+ All Categories
Home > Documents > FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

Date post: 03-Oct-2016
Category:
Upload: laszlo
View: 214 times
Download: 1 times
Share this document with a friend
14

Click here to load reader

Transcript
Page 1: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAMEGRAMMATICAL CATEGORY

LÄSZLO ANTAL

1. It is the aim of this paper to show the extent to which the generalaspect of a grammatical category is affected by the type of grammarused to establish it.

By grammar I understand any particular system of hypotheses, the aimof which is to describe linguistic reality. It is evident that the number oftheoretically possible grammars is endless. Historically, however, therehave been four different kinds of grammar, namely:

a) traditional grammar, which originated in the period of the classicalGraeco-Latin civilization;

b) the grammar, often known today as "constituent analysis", i.e., the"immediate constituents" method;

c) string analysis, which is an independent kind of grammatical des-cription, and

d) transformational grammar.The three latter types of grammar are commonly called structural and

are different from traditional grammar. A well-known feature of struc-tural grammars is their formal character, whereas traditional grammar isoften interpreted as a grammar which - in contrast to the structuraltype - is built up upon another, i.e., a non-formal, principle. In currentlinguistic literature contradictory views are often expressed about theformal or non-formal character of grammar. The following quotationsillustrate the dissatisfaction felt with the formal procedures of the struc-turalists :

"The real trouble is not that structuralists are avoiding semanticcriteria in their phonological and morphological studies, but that theyseldom venture beyond phonology and morphology. As a result, lin-guistics tends to become a purely formalistic discipline, a kind of 'pseudo-mathematics' devoid of any humanistic content."1

"Welchen Aufschluss kann uns das alphabetische Wörterbuch und die

1 Ullmann, St., The Principles of Semantics (New York, 1957), pp. 318-19.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 2: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

52 LÄSZLO ANTAL

formale Grammatik geben über den inhaltlichen Aufbau der Sprache?"2

"Die formale Analyse kann demnach weder den vollständigen Zu-sammenhang, noch den Grund der von ihr nachgewiesenen Strukturerschliessen. Um diese Unzulänglichkeit zu beheben, müssen wir ihreSchranken überschreiten, den grammatischen Apparat als die Einheitvon Form und Inhalt betrachten."3

"Nun ist die Frage, ob eine streng formale Grammatik möglich sei,stark umstritten 4 Während wir bei einer formalen Charakterisierung,durch mechanische Anwendung der Formationsregeln - sobald dieseverstanden sind - Sätze konstruieren können, erlauben die grammatischenRegeln einer gewöhnlichen Grammatik im allgemeinen kein mechanischesKonstruieren von Sätzen... .5 Aber selbst die Gesamtheit aller gramma-tischen Regeln genügt im allgemeinen nicht, um einen einzigen, gramma-tisch korrekten Satz mechanisch, ohne Bezugnahme auf die Bedeutung,zu konstruieren."6

On the other hand, there are authors who maintain the opposite view.For example, here is a quotation from Chomsky: "A great deal of efforthas been expended in attempting to answer the question: 'How can youconstruct a grammar with no appeal to meaning?' The question itself,however, is wrongly put, since the implication that obviously one canconstruct a grammar with appeal to meaning is totally unsupported. Onemight with equal justification ask: 'How can you construct a grammarwith no knowledge of the hair colour of speakers?' The question thatshould be raised is: 'How can you construct a grammar?' I am notacquainted with any detailed attempt to develop the theory of grammaticalstructure in partially semantic terms or any specific and rigorous pro-posal for the use of semantic information in constructing or evaluatinggrammars."7

On the basis of the above-mentioned quotations we have to investigatethe following problems:

a) Is a non-formal grammar possible?b) If our answer to the above question is in the negative, how can we

explain the alleged non-formality of traditional grammar?

