Top Banner
Pergamon LanguageSciences, Vol. 18, Nos 1-2, pp. 139-152, 1996 Copyright© 1996 ElsevierScienceLtd Printed in Great Britain. All rights reserved 0388-0001/96 $15.00 + 0.00 S0388-0001(96)00012-5 Hungarian syntactic structure and metrical prosody IAszl6 Hunyadi Department of General and Applied Linguistics Lajos Kossuth University H-4010 Debrecen, Hungary e-mail: hunyadi @llab2.arts.klte.hu ABSTRACT The present paper is an attempt to capture the prosodic features of Hungarian syntax as related to word order and logical interpretation. It is assumed that a) the grammaticality of the various linear orders of words in a sentence is determined by proper stress/accentuation and b) the interpretation of scope also follows the prosodic structure of the sentence. This model of the Hungarian sentence structure is presented as an extension of the framework of metrical phonology. The proposed metrical syntax is based on stress reduction rules similar to those of metrical phonology but it also uses semantic-logical information from the lexicon to account for word order variation and peculiarities of scope assignment. KEYWORDS Syntax, semantics, scope, prosody, stress, metrical phonology WORD ORDER IN HUNGARIAN Languages, characterized by an elaborate inflectional morphology have the general syntactic property of being flexible in the relative ordering of grammatical functions. This flexibility means that the grammatical functions, usually expressed by word order in other languages, are expressed by the inflectional system, and, in turn, word order can be used for the expression of other functions (cf. 1~. Kiss, 1978, for a discussion of (non)configurationality and its relation to Hungarian syntax see 1~. Kiss, 1987). Hungarian, one of such languages, has the property that the order of the arguments with relation to the verb is syntactically free, whereas the NP has a set inner structure. The examples below show possible linear order variants: (1) a. J~inos kOszOntOtte Katit a koncerten. John (-NOM) greeted Kate-ACC the concert-on 'John greeted Kate at the concert' b. Katit ktisztintOtte J~inos a koncerten. c. Katit J~inos k6sztint/Stte a koncerten. d. J~inos Katit ktisz/SntOtte a koncerten. e. K6sz~nttitte Jfinos Katit a koncerten. f. K6sztinttitte Katit J~nos a koncerten. 139
14

Hungarian syntactic structure and metrical prosody

Dec 15, 2022

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Hungarian syntactic structure and metrical prosody

Pergamon Language Sciences, Vol. 18, Nos 1-2, pp. 139-152, 1996

Copyright © 1996 Elsevier Science Ltd Printed in Great Britain. All rights reserved

0388-0001/96 $15.00 + 0.00

S0388-0001(96)00012-5

Hungarian syntactic structure and metrical prosody

IAszl6 Hunyadi Department of General and Applied Linguistics

Lajos Kossuth University H-4010 Debrecen, Hungary

e-mail: hunyadi @llab2.arts.klte.hu

ABSTRACT

The present paper is an attempt to capture the prosodic features of Hungarian syntax as related to word order and logical interpretation. It is assumed that a) the grammaticality of the various linear orders of words in a sentence is determined by proper stress/accentuation and b) the interpretation of scope also follows the prosodic structure of the sentence. This model of the Hungarian sentence structure is presented as an extension of the framework of metrical phonology. The proposed metrical syntax is based on stress reduction rules similar to those of metrical phonology but it also uses semantic-logical information from the lexicon to account for word order variation and peculiarities of scope assignment.

KEYWORDS

Syntax, semantics, scope, prosody, stress, metrical phonology

WORD ORDER IN HUNGARIAN

Languages, characterized by an elaborate inflectional morphology have the general syntactic property of being flexible in the relative ordering of grammatical functions. This flexibility means that the grammatical functions, usually expressed by word order in other languages, are expressed by the inflectional system, and, in turn, word order can be used for the expression of other functions (cf. 1~. Kiss, 1978, for a discussion of (non)configurationality and its relation to Hungarian syntax see 1~. Kiss, 1987).

Hungarian, one of such languages, has the property that the order of the arguments with relation to the verb is syntactically free, whereas the NP has a set inner structure. The examples below show possible linear order variants:

(1) a. J~inos kOszOntOtte Katit a koncerten. John (-NOM) greeted Kate-ACC the concert-on 'John greeted Kate at the concert'

b. Katit ktisztintOtte J~inos a koncerten.

c. Katit J~inos k6sztint/Stte a koncerten. d. J~inos Katit ktisz/SntOtte a koncerten. e. K6sz~nttitte Jfinos Katit a koncerten. f. K6sztinttitte Katit J~nos a koncerten.

