constraints on intonational prominence of focalized constituents

20
GORKA ELORDIETA CONSTRAINTS ON INTONATIONAL PROMINENCE OF FOCALIZED CONSTITUENTS* Abstract. In this paper we will provide a description of the prosodic constraints that apply on focalization in a pitch-accent dialect of Basque. In this dialect only words that bear a lexical or derived pitch accent can be singled out prosodically. In contexts of narrow non-corrective focus, some speakers reveal a more restrictive constraint, which limits intonational highlighting to words which constitute Accentual Phrases (APs) by themselves. Words which do not fulfill these intonational properties are not singled out intonationally but are pronounced with the rest of the words in the AP they form part of, despite being narrow focus constituents from an informational point of view. To the best of our knowledge, these constraints are unattested in other languages. 1. INTRODUCTION Across languages, in narrow contrastive focus constructions one or more cues (morphological, syntactic, intonational) are used by speakers in order to express the intended meaning correctly, singling out the focalized element or constituent from the rest of the elements in the sentence. However, in this article I will provide evidence that in the pitch-accent dialects of Basque classified as Northern Bizkaian Basque (NBB, Hualde, Elordieta, Gaminde and Smiljanić 2002) narrow focus expressions may be left unexpressed through these cues. There are cases in which focalized words cannot be identified on the basis of syntax or intonation alone (morphology does not play a role as a focus cue in Basque). They may satisfy the necessary syntactic conditions, but they do not satisfy the necessary conditions imposed by the intonational grammar of these dialects. There is a constraint on intonational focalization limiting main intonational prominence to focalized words that bear a lexical or derived pitch accent, and more radically to words that constitute a separate intonational unit on their own, an Accentual Phrase (AP). A word forms an independent AP if it has a H*+L pitch accent and the word to its left ends an AP. 2. BACKGROUND It is well known that languages differ in the overt cues they use to make the hearer identify clearly the focalized constituent. On the one hand, there are languages which signal focalized elements intonationally, without overt syntactic or morphological cues. These are languages of the so-called English type, in which focalized elements receive main prosodic prominence in-situ, with no movement from their base position. 1

Upload: ehu

Post on 13-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

GORKA ELORDIETA

CONSTRAINTS ON INTONATIONAL PROMINENCE OF FOCALIZED CONSTITUENTS*

Abstract. In this paper we will provide a description of the prosodic constraints that apply on focalization in a pitch-accent dialect of Basque. In this dialect only words that bear a lexical or derived pitch accent can be singled out prosodically. In contexts of narrow non-corrective focus, some speakers reveal a more restrictive constraint, which limits intonational highlighting to words which constitute Accentual Phrases (APs) by themselves. Words which do not fulfill these intonational properties are not singled out intonationally but are pronounced with the rest of the words in the AP they form part of, despite being narrow focus constituents from an informational point of view. To the best of our knowledge, these constraints are unattested in other languages.

1. INTRODUCTION

Across languages, in narrow contrastive focus constructions one or more cues (morphological, syntactic, intonational) are used by speakers in order to express the intended meaning correctly, singling out the focalized element or constituent from the rest of the elements in the sentence. However, in this article I will provide evidence that in the pitch-accent dialects of Basque classified as Northern Bizkaian Basque (NBB, Hualde, Elordieta, Gaminde and Smiljanić 2002) narrow focus expressions may be left unexpressed through these cues. There are cases in which focalized words cannot be identified on the basis of syntax or intonation alone (morphology does not play a role as a focus cue in Basque). They may satisfy the necessary syntactic conditions, but they do not satisfy the necessary conditions imposed by the intonational grammar of these dialects. There is a constraint on intonational focalization limiting main intonational prominence to focalized words that bear a lexical or derived pitch accent, and more radically to words that constitute a separate intonational unit on their own, an Accentual Phrase (AP). A word forms an independent AP if it has a H*+L pitch accent and the word to its left ends an AP.

