jakobson7' for Paleosiberian and by Capell8' for Kiwai, or, for that purpose by Swadesh (orally) for Gulf-Indian Chitimacha in other grammars I always found separate pronouns in existence which could be used with the verb to specify, if needed. Benveniste9) also checked the Paleo-Asiatic forms and concludes: "Mais toutes ces lan'gues possedent des pronoms personnels. Au total, il ne semble pas qu'on connaisse une langue dotee d'un verbe ού les distinctions de personne ne se marquent pas d'une maniere ou d'une autre dans les formes verbales. On peut done conclure que la categorie de la personne appartient bien aux notions fondamentales et necessaires du verbe. C'est lä une constation qui nous suffit, mais il va de soi que l'originalite de chaque systeme verbal sous ce rapport devra etre etudiee en propre." In a language that must use a pronoun with the verb for clarity, this pronoun is, for the purpose of the above analysis, apparently considered part of the verbal paradigm. ') Schlauch, Margaret, The Gilt oi Tongues, New York 1942, pp. 11—12 ) Balken, Eva Ruth, „Language and Psychology", The Encyclopedia oi Psychology, New York 1946, p. 321 ') Schmidt, Pater Wilhelm, Das Eigentum in den Urkulturen, Münster 1932 4 ) Porteus, S.D., "Primitive Mentality", The Encyclopedia oi Psychology, New York 1946, p. 572 2
21
5 ) Buehler Carl, Die geistige Entwicklung des Kindes, 5th ed. Jena 1929, p. 231 '} Leopold, Werner F., Speech Development of a Bilingual Child, vol. I., Evanston-Chicago 1939, footnote p. 87 7 ) Jakobsen, Roman, „The Paleosiberian Languages", American Anthropologist NS 44, pp. 617 ff., 1942 8 ) Capell, Α., Language Study for New Guinea Students, Sydney 1933, p. 20 ·) Benveniste, Emile, „Structure des relations de personne dans le verbe", Bulletin de la Societe de Linguistique de Paris, XLIII, 1946, p. 3.
VII. A Fourth Person? Some writers on grammar and linguistics use the term "fourth person." Jesperson1' takes it up and uses it in the sense of A telling Β what C, the third person, told D, a fourth person. He concludes himself that such a system would have no end. According to our definition the first and second person, as actors, have been singled out from the great pool of the impersonal. Whoever is neither speaker nor addressed is eo ipso a third person, grammatically. Thalbitzer2' calls the reflexive third person in Eskimo a fourth person. The temptation is strong. For a first person, oblique forms can only refer to the subject, same for a second person, although in the plural the situation can become more complicated, e. g. "you hurt yourselves" and "you hurt each other." In the second case, however, "each other" is no longer a second person, and the impression of a plural of the second person is destroyed by opposing individual third persons, emanating from the second person plural, to individual second persons. In the third person, two entirely different forms are possible, each truly a third person: "He hurts himself," "he hurts him." For convenience of description, one can invent here the term "fourth person." Again, as some American Indian languages have definite and indefinite pronouns of the third person, the latter is occasionally referred to as "fourth person." Hall") even goes so far as to call the French generic on a fourth person, and then announces the rule that the fourth person takes the verb in the third person. Again, the many different demonstrative distinctions possible in the third person have moved grammarians to use higher numbers. Nevertheless, it must be said that all languages separate speaker and addressed from everybody and everything else, and all the subdivisions of the third person follow from its place on the deictic scale. There is no room, in our definition, for a fourth person. *) Jespersen, Otto, The Philosophy of Grammar, London 1924, p. 220 ) Thalbitzer, William, „Eskimo", Handbook ol American Indian Languages, Part 1, Washington 1911, p. 1021 *) Hall, Robert Α., Jr., Structural Sketches, I, French, Baltimore 1948, p. 24 2
22
VIII. Some Remarks on Personal Pronouns Personal pronominal elements are found in different forms. They may occur either as enclitics (affixes), or as so-called free, separate, independent, disjunctive, or absolute pronouns. Many of the older writers considered the enclitic forms contracted, abbreviated, or corrupted forms of the free or "full" pronouns. For some languages this may be true, as Ray 1 ' claims, e. g. for Indonesian and Melanesian dialects. In general, however, it is not borne out. On the contrary, after studying hundreds of systems, I became convinced that the affix-pronouns often represent the pure pronominal element while the absolute forms usually consist of a nominal or, perhaps, demonstrative base and a pronominal affix. Sometimes pronominal elements may be combined, as Schmidt2) has shown for some Australian languages, and Hoijer3) for Texan Tonkawa. Wright makes the following interesting statement about pronouns in his Gothic Grammar4': "The most difficult chapter in works on comparative grammar is the one dealing with the pronouns. It is impossible to state with any degree of certainty how many pronouns the parent Indg. language had and what forms they had assumed at the time it became differentiated into the various branches which constitute the Indg. family of languages. The difficulty is rendered still more complicated by the fact that most of the pronouns, especially the personal and demonstrative, must have had accented and unaccented forms existing side by side in the parent language itself; and that one or the other of the forms became generalized already in the prehistoric period of the individual branches of the parent language. And then at a later period, but still in prehistoric times, there arose new accented and unaccented forms side by side in the individual branches, as e. g. in prim. Germanic ek, mek beside ik, mik. The separate Germanic languages generalized one or other of these forms before the beginning of the oldest literary monuments, and then new accented beside unaccented forms came into existence again. And similarly during the historic period of the different languages." The observation that the affix pronouns, not the free pronouns, constitute the original personal elements probably applies as a rule, while the exception also occurs. This statement is of great importance for our method. If we observe that a number of personal pronouns in a language are almost alike, but for a variation in suffix or in prefix, then the chances are that just this variation represents the personal element. Thus, if we compare Tigrigna5) nessatna "we," nessatkum "you," and nessatom "they," we notice 23
immediately a common base with different endings. The base can be indentified as ness from neis "person", plural nessat, and its function is to give "body" to the suffix pronouns. Not considering, at this point, psychological implications for the delimitation and definition of pronouns, we disregard nessat and only consider -na, -kum and -om. It is often impossible, without more historical and comparative knowledge, to link etymologically enclitic and absolute or object and subject pronouns. There is a possessive form, sometimes different from the subject form and from the oblique form. Often the object form also serves as subject of intransitive verbs. In many languages which inflect the free pronoun, subject form and oblique form can not be linked. A further complication may arise when more than one possessive form appear. Thus Waterhouse6* writes about the Blanche Bay Melanesian dialect of New Guinea: "Still another difficulty is the changed form of certain possessive pronouns relating to one's articles of food or drink or tobacco. In certain cases this is extended to poisons, charms, weapons, etc." Concerning Australian languages, Capell7) mentions: "Besides suffixed pronouns, the Northern Kimberley languages have prefixed possessives, limited, however, to parts of the body . . . Some other Worora nouns, namely terms of relationship, take suffixed pronouns, while the majority of nouns have independent possessives after t h e m . . . . Similarly, in Kokoyimidir, suffixed pronouns are used with the words for father, mother, and brother only. Narrinyeri extends the suffixed pronouns to all relationship terms." Boas8' reports a similar distinction in Dakota: "The same phenomenon (i. e. contraction to express closer semantic union, and formation of compound words) occurs in the possessive pronoun, intimate possession being expressed by contraction: t'a owi 'his earring,' i. e. those he made or happened to wear; fowi 'earrings he always wears and that nobody else has a right to wear'; t'a wowashte 'his occasional good acts,' towashte 'his goodness' as permanent quality." Quite often, the suffix (or prefix) pronouns give a clearer picture of the person system of a language than the absolute forms. On the other hand, these affixes may be so worn down that several forms come to coincide and must be analysed from the context or reinforced with the full form, which then gives a better picture of the oppositions in the particular language. Thus there is no strict rule and discretion must be used for every language to select those sets of pronouns that best show the pattern. This can, of course, only be done after all available forms have been scrutinized. Not all sets 24
may show all distinctions. Sierra Popoluca free pronouns, for instance, show only one plural of the first person, while the verbal affixes distinguish three degrees of inclusion. *) Ray, Sidney Herbert, A Comparative Study oi the Melanesia!! Island Languages, Cambridge and Melbourne 1926, p. 69 2 ) "Personalpronomina" (vide supra) p. 33 3 ) Hoijer, Harry, "Tonkawa", Linguistic Structure oi Native America, New York 1946, p. 303 4 ) Wright, Joseph, A Grammar oi the Gothic Language, Oxford 1910, p. 129 5 ) Leslau, Tigrigna (vide supra) p. 45 6 ) Waterhouse, J. H. L., A New Guinea Language Book, Sydney 1939, p. 12 7 ) Capell, Α., "The Structure of Australian Languages", Oceania VIII, p. 78 8 ) Boas, Franz, "Some Traits of the Dakota Language", Language vol. 13, no. 2, pp. 137 ff.