2 Weisgerber, L., Vom Weltbild der deutschen Sprache, I. Halbband: Die Inhalt-bezogene Grammatik (Düsseldorf, 1953), p. 99.3 Telegdi, Zs., "Über die jüngere Entwicklung der Sprachwissenschaft", Acta Linguist.Hung., XII (1961), p. 246.4 Hiorth, F., Zur formalen Charakterisierung des Satzes (The Hague, 1962), p. 10.5 Ibid., p. 99.6 Ibid., p. 100.7 Chomsky, N., Syntactic Structures (The Hague, 1957), pp. 93-94.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 3: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 53

2. In every linguistic community a great many sentences are spoken andheard daily. Although the number of these sentences is not infinite, itcannot be given with any exactitude. These sentences lend themselves tothe most varied types of classification. For our purposes here, I proposeto divide them into two large classes along the following lines:

A considerable part of the sentences that are spoken and heard dailyconsist of those which "cannot be addressed by everybody to everybody".To this class belong the sentences that occur in astronomy and can beused by astronomers only among themselves because lay people wouldnot understand them. Similarly, the sentences used in the carpenter'strade also belong to this class. These sentences, i.e. sentences that aretypical of a particular science, trade, body or community (such as thearmy, etc.), may be termed professional or particular sentences.

As against these particular sentences we have those that "can be address-ed by everybody to everybody", for example, by the astronomer to thecarpenter or vice versa. Examples of such sentences are It looks like rain,I'm hungry, Good morning, etc. To distinguish these sentences from theformer class we may call them universal sentences. What is a universaland what is a particular sentence in a given linguistic community dependson a number of factors. At the present level of our general culture thesentence The earth revolves around the sun can be regarded as universal,despite the fact that originally it was a particular sentence of the astro-nomer. The spread and advance of general culture can be gauged - fromthe linguist's point of view - by the increasing number of originallyparticular sentences which have become universal.

Let us now ascertain which of the two types of sentence is the mostsuitable for linguistic analysis of content. In my opinion, it can hardlybe disputed that the investigation of the content of particular sentencesis out of the question. Sentences characteristic of particular sciences oroccupations can be understood and analysed only by the specialistsconcerned: sentences on the subject of astronomy only by astronomers,and so on. Thus, linguistics is unable to undertake the task of investi-gating the content of a considerable number of kinds of sentence.

Hence only universal sentences now demand our attention. Thecontents of these are common knowledge to everybody. They couldconceivably be subjected to content analysis by linguistics, but a goodmany other sciences would be equally competent to do this. Why shouldlinguistics investigate them in preference to any other science? If, inspite of this, we should allow that linguistics is competent to investigatethe content of universal sentences, a peculiar situation would arise in

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 4: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

54 LÄSZLO ANTAL

that the content of certain sentences would be analysed by linguistics,while that of others would not. Thus we cannot formulate statementswhich are valid for every sentence. It is evident that the legitimatesubject-matter of linguistics can only be that which

a) can be investigated in the case of all sentences, andb) it can be investigated only by linguistics.Since linguistics cannot investigate the content of particular sentences,

and since, furthermore, the content of universal sentences can be inves-tigated by other sciences as well, we have to conclude that the analysis ofcontent is outside the scope of linguistics in the case of both types ofsentence.

Let us again divide into two groups sentences that are spoken andheard daily. One group will comprise sentences that are not concernedwith language and the other group will include all those that are. It isclear that the latter group contains the particular sentences of linguistics.(Any science can be conceived of as the sum of a certain number ofparticular sentences.) The question arises as to whether the particularsentences which refer to linguistics have content. They certainly have.What does this content consist of and what does it refer to? We have seenpreviously that linguistics cannot analyse sentences on the basis of theircontent. It follows from this that the content of the sentences of lin-guistics cannot refer to the content of the sentences of other sciences -it can refer only to their form.

Linguistics, therefore, can only examine the content of its own parti-cular sentences; its own sentences, however, differ radically from allother particular and universal sentences in that their content is the formof the other sentences. Thus, that which is form in the case of the othersentences becomes content in the case of the sentences of linguistics. Wemay therefore conclude that linguistics is composed of sentences which- in common with all other sentences - have form and content, but thecontent in this particular case relates to form, the form of other sentences.

The content of these sentences is no concern of linguistics, because itreflects reality - naturally, at various levels. Linguistics, on the otherhand, is concerned only with one part of reality, namely, language itself.Content is no more part of language than are occasional travellers part ofthe bus in which they happen to travel.8

As we have seen in the short discussion above, linguistics cannotinvestigate the content of sentences, with the exception of its own sen-8 These thoughts and their consequences are outlined in my Content, Meaning,Understanding (The Hague, Mouton & Co., 1964).