2. BACKGROUND

It is well known that languages differ in the overt cues they use to make the hearer identify clearly the focalized constituent. On the one hand, there are languages which signal focalized elements intonationally, without overt syntactic or morphological cues. These are languages of the so-called English type, in which focalized elements receive main prosodic prominence in-situ, with no movement from their base position.1

Other Germanic languages such as Dutch and German also have this strategy of English for signaling narrow focus. However, in some cases these languages may also resort to syntactic movement operations to cue focus. When the verb is the focus of the sentence and a definite object is used, scrambling of the object may take place so that the verb is interpreted as narrow focus (Reinhart and Neeleman 1998). The verb receives main prosodic prominence by virtue of being in clause-final position.2

Some languages display two kinds of strategies for narrow focus manifestation: one in which words are assigned main prominence in their base-generated syntactic position (a strategy of the English-type), and another one in which syntactic displacement operations are produced such that these words or constituents end up occupying a syntactically specified position for narrow focus, by means of scrambling or fronting, or some other means. Unlike Dutch or German, the latter option is available for all constituents and is not subject to definiteness constraints, and perhaps most importantly, in these languages focalized words which are syntactically displaced are also assigned main prosodic prominence in the sentence (cf. Bolinger 1954, Ladd 1980, Culicover and Rochemont 1983, Vallduví 1990, Cinque 1993, Reinhart 1995, Selkirk 1995, Zubizarreta 1998, Frota 2000, among others). Spanish and Italian constitute examples of this type of languages. But in Spanish focalized words can also occur in non-in-situ positions, such as sentence-end position or also a fronted position, in both cases accompanied by main sentence stress (cf. cf. Bolinger 1972, Contreras 1978, 1980, Uriagereka 1995, Zubizarreta 1998 among others, for discussion of the different options).

Then, there are languages which signal focus morphologically, by the addition of a suffix, a prefix or some other overt marker that indicates focalization. This strategy can be combined with syntactic displacement, intonational marking, or a combination of both. In Wolof, for instance, a so-called ‘emphatic marker’ inserted before the verb indicates which constituent is being focalized, whether it is the subject, a complement, or the verb (cf. Rialland and Robert 2001). Narrow focus is also cued syntactically in Wolof, as focalized constituents have to appear in sentence-initial position.3 It is important to point out that in this language no intonational prominence or phrasing effects are manifested on the focalized element. English Creoles could be similar to Wolof in this respect, as focus is marked morphologically and also syntactically, by fronting the focalized constituent, and prosodic marking may be absent (cf. Bickerton 1993).

In Japanese, prosodic prominence and phrasing effects leave clear which constituent(s) have to be interpreted as narrow focus, but a focalized subject also appears followed by the particle –ga. Thus, in this specific context morphological marking may combine with prosodic marking to indicate focus (cf. Pierrehumbert and Beckman 1988, Haraguchi 1991, Kubozono 1993, among others).4

In other languages, focus is both syntactically and intonationally identified. That is, narrowly focalized elements not only receive main prosodic prominence and/or are accompanied by intonational phrasing boundaries, but they also occupy a syntactic position structurally defined for focalized expressions, be it Spec-CP, Spec-FocusP, the most embedded position in the sentence, the position immediately preceding the verb, or some other position. Hungarian, Turkish, Quechua, Basque and Hausa are examples of this type of language (cf. among others Horvath 1986, Kiss 1995, 1998

for Hungarian; Vogel and Kenesei 1987, 1990 for Turkish; Ortiz de Urbina 1989, 1999, Hualde et al. 1994, Elordieta 2001, Arregi 2001, Etxepare and Ortiz de Urbina 2003 for Basque; Inkelas & Leben 1990 for Hausa). The following paradigm from Basque illustrates this pattern in which focalized elements must appear immediately preceding the verb. Thus, examples (1e-g) are ill-formed because they contain focalized constituents which are either postverbal or not immediately preverbal. As in all the examples above, sentence (1a) represents a neutral declarative sentence, and the rest are sentences with focalized constituents (capitalized):5

(1) a. Jonek Mireni liburua eman dio

John-erg Miren-dat book-abs give aux ‘John has given the book to Miren’

b. Jonek liburua MIRENI eman dio

John book-abs MIREN-DAT give aux ‘John has given the book TO MIREN’

c. Mireni liburua JONEK eman dio book-abs Miren-dat JOHN-ERG give aux

‘JOHN has given the book to Mary’

d. Jonek Mireni LIBURUA eman dio John-erg Miren-dat BOOK-ABS give aux

‘John has given THE BOOK to Miren’ e. *Jonek liburua eman dio MIRENI John-erg book-abs give aux MIREN-DAT f. *JONEK Mireni liburua eman dio

JOHN-ERG Miren-dat book-abs give aux g. *Jonek Mireni eman dio LIBURUA

John-erg Miren-dat give aux BOOK-ABS

These examples show that, although Basque is a language with flexible word order, there is a syntactic constraint in this language on the relative word order between focus constituents and the verb, namely that they must be left-adjacent to it (cf. the references mentioned in the previous paragraph for details on syntactic analyses that could explain this constraint).6 But apart from this syntactic restriction, in Basque the focalized expression receives main prominence in the sentence, that is, focus is cued both syntactically and intonationally.