IX. Non-Pronominal Expression of Person Many languages have no personal inflexion. They just juxtapose the personal pronoun with the proper tense and mood of the verb. (I cannot see what we gain, for this discussion, by distinguishing "true" inflexion and "verbal nouns" used with pronominal elements for exactly the same purpose, though I am conscious of the fact that in some languages nouns and verbs are hardly distinguished, while in others they form two separate categories. For all practical problems of this chapter I overlook the distinction, often exaggerated, between so-called true inflexion and equivalent processes.) The verbal form itself then only indicates aspect and, perhaps, number, but not person. In other languages pronominal elements fuse with the verb and thus create an inflexion for person. The choice of pronominal elements, where more than one set are available, and its position may also serve to indicate aspect, as in Semitic and Maya. In such systems, usually only the first and second person are indicated by pronouns, while the absence of a pronoun indicates automatically third person; a noun or demonstrative often represents the subject, so that a pronoun besides it would be pleonastic. These systems are not essentially different from each other and often alternate between closely related dialects, e. g. in Australia. In addition to that we find an essentially different inflexional system, the Indo-European one. There are also, as in many other languages, two sets of suffixes. But these suffixes do not represent pronominal elements. Scholars have abandoned all efforts to identify these personal endings with pronominal elements. They never succeeded. And yet, there are definite forms that stand for definite persons in definite numbers. 25
Both Hirt 1 ' and Benveniste 2 ' have attempted quite convincingly to show that the different verbal forms are originally nominal elements which, somehow, specialized to form verbal "nouns" in conjunction with specific persons only, and thus became personal endings. Hirt writes 3 ): "'Zwischen der nominalen Flexion und der nominalen Stammbildung auf der einen und der verbalen Flexion und der verbalen Stammbildung auf der anderen Seite besteht eine unzweifelhafte Ähnlichkeit." He compares various examples, e. g. α in the first Greek declension dorä and in the Latin subjunctive ierämus, je in the Latin fifth declension fades and in the optative siem. He also equates the s of the s-aorist and of the neutral s-stems, etc., etc. I quote further 4 ': "Das indogermanische Verbalsystem ist durchaus nominalen Ursprungs. Eine Reihe von Nominalformen, teils der reine Stamm, der sogenannte Kasus indefinitus, teils Partizipia, werden zunächst in verbalem Sinne gebraucht, so das Partizipium auf -nt, das sich als 3. Plur., und das Partizium auf -to, das sich als 3. Sing, und 2. Plur. festsetzt. Die Übereinstimmung zwischen nominaler und verbaler Flexion ist, wie ich glaube, vollständig. Die Hauptfrage bleibt nun noch, wie sich die einzelnen indifferenten Nominalformen als bestimmte Verbalformen fixiert haben." Benveniste makes a similar statement 2 ': "Les desinences verbales ne comptent aucun element qui ne soit represente parmi les suffixes. While both scholars agree on the main issues, the system of distribution of the formatives has not been explained. As a matter of fact, it is not uniform in the various branche of IE, and too little is still known about the origin of these formative elements. Among the outstanding features, however, it may be noted that third person and participle usually go together. Imperative and vocative are kindrpd, and as the infinitive is often used for a command, a connexion between infinitive and second person plural is also seen. Yet the infinitive in men seems also to have a cross-connexion to the first person plural. Should the first person, the most subjective of all, really be connected with -men in Meillet's sense of "mente agitaie?"5' But this would lead to speculation. Why is the suffix of the first person singular in the subjective aspects identical with that of the object case -m? We cannot answer these questions now, but leads may result from the following analysis of the Mongol system and the paragraph on objective-subjective opposition. In Mongol, where we have originally only an aspect flexion with person indicated by pronouns, a system such as the IE one seems to be in the making if my analysis of Hambis' data is correct. It does 26
not yet exist in all tenses and moods but it makes its debut in the most subjective ones, especially in what Hambis calls the future. The two alternative forms in the following paradigm depend on sound harmony. Hambis describes the "future" aspect"): "Le... futur... exprime soit les actions projetees et meme probables, soit les actions simplement futures, soit les actions qu'on est dans l'obligation d'accomplir, qui sont necessaires et meme probables." He gives the following suffixes for the future: st
1 person alternative 2nd person 3 rd person alternative
SINGULAR -suqai
-sükäi
-qu -qu -ju
-kü -kü -jü
PLURAL -qu -kü -ya -yä -qu -kü -qu -kü -ju -jü
The suffix -qu, -kü really is an infinitive suffix, used for the future. This has nothing strange for the European, especially the Romance linguist. The alternative for the third person is really the gerundive ending, reminding us of the connection between1 IE third person and participle. If the basic future form in -qu, -kii is used for the third person, it is often re-inforced with a future of the auxiliary bol- "to become," e. g. törükü boloyu "he will be reborn." (cp. German future — infinitive plus werden "to become.") If the infinitive with the first or second person pronoun acquires future meaning, this may not be so clear in the third person when no pronoun is used with the infinitive. This explains the use of the auxiliary in the third, person or of the gerundial suffix. As to the first person, the singular, which is most subjective, cannot take the normal future form (infinitive). I have, however, not found any other form in the paradigm that would allow an analysis of -suqai, -sükäi. The alternative suffix for the first person plural, which can also take the regular form, being closer to the other persons than the singular, is closely related to the optative. The future describes an imperfective state, existing in the mind rather than in reality. For this reason the difference in subjective or objective nature of the persons is also brought out better, e. g. also in conservative English shall and will. But whatever the reason may be, we have here definitely a case where different so-called verbal nouns are used for different persons in building up a paradigm. While Altaic languages, closely related to Mongol, have built up an inflexion by addition of pronominal elements (although some suffixes are hard to explain), Mongol, which did 27
not inflect for person at all, is thus on the way to a system of the Indo-European kind. There is one more tense or aspect in Mongol where not all persons take the same form. While Hambis calls it surprisingly the passe indeiini, he explains that it also serves for the imperfect. (And it is in the imperfect aspects that we would first look for such a differentiation, due to their greater subjectivity.) In the third person the suffix of that tense, -lu'a, lü'ä, is often replaced by the compound suffix -ju'ui, -jü'üi (cp. future). To a smaller extent this Mongol development is also paralleled in Mandchu, as described recently by Sinor7); "L'optatif, qui sert ä exprimer le desir du sujet parlant de voir s'accomplir le proces enonce, a pour suffixes -ki et -kini. II s'emploie comme predicat de la proposition principale: ejen-i doro-be akumbuki 'je desire accomplir les devoirs de l'empereur.' II est assez difficile d'etablir avec exactitudes les regies qui president ä l'emploi respectif de -ki et de -kini. II semble cependant que le premier soit employe pour la l r e , et -kini pour les 2 e et 3® personnes, singulier ou pluriel". A little later, Sinor states: "Le verbe mandjou ne connait pas les formes personnelles et ne distingue pas non plus entre singulier et pluriel. II ne faut pas se laisser meprendre (sic) ä l'emploi occasionel des pronoms personnels. Tout d'abord celui-ci est limite aux l r e et 2e personnes, et deuxiemement les pronoms personnels ainsi employes sont des veritables sujets des propositions ä verbes actifs... Pour la troisieme personne, aucun pronom n'est employe, mais nous devons insister sur le fait que cette absence ne saurait etre consideree comme un formatif zero de la 3e personne; elle indique simplement que le sujet de la phrase est la 3e personne." This points again in the same direction as the Mongol development and may be the beginning of an inflexion for person. It is also interesting that again the most subjective first person has a form of its own in the subjective optative. The second and third person coincide but may, if necessary, be distinguished by absence or presence of the pronoun of the second person. Eastern Suketi, a Koci dialect of Western Pahari (Himalayan Indo-Iranian dialects with Tibeto-Burman and Munda [?] substrata) has only one form in the past for all persons, but distinguishes not only singular and plural, but also masculine and feminine. (The non-distinction of gender in the typical IE paradigm is another argument against its pronominal and for its nominal origin.) In the future, the suffix -ang serves for all persons and numbers in 28