139

Page 2: Hungarian syntactic structure and metrical prosody

140 Lfi~SZI,6 HUNYADI

g. A koncerten k6sz6nt6tte J~inos Katit. h. A koncerten k6sz6ntiStte Katit J~inos.

h. A koncerten J~inos Katit k6sz6nt6tte. i. A koncerten Katit J~inos k6sz6nt6tte.

j. A koncerten J~inos Katit k6sz6nt6tte. k. A koncerten Katit J~inos k6sz6nt6tte.

The above sentences express the same grammatical functions, regardless of the relative position of the arguments (Jdnos = grammatical subject and agent, Katit = grammatical object and theme).

Whereas linear order in the above sentences is not syntactically constrained, they are only considered grammatical (accepted by native speakers) if the given linear order is accompanied by a proper prosody. From the fact that not all prosodic variants are grammatical, it follows, that, along with an elaborate inflectional morphology, Hungarian also has an elaborate prosodic system (a system of intonational and accentual patterns) and that in Hungarian there is a strong relation between syntactic structure and sentence prosody.

In addition to their prosody-dependent grammaticality (as seen from variants with *) , even the grammatical sentences in (1) a.-k. are not synonymous: they represent various sentence perspectives; e.g.

(1) c. Katit "J~inos k6sz6nt6tte a koncerten. 'As for Kate, it was John who greeted her at the concert.'

c'. * "Katit J~inos k6sz6nt6tte a koncerten.

e. "K6sz6nt6tte Jfinos Katit a koncerten. 'John did greet Kate at the concert.'

e'. * K6sztint6tte "J~inos Katit a koncerten.

g. A koncerten "k6szOnt6tte J~inos Katit. 'As for the concert, John did greet Kate.'

g'. "A koncerten k6sz6nt6tte Jfinos Katit. 'It was at the concert that John greeted Kate.'

g". * A koncerten k6sz6nt6tte "J~inos Katit.

Along with its communicative functions, Hungarian also has an elaborate system of logical scope assignment, based again on word order and prosody; cf. (2):

(2) a. 'J~inos 'k6szi3nt6tt 'mindenkit. John (-NOM) greeted everyone-ACC 'John greeted everyone'

a'. "J~inos k6sz6ntOtt mindenkit. 'It was John who greeted everyone (others may have greeted fewer people).'

a". "J~inos ktJsz6nt6tt "mindenkit. 'It was John who greeted everyone (everyone was greeted and each of them by John).'

b. "Mindenkit k6sz6nt6tt J~inos. 'John greeted everyone (not just some people)'

c. J~inos "mindenkit k6sz6ntOtt. 'As for John, he greeted everyone (not just some people)

d. "Mindenkit J~inos k6sz6nt6tt. = (2a") e. Mindenkit "J~inos k6sz6nt6tt.

'It was John who greeted everyone (others may have greeted fewer people)' The prosodically possible and communicatively different variants in (2) a.-e. also have different scope interpretations: in (2a) the widest scope is expressed by the quantifier mindenkit, in (2a') by the focused John, in (2a") by the quantifier, in (2b) by the focused John, in (2c) and (2d) by the quantifier, and in (2e) by the focused John again.

Page 3: Hungarian syntactic structure and metrical prosody

HUNGARIAN SYNTACTIC STRUCTURE AND METRICAL PROSODY 141

With the above elaborate morphological, prosodic and semantic systems, we can summarize that the fundamental role of Hungarian syntax is to express a) communicative, rather than grammatical functions and b) logical (scope) relations. In both of these functions prosody plays an important role.

The aim of the present paper is to offer a systematic account of the syntactic and semantic (logical) functions of Hungarian sentence stress by presenting the outlines of a model of prosodic syntax.

SYNTACTIC STRUCTURE, STRESS AND SCOPE

In earlier work based on empirical observations (Hunyadi, 1981a, 1981b) it was found that sentential stress has important logical functions in Hungarian. According to it, the relative scope of operators is expressed by heavy stress in a way that

(3) Wide-scope operators have heavy stress over their scope.

Following this observation, if a verb or any other constituent is negated, the negative nem 'not' is stressed and the negated constituent is unstressed. If negation has narrow scope, the negative nem will be unstressed, too; cf. (4):

(4) a. 'Nem Ifittam Katit. Not I-saw Kate-ACC 'I did not see Kate' (wide scope of neg)

b. 'Nem Katit lfittam. 'It was not Kate whom I saw.' (wide scope of neg overJocus )

c. 'Katit nero lfittam. 'It was Kate whom I did not see.' (wide scope of focus over neg)

Whereas in most cases left-to-right linear order of wide-scope (stressed) operators also corresponds to their relative scope (as in (6a-c)), apparently the stress-scope rule in (3) is more general. Namely, there are two cases where linear order does not correspond to relative scope, cf. (5a)=(5b), (6a)=(6b)*(6c), yet scope rule (3) applies:

(5) a. 'Nem hittam mindent. not I-saw everything-ACC 'I did not see everything.'

b. Mindent 'nem lfittam.