Serbo-Croatian offers a particularly rich case in focus marking possibilities (Godjevac 2000, Frota 2002). Like in English, prosodic phrasing and prominence with canonical word order serves to cue narrow focus. Another strategy to signal narrow focus is to produce a marked word order by scrambling operations, assigning at the same time prosodic phrasing and prominence cues to the constituent that is focalized (i.e., the Hungarian-Basque type). Finally, it is also possible to mark narrow

focus by scrambling operations under a neutral intonation, leaving the focalized constituent in sentence-final position, so that it receives default sentence stress (like in Dutch or German for verb focus). Thus, three different strategies or options are available in Serbo-Croatian to signal narrow focus. The different possibilities for signaling narrow focus discussed above might not constitute an exhaustive typology, although they might suffice for expository purposes. The table in (2) summarizes this typology of the different possibilities for signaling focus by means of syntax, morphology or prosody, or a combination of more than one of these strategies. A few representative languages are also included. Slots with a ‘?’ are those that to my knowledge do not have representatives.

(2) Strategy for focus marking Sample languages (a) Prosody alone - Only strategy: English, European Portuguese

- One strategy: Dutch, German, Spanish, Italian (b) Morphology alone ? (c) Syntactic displacement alone

- Only strategy: ? - One strategy: Serbo-Croatian, Dutch, German

(d) Prosody and Morphology - Only strategy: Japanese (subjects)? - One strategy: ?

(e) Morphology and syntactic displacement

- Only strategy: Wolof - One strategy: ?

(f) Prosody and syntactic displacement

- Only strategy: Hungarian, Basque, Turkish - One strategy: Serbo-Croatian, Spanish, Italian

(g) Prosody, syntactic displacement and morphology

?

Despite all these possibilities of marking focus, we will show that in pitch-accent

dialects of Basque (i.e., Northern Bizkaian Basque, NBB) there are cases in which words which constitute the narrow focus of the utterance are not singled out by syntactic, morphological or intonational means. In these dialects, intonational highlighting of narrow focus is restricted to words which bear a lexical or derived accent, or for some speakers, to words that constitute Accentual Phrases (APs) by themselves. That is, not any independent word can bear intonational prominence even though it may be the pragmatic focus of the utterance. We discuss these cases in the following section.

3. SYNTACTIC AND PROSODIC CONSTRAINTS ON FOCUS IN NBB

3.1. Lexically and morphologically conditioned accentual classes in NBB In order to understand the syntactic and prosodic constraints on focus in NBB, it is necessary to provide an overview of the prosodic features of these dialects. NBB dialects are pitch accent varieties of the Bizkaian dialect of Basque, and are spoken in the northwestern Basque-speaking area, along the coast and in a band of around 15

kilometers inland from the coast. A noteworthy feature of these dialects is the lexical distinction between unaccented and accented roots, stems and affixes, like in Japanese (cf. Poser 1984, Pierrehumbert and Beckman 1988, Haraguchi 1991, Kubozono 1993 among others for details on Japanese tone and intonation structure). An accented root or affix is sufficient to render an accented word, which surfaces with prominence on a non-final syllable in all contexts.7 In most NBB varieties, the syllable preceding the leftmost accented morpheme surfaces with main prominence, as illustrated in (3) below for the Gernika variety (accented morphemes are indicated by an apostrophe). In a few others, it is always the penultimate syllable that is accented (as in the Lekeitio variety, cf. Hualde et al. 1994, Hualde 1997, 1999, Elordieta 1997, 1998):

(3) a. sagar -‘ata - ‘tik → sa.gá.rra.ta.tik ‘from the apples’

apple-plur.loc.-abl. b. léku -‘ata - ra → lé.ku.e.tara ‘to the places’ place-plur.loc.-all.