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 5: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 55

tences. It follows from this that every description of language - or everygrammar - can only be formal. A grammar which is based upon formalprinciples is not just one possible grammar and certainly not an imper-fect grammar; but it is, in fact, the only expert form of grammar. Agrammar based on content is not only impossible in practice, but -because of the special nature of language - is also impossible in theory.9The truth of this statement is supported by a further consideration.

The fundamental task of language is to give a description of reality.("Description" is used here in the sense of "Darstellung", a term whichwas introduced by Bühler.) Hence the content of language is a descript-ion of reality. Now, if we wanted to investigate the content of sentences,then our description would be a description of a description. In this casetherefore we would be investigating not language itself but the descriptionof reality presented by language. However, we would not be able toconfirm whether this description were true or false by comparing it withlanguage; we could only do this by comparing it with reality itself.

It follows from the above that every kind of grammar has to be formal,no other kind of grammar being even conceivable. In this case, one couldask, what is the status of traditional grammar, which - at least accordingto the common view - is a grammar based upon non-formal principles.Traditional grammar is not an independent type of grammatical des-cription but is an imperfect kind of structural grammar. In other words,it is a grammar only to the extent that it is formal. Now, we know thattraditional grammar consciously operates on the basis of semanticcriteria and that it sums up its results in semantic terms. Traditionalgrammar could, therefore, only make a true statement if it became un-faithful to its own principles.

3. After this short theoretical introduction, we will examine a particulargrammatical category, the Hungarian attribute, in the light of fourdifferent grammatical conceptions. However, we must first list thephenomena which various authors have assigned to the "attribute"category in Hungarian grammatical literature. The chief types are asfollows:9 Therefore Weisgerber errs when he maintains that formal grammar is only an inter-mediate phase and not the final aim of linguistics; see, for example, the followingquotation: "Nun ist wiederum zu sagen, dass gegen eine lautbezogene Ordnung undVerarbeitung sprachlicher Beobachtungen an sich nichts einzuwenden ist; solcheGrammatik ist eine notwendige Zwischenstufe, und wir könnten sie in keiner Wieseentbehren. Die Fehlentwicklung lag aber darin, dass diese lautbezogene Grammatikweithin zum Endziel wurde." Op. c//., p. 98.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 6: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

56 LÄSZLO ANTAL

a feher lo

a dolgozo no

a szabo Bencsik

egy pohar bor

afiu könyve

afiunak a könyve

(= the white horse; a — the; feher = whiie; =horse; the attribute is the word feher)(= the working woman; dolgozo = working; this isthe present participle of the verb dolgozni 'to work',which is formed by adding the suffix -o to the rootdolgoz-; no = woman; the attribute is the worddolgozo)(literally = the tailor Bencsik; the attribute is theword szabo)(literally = a glass wine, i.e. a glass of wine; egy = a,pohar = glass; bor = wine; the attribute is the wordpohar)(= the boy's book or the book of the boy; fiu = boy;könyv — book; and the -e at the end of könyve is thepossessive suffix of the third person: the attribute isthe word fiu)(= the boy's book or the book of the boy; this con-struction is identical in all respects to the previousone apart from the fact that the suffix -nak, which isthe ending for the dative, occurs at the end of theword fiu; the attribute is the wordfiunak)(=my fatherland; en = /, haza = fatherland and the-m at the end of the word hazam is the possessivesuffix of the first person; the attribute is the worden)(=[the] three girls; harom — three; kislany = girl;the attribute is the word harom)(= as in lattam egy lovat, feher et = / saw a horse,a white; lovat = horse in the accusative; feher et =white in the accusative; -/ is the ending for the accu-sative, modified in the quoted example in accordancewith the morpho-phonemic rules; the attribute is theword feher et)

We have enumerated the chief types of Hungarian attribute as hithertoset out in various grammars. On the following pages we will examinehow the general aspect of the Hungarian attribute is modified in the lightof different grammars.

az en hazam

a harom kislany

egy lovat, feher et

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 7: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 57

4. THE ATTRIBUTE IN THE TRADITIONAL GRAMMAR. Every traditional Hun-garian grammar - and with two exceptions they are all traditional10 -stubbornly states that "the attribute modifies, makes more precise andenriches the meaning of the word it qualifies".11 A similar statement canbe found in every non-structural grammar of every other language.Therefore, we have to investigate the question as to whether an attributecan modify the meaning of the word it qualifies.