(6) a. 'Mindent Jfinos lfitott. everything-ACC John (-NOM) saw. 'for every x, J. saw x'

b. 'Jfinos lfitott 'mindent.

c. 'Jfinos lfitott mindent. 'It was J., who, for every x, saw x'

In (5b), the universal quantifier precedes the negative operator, yet it has narrow scope with respect to the latter; on the other hand, in (8b) the focused Jdnos precedes the universal quantifier, yet the universal quantifier has the wide scope. In one of the theoretical accounts of these partially controversial scope phenomena, within the frameworks of GB-theory 1~. Kiss (I~. Kiss, 1987, 1994) introduces the scope principle:

(7) the scope principle: an operator c-commands its scope

Page 4: Hungarian syntactic structure and metrical prosody

142 LASZL0 HUNYADI

According to this account, the scope principle is observed in Hungarian at S-structure: in multiple quantification, the quantifier on the left asymmetrically c-commands, and has scope over, the quantifier on the right.

The derivation of narrow scope of the type in (5b) is explained in the following way: the narrow scope universal quantifier is in Left Dislocation and is interpreted as if it were in V', in the position of the gap associated with it.

The derivation of wide scope of the type in (6b) takes place in two steps: first, the Quantifier Raising is applied to the universal quantifier in V' and the precedence principle applies, second, the wide scope universal quantifier undergoes Quantifier Postposing in the phonological component.

This model can in a straightforward way account for the proper generation of scope when scope assignment follows linear order (the precedence principle). Its account of the derivation of narrow scope assignment of the type (5b) also corresponds to the basic assumptions of GB-theory regarding movement, but the derivation of narrow scope of the type (6b) with both Quantifier Raising and Quantifier Postposing of the same quantifier is less intuitive. In addition, the model can only descriptively handle the systematic nature of the relation between syntactic structure and sentential stress.

In another model of the theoretical status of stress and logical scope, Vogel and Kenesei (Vogel and Kenesei, 1987, 1990) present a metrical phonological account. According to this approach logical scope is expressed in the relation of intonational phrases (IPs).

Vogel and Kenesei maintain that intonational phrases, derived from phonological phrases can be identified at the level of S-structure.

According to it, certain elements are marked as [+SCI for wide scope quantifiers and others as [+OSI (for 'operator status'). Phonological phrases are grouped into intonational phrases from left to right, starting from the element with the widest scope until another element with a logical function (either marked I+SC1 or I+OS I) or the end of the sentence is reached.

From their hypothesis it follows that if two sentences consist of the same IPs but with different respective order, their logical representation will also differ:

(8) a. [ip"minden 'nytiL]hp "Jdli~it szereti a 'legjobbanl every rabbit-(NOM) Julia-ACCloves the most

'For every rabbit, it is Julia that it likes best'

b. lip "Jtili~it szereti]hp "minden 'nytil a 'legjobbanl 'It is Julia that every rabbit likes best'

(Vogel and Kenesei, 1990, 360)

According to the stress rule in Hunyadi, 1981a, 1981b and the rule of quantifier-postposing in I~. Kiss, 1987, 1994, however, the above pair of sentences do not differ in logical scope. In both (8a) and (Sb) the universal quantifier has wide scope over focus, interpreted as (9):

(9) 'For every rabbit, it is Julia that it likes best'

In addition, in the Vogel and Kenesei approach, the grammaticality (or logical interpretation) of sentences with unstressed quantifiers (i.e. their relative narrow scope), as in (8) c. is not accounted for:

(8) c. "Jtili,'tt szereti minden nytll a legjobban

In yet another approach on the relation of syntax and prosody Kornai and K~ilm~in have presented a model of Hungarian intonational system in a model of autosegmental phonology, in which they also describe accent rules (Kornai and K,'tlm~in, 1988).

Page 5: Hungarian syntactic structure and metrical prosody

HUNGARIAN SYNTACTIC STRUCTURE AND METRICAL PROSODY 143

Kornai and K~ilm~in assign a diacri t icf (focus, or in their reading: Foykes) marking to certain syllables. In their account the formation of intonational patterns is the result of the Eradication rule, a rule which deletes the accent and word boundaries of the subsequent words up to the next diacri t icf marking. As a result of the Eradication rule segments from o n e f to the next or to the end of the phrase form one phonological word with one accented syllable.