A combination of unaccented roots and affixes produces unaccented words (except in compounding, where even if the members are unaccented the compound word is accented). Unaccented words will only receive prosodic prominence if they occur immediately preceding the verb or are pronounced in isolation. In these cases, in most NBB varieties they display prominence on the final syllable, and in a few dialects they show penultimate prominence (e.g., Ondarroa and Markina, cf. Hualde 1997, 2000). This kind of prominence is called derived accent by Jun and Elordieta (1997), to distinguish it from the lexical accent of accented words. In all other contexts, unaccented words do not surface with any kind of prosodic prominence on any syllable. Thus, observe the behavior of the unaccented word laguna ‘the friend’ in (4), corresponding to the Lekeitio variety (henceforth Lekeitio Basque, LB). This word is composed of the unaccented root lagun ‘friend’ and the unaccented singular determiner –a. Prosodic prominence is indicated by an acute accent mark. The different word orders in (4a-d) are due to the flexible word order of Basque, constrained by topic and focus or theme-rheme structures. That is, (4a-d) differ in information structure (rheme constituents are underlined).

(4) a. umiágas laguná etorri da child-com friend-abs come aux ‘The friend has come with the child’

b. laguná etorri da umiágas friend-abs come aux child-com ‘The friend has come with the child’

c. laguna umiágas etorri da friend-abs child-com come aux ‘The friend has come with the child’

d. umiágas etorri da laguna child-com come aux friend-abs ‘The friend has come with the child’

e. *laguná umiágas etorri da

f. *umiágas etorri da laguná

The unaccented/accented distinction is directly relevant for the intonational phrasing in NBB. Prominence is realized as a H*+L pitch accent, on the syllable that is phonologically associated with accent. As already mentioned above, accented words will always bear stress in any position in the sentence, whereas unaccented words only display a H*+L pitch accent if they are immediately left-adjacent to the verb, i.e., when they bear derived accent. The intonational pattern that arises is the following: the sentence starts with an initial low tone (%L), immediately followed by a rise phonetically associated to the second or third syllable of the first word. The pitch level is maintained until a H*+L pitch accent, whether of an accented word or an unaccented word with derived accent. If after that H*+L pitch accent there is another word, the contour that is observed is one in which again there is an initial low tone on the first syllable of that word, the pitch level rising again on the second or third syllable of the following word, and the high tone level plateau being maintained on all syllables until another H*+L accent, corresponding to an accented word or an unaccented word preceding the verb, i.e., with derived accent. And if another word follows, the same pattern is observed. Thus, a cycle of low tone, rise, plateau and H*+L pitch accent is observed. The intonational units or constituents with this shape are identified by Elordieta (1997, 1998) as Accentual Phrases (APs). Jun and Elordieta (1997) and Elordieta (1998) show that APs consist of an initial %L boundary tone, a phrasal H tone (H-) on the second syllable,8 and a H*+L pitch accent. The phrasal H tone spreads phonologically onto all syllables between the second one and the one with the pitch accent. Schematically, the tonal structure of an AP is %L H- H*+L (cf. also Hualde et al. 2002).9 Figures 1-2 illustrate the general shape of APs in NBB, corresponding to (5a-b), respectively. Figure 1 is an example of a sentence containing three unaccented words before the verb; from an IP-initial %L there is a rise on the second syllable, reaching the peak on the third syllable, and the H tone continues until the H*+L pitch accent on the final syllable of the third word (i.e., the one immediately preceding the verb, with the derived accent). The pitch drops on the verb until the end of the utterance. Fig. 2 contains two accented words, each of them with their corresponding H*+L pitch accent. Due to downstep, the second phrasal H- does not rise as much as the first one, and the second peak is smaller than the first one (cf. Elordieta 1997, 1998, Jun and Elordieta 1997 for details and more pitch tracks):

(5) a. AP{%L H- H*+L} | | | alargunen nebien diruá galdu dot widow-gen brother-gen money-abs lose aux

‘I have lost the widow’s brother’s money’

b. AP{%L H*+L} AP{%L H-H*+L} | | | | | amúmen liburúak biar doras

grandmother-gen books-abs need aux ‘I need grandmother’s books’