Let us compare the following two sentences: A dog walks along thestreet; A white dog walks along the street. As we can see, the word dogis used in both sentences. The two sentences are not identical, since inthe second sentence the word dog has the attribute white. Now, accordingto the view of traditional grammar referred to above, the meaning of theword dog cannot be identical in the two sentences quoted, for in thesecond sentence its meaning is modified by the word white. Is thistrue?

In the case of A white dog walks along the street, can we say that theword white complements the meaning of dogl Can the meaning of dogbe white? No! It only means that the actual dog to which the sentencerefers is white. Whether we adopt the correct procedure by accepting themeaning as a rule or regard it as an image or concept, it is obvious thatneither an image nor a rule can be white. In the sentence A white dogwalks along the street the word white qualifies not the meaning of theword dog but its denotatum.12

This means that the view held by traditional grammar that the attri-bute modifies the meaning of the qualified word, is false. As we haveseen, the attribute does not modify the meaning but helps us to becomebetter acquainted with the denotatum of the qualified word. It followsfrom this that the qualified word is qualified not with regard to meaningbut with regard to denotatum.

We can now argue as follows: The definition of the attribute in tradi-tional grammar was, in fact, erroneous, but we can correct it by sayingthat the attribute is that part of the sentence which modifies, makes moreprecise and enriches our knowledge of the denotatum of the qualifiedword. Nevertheless, we can easily see that this definition of the attributeis also untenable. Let us compare the following two structures:10 These exceptions are: Das Ungarische Sprachsystem by J. Lotz (Stockholm, 1939),and Hungarian Grammar by R. A. Hall (Baltimore, 1944).11 It gives, for example, a very detailed summary of the traditional views in the so-called "academic grammar": A mai magyar nyelv rendszere (System in Present-dayHungarian) (Budapest, Part I, 1961; Part II, 1962).12 See my Questions of Meaning (The Hague, 1963), pp. 53-59.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 8: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

58 LÄSZLO ANTAL

The white dogThe dog is white.

In both constructions the word white performs the same function, i.e.,it describes the denotatum of the word dog in more detail. Nevertheless,it is an attribute in the first construction, but not in the second. There-fore, the traditional definition cannot be maintained even if we replacethe word "meaning" by the term "denotatum". (To say nothing of thefact that the denotatum is a part of extra-linguistic reality and as such isunsuitable for the determination of linguistic elements.)

Thus we have to conclude that the definition of the attribute proposedby traditional grammar is erroneous and serves no purpose. If we wishedto determine the concrete attributes in concrete sentences on the basis ofthis definition, our efforts would not meet with success. The fact, how-ever, that traditional grammars - in a great majority of cases - haveselected the attribute correctly shows that grammarians have used other(i.e., formal) methods intuitively, despite their statements to the contrary.

It is a well-known fact of the Hungarian language that the attribute -if it precedes the word it qualifies - remains unaltered in every numberand case, whereas if it follows the word it qualifies, then it receives theappropriate suffixes. The importance of this latter fact has been over-looked by traditional grammar, and apposition - the term by which theattribute was denoted in Hungarian national grammar if it followed theword it qualified - has been treated as a sub-type of attribute. As weshall see in the following sections, apposition does not fall into the cate-gory of attribute.

5. THE ATTRIBUTE IN THE CONSTITUENT TYPE OF GRAMMAR. In a structuralgrammar which is based upon the concept of the immediate constituents,the Hungarian attribute can be defined as follows: In terms of the wordclasses, an attribute can only be nominal, i.e., an adjective, a substantive,a numeral or a pronoun. Furthermore, the attribute cannot be the headof a construction, only its complement, but a complement which isalways added to a nominal part of speech. Hence we can make thefollowing definition: The attribute is a nominal complement of a nominalpart of speech. This definition is, however, not strict enough for thefollowing reason: There are nominal parts of a Hungarian sentence whichare complements of nominal parts, in spite of the fact that their propertieshave nothing in common with the properties of the attribute. Here aresome examples:A kabatban dolgozo no (literally = The in coat working woman; i.e.,

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 9: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 59

The woman who works in a coat; kabat = coat;-ban = the suffix of inessivus)

A sokat dolgozo no (literally = The much working woman; i.e., Thewoman who works very much; sok = much; -at =the accusative suffix)

If we were satisfied with the definition that an attribute is any nominalpart of a sentence which is a complement of another nominal part, thenthe words kabatban and sokat would also be attributes, for they arenominal and they are complements of a nominal part, namely dolgozo.Nevertheless, we know that the part kabatban is an adverbial complementand the part sokat is an object. Therefore, we have to limit our definitionin the following way: An attribute in Hungarian is any nominal part ofa sentence which is in the nominative and which is a complement of anominal part.