Kornai and K~lm~in are not concerned with a difference between quantified and focused constructions, and so they do not make a distinction between a heavy stress on the quantifier and one on a non-quantified focused element as far as their mutual relation is concerned. Consequently, following their account, a construction, consisting of a universal quantifier and a subsequent element with a f (focus) marking, can only have equal stress on both of the elements and so they cannot generate the otherwise grammatical construction (6a) [Q"Mindent [FJdnos ldtott I1.

PROPOSAL: A CATEGORY BASED PHONOLOGICAL APPROACH

Before discussing the formal properties of a model aimed at capturing the principled relations between syntactic and prosodic structure, we shall specify what we mean by stress, the concept most often referred to in the description of prosody.

Indeed, since there is an obvious relation between syntactic structure and stress in Hungarian, the various--both theoretical and descriptive--approaches heavily rely on this term. In particular, we cannot talk about focus without mentioning that it carries main stress (according to 1~. Kiss ibid. the focused element will always have stress even if preceded by a stressed universal quantifier), or topic without discussing its stress (and intonational) properties with respect to those of the subsequent focus. Again, acoustically, focus in a sentence without a verbal prefix, as well as wide scope operators are identified by their stress. However, there is no agreement even among phoneticians whether stress is a property that can fully be identified acoustically or whether it is a more complex phenomenon involving both formal and semantic/pragmatic properties.

On the basis of experiments we have carried out (cf. Hunyadi 1995a) we shall consider stress as follows:

(10) A definition of stress

Generally speaking, stress is a relation of fundamental frequency and intensity through time expressed by an index of relative relation, according to which subsequent indices form a significant tendency on a stressed syllable as compared to those of the neighboring one(s). In these terms, a word has sentential stress if the relational indices of the relative relation of fundamental frequency and intensity form a significant tendency spreading over to the subsequent word(s).

Thus, since fundamental frequency and intensity are taken as inputs to a complex index of the relative relation of the two parameters, this index does not only represent "loudness", often considered to be a main perceptual constituent of accent, but also "tone", a main perceptual constituent of intonation. Consequently, talking about stress we shall mean a significant relational phenomenon, that is both represented in accent and intonation.

According to it, lexical items can be of two kinds: a) those whose stress tendency can spread over to the subsequent word(s) and b) whose stress tendency cannot. Whereas the majority of lexical items belong to the first kind, existential quantifiers belong to the second one; cf.

(11) a. "Jtinos lhtta a ki~illft~st. John saw the exhibition. 'It was John who saw the exhibition.'

b. "Mindenki l~itta a ki~illft~st. everyone saw the exhibition 'Everyone saw the exhibition.'

Page 6: Hungarian syntactic structure and metrical prosody

144 L,h, SZLO HUNYADI

c. * "Valaki l~itta a kifiilRfist. someone saw the exhibition

d. Valaki "lfitta a kifillft,4.st. 'Someone saw the exhibition.'

Considering the prosodic structure of Hungarian sentences and their logical function from a descriptive point of view (Hunyadi 1995b) we can maintain the following:

(12) a. A sentence (utterance) consists of distinct phonological phrases (PhPs) each of which is based on one intonational contour with one obligatory main stress;

b. Main stress indicates the function of an operator having wide scope over the rest of the PhP; c. The relative scope of the corresponding PhPs of a sentence is determined by the category of the

stressed operators.

Furthermore, we shall make the following assumptions:

(13) a. There is a systematic relation between neutral and focused sentences, as well as sentences with multiple foci practically undergoing the same derivation;

b. The grammaticality of a sentence requires both proper syntactic and prosodic derivation; c. In a language like Hungarian derivation essentially means the derivation of prosodically acceptable

configurations and thus these rules should essentially be phono-syntactic; d. There should be a direct relation between prosodic derivations and their logical (scope) interpretation.

THE ORGANIZATION OF THE GRAMMAR

We are going to apply to Hungarian syntax a framework of metrical phonology developed in Liberman and Prince, 1977 and in subsequent work. According to this theory stress is not considered to be a phonological feature, it is rather captured in a rhythmic structure. This structure is hierarchically organized and the various patterns are a result of stress reduction rules. Although significant subsequent work has been done on the syllable (word) level (e.g., Selkirk, 1980, ) and there are also investigations into the phrasal level as well (cf. Selkirk, 1984, Halle and Vergnaud, 1987, or most recently Hayes, 1995), these studies have not been directed to the description of the relation of the syntactic, phonological (prosodic) and semantic (logical) properties of the language. Since, in Hungarian, there is an apparent relation between these aspects of the language, and since stress appears to have an important share in all of them, a metrical model of syntax may capture their relation in one single approach. In what follows, we are going to describe the main properties of such a metrical model of Hungarian syntax.