Figure 1. alargunen nebien diruá galdu dot

Figure 2. amúmen liburúak biar doras

3.2. Intonational restrictions on the assignment of prominence to focalized words As explained in section 2 above and illustrated in (1), in NBB only words contained in an immediately preverbal syntactic constituent can be focalized. The focalized word does not have to be the one immediately preceding the verb, but it has to be contained in a syntactic constituent that is immediately preceding the verb. Thus, in the following examples, (6b) is grammatical, as well as (6a). (6c) is ungrammatical, however, as the syntactic constituent it is contained in is not immediately preverbal (syntactic constituents are separated by square brackets): (6) a. [maixuári] [lagúnen LIBURÚAK] emon dotzaras. teacher-dat friends-gen BOOKS-ABS give aux

‘I have given the friends’ BOOKS to the teacher’ (responding to stimuli such as: ‘Which of the friends’ things have you given to the teacher?’)

b. [maixuári] [LAGÚNEN liburúak] emon dotzaras.

teacher-dat FRIENDS-GEN books give aux ‘I have given THE FRIENDS’ books to the teacher’

(responding to stimuli such as: ‘Whose books have you given to the teacher?’)

c. *[MAIXUÁRI] [lagúnen liburúak] emon dotzaras. TEACHER-DAT friends-gen books give aux ‘I have given the friends’ books TO THE TEACHER’

(erroneously responding to stimuli such as: ‘Who have you given the friends’ books to?’)

However, in cases of utterances where one of the words constitutes the narrow

focus of the utterance, even if that word is contained in the immediately preverbal constituent, there is a further constraint it must obey in order to be intonationally singled out. In the variety of NBB we have investigated, LB, a focalized word can be the most prominent intonationally if it has a lexical pitch accent (i.e., if it is a lexically accented word) or if it has a derived accent (i.e., it is an unaccented word immediately preceding the verb). Let us illustrate this constraint with sentence (7) (repeated from (5b)), containing only one preverbal constituent with two accented words, amúmen ‘grandmother’s’ and liburúak ‘books’. The intonational structure corresponding to this constituent is thus the following:

(7) AP{%L H*+L} AP{%L H-H*+L} | | | | | amúmen liburúak biar doras

grandmother-gen books-abs need aux ‘I need grandmother’s books’

That is, in the immediately preverbal syntactic constituent there are two APs, each of them containing one accented word. Let us now describe the main patterns observed in contexts of narrow focus, that is, in cases in which the focalized word replaces the variable introduced by a wh-word in a previous question. The two words in (7) would become the narrow focus of an utterance if they formed part of a response to the preceding questions in (8a,b), respectively:

(8) a. Nóren liburúak biar dósus? whose books-abs need aux ‘Whose books do you need?’

b. Sér biar dósu amuména? what need aux grandmother-gen ‘What do you need of grandmother’s?’

Since amúmen and liburúak have lexical H*+L pitch accents, they can be

pronounced standing out as the most prominent words in the utterance. An interesting aspect worth mentioning is in narrow focus cases in which the first word is focalized the pronunciation of such utterances is not usually distinguished from cases of broad focus. That is, the first word will not necessarily show a boosted pitch level and/or a following decreased pitch level. In the data we have analyzed from five female speakers of LB, only one speaker produced some utterances in which the first word was pronounced with a higher pitch followed by a lower level on the following word. This might be due to the fact that in broad focus cases the difference in pitch between the first peak and the following peaks is already quite big (cf. Fig. 2). However, when the second word is focalized, there are more instances in which the word is made more prominent intonationally and perceptually distinguishable from broad focus cases. The focalized word may present a higher pitch level (although the peak is still lower than the first peak, due to downstep), followed by a decreased pitch level. Quite often there may also be a displacement of the peak of the first word to the posttonic syllable. This strategy signals old information or topic status for that word.10 For sentence (9), which would be an answer to (8b), Figure 3 illustrates a case without peak delay at the end of the preceding word, and Figure 4 illustrates a case with peak displacement, indicated in the tone tier with a ‘>’ sign:

(9) amúmen LIBURÚAK biar doras. grandmother-gen BOOKS-ABS need aux ‘I need grandmother’s BOOKS’

A similar scenario would apply for a constituent preceding the verb which contained two words, the first one accented and the second word unaccented. The accented word has a lexical H*+L accent, and the unaccented word receives a H*+L pitch accent by virtue of preceding the verb (i.e., it has a derived accent on its final syllable). The sentence in (10) is an example:

Figure 3. amúmen LIBURÚAK biar doras.

Figure 4. amúmen LIBURÚAK biar doras.