In Section 3 I have listed those phenomena which have been describedby traditional Hungarian grammar as attributes. Now, if we comparethis list with our latter definition, we see that there are two such "attri-butes", which - according to our new definition - do not seem to be realattributes. I am referring to the construction afiunak a könyve (= theboy's book), where the attribute, the wordfiunak, is not in the nominativebut in the dative, and the construction egy lovat, feheret, where theattribute, or - if you like - the apposition, feheret, is in the accusative.Moreover, the apposition is generally endowed with the same suffixes asthe word it qualifies. How can we solve the problem of these phenomena?

According to the above definition and to my own personal conviction,apposition in Hungarian has nothing in common with the attribute. Thefact that the apposition has the same suffixes as the word it is alleged toqualify shows that it has the same relation to the verb as has the word itqualifies. For example, in the construction egy lovat, feheret, the partfeheret is an object to the same extent as the part egy lovat. There isaccordingly a coordination between the parts egy lovat and feheret.Naturally, there is a very remarkable relation between the constructionsegy feher lovat (= a white horse; horse in the accusative) and egy lovat,fehoret (= a horse, a white; both horse and white in the accusative).This relation can be explained only in terms of transformational grammar,in that the construction egy lovat, feheret is a transform of the construc-tion egy feher lovat.

However, how can we explain the status of the construction afiunaka könyve? In constituent grammar there is only one possibility. Thefundamental definition of the attribute to the effect that it is a nominal

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 10: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

60 LÄSZLO ANTAL

part of the sentence, being in the nominative and being a subordinate of anominal part, has to be modified to accord with a new rule, according towhich the construction afiunak a könyve contains an exceptional kind ofattribute.

Hence, in constituent grammar the definition of the attribute takes thefollowing form: An attribute is the nominal part of a Hungarian sentencewhich is in the nominative or dative and which is always a subordinatemember of a nominal part. However, we know that apposition can occurin every case and number, even in the nominative and dative. Neverthe-less, no confusion can occur, since apposition, even if it is in the nomina-tive or dative, is not a subordinate member of a nominal part. It isadvisable, however, to disregard the construction afiunak a könyve as akind of attribute, and it is simpler to treat it as a member of a particularcategory.

6. THE ATTRIBUTE IN STRING ANALYSIS. I use the term "string analysis"as it was defined by Harris.13 "String analysis," writes Harris, "character-izes the sentence of a language as follows: Each sentence consists of oneelementary sentence (its center), plus zero or more elementary adjuncts,i.e. word-sequences of a particular structure which are not themselvessentences and which are adjoined immediately to the right or to the leftof an elementary sentence or adjunct, or of a stated segment of an ele-mentary sentence or adjunct, or of any one of these with adjuncts adjoinedto it— Each word is assigned (on the basis of its affixes or its positionin elementary sentences and adjuncts) to one or more word-categories(rarely, word-sequence categories) 14 We now consider each sentenceS as a sequence of morphological word-categories (or sub-categories ordisjunctions of categories or, rarely, sequences of categories) st. Whenwe are given an arbitrary sentence S, we isolate the elementary partS0 by asking what contiguous sequences of the Si can be excised, onesequence at a time, by operations of general and nearly general applica-bility, the residue of S after each excision being still a sentence of thelanguage."15

In Hungarian - on the basis of the string analysis - we can define theattribute as follows:

13 Harris, Z. S., String Analysis of Sentence Structure (=Papers on Formal Linguistics,1) (The Hague, 1962).14 Op. cit., p. 9.15 Ibid., p. 22.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 11: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 61

The attribute is always a left-hand adjunct of the category N. Its mostimportant sub-types are the following:Adjective (^4) e.g. a feher Ιό (= the white horse) = ANParticiple (Vo) e.g. a dolgozo no (= the working woman) = VONSubstantive (TV) e.g. a szabo Bencsik (== 'the tailor Bencsik')',

egy pohar bor (= a glass of wine); afiu k nyve (= theboy's book). This last type of the TV adjunct differs fromthe preceding ones in that it can appear to the left of anN only if the N to which it is adjoined contains apossessive suffix. Every construction in which theattribute is a substantive is roughly equal to NN.