The grammar operates upon the output of a propositional component which has already taken care of the propositional composition of the sentence, including the formation of the (non-configurational) NPs. The order of the arguments is left undefined. The task of the grammar is to identify all and only the grammatical permutations of the arguments in the given sentence. It is assumed that the grammaticality of a permutation depends on whether a phonological structure can be assigned to it. Only those with a proper phonological structure can be grammatical. Thus the derivation is essentially based on phonological rules.

The main characteristics of this phonological derivation are as follows:

(14) a) In the underlying structure every word has its own stress and represents a single phonological word. Rules of generation are reduction rules describing how the stress of a given word is reduced.

b) In the course of stress reduction the phonological words are combined into phonological phrases and on higher levels reduction takes place between phonological phrases and phonological words or further phonological phrases.

c) Stress reduction only takes place between neighboring phonological components so that, in case of two neighboring phonological components C1 and C2 (C1 preceding C2), C2 reduces its stress.

d) On the first level of derivation two neighboring phonological words are combined into one phonological phrase.

e) The combination of C1 and C2 into one phonological phrase (the reduction of the stress of C1) depends on the hierarchy of the categories C1 and C2.

Page 7: Hungarian syntactic structure and metrical prosody

HUNGARIAN SYNTACTIC STRUCTURE AND METRICAL PROSODY 145

The hierarchy of stress reduction is based on a) information from syntax and b) information from the lexicon. In the first case, complex NPs consisting of two or more phonological words have either equal stress on their components (on the underlying level) or main phrasal stress on the left side of the NP in which case an NP with one phrasal stress is considered as one phonological word.

In the latter case, that of simple NPs and NPs with one phrasal stress, however, stress reduction takes place following categorial information from the lexicon. According to it, the reduction of the stress of (simple) lexical/phonological words depends on their operator status. In what follows we shall mainly be concerned with this latter option.

Operators are organized into a hierarchy which fundamentally determines the order and direction of stress reduction; cf. (15):

(15) Hierarchy of categories for stress reduction (in increasing order of ranking): focus, quantifier and sentential operator (such as negation or modal operations; the latter often, but not exclusively expressed on the verb).

The condition of stress reduction is formalized in (16):

(16) Conditions of hierarchical stress reduction: a. On the first level of derivation, in a configuration of phonological components C1 and C2, C1

preceding C2, C2 reduces its stress if it is lower on the hierarchy than CI. b. On higher levels of derivation there is no hierarchical constraint of stress reduction. c. Stress reduction of phonological words is obligatory in cases of cliticization and in other cases it is

optional.

As the following data suggest, stress reduction does not directly follow from the Nuclear Stress Rule (Chomsky and Halle, 1968), instead, it follows the above hierarchical rule of stress reduction.

(17) x Level 4

(B~r) x x Level 3 (bdr ='although') *x x Level 2

x x x Level 1 L~itta J~inos a ki~illit~st. Saw John the exhibition 'John saw the exhibtion'

On the first (underlying) level of derivation, each word has its own stress and forms a single phonological word. On the second level, "Ldtta "Jdnos a kidllitdst i.e. the reduction of the stress of kidllltdst is ungrammatical, because the condition of hierarchical reduction is violated (both Jdnos and kidllltdst, being potential foci, are on the same level of hierarchy). On the third level, however, the stress of Jdnos is reduced correctly, since the preceding constituent is a verb, carrier of a sentential operator and as such is higher on the hierarchy than the potential focus Jdnos. On the uppermost level the stress of kidllltdst is properly reduced. (NB.: we consider each level of derivation an independent horizontal structure, i.e. the vertical amount of stress marks in the columns does not play a role, in contrast to other models, such as Liberman and Prince, ibid., Selkirk, 1984 or Hayes, ibid.)

(18) x

(B~ir) x x X X

X X X

LAtta mindenki a ki~illi~st Saw everyone the exhibition 'Everyone saw the exhibition'

Page 8: Hungarian syntactic structure and metrical prosody

146 I,ASZL0 HUNYADI

(17) and (18) differ on the second level of derivation. Whereas in (16) we had an ungrammatical derivation due to the violation of the hierarchy of stress reduction, in (18) this constraint is not violated, since the quantifier mindenki is higher on the hierarchy than the subsequent kidllftdst.