(10) AP{%L H*+L} AP{%L H-H*+L} | | | | |

Amáien alabiá topa dot Amaia-gen daughter-abs find aux

‘I came across Amaia’s daughter’

If the first word were the narrow focus of the sentence, most commonly it would not receive more prominence than in broad focus cases. If the second word were the narrow focus, however, it would be made more prominent by presenting a higher

pitch level than in broad focus cases, accompanied or not by peak delay in the first word (interestingly, when there is peak delay in the previous word a bigger pitch level on the focalized word is not necessary). An example with peak delay in the first word is illustrated below in Figure 5, corresponding to (11). As described above, however, this pattern does not necessarily have to arise, and it is also quite normal to find cases which are intonationally very similar to broad focus utterances.11

(11) Amáien ALABIÁ topa dot. ‘I came across Amaia’s DAUGHTER’.

Figure 5. Amáien ALABIÁ topa dot

However, in the case of preverbal constituents containing one or more

unaccented words in nonfinal position (i.e., not immediately preceding the verb) the situation is different. An unaccented word will only get a derived accent if it is left-adjacent to the verb, and hence an unaccented word which is the narrow focus of an utterance but which is not in the position that grants a derived accent cannot be made more prominent intonationally. From a neutral sentence such as (12), the leftmost unaccented word, nebien ‘the brother’s’ would not receive main prominence even though it were the narrow focus of the sentence (as an answer to ‘Whose money have you lost?’, because it does not have a pitch accent, lexical or derived. A crucial aspect of this pattern in NBB is that focus does not insert accents that are not already there lexically or by virtue of a preverbal position. The first word is lexically unaccented, and even if it is focalized, it remains unaccented, that is, no accent is associated to it, as it is not left-adjacent to the verb and hence does not receive a derived accent. This impossibility does not depend on the accentual nature of the following word, as the same impossibility occurs with accented words following the unaccented word. Thus, out of a sentence such as (13) it would not be possible to highlight the first word. The type of contours that surface in these instances is one in which the leftmost word has to be pronounced in the same pitch level as the following word, in the same AP.

Figure 6 serves to illustrate such a contour, corresponding to narrow focalization of the word nebien in (12):

(12) AP{%L H- H*+L} | | |

nebien diruá galdu dot brother-gen money-abs lose aux

‘I have lost the brother’s money’

(13) AP{%L H- H*+L} | | |

lagunen liburúa biar dot friend-gen book-abs need aux

‘I need the friend’s book’

Figure 6 nebien diruá galdu dot

As for the second word in sentences such as (12)-(13), we do not find a uniform

behavior among speakers. However, such speaker variation reveals important facts about constraints on the intonational realization of main prominence in contexts of narrow focus. For two of the five speakers recorded, the second words in those cases would be able to receive main prominence if they were the narrow focus of the utterance, as in (15b), responding to a question such as (15a). An observed strategy in these cases is a continuation rise at the end of the preceding word, signaling old or known information. This rise cannot be due to an accent in the first word, so it must be due to H- (cf. Fig. 7). Another possibility is to have a sustained pitch at the end of the preceding word followed by a rise in pitch level on the focused word (other non-intonational features such as higher intensity may also surface). In both cases, a decrease in pitch level follows the focalized word. The same pattern is observed in

cases in which the second word is lexically accented, as in (16):

(15) a. Ser galdu dósu nebiena?

What lose aux brother-gen ‘What have you lost of the brother?’ b. nebien DIRUÁ galdu dot brother-gen MONEY-ABS lose aux

‘I have lost the brother’s MONEY’ (16) a. Ser biar dosu lagunena?

what need aux friend-gen ‘What do you need of the friend?’ b. lagunen LIBURÚA biar dot friend-gen BOOK-ABS need aux ‘I need the friend’s BOOK’