Suketi, but there are two alternatives in -ma and -me for the first person singular and plural respectively. These, however, may be of pronominal origin. In any case, they provide again for a personal inflexion. The present conditional, also called indicative by Bailey8), yet certainly not an objective or perfective aspect, has one suffix (-ΰ) for the first person sg. and pi., one (-e) for the second sg., and one (-o) fo:r the second pi. and third sg. and pi. The third sg. can also take (-a). I have no explanation for these forms. The perfective aspects all have only one form for the three persons. Thus the development of the Eastern Suketi verb, which can by no means be called an IE type (while pronouns and numbers are Indo-Aryan), seems to follow the same path as Mongol. ') Hirt, Hermann, „Uber den Ursprung der Verbalflexion im Indogermanischen", Indogermanische Forschungen 17, 1904, pp. 36—84 do. Indogermanische Grammatik, 7 vols., Heidelberg 1927/37, vol. 7, pp. 19 ff. do. Die Hauptprobleme der Indogermanischen Sprachwissenschaft, Halle 1939, pp. 56 ff. do. Indogermanica, posthumously edited by Arntz, Halle 1940, pp.411—31 2 ) Benveniste, Emile, Origines de la formation des noms en Indo-europeen, Paris 1935, p. 173 3 ) Hirt, Indogermanica, p. 386 4 ) op. cit. p. 431 5 ) Meillet, Antoine, de indo-europaea radice men- 'mente agitare", Paris 1907, p. 56 e ) Hambis, Louis, Grammaire de la langue mongole ecrite, Paris 1946, pp. 49 and 51 7 ) Sinor, D., "Le verbe mandjou", Bulletin de la Soci0te de Linguistique de Paris, XLV, 1949, pp. 146—56, pp. 152—3 8 ) Bailey, Rev. Τ. Grahame, Linguistic Studies from the Himalayas, London 1920, pp. 208—9
X. Objective — Subjective Persons The first person, referring to the speaker, is subjective, even though one might make a factual statement about oneself. It refers to the subject, i. e. actor of the act of speech. If, at the same time, he is the grammatical or logical subject of the statement, then his announcement about himself is even more subjective. He originates the action. If the speaker is the object of the action, while still talking about himself, he is not describing what he does but what is done to him. The action is decided by someone else. The form, therefore, is less subjective. That this form then is often also used with intransitive verbs — the usage varies between different languages — may be connected with the structure of the verb which may, in some language, rather be without subject than without object. Or some kind of analogy may be at work. 29
The second person is inbetween. It is not the speaker, therefore the speaker may be objective about it. This would be brought out by Boas' and de la Grasserie's distinction of sell and not-self, moi and non-moi. On the other hand, the speaker takes a more personal attitude towards the one he is speaking to. Grammatically, as the conversation shifts between A and B, they are alternatingly I and you. Thus they are definitely no longer impersonal, i. e. third personal. The second person is included in the personal or subjective class, though to a lesser degree than the first. To analyse usage from this point of view would take large scale stylistic studies in most languages and tie up with many other problems. I am thinking, for instance, of one teacher in the Old Gymnasium in Nuremberg, who, when scolding a student and becoming increasingly angry, i. e. personal and subjective, shifted from the polite to the familiar address. He was famous for his: "Sie Lausbub, du dummer!" Similar instances can probably be found in many languages. But there are also languages, though few and far between, which have formal grammatical expression for this dual nature of the second person. One of them is Abyssinian Tigrigna which we had already occasion to quote. It has for the first and second persons forms of the common Semitic pronouns, and for the second and third persons ness 'person' or nessat 'persons' with possessive suffixes. The forms given by Leslau 1 ', rearranged for the purpose of our discussion are: singular 1 2 nd 2 nd 2nd 2nd 3rd 3 rd st
person person person person person person person
m. subjective f. subjective m. objective f. objective m. f.
ane ant a anti nesseka nesseki nessu nessa
plural nekhna antum ant en nessatkum nessatken nessatom nessatän
There is also an infrequent alternative for the plural of the first person, nessatna. This can only be explained by the fact that the plural incorporates third persons, or at least second persons together with the speaker. "I" thus is the only form without an objective form, while the third personal free pronouns have no subjective alternative. Another language with a similar distinction, even a threefold one, is North-American Quileute. Andrade 2 ' gives a vocative pronoun which is used in polite imperative and in general pronounce30
ments of a subjective nature. Besides it, he quotes an even more personal "imperative" pronoun. On the other hand, there is a more objective set, called "indicative" pronouns. Yet Andrade's table shows opposition of vocative and imperative only for the second person fem. Unfortunately, Andrade does not analyse the forms for us. They are: suffix pronouns sing. 1 pers. -Ji 2 nd per. m. comm. -lite 2 nd pers. f. st
free pronouns indicative vocative imperative plur. sg. pi. singular sing. plur. -lo lab lubaa -sto -ka tche heka'a tca-(li) -αχοί da-(li) ax
The vocative forms seem to contain, optionally, a first person suffix with a second person root, thus subjectivizing the addressee as "my thou" on the order of "my good friend, mi Brute". Maybe vocatives of the possessive pronouns, such as the just mentioned Latin mi Brute, or Sierra Popoluca (Mixe Zoque) man-casi "(thou) my child" should also be mentioned here. Other objective forms of the first and second person can be found in formal speech and court speech, especially iii Southern and Far Eastern Asia. In reality, however, the first and second persons, there are really replaced by nouns, i. e. third personal forms, in order to satisfy modesty by eliminating the personal element from speech. These forms have been discussed under "substitutions". While discussing the Indo-European and the Mongolian nonpronominal inflexion' for person, I have found it necessary to extend the distinction of objective and subjective also to aspects. Objective more or less coincides with perfective, subjective with imperfective. By using the same terms here and there, further insight is gained into the connexions between person and aspect. Person, being a subjective distinction by separating speaker and addressee from everybody and everything else, also occurs first and more frequently in subjective aspects. This is strikingly illustrated in Maya3). While the subject of transitive verbs is always expressed by the subject pronouns, the subject of intransitive verbs is expressed by the subject forms only in the subjective, i. e. imperfective aspects (present and future), but uses the object forms of the pronouns to indicate the subject in the objective, i. e. perfective aspect (past). It is also of interest that 31
the object forms in Maya are suffixes, while the subject forms are prefixes. In Semitic the perfective aspect is also formed with a pronominal suffix and the imperfective one with a prefix. Maya further illustrates the tendency of more personal distinction in the subjective form than in the objective one. Inclusive and exclusive first person plural are only distinguished in the subject pronouns. In English w e also have more personal inflexion in the imperfective tenses. The present indicative has the -s suffix for the third person singular, and the future and conditional, at least in conservative use, have the auxiliaries shall for the first person and will for the second and third. This tendency has also been observed by Jakobson. He mentions of Russian4): "Ainsi, en russe, l'aspect perfectif enonce la firi absolue d'un proces, par opposition ä l'imperfectif qui laisse la question de terme hors c a u s e . . . L'opposition des verbes determines et indetermines est seulement dans l'aspect imperfectif. Le present distingue les personnes contrairement au p a s s e . . . " In another context5), the same author says: "In Gilyak the neutral moods alone possess a person and number indication,...". We find a similar state of affairs in Sumerian"' where only the present-future (imperfective, subjective) distinguishes person and number, while the preterit (perfective, objective) distinguishes only number. While classical Arabic had a great wealth of forms, fewer are distinguished in modern colloquial Arabic. Again, more distinctions have been preserved in the imperfective aspect than in the perfective aspect. In those Athapaskan dialects where the plurals of the first and second persons coincide, this is only for object and possessive suffixes. Subject suffixes are distinguished. A connexion between the objective or subjective nature of tenses and their different formation has already been recognised by de la Grasserie. His classification of objective and subjective tenses differs markedly from the one used in this study, yet his basic conclusions can be fully accepted. They are first 7 ': " . . . le point de repere du temps subjectif est dans la personne, le point de repere du temps objectif est dans Taction elle-meme." And 8 ': "A cette distinction profonde des temps objectils et des temps subjectiis dans leur idee correspond une difference aussi complete dans leur expression..." A t the end of this discussion it may be worthwhile to consider the special treatment of the group of names which comes closest to personal pronouns both from the deictic and the subjective point32