(19) X

(B~ir) x x * X X

XXX

LAtta a kitilli~st mindenki Saw the exhibition everyone 'Everyone saw the exhibition'

Again, in (19), the reason for the ungrammaticality of the derivation on the second level is the violation of the hierarchy of stress reduction: contrary to the actual derivation, the stress of the quantifier cannot be reduced to the (preceding) potential focus, kidllftdst.

(20) * X

* X X

X X

X X X

J~inos elment a kitilli~sra John went the exhibition-to 'John webt to the exhibition'

(20) accounts in this model for the known fact that a noun with main stress cannot be followed by a verbal prefix (or verbal modifier): here, too, the hierarchy of stress reduction is violated, both John and the prefix el being on the same level.

In contrast, (21) demonstrates that if the hierarchy constraint is not violated, the derivation is grammatical (the universal quantifier is higher on the hierarchy than the verbal prefix):

(21) X

(B~r) x x (?) x x

X X X

Mindenki elment a kifdlftfisra Everyone went the exhibition-to 'Everyone went to the exhibition'

The fact, that stress reduction should always follow from a hierarchically lower element, consequently equal levels are not allowed, is demonstrated in (21):

(22) * X

X X

X X X

Mindenki mindent megn6zett Everyone saw everything-ACC 'Everyone saw everything'

It is interesting to observe, that the conjunction is 'too', often considered to be a form of universal quantification, does not behave in the way an overt universal quantifier, such as mindenki 'everyone' does:

Page 9: Hungarian syntactic structure and metrical prosody

tlUNGARIAN SYNfACTIC STRUCTURE AND MFTRICAL PROSODY 147

(23) X

X X

X X X

P6ter is mindent megn6zett Peter too everything-acc saw 'Peter, too, saw everything'

Since (22) and (23) differ on the third level of derivation, this difference should be accounted for by the difference between is and mindenk i . From this we conclude that is is to be listed among the sententiai operators in the hierarchy allowing for the reduction onto them of the stress of a subsequent universal quantifier. This behavior can logically be accounted for by the fact that with the conjunction is two propositions are conjoined thus representing a sentential operation.

In the case of negative sentences the negative nero 'no' or ne 'do not [imp.]' is stressed and this stress pattern follows the hierarchy principle, cf. (24) and (25):

(24) "Nem P6ter ment el a ki~ilHt,'isra 'Its was not Peter who went to the exhibition.'

(25) "Nem mindenki ment el a ki~illf~sra 'Not everybody went to the exhibition.'

In both cases, the neg operator has a higher hierarchical status than either the focus or the universal quantifier. In (26), however, this is not the case:

(26) "P6ter nem ment el a ki~illlt~isra 'It was Peter who did not go to the exhibition.'

(26) is, nevertheless, a grammatical derivation, since it is not n e m that reduces its stress onto P ( t e r , but the negative phrase formed with nem, and--following the rules of hierarchical stress reduction--hierarchy only constrains reduction on the underlying level of derivation.

The fact that (27) is ungrammatical, however,

(27) * "Mindenki nem ment el a ki~llit~isra everyone not went CONV the exhibition-to

follows a possibly universal constraint: most universal quantifiers of the type minden- must have narrow scope in relation to negation. (For an analysis of sentences comparing the quantifier minden- and similar constructions with the conjunction m i n d cf. Hunyadi, 1981b.)

The case that, with the exception of cliticization, stress reduction is not obligatory, can be seen in (28) where, in contrast to the more general case of (29) cliticization of is + nero does not take place.

(28) * X

X X

?x x x X X X X

X X X X X

P6ter is nem ment el a ki~illit~sra Peter too not went CONV the exhibition-to 'Peter, too, he did not go to the exhibition'

Page 10: Hungarian syntactic structure and metrical prosody

148 LASZLd) HUNYADI

(29)

X

X X

X X X

? x x x X X X X X

P~ter sem ment el a ki~illft,'tsra. Peter neither went CONV the exhibition-to 'Peter did not go to the exhibition either'

The grammaticality of (28) also underlines our previous analysis of the conjunction is attributing to it sentential operator status.

As it was pointed out earlier, the stress tendency of existential quantifiers does not spread over to the subsequent word(s). In this way, they do not take part in the process of stress reduction as triggers of stress reduction. According to it, (30) cannot have a third level derivation:

(30)

* X

X X

X X X

Valaki elment a kifillftfisra. Someone CONV-went the exhibition-to 'Someone went to the exhibition'

But this restriction only refers to constraints of the quantifier as trigger of stress reduction, and its stress can be reduced, cf.:

(31)

X

X X

X X X

Mindenki hitott valakit. Everyone saw someone 'Everyone saw someone'

Stress reduction is again made possible by the fact that the hierarchical constraint only affect first level derivation.