Figure 7. nebien DIRUÁ galdu dot

Importantly, three of our five speakers did not produce utterances like (15b), or

could not pronounce the second word in (16) with main intonational prominence. That is, they cannot highlight a word intonationally if it is preceded by an unaccented word. For these speakers, not only the leftmost word but also the second word cannot be prosodically highlighted. Regardless of which word is the corrective focus of the utterance, the whole AP (i.e., the two words) would have to be pronounced together. The explanation for this pattern would be that these speakers have a stricter constraint on the intonational highlighting of focalized words. This constraint would state that only words which constitute APs by themselves can be made intonationally prominent. In cases of two words with accent, such as the ones in (7)-(10), each word

constitutes its own AP, and can thus be singled out intonationally. But in cases in which the first word is unaccented, the second word does not constitute an AP by itself. Rather, it continues the AP that the first word started. As the intonational schemas in (12)-(13) show, the unaccented word starts an AP, with the initial %L H- tone sequence, but since it does not have a pitch accent, the phrasal H- tone spreads onto the next word, until the H*+L accent (lexical or derived) of the following word ends the AP (cf. Elordieta 1998). There is thus only one AP before the verb, containing the two words. Since neither word forms an independent AP, they cannot be made intonationally prominent on their own. The two words have to be pronounced in the same pitch level, in the same AP. The contour observed in these instances is similar to the one illustrated in Figure 6, which showed the impossibility of having the leftmost word as the most prominent word in the utterance. The important issue at work here is that no pitch accent is specially inserted to the first unaccented word, even if it is the narrow focus of the sentence from a pragmatic or information-structure point of view, as already mentioned above. Hence no AP boundary can be inserted at the right edge of the first word. That is, the lexical association of pitch accents is respected by focus in NBB.

Thus, a mismatch between semantics and intonation arises in cases where a word which does not constitute an AP by itself is the corrective focus of an utterance. No intonational cues are used within the utterance containing the contrastively focalized word alone in order to convey the intended meaning. There is no way to single out the focalized word syntactically, as the word occurs with other words in the preverbal constituent. Disambiguation can only come from the preceding linguistic context. This mismatch situation between semantics and prosody does not arise in languages surrounding NBB (Spanish and French) or in Indo-European languages. And an insufficiency of syntax and/or morphology to mark focalized words is unattested in the languages for which there are descriptions of focus realization, a summary of which was provided in section 2. Thus, this property of NBB is interesting from a typological point of view as well.

The patterns of realization of intonational highlighting change slightly when corrective focus is considered. Corrective focus refers to those instances in which the speaker corrects one of the words or syntactic phrases that her interlocutor has stated incorrectly. For instance:

(17) a. Nóren alabia topa dosula? Alaznena? whose daughter-abs find aux Alazne-gen ‘Whose daughter did you come across? Alazne’s?’

b. Es, AMÁIEN alabiá topa dot. no AMAIA-GEN daughter-abs find aux ‘No, I came across AMAIA’s daughter.’

In (17b) above, the first accented word Amáien can be made more prominent,

usually by having a boosted pitch level followed by a decreased pitch level in the rest of the material in the sentence. Thus, in corrective focus the first word is distinguishable from cases of broad focus, unlike in narrow non-corrective focus. The

second word in (17b) would also be made more prominent, by means of a delayed peak in the preceding word, signaling the character of topic or old information of that word. This type of contour is illustrated in Figure 8, for a sentence such Es, Amáien ALABIÁ topa dot ‘No, I came across Amaia’s DAUGHTER’. Another option is to have simply a higher pitch level on the focalized word, without a preceding peak displacement. Quite often, the focalized word is accompanied by higher intensity levels and longer duration.12 As already described above, the same options would be available for sentences in which the second word were lexically accented.

Figure 8 Es, Amáien ALABIÁ topa dot

But the interesting cases are those in which the first word is unaccented, forming

an AP with the following word. As described above, in narrow non-corrective focus some speakers could not highlight intonationally neither of the two words, because for those speakers the constraint at work is that a word has to constitute an AP by itself in order to be the most prominent word in the utterance, rather than simply having a pitch accent. In corrective focus, however, these speakers can make a word have main intonational prominence even if it does not constitute an AP by itself. The sufficient condition is that the word has an accent, lexical or derived, like in narrow non-contrastive focus for the other speakers. Words bearing an accent and following an unaccented word may surface with main prominence, cued by a rise in pitch on the focalized word coming from a sustained pitch of the unaccented word, or by a rise at the end of the prefocal unaccented word. In both cases, usually the focalized word displays higher intensity and duration (cf. Elordieta and Hualde 2001, in press). It is important to bear in mind, however, that these type of prosodic realizations are scarce in the production of the most restrictive speakers, that is, those for whom a word has to constitute an AP by itself in order to stand out as the most prominent word.13 Figure 9 illustrates an F0 contour for a sentence such as (18b), in which the first option is realized, and Figure 10 illustrates the second possibility, with a rise at the end of the first word.