of-view, the names of persons. When Latin declensions became more and more levelled in Romance, they were longest retained with proper names. Pei") writes on this subject: .. quelli italiani dimostrano all' opposto una torte tendenza alia conservazione del genitivo classico, specie per quanto si referisce si nomi propri". The English so-called Saxon genitive is used, preferably, for persons. In modern French and German the colloquial expressions for possession also seem to vary for persons and objects, e. g. le chapeau ά papa, but la poite de la maison, dem Vater sein Hut, but die Tür vom Haus. Old French: Ii fils le reis, but Ii murs de la maison. Modern Spanish has the "personal" α when a person is direct object. It has been mentioned before that many languages express number only for persons, or use separate pluralizers for them. ') Tigrigna, loc. cit. and p. 46 — Leslau lists the "subjective" forms on page 46 as "pronoms vocatifs". The coexistence of two forms for the 1st person plural, however, makes the distinction of objective and subjective forms more satisfactory. 8 ) Quileute, loc. cit. *) Tozzer, Alfred M., A Maya Grammar, Cambridge, Mass., 1921, pp. 42—3. also Gates, William, A Grammar of Maya, Baltimore 1940, p. 160 4 ) Jakobson, Roman, "Le signe zero", Melanges Bally, Geneva 1939, p. 145 5 ) do. "The Paleosibirian Languages", Am. Anthropologist, NS 44, 1942, p. 617 e ) Jestin, Raymond, Le verbe sumirien, Paris 1943, quoted after Marcel Cohen's review in Bulletin de la Societ6 de Linguistique de Paris, XL1I, 2, 1946, p. 166. — I accept this later and more spezialized authority against contradictory, but vague information on this point in Gadd, C. J., A Sumerian Reading Book, Oxford 1924, p. 35 7 ) de la Grasserie, Rauol, De la categorie du temps, Etudes de grammaire compare, Paris 1888, p. 3 8 ) op. cit., p. 4 9 ) Pei, Mario Α., "La costruzione 'In casa i Frescobaldi'", Lingua Nostra I, Florence, August 1939
XI. Gender, Class, and Location Distinctions There are a number of distinctions within person, the best known ones being sex or gender, class, and animate-inanimate with varying limits between men and animals or within the group. Animateinanimate opposition is often fused with a class system. Sex and gender often are closely connected, but sometimes gender comes closer to class than to sex. Other oppositions are those of location (varying degrees of proximity and remoteness) and of visibility. All these oppositions belong primarily to the third person, yet they may, occasionally, also be found in the second and even in the first person. Forchhelmer,
Language
33
Gender and class have been discussed from many points of view. Anthropologists, psychologists, cultural historians, philologists and linguists have cooperated to account for the origin of these categories, which, according to Bonfante, are vanishing as language develops. Bonfante has summarized our present knowledge of gender from the linguistic point of view1). As far as person is concerned, it must be said that gender still clings to pronouns, especially in the singular, where it is already lost in nouns. In the singular pronouns, however, I cannot see any sign that gender is losing ground, except, perhaps, the neuter, though languages that do not possess gender do not seem to lack anything. The distribution of gender seems to be quite independent of that of person, and historically I found hardly ariy connexion between the development of person patterns and gender in any way comparable to the connexion established between person and number. I also have not used gender as a criterium for classifying person oppositions. The sporadic existence of gender in individual dialects of larger families would warrant special investigation, but remains outside the scope of the present subject. I am only concerned with supplementing my survey of person patterns by a summary survey of the distribution of gender and of the various resulting enrichments of person oppositions. Gender exists in Indo-European, Semitic, Hamitic, and sporadically in Australian, Papuan, and American Indian languages, also in Balti (Tibetan) of the Sino-Tibetan languages and in some other ones. Most Indo-European languages have three genders, yet English has masculine and feminine for persons and personified animals and for ships only (extended to all kinds of machines and pets in substandard language). Romance, on the other hand, has completed the absorption of the Latin neuter into the masculine, and has thus created a two-gender system like that of Semitic and Hamitic. The Indo-European first and second persons do not distinguish gender, except in compound verbal forms that include a past participle, e. g. Fr. es-fu venue, Marie? These forms, however, are no real exception, as they have nominal (i. e. third-personal) character, the past participle being little different from an adjective. Semitic has two genders throughout, and includes the second person in the distinction. Only the Hadramaut-dialect developed a gender-distinction in the first person2'. In Hamitic3', Shilh distinguishes masculine and feminine only in the plural of the first peron. This, like the objective form for the first person plural in Tigrigna, can be explained with the fact that this form really includes other 34
persons as well. Usually, however, gender distinctions are lost in the plural first, cp. in the conjugation of Modern Arabic and Rabbinic Hebrew verbs. Modern English and Modern German also have only one plural for three different singular forms of the third person pronoun. Balti 4 ', too, distinguishes gender in the third person singular pronoun only. On the other hand, the Kakadu language of Australian Arnhem Land 5 ' has one form only for the singular and dual of the first person and the singular of the second person, while it distinguishes gender in the plural of the first, the dual of the second, and all numbers of the third person. Similarly, Spanish distinguishes gender in the singular in the third person only, but in plural in all persons, except the polite second person. The reason, of course, is composition of nos and vos with -otros, -otras, an element of the third person. Gumulgal and Saibalgal of the Cape York group, also North Australia, show gender in the singular of the first and in the third person, but not in the dual and plural of the first person, nor in the second person"). The majority of Australian languages possessing gender, however, restrict it to the third person, especially its singular. This may be considered a general linguistic trend, due to the nominal nature of the third person and the distinct nature of the singular. Representative of the American languages with gender is Tunica, spoken in the lower Mississippi region7)1. It has gender in the second and third person, singular, dual, and plural. The related dialects of Chitimacha and Atakapa do not show it. Class-distinction is predominant on the African continent, and also found in many American' and groups of Australian and Papuan languages. There is usually one class for human beings (in Bantu Class I for the singular, Class II for the plural), or for human beings and higher animals, etc. First and second person then belong entirely to these classes, as they are personal. In Australian classifying languages, e. g. Wunambul8), only this class has a plural. This is found in many other languages too. The third person changes its forms to conform with the class of the object. This, again, emphasizes the objective, nominal character of the third person. Very interesting is here to note that Sierra Popoluca Speech*), as previously quoted, has two plural suffixes, one for human beings, and one for objects. The former is used for pronouns and verbs in the first and second person, the latter in the third person, even though referring to human beings. 35