Pronoun (P) e.g. az en hazam ( — my fatherland) = PN.Number (L) e.g. a harom kislany (= three girls) = LN.

We have so far obtained the following constructions, in which an attri-bute occurs as the left-hand adjunct of an N: AN, VON, ΝΝ,ΡΝ, and LN.Naturally, the first W in the construction NN can have a further left-handadjunct A or N9 and thus the constructions ANN or NNN are alsopossible.

What place is to be assigned to apposition? Let us consider the follow-ing example: Bencsik, a szabo, beteg (= Bencsik, the tailor, is ill). If wetake out the word Bencsik, we obtain a normal sentence: A szabo beteg(=The tailor is ill). However, we also obtain a normal sentence if weleave out the part a szabo (Bencsik beteg; Bencsik is ill). This factclearly shows that neither is the part Bencsik an adjunct of the part aszabo nor the part a szabo an adjunct of the part Bencsik. Hence, bothBencsik and a szabo are equal components of the centre string.

As we have akeady noted, there are three forms of the constructionNN. In string grammar, two of these - the form a szabo Bencsik and theform egy pohar bor1* - cannot be distinguished from each other, since inboth constructions the adjunct N lacks any formal index. The third formof this construction - that represented by afiu k nyve since in this case,the second TV always has a possessive suffix - can, however, be isolatedin string grammar. In transformational grammar, on the other hand, it16 The egy pohar bor form has been isolated by traditional grammars on a semanticbasis; the authors have stressed that in this construction the attribute is always a wordwhich denoted quantity. In constituent and string analysis the peculiarity of this formcan only be demonstrated by means of a very complicated procedure. In the construc-tion quoted both parts can receive further attributes and therefore the distributionbetween attribute and head is different from the distribution between the correspondingparts of other attributive constructions.

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 12: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

62 LÄSZLO ANTAL

is possible to separate clearly the a szabo Bencsik form from the egypohar bor form.

7. THE ATTRIBUTE IN TRANSFORMATIONAL GRAMMAR. Before we begin anexamination of the Hungarian attribute in the light of transformationalgrammar, we must briefly refer to some peculiar features of Hungariansyntax and morphology.

A well-known feature of present-day Hungarian is the "nominal"sentence, i.e. a sentence where the predicate has no verb, for exampleHugo beteg (= Hugo is ill). We should note, however, that the omissionof a verb in the predicate is permitted only in the present tense and onlyin the third person, cf. Hugo beteg volt (= Hugo was ill) and the followingsentences: £n beteg vagyok (= I am ill); Te beteg vagy (= You are ill)and 0 beteg (= He is ill). There is a special case where the correspondingform of the verb lenni (to be) can occur but here, too, it can be omitted.The case in point is a question and the answer to this question: Hoi akulcs or Hoi van a kulcs! ( Where is the key!); Itt a kulcs or lit van a kulcs(Here is the key). Hence we have to distinguish between the followingcases:

a) there is only one type of sentence, with a particular subject and time,in which the verb lenni cannot occur;

b) there are only two types of sentence in which the use of the verblenni is optional;

c) in every other case the use of the appropriate form of the verb lenniis compulsory.

Therefore, a sentence such as Hugo beteg will be evaluated by us not asN A but as N A K, where the K is a K0, just as in a sentence where the verbis intransitive we have to allow for a zero-object.

There are two further suffixes, with which we have to deal: - and-nyi. According to Lotz, the suffix, -e is a derivative suffix; the author ofthe present paper considers it to be an element with the same morpho-logical status as the possessive suffixes, although it does not belong to thepossessive paradigm. The following examples can illustrate its meaningand role: Ez a haz Hugoe (='This house is that of Hugo' or rather Thishouse belongs to Hugo). The case suffixes can occur after the -e, and itcan be combined with the possessive suffixes. However, this latter factis of no importance to us.