Although the direction of stress reduction is generally right to left, there is one case in which this is reversed: in the case of topicalization:

(32)

X X

X X X

X X X X

Tegnap 'tftz tittitt ki Yesterday fire broke out 'A fire broke out yesterday'

Page 11: Hungarian syntactic structure and metrical prosody

HUNGARIAN SYNTACTIC STRUCTURE AND METRICAL PROSODY 149

(33) X

X X

X X X

S X X X

X X X X

Tegnap 'tfiz tit6tt ki ( ,ma 'vihar volt.) Yesterday fire broke out (today storm was) 'Yesterday, a fire broke out (, today it was stormy)

The last level of derivation in (33) expresses contrastive topic. However, it should be observed that this (exceptional) stress reduction can only take place at this last level of derivation. An earlier derivation (as on the second level) is ungrammatical.

In this model of metrical syntax several complex phenomena related to word order, scope assignment and stress receive a straightforward account.

Sentences with the so-called counter focus as well as multiple foci can be generated in a unified fashion. Since this model is based on the rule of stress reduction, it is not necessary to account for multiple (heavy) stresses: they simply follow from the fact that those syllables (and phonological words) have not undergone stress reduction. On the contrary, what is to be accounted for is the lack of stress on certain syllables(and phonological words). This is what the categorial hierarchical stress reduction rule is to do.

Here is a sentence with double focus:

(34) X X

X X X

J~inos l~itta Katit. John saw Kate-acc. '(app.) It was John who saw Kate and not that it was Kate who saw John'

On the second level of derivation the stress of the verb Idtta is reduced onto the focus Jdnos.

It is also straightforward to derive (6b) "Jdnos ldtott "mindent by stress reduction:

(6b) X X

X X X

Jhnos l~itott mindent. John saw everything-acc. 'John saw everything'

As we can see, the same rule applies to the generation of (6b) as to that of (33).

We can also account for the several stresses in a (non-) neutral sentence, such as (35):

(35) X X

X X X

X X X X X

J~inos Katinak telefon~ilt minden h6ten. John Kate-DAT. called every week 'For John, every week (and not just, say, every second week) it was Kate who he called.'

LSCI8/I-2--F

Page 12: Hungarian syntactic structure and metrical prosody

150

(36)

I~ASZL0 HUN YADI

X X X

X X X x

X X X x x

J~inos Katinak telefoMlt mindenh6ten. John Kate-DAT. called every week 'For John, every week (so often[) it was Kate who he called.'

Similarly, the fact that in the case of two (or more) neighboring universal quantifiers no stress reduction can take place can easily be accounted for by the hierarchical constraint: it is simply not possible to reduce stress if CI and C2 are on the same hierarchical level; cf. (37)-(38):

(37) * X

X X

X X X

"Mindig "b~irkinek segitek. always anyone-to l-help '(appr.) I am always ready to help anyone'

(38) * X

X X

X X X

"B~irkinek "mindig segitek. always anyone-to l-help

THE INTERPRETATION OF SCOPE

The scope interpretation of phonologically grammatical sentences directly follows the given prosodic structure:

(39) a. The stress-bearing element in each phonological phrase has wide scope over the rest of the operators in the same phonological phrase;

b. The relative scope of phonological phrases is determined by the category based hierarchy of stress reduction; i.e. the higher on the hierarchy a stressed operator of a phonological phrase, the wider its scope over a phonological phrase with its main operator on a lower level in the hierarchy;

c. The relative scope of phonological phrases with equal ranking wide scope operators is ambiguous.

Thus, for illustration:

In (2a') "Jdnos kOsz6nt6tt mindenkit the focus has wide scope over the universal quantifier: since this sentence consists of one phonological phrase with Jdnos having main stress, Jdnos has wide scope over the universal quantifier.

In (2a") "Jdnos kO'sz6ntO'tt "mindenkit there are two phonological phrases ("Jdnos kO'sz6nt6tt and "mindenkit). It is the hierarchical position of the category of the wide scope operators of each phrase that determines their relative scope. Since Jdnos with its focus function is lower in the hierarchy than the universal quantifier mindenkit, the quantifier has wide scope over the focus.

The reliance of scope assignment on hierarchical stress reduction can also account for the logical equivalence of (2a") "Jdnos kb'szb'ntOtt "mindenkit and (2d) "Mindenkit Jdnos kb'szb'ntb'tt. Namely, (2d) consists of one phonological phrase, and it is the stressed universal quantifier that must have wide scope over the rest of the sentence, including Jdnos with its focus status.