Apart from these distinctions which oppose a number of parallel third personal pronouns and, in some classifying languages or languages with gender, even verbal forms, to the simple forms for the first and second persons, there are many other distinguishing features. Many languages do not have a proper third person pronoun at all, e. g. most Indo-European languages, especially the older ones, have it only in oblique cases. The same is true for Turkic, Mongol, and Tungus, and many other dialects, e. g. American Indian Coos10) which has only a plural form. The reason is clear. If I am not talking about me or you I can use the name or a demonstrative. In many languages one demonstrative has practically become the pronoun of the third person, which, therefore, is less often cognate in related languages than those of the first and second, but in many other languages there may be a whole range of demonstratives, none of which is a third personal pronoun more than the others, e. g. Kwakiutl or Aleut. On the other hand, the accepted personal pronoun often can still do for a demonstrative, e. g. English he who lives in a glass house..., or Hebrew ha-ish ha-hu "that man." In Kwakiutl11', we find the following personal pronouns: l &t person Inclusive combinatory Exclusive combinatory 2 nd person 3 rd person near 1 st 3 rd person near 2 nd 3 rd person near 3 rd 12 Aleut ) shows even a greater range in the third person: ting "I," ixin "thou," wan "the first one nearest the speaker, counting towards the entrance of the dwelling," ingan ''the second from the speaker," ikun "the third one," akan "one far off," qagan "the last but one," qakan "the last one, i. e. nearest the door," qikun "one sitting in front of the speaker," qakun "one directly opposite," ikan "one above," axan "exactly above," unknan "one below," unan "one still lower," sakan "the lowest," ikun "the nearest one to the speaker in standing position," akun "the one standing farthest away," awan "the nearest one walking," udan "one lying near," sadan "one outside the house," uJcan "one inside the house," agan "one on this side." There are further variations, too. All these demonstratives together are alternatives for the third person pronoun, though Veniaminov prefers to mention' only ingan in the paradigm. These distinctions are clearly the realm of the third person. A sergeant or instructor facing a group whose names he does not 36
know, may use similar concepts in the 2nd person, such as you there, you with the ied hair, and so on, but, besides, being compounded with the unchanged "you," this is almost equivalent to "the one there," "the one with the red hair," really an objective description in the third person. If somebody then should try to speak and, not being recognised, refer to himself by a description of his whereabouts or looks, this, again, would rather be third personal. *) Bonfante, Guiliano, "Semantics, Language" Sec. 16, "Gender", Encyclopedia of Psychology, New York 1946, pp. 847—51 2 ) Grundriß I, par. 104, p. 297 3 ) Meinhof, Carl, Die Sprachen der Hamiten, Hamburg 1912, p. 27 f. 4 ) Read, R.F.C., Balti Grammar, London 1934, p. 12 8 ) Capell, Α., "The Structure of Australian Languages", Oceania VIII, p. 38 "The Languages of Arnhem Land, North Australia", Oceania XII, no. 4, June 1942 e ) Schmidt, Personalpronomina, p. 83 7 ) Haas, Mary R., Tunica, Reprint from Handbook oi Am. Indian Languages IV, New York 1940, pp. 37—8, "Tunica", Linguistic Structures, pp. 348 f. 8 ) Capell, Α., "Notes on the Wunambul Language", Oceania XI,3, Mardi 1941 ') Foster and Foster, Sierra Popoluca Speech, Washington 1948 p. 24 10 ) Frachtenberg, Leo, "Coos", Handbook oi American Indian Languages II, Washington 1922 p. 396 ") Boas, Franz, Kwakiull Grammar, Philadelphia 1947, p. 252 12 ) Veniaminov, Aleut Grammar, p. 31
XII. Substitution of One Pereon for Another Substitution of one person for another is found primarily if the "subjective" ("personal") person is to be replaced by an "objective" ("impersonal") person. This can provide a certain distance between the speakers which may be desirable for majesty or modesty. Already a change in number has this effect, as it really changes person as well by inclusion of other persons. Thus we have, in the first person, the pluralis majestatis or modestiae, or the polite plural of the second: English you, French vous, with parallels in Slavic, Arabic, Hebrew, Turkic, and elsewhere. But quite often nouns are used to describe either one's humble nature, not daring personal address, or the other's excellence, not suffering a personal relationship. The personal relations are expressed impersonally, but as new nominal forms begin to shift between speakers as do "I" and "you," the new terms become themselves personal pronouns of the first and second person. In some Asiatic languages we have long social scales of such formal expressions as "your servant," "my older brother," but they function as little more than pronouns. Only, we can mostly single out the real historical pronouns that still live on in the simple everyday use of certain social classes. 37
Before we have to consider nouns, we can find substitution by other pronouns. Thus older German used the polite plural of the second person, as Yiddish and Hessian dialects still do. Later the more impersonal singular of the third person was used, and finally modern German adopted the third person plural for polite address. Then, any number of nouns can be used instead of the pronoun, often with the possessive pronoun of the first or second person, depending on the sense, e. g. German meine Wenigkeit "my humbleness," Syrian Arabic hoditak "your honour," English Your Highness, etc. Often, when addressing people by title, the third person is used for the second: French monsieur desire?, German die gnädige Frau wünscht, Danish1' Hved mener professoren? But German Kellnerdeutsch has also a hybrid form which I have not yet found in literature, using the verbal form of the polite address (3rd person plural) with the title (3rd person singular). Whether we call it an anakoluthon or a mixed form, it does exist, e. g. Was wünschen der Herr Leutnant? Der Herr Direktor belieben... Haben der Herr schon bestellt? In some languages, e. g. Spanish Usted - Ustedes, this substitution has yielded one polite form that became the one polite pronoun. In Italian, we have a choice of three, familiar: tu, intermediate: voi (plural of familiar tu), polite: Lei (grammatically third person feminine), Plurals voi and Lore. In some other languages a whole list developed, e. g. Malay2' 1 st person: aku "I" (real pronoun), generally used by natives among themselves, implying familiarity and equality, sahaya ("companion, slave") used by Europeans with natives of all classes and by upper class natives with Europeans, hamba ("slave") or hamba tuan ("master's slave") used by person of inferior rank to those of superior rank, or by one of high rank to imply affectation or modesty, perhamba ("lowest slave") similarly used. beta ("slave") literary only, patek ("slave") used by natives when addressing a royal person, tean ("companion") used between upper class natives of equal rank. There is a similar list, of course, for address in the second person. In order to make a less personal and more objective impression, our journalists also use such expressions as your reporter saw, in this writer's opinion, etc. The former example tries to establish liaison with the reader and to bring his person in, while eliminating the reporter's person, so as to interest and convince the reader. On the other hand, modesty or fear often dictates the substitution of a noun for the speaker or hearer, e. g. Aramaic 3 ': lo bo'ino deliqberuho lehahu gabro, "I do not require that that man be buried," really 38
meaning "I do not want to be buried," or Arabic4': naqati warakibuha "my camel and her rider." On the other hand, the second person can also be substituted for the third person in general use, e. g. when the wind blows, you hold γ our hat, Latin diceris, Aramaic i bo-ith emo "if you want to, you can say," a formula introducing an alternative argument. When giving directions, all persons can be confused. Thus, Sommer5' quotes Hittite priests as constantly changing between the three persons when giving directions as to how to perform rituals. This phenomenon, however, is also found elsehwere. Jespersen"' mentions also the substitution of a paternal "we" for the address, e.g. a doctor's or teacher's "how are we today3'." It probably shows the real or pretended fusion of the interest of the first person with that of the second person. ') Jespersen, Philosophy oi Grammar, p. 218 j Winstedt, R. O., Malay Grammar, 2nd ed., Oxford 1939, pp. 106—13; also Hendershot, Vernon E., A First Year of Standard Malay, Mountain View, Cal., 1943, pp. 59—60, and Maxwell, William Edmund, A Manual oi the Malay Language, London 1907, pp. 48—50 ») Talmud Babli, Sanhedrin 46b 4 ) Brockelmann, Grundriß II, par. 26, p. 72 e ) Sommer, F., „Die Ahhijava-Urkunden", Abhandlungen der Bayrischen Akademie der Wissenschaften, philsophisch-historische Abteilung, N. F. 6, Munich 1932, pp. 89 ff. β ) op. cit., p. 217 2