-nyi is a derivative suffix; with it we can derive adjectives from sub-stantives, giving them the meaning of "as much as" or "as great as".For instance, tengernyi, derived from the substantive tenger (=sea)9

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 13: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY 63

means "as much as a sea" or "as great as a sea".Now we have to complete our list of symbols. In addition to N, A, V,

etc., we will use the following symbols:ppe = personal pronounppo == possessive pronouncs = case suffixps — possessive suffixThe elements -e, -nyi and -nak (the suffix of the dative) occur in the

formulae in their ordinary form.We can now define the Hungarian attribute in transformational gram-

mar in the following way: An attribute is a part of a construction whichcan be transformed into the predicative part of a corresponding sentenceby means of one rule, as follows:1) a feher Ιό-*Α Ιό feher (= the white horse - The horse is

white)2) a dolgozo no-+A no dolgozik (— the working woman - The

woman works)3) a szabo Bencsik-+A Bencsik szabo (—''the tailor Bencsik' - Bencsik

is a tailor)4) egy pohar bor^A bor poharnyi ( = a glass of wine - 'The wine is

as much as a glass')5) afiu k nyve-*A k nyv afiue (= the boy's book - The book

belongs to the boy or The book ispossessed by the boy)

6) afiunak a k nyve-*A k nyv afiue (= the boy's book - The bookbelongs to the boy)

7) az έη hazam-+A haza az enyem (= my fatherland - The father-land is mine)

8) a harom kislany-+Kislany van harom (= three girls - There are threegirls)

We can represent the way in which each of the above sets of constructionsare related by means of the following formulae:

1) AN-* N AV2) VON -> N V3) ΝλΝ2 -> N2 NiV5) NiNtfs -» N2 Ν! έΥ6) N^nakN^ps -> N2 N^V7)PpeNps -+ NPpoV$)LN -* NLV

The construction egy lovat, fehoret cannot be transformed by means of

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM

Page 14: FOUR APPROACHES TO THE SAME GRAMMATICAL CATEGORY

64 LÄSZLO ANTAL

one rule into the predicative part of a corresponding sentence, and there-fore it also has to be isolated from the attribute in transformationalgrammar. The reason for this is that this construction is derived -according to the rule ANcs-+NcsAcs - from a construction which con-tains an attribute, so that we cannot move from it straight to a predi-cative part; we have to apply not one but two rules before we can dothis, namely the rules NcsAcs-+ANcs-*NA V.

Traditional grammar related apposition - on the vague basis of thesense - to the attribute.

As we have seen, apposition is isolated from the attribute in consti-tuent grammar, because the behaviour of apposition is the behaviournot of a complementary part but of the main part of the construction.

In string analysis the attribute is always a left-hand adjunct adjoinedto the nominal part of the centre string. Therefore, if we leave out theattribute, we should always obtain a normal sentence as the residue.However, if we leave out the part of the sentence to which the attributeis adjoined, then the residue cannot be a normal sentence. If, on theother hand, in string analysis we leave out of the sentence the othernominal part, and the apposition remains alone in the position of thesubject, the sentence will be as normal as it was before the omission.Hence we have to conclude that apposition cannot be an adjunct i.e. anattribute, at least not in the light of string analysis and not in the case ofpresent-day Hungarian.

Transformational grammar goes further; it shows that apposition isnot immediately identical to the attribute but is the transform of aconstruction which contains an attribute.

As we have already seen, there are four types of N adjunct which areadjoined as attributes to the left of an N. In two case (a fiu kony\e anda fiunak a könyve) we can define these constructions and isolate themfrom the others at a lower level of grammatical description, because theyhave special formal (morphological) indices. However, there are twosub-types (a szabo Bencsik and egy pohar bor) the difference betweenwhich can be demonstrated in the most simple way in transformationalgrammar, since these constructions have a different transformationalhistory; in the case of the first construction the transformation is NiN2-+

2 , but in the case of the second it is N^N^-^N^

Eötvös University, Budapest

Brought to you by | University of Massachusetts - Amherst W.E.B. Du Bois LibraryAuthenticated | 128.119.168.112Download Date | 8/19/12 6:18 PM


Recommended