Page 13: Hungarian syntactic structure and metrical prosody

HUNGARIAN SYNTACTIC STRUCTURE AND METRICAL PROSODY 151

At last, scope relations between a universal and an existential quantifier can also be accounted for in a similar fashion:

(40) X X

X X X

"Valamikor "mindenki hazament. Sometime everyone went-home 'Everyone went home at some time'

The sentence is ambiguous due to the fact the quantifiers belong to two different phonological phrases and they are on the same hierarchical level from the point of view of stress reduction (hence both should be stressed) and scope assignment.

(41) X

X X

X X X

"Mindenki hazament valamikor. 'Everyone went home at some time'

The second level derivation ("Mindenki hazament "valamikor) is ambiguous for the same reason as the previous (40). The third level derivation, however ("Mindenki hazament valamikor) is unambiguous: with its single main stress it forms one phonological phrase and, following the scope rule, the main stress bearing universal quantifier has wide scope over the unstressed existential quantifier.

CONCLUSIONS

a) According to a purely syntactic account (like the one in I~. Kiss's framework, cf. I~. Kiss, ibid.), Hungarian displays the characteristic features of two different syntactic types. It has a highly structured, configurational pre-verbai field with specific syntactic positions, on the one hand, and a flat post-verbal field with no specific argument positions, and on the other.

Whereas the given syntactic positions predict the prosodic properties of the constituents they contain, it is only the prosodic properties of the constituents in the post-verbal field that determine grammaticality.

There is a certain symmetry between the pre-verbal and the post-verbal field: logically equivalent sentences of the two types of constructions have equivalent prosodic properties (the constituents with wide scope also have main stress). This symmetry, based on the equivalence of prosodic (stress) properties offers the opportunity to describe the sentences in the two fields in one unified model.

b) In this model of metrical syntax structural relations are expressed purely prosodically in phonological phrases. Phonological phrases are derived by the hierarchical stress reduction rule. The head of the phonological phrase is determined by the category/operator status of the given constituent. The hierarchical stress rule also determines scope relations.

Thus, this metrical model of Hungarian syntax can capture important structural relations between the generation of the grammatical sentences, their phonological properties and their semantic (logical) interpretation.

REFERENCES

Chomsky, N. and M. Halle (1968). The sound pattern of English.. Harper and Row, New York Halle, M. and J-R. Vergnaud (1987). An Essay on Stress. MIT Press Hayes, B. (1995). Metrical Stress Theory. Univ. of Chicago Press Hunyadi, L. (1981 a). A nyelvi polaritds kifejez~se a magyarban. Debrecen.

Page 14: Hungarian syntactic structure and metrical prosody

152 I.~SZLO HUNYAI)I

Hunyadi, L. (1981b). Remarks on the syntax and semantics of topic and focus in Hungarian. Acta Linguistica Acad. Sc. Hung., 31, 107-136. Hunyadi, L. (1995a, to appear). Acoustic cues to sentential stress in Hungarian and their measurement. Studies in Applied Linguistics, Vol. 2. (L. Hunyadi, M. G6sy, G. Olaszy, eds.). KLTE, Debrecen Hunyadi, L. (1995b, ms.). A Fonetikai Forma Szemantikai Szerepe IThe semantic role of Phonetic Forml E. Kiss, K. (1978). A magyar mondat egy transzformtlci6s generatfv leff~isa. Nyelvtudomdnyi K6"zlemdnyek, 87. Akad~miai Kiad6, Budapest. I~. Kiss K. (1987). Configurationality in Hungarian. Akad6miai Kiad6, Budapest. 1~. Kiss, K. (1994). Sentence Structure and Word Order. In: Syntax and Semantics, Vol. 27 (F. Kiefer and K. I~. Kiss, eds.), 1-90. Kornai, A. and L. K~lm~in. (1988). Hungarian sentence intonation. In: Autosegmental studies on pitch accent (H. v. d. Hulst and N. Smith, eds.), Foris Pub., Dordrecht Liberman, M. and A. Prince. (1977). On stress and linguistics rhythm. Linguistic Inquiry, 8. 249-336. Selkirk, E. O. 1980: The role of prosodic categories of English word stress. Linguistic Inquiry 11. 563-605. Selkirk, E.O. 1984: Phonology and Syntax: The relation between sound and structure. MIT Press Vogel, I. and 1. Kenesei, 1987. The interface between phonology and other components of grammar: The case of Hungarian. Phonology, 4. 243-263. Vogel, 1. and 1. Kenesei, 1990. Syntax and semantics in phonology. The phonology-syntax connection (Sh.

Inkelas and D. Zec, ed.), 339-363. The Univ. of Chicago Press. Chicago.