Β. TYPES OF PERSON PATTERNS I. General Remarks The distinction of speaker, addressed, and neither speaker nor addressed is universally found. We refer to these categories as to the first, second and third person, respectively. Different types of person patterns arise, however, when references to more than one person at a time are made, and the resulting expressions are formalized into categories. In languages without number, expressions for the second person stand often, and those for the third person practically always for one or more individuals or (in the third person) objects. In the first person, however, a speaker referring to himself plus one or more persons belonging to the categories of the second or third person, rarely says "I." He either uses "I a n d . . . , " or he needs a new word. Some languages use a modification' of the word for "I," which is often the beginning of a regular plural hereinafter referred to as "morphological plural," 39
while others use an entirely different word, a lexical plural designating the self plus others as a unit. The lexical plural is sometimes restricted to designate a dual, "I and thou," e. g. in Dakota. If a distinction is made between including the adressee (inclusive form) and excluding the addresse (exclusive form), the lexical plural is mostly used for the inclusive. There may also be two such forms. Indonesian, for instance, has one form for "I plus thou or you" and one for "I plus he or they," and also a form for "you" as distinguished from "thou." Otherwise, the language is without number. Besides simple forms, their morphological and lexical duals and plurals, we also find "composite forms", i. e. pluralic forms that are composed of two or more simple elements, e. g. Melanesiari Pidgin English jumi ("you plus me") or Kele benonei "we-you-they". The majority of composite forms, again, is used to designate inclusive plurals. Composition, however, occurs also with deictic elements, e. g. in Kwakiutl, a language without number, where the inclusive and exclusive plural of the first person are expressed by "I plus visible" and "I plus invisible" respectively. In the following pages I start with the languages that use only three elements. Next come those that have lexical plurals, and third those with composite forms. Finally, I add illustrations of other systems that do not fit too well into the three main groups. In each group I start with languages without number, the ones with least forms, and proceed to those with more different forms. For this classification, I pay less attention to whether there is one plural, a dual and a plural, dual, trial, quatrial, and plural, or two or more plurals (paucity and abundance). I also do not use as a criterion of classification whether or not gender or class is distinguished, and in which persons. By this, I do not mean to minimize these important categories. I only found the morphological analysis more fruitful for the present typology and reserve other criteria for subclassification. II. Languages with Morphological Plural of Pronouns 1. A l l languages discussed in this section have in common that they form morphological plurals of the singular pronouns. They differ, however, when analysed against the structural background of the language as a whole. There are three main types of this class, distinguished by whether or not there is a nominal plural, and if there is, by whether or not it is formed in the same w a y as the pronominal plural. 40
The following schemes illustrate these types: Type A No plural of nouns SINGULAR PLURAL 1st person
I
2nd person
thou
thou-many
he
he(-many)
3
rd
person
I-many
Instead of -many, we may have -several, -class, or similar words, often also endings beyond analysis. Type Β Same pluralizer for nouns and pronouns SINGULAR PLURAL st
person
I
nd
person
thou
rd
person
he
1 2 3
I-s thou-s he(-s)
Type C Different Pluralizer for Nouns and Pronouns Pronominal Pluralizer, usually older, symbolized by -z SINGULAR PLURAL 1®' person
I
2nd person
thou
3
rd
person
I-z thou-z
he
he-z
A. LANGUAGES WITHOUT FORMAL PLURAL OF NOUNS OR WITH OPTIONAL PLURALS ONLY a. C h i n e s e ( P e k i n e s e ) The modern Chinese personal pronouns are pluralized by means of the suffix -men. (All references are to the Peking dialect1'). The forms of the pronoun are: 41
singular 1 st person
uo*
plural uo*-men
2
nd
person
ni5
ni'-men
3
rd
person
t'a1
t'al-men
Polite usage, as in Japanese, Korean, and Malay, demands omission of pronouns, i. e. an impersonal style wherever possible. But these pronouns definitely exist and can be used if and when clearness requires it.2' men1 as an independent noun means "class," 2 ' as a toneless suffix it indicates plural. Reduplication of nouns is used only for a plural of totality, i. e. to indicate "all of a kind."3' If the idea of plurality follows clearly from the context, no suffix is used with nouns, e. g. in connexion with numerals or expressions of quantity. The absence of a category of number in Chinese is illustrated also by the fact that the words ye8 '"man" and niar "woman" may, with the socalled plural-suffix -men, still be used as singulars as well as plurals 2 '. The formalization of pronominal plural while nominal plural still is avoided and felt as "Westernism" is probably not old. As a matter of fact, Ruedenberg's dictionary4'1 still translates uoS) ich, mein, wir, unser," and ni' "du, dein, ihr, euer, Sie, Ihr." That this corresponds to the popular Chinese "Sprachgefühl," though it may by now be archaic, is evident from Chinese Pidgin English. This language really consists of English vocables, picked up and moulded by native Chinese and used 5 ' "since the early eighteenth century in the Treaty Ports and in central and southern China." Hall states that only singular pronouns are used, wi "we," he states, is an Anglicism 6 '. It must also be said, in this connexion, that, while common roots have been found for the Sino-Tibetan pronouns, their pluralizations are local innovations. There are Sino-Tibetan languages where the word for "I" still can also be used for we, e. g. the Tibeto-Burman Kanawari dialect of Chitkuli. b. B u r m e s e The Burmese system, similar to Chinese, is even more remote from a notion of number, as it has no standardized pluralizer. Maspero7' described it in a posthumously published article: "La notion de nombre n'est pas exprime, mais en cas de necessite on fait suivre le nom d'un nom collectif myä3 "beaucoup," -to4 (-doi) qui 42
doit avoir une valeur analogue. Un trait curieux est que le verbe prend un affixe special krai (kyai) quand on croit utile d'indiquer que le sujet non exprime serait au pluriel.... influence du pali? . . . " The pronouns are: singular plural l3t person 2nd person 3rd person
ija koh
ijado kohio thu
Latter 8 ', who gives these forms, explains kohio < kohdo and gives for do the meaning "increase." The pronominal plural seems more formalized than that of nouns. Like nouns, the third person does not change for number. A s do can also be used to indicate a plurality of nouns, where the need for such an indication arises, we can probably say that the word for " w e " is felt as a plurality of "I." W e have a choice of plural suffixes, often also independent words in their own right, to indicate "more than one." The Burmese optional plural is therefore still less formal than the Chinese one. It became only formalized for pronouns as evidenced by a phonetic change in kohio. In pronouns, especially in the first person, a change in quantity can mean a change in quality, " w e " is different from "I." Thus number becomes formalized here first, but may almost have a lexical interpretation as long as it has not spread to nouns. The verbal pluralizer, however, even though it is rather restricted in its application, is a step ahead of Chinese towards number, as is the formalized standard plural of pronouns. c. J a p a n e s e Japanese also forms morphological Japanese pronouns are: singular 1 st
person
plurals of
pronouns. The plural
wata(ku)shi(domo)
wata(ku)shi-domo wata(ku)shi-tachi 2nd person anata anata-gata 3rd person cino-hito ano-hito-tachi -domo') is a plural suffix for persons only, etymologically connected with tomo "together." It has a connotation of humbleness. Henderson only lists wata(ku)shi-domo for "we,"yet I could not understand that there should only be an outspokenly modest form for "we". Logically, it would be out of place in an inclusive sense where, instead of expressing common modesty, respect should be shown to the other members of the group. A Japanese informant, Miss M. Kai 43
of the Columbia Far Eastern Library, confirmed that in such cases a neutral "we," wata(ku)shi-tachi, would be used. Most Western grammars of Japanese emphasize that wata(ku)shi means "servant" and is an expression of modesty. If this would still be felt, as in many substitutes for the pronouns in Malay, we would have to look for a real pronoun, which still exists in Malay. The fact is that watakushi, to the Japanese, means just plain "I" and nothing else, as confirmed by Miss Kai who had never heard of the meaning "servant" before. This, she said, is not known to non-linguists. -fachi10) is used to express nominal or pronominal plural, again with persons only. It is neuter as to politeness, modesty, etc. -gatan\ the plural suffix of the second person, is again a pluralizer for persons only, and implies respect. It is the only form used with the second person. There are few other pluralizers in Japanese. -nado12) means "such as." It is used with names of persons and of objects. As there is no real plural category, the pluralizers can be combined, watakushi domo nado "such people as we." With names of objects -nado is used without another plural suffix, while watakushi domo also is not necessarily a plural 9 ). Suffixes that clearly indicate plurality are only used with persons. Also used with nouns and pronouns is -ra 13 \ but preferredly only used with words denoting people. It extends the meaning of the word to "and others" and may even approach the meaning of English "-s etc.". Of Chinese origin is the more bookish suffix -teu\ originally meaning "class, grade, degree". It means, "such as", and is practically synonymous with -nado. The same Chinese ideogram, used in Japanese, may be read -nado, -τα, and -to, also -tachi, as the occasion may warrant"). From a discussion of these pluralizers one easily derives the impression that there is no formal plural in the European sense to be found in Japanese. Verbs are not inflected for number. An abstract plural can only be expressed for persons. But even here -domo, which usually expresses plural plus humility or derogation, can have the second meaning without the first, ko domo has become synonymous with ko "child"®) 14a'. Thus watashi domo can be a polite, humble expression for "I" as well as for "we"'). wata(ku)shi alone is a strictly neuter, colourless word for "I" and can be pluralized in several ways, as mentioned previously. d. M a i d u Other examples of this type can be found in North America, e. g. Maidu15), a Californian Penutian Language. The language is not 44
incorporating, but has both suffixed and "free" pronouns. The suffix pronouns distinguish three numbers in the first person, but none in the second and third. Maidu Suffix Pronouns dual plural -es -as -no -η Dual and plural seem closely related. The second and third person also seem closer to each other than to the first. There is no other indication of person or number in verbs. For a distinction of number in the second or third person it would be necessary to add forms of the independent pronoun, which does have a full system of number inflexion. Only a handful of nouns, such as "man, woman, husband, child" have number, and dual is more frequent than plural. It is indicated by the suffix -tso, which also appears in the dual of the second and third person free pronouns. With verbs, -tso has an iterative function, indicating "round and round, over and over". The nominal plural suffix, which also serves for the independent pronoun, is -se or -so, very similar (metathesis?) to the suffix pronouns of the first person. Perhaps the essential element of plurality is the s which both forms share, if not derived from another by metathesis. The absolute (composite) pronouns of Maidu are: 1st person 2nd person 3rd person
singular -a
singular
dual
plural
1st person
ni
ni-sa
ni-se
2nd
person
mi
mi'ntse
mi'nsö
3rd
person
mö'
möi'tso
mö'se
If we assume metathesis, the dual and plural of the first person are best explained as composed of the singular ni plus the dual and plural of the suffix pronouns. It is, however, hard to understand that a form meaning only "we" should be extended to pluralize nouns of relationship, excluding the speaker. Nevertheless, a common element ί exists, whether or not it is of common origin', and it has a vowel preceding or following, distinguishing it from the singular suffix of the first person. Thus, we have number developed primarily in the first personal pronoun, considering that the suffix pronouns are much more frequent than the free pronouns. Number is then extended to the 45
other pronouns and to nouns of close relationship plus words for man, woman, child. The plural, less frequent than the dual, morphologically reminds of the first person. The dual corresponds in form to the iterative of verbs. It is, therefore, not extended morphologically to the first person, but represented there semantically by a variation of the plural form.
e. Y u m a Another interesting illustration is offered by the Yuma18) dialect, spoken in the lower Colorado Valley, and belonging to the Yuman subgroup of Hokan languages. The possessive prefixes, which also serve as verbal prefixes, do not distinguish number. Nouns have no real plural, but do have collective and distributive plurals, which can both be indicated by the same two suffixes -c and -t. The former one we also find with the free pronouns. Verbs indicate number by vowel gradation in the theme, so that a finite verb had expression, though separately, of person and of number. The pronominal prefixes, with a basic (I) form used only for body parts and kinship terms, and an extended (II) form for natural and artificial objects, are: 1 st 2Dd 3 rd do.
person person person indef.
I — m— kw-
II αη^manyn7kwan?·
While these prefixes do not distinguish number, the free pronouns do form a plural by means of the suffix -c, accompanied by slight modifications in the theme.
Is* person 2 nd person
Is* person 2 nd person 46
Free pronouns 1. absolute form singular Ραπ7έρ, ?αη?άρ mär? 2. nominative singular ?an7ä'c män^c
plural Partie (missing)
plural Pan^cac mäcac
The Yuma system thus shows the following: Number and person are independent. We have three persons, but do not primarily distinguish singular and plural persons in the sense in which we distinguish them in IE. The same prefixes serve for both numbers, and, thus, it is impossible to distinguish whether the possessor, in case of possessive prefixes, is a singular or a plural. A plural suffix on the noun would only express a distributive or collective plural of the possessed objects. Thus, we must conclude, that the idea of person is independent of the idea of number. Distributive plurals really refer to recurrence of similar items, while collective plurals conceive of them as a unit, a singular. Both may be steps toward a formal plural, but are still more concrete than real formal plurals, which require a definite abstraction. This formal plural is absent for objects. Verbs, on the other hand, do show a separate expression of plural, indicating more than one actor. This indication, vowel gradation, seems quite unrelated to the nominal plural suffix. As plural in verbs refers to a plurality of actors which need not have anything in common but to share this action, the abstraction of a formal plural follows easily, almost automatically. It is not the distributive kind, referring to repeated actions, nor the collective kind which would require a singular verb, but a simultaneous, multiple action of a number of subjects. Free pronouns, on the other hand, are mainly used to re-inforce the expression of subject. Thus the formal plural of verbs could easily be transferred to pronouns, though it is expressed by the same suffix that stands for collective or distributive plural in nouns. It may well have started from pronouns and spread to nouns, as the individuality of persons stands out more than that of objects, and thus stimulates more than objects do an expression of non-singularity. As a plural of person indicates, first of all, a number of individuals, it may have started with a distributive connotation which, after spreading to nouns, developed further into a formal category. The absolute forms of the free pronoun show the morphological structure more clearly than the nominative forms, which contain another c, besides the plural suffix, in the function of case determinant. But, as the 2Dd person plural of the absolute form is not verified in my source, I prefer to add the nominative forms, a full paradigm of which was available. Considering the formalized distributive plural in Yuma, this language is, in a way, a transition to the next sub-group to be discussed, i. e. languages with a formal nominal plural, still showing identical formation of nominal and pronominal plural. 47
Β. LANGUAGES WITH FORMAL PLURALS OF NOUNS a. K o t t i s h Kottish, as described by Castren, was spoken along the Siberian rivers Kungus, Kan, especially also around the city of Kansk, and Ulijka, a left tributary of the Agul river, into which the Kungus empties from the right side. The Kottish pronouns are 17 ': st
1 2 nd 3 rd 3 rd
person person person m. i person f.
singular ai au uju • uja
plural ajoy auorj ] < j
uman
Verbs are inflected for person and number, i. e. for each tense and mood they have six forms, like Latin. The plural forms end in -/? or -n18>. The same consonants also serve as general noun pluralizers, preceded by connecting vowels if the noun ends in a consonant. Thus there is a formal plural for nouns as well as for pronouns, using the same formatives. Possessives are derived from the personal pronouns"). Thus a possessive indicating a plural possessor and a plural possessed, e. g. ajonsin "nostri". This form is composed of the genitive plural ajot) "of us", the possessive formative Se and a second plural suffix -n. Thus plural is indicated separately, in its proper places, for possessor and for possessed. Historically, the picture is more complicated. Castren states 80 ': "(Es)... gibt im Kottischen keine bestimmte Pluralendung. Im allgemeinen ist der Plural in dieser Sprache wenig im Gebrauch gewesen und die wenigen Wörter, welche diesen Numerus nicht entbehren konnten, haben zum Unterschied vom Singular einzelne Laute des Singulars einer Umgestaltung unterworfen. Fast alle einsilbigen Wörter haben einen solchen Plural. Die übrigen haben, wie es scheint, ihre Pluralendung dem Samojedischen entlehnt. Die Pluralendung ist teils n, teil η, welche die vocalisch auslautenden Nomina unmittelbar an den Stamm fügen, während die consonantisch ausgehenden einen Bindevocal zu Hilfe n e h m e n . . . " If we admit that the general formal plural which has the same suffix for nouns, pronouns, and verbs, is a late borrowing, we would have to assume that there was an older plural in existence for some nouns before there was a plural of pronouns. Such an assumption is very difficult, indeed. On the other hand, on the basis of the language as described in Castren's grammatical essay, we can talk 48
of a language which, besides irregular internal plurals in some nouns, uses a general pluralizer (with a few alternating forms) for pronouns, nouns, and verbs. b. H u r r i a n Other examples of this group can be found in ancient Asia Minor, e. g. Sumerian and Hurrian. Our knowledge of Hurrian is still very limited; but the possessive suffixes, as described by Speiser41', illustrate its system of these pronominal plurals well. They are: singular
plural of possessor
1 st person -iwwa -iwwas 2 nd person -p^ (not ascertained) 3 rd person -i -as(e) Though still subject to revision, these forms are worth considering. It would be surprising, indeed, if the second person plural possessive did not fit in with the forms so far known. The third person plural suffix is added to the first person to pluralize it. There are various nominal plural suffixes, but the element s certainly is a prominent part of them. Speiser calls it a merit of Friedrich's to have shown that -sus marks the plural of nouns the singulars of which end in -s2i>. c. S u m e r i a n A similar scheme can be observed in Sumerian for which we have more information. When trying to analyse Sumerian forms, we have to consider that this language has vowel harmony. Besides, it also shows some vowel gradation. Thus the suffix of the first person is -DIU"', while the free form is ma. The oldest form of pluralization in Sumerian seems to be reduplication which is still practiced in classical times, but accounts for a minority of forms only"'. The majority is formed with suffixes, mostly with -ne. But this suffix seems to be a contraction of the suffix pronoun -ne-ne, "they", formed by reduplication of -ni "he, she, it." It also serves to pluralize pronouns. In the first person, where there also occurs a vowel change to e, perhaps under the influence of the following ne, the η is, however, optional. The suffix pronouns in Sumerian are"': singular -mu -zu (-a)-ni -bi
1 st person 2 nd person 3 rd person Forchheimer,
Language
plural -me(n) -zu-ne (-a)-ne-ne bi-ne(-ne) 49
Whether or not the plural suffix -ne stands originally for "they", a development known elsewhere too (e. g. in Central Australian Aranda), the fact remains that pronouns and nouns share the same plural suffix, though there are other suffixes (e. g. -mes)24) and other methods for pluralization available. The suffix shared with pronouns, however, is the most used with nouns. There are several forms of the independent pronoun, but that table would confuse here rather than help. It may suffice to mention that all plurals are formed somehow from the singulars, and that, besides other formatives, and in varying composition, -ne- is always present in the plural, except that it may be omitted in the first person. In the third person, the singular e-ne may occasionally also serve for the plural"). d. S h i l h In Africa, this type is represented in Shilh, a Berber language, The pronouns of Shilh are2«): singular plural 1 st person m. nkunne nki(n) f. nukenti nd 2 person m. kii(n) kunne f. kimi(n) kunemti nta, nten nitni 3 rd person m. f. ntet nitenti -un is the Shilh plural suffix. Meinhof analyses the pronominal plurals as having a plural suffix -une, and, for the feminine, the feminine suffix -(e)nti. The third person feminine shows the element -t to distinguish it from the masculine. It is of interest that the first person distinguishes gender in the plural, though not in' the singular. This can be more easily understood in languages where the plural of the first person is lexically distinct (cp. Spanish: yo, nosotrosnosotias), but if the word for "we" is felt grammatically as "I's", and if the singular does not distinguish gender, this form is worth mentioning. e. W e s t G r e e n l a n d
Eskimo
This type is also illustrated by Eskimo dialects. For the WestGreenland dialect, the personal pronouns27) are: 1 st person 2nd person 3 rd person 50
singular u w ana iL'it una
dual u^aguk ilit'ik
plural ifagut ilis'e uko
While the second person has a regular dual, but a special plural, and the third person has no dual, and again an irregular plural, the first person has a regular dual and a corresponding plural in -t, also the nominal plural suffix. (-K is also the nominal dual suffix.) Birket-Smith, who lists the above forms, gives no grammatical information, but Thalbizer28' mentions that Greenlandic Eskimo dialects have singular and plural, rarely dual, in nouns and verbs, but do have a dual in the possessive suffix, or rather separate duals for possessor and possessed. Thus, as we have a combination of two, possessor and possessed, and each can be in one of three numbers, singular, dual, or plural, we have 3 X 3 , or nine possible forms. Though some forms overlap in practice, these three numbers can definitely be said to be distinguished29). C. LANGUAGES WITH DIFFERENT PLURALIZATION FOR NOUNS AND PRONOUNS It is hard to draw an exact line between languages with identical and languages with different pluralization of pronouns and nouns. While these two groups certainly stand out and justify this division, there are transitions and cases that are hard to classify. In Sumerian, we saw that the pronominal plural suffix was really optional in the first person and, while predominant with nouns, still not the only such suffix, and suffixation not the only nominal pluralization. Yet, as this suffix did exist, though optionally, for the first person also, and as it was the most frequent noun pluralizer, we were justified in stating that there is analogy between nominal and pronominal pluralization. The crucial point, psychologically, was the morphological justification that "we" seems to have been felt as "I's". Considering that the third person comes closest to general nouns, while the first person is the most subjective and "personal," with the second person somewhere in between, the fact that the first person in Greenlandic (and most other) Eskimo dialects takes the nominal pluralizer, was taken as sufficient to classify these languages with those having identical pluralization. The group we are going to examine now still has the same elements in singular and plural of the same person, usually with analogous pluralization in first and second person. The third person may follow this analogy, as in Jennisei-Ostyak, or may follow nouns, as in Turkic. Pronouns, i. e. at least those of the first two persons, and nouns exhibit diverging inflexion for number. Due to the different regular pluralization of nouns, "we" is most likely less strongly felt as a plural of "I". This group is of great importance as it comprises languages in the 4