(18) a. Ser biar dosula lagunena? Kuadernúa?

what need aux friend-gen notebook-abs ‘What do you need of the friend? His notebook?’

b. Es, lagunen LIBURÚA biar dot. ‘I need the friend’s BOOK’.

Figure 9. lagunen LIBURÚA biar dot

Figure 10. lagunen LIBURÚA biar dot

We will finish our presentation of the intonational constraints on the prosodic

realization of focus in NBB by summarizing in a table the focus realizations for all logically possible two-word combinations in a preverbal phrase. The left-hand

column summarizes the patterns in narrow non-corrective focus, and the right-hand column those of corrective focus. When the two types of constraints for intonational highlighting (having an accent or being an independent AP) produce different outputs, they are distinguished as (a) and (b).14

5. SUMMARY AND CONCLUSION

In this paper we have described the main constraints on the realization of prosodic prominence on focalized words in a pitch accent dialect of Basque. It has been shown that the minimum condition a word has to satisfy to receive main prosodic prominence if pragmatically focalized is that it has an accent, whether lexical or derived. However, in cases of narrow non-corrective focus some speakers reveal the existence of a more restrictive constraint, which demands that a word has to constitute an AP by itself in order to surface with main prominence. Words which do not fulfill these conditions have to be pronounced in the same AP with the other words contained in it. In corrective focus the sufficient condition for the five speakers recorded is that a word has an accent. In either case, the interesting fact is that an

Narrow (non-corrective) focus Corrective focus

H*L H*L | | AP[Accented]–AP[Accented] – Verb Each word can be highlighted

H*L H*L | | AP[Accented]–AP[Accented] – Verb Each word can be highlighted

H*L H*L | | AP[Accented]–AP[Unaccented] – Verb

Each word can be highlighted

H*L H*L | | AP[Accented]–AP[Unaccented] – Verb

Each word can be highlighted

H*L | AP[Unaccented–Accented] – Verb

a. Only the accented word can be

highlighted b. The two words are uttered in the

same AP

H*L | AP[Unaccented–Accented] – Verb

Only the accented word can be highlighted

H*L | AP[Unaccented–Unaccented] – Verb

a. Only the accented word can be

highlighted b. The two words are uttered in the

same AP

H*L | AP[Unaccented–Unaccented] – Verb

Only the accented word can be highlighted

unaccented word which does not have an accent cannot receive an accent even if it is pragmatically focalized. The context seems to prevent possible ambiguities between neutral and narrow focus readings of unaccented words without an accent. To our knowledge, these are unattested constraints crosslinguistically. In Tokyo Japanese, which has a lexical distinction between accented and unaccented words, any unaccented word can be prosodically highlighted (Pierrehumbert and Beckman 1988).

Dept. of Spanish Philology, University of the Basque Country, Vitoria-Gasteiz, Spain

6. NOTES

* Many thanks are due to Matthew Gordon and José Ignacio Hualde for comments on earlier versions of this article, as well as to Sónia Frota, Carlos Gussenhoven, Kiwako Ito and Sun-Ah Jun for help with section 2. Of course, this article would not have been made possible without my native informants, to whom I am indebted immensely. This work is part of two wider research projects on interfaces and the architecture of grammar (refs. 9/UPV 00033.130-13888/2001 and BFF2002-04238-C02-01), funded by grants from the University of the Basque Country and the Ministry of Science and Technology of Spain/FEDER, respectively. 1 For the sake of expository purposes, we exclude cleft and pseudo-cleft sentences from the discussion, as we will compare this type of languages with another type of languages which mark focus constituents syntactically without clefting, by having focalized constituents occupy a certain syntactic position (below in the text). Thus, we want to distinguish languages which have a structural position for focus from languages such as English that do not, although they may make use of cleft sentences to mark focus. 2 Scrambling is disfavored or does not apply with indefinite objects. In those cases, there is simply main prosodic prominence on the verb. 3 However, when an object is focalized and there is a nonpronominal subject, the focalized object has to follow the subject, which obligatorily appears thematized (i.e., topicalized, cf. Rialland and Robert 2001:897-898). 4 The particle –ga cannot be considered a focus particle in Japanese, however, because it only attaches to subjects and because it can follow subjects which constitute old information in some contexts, such as relative clauses (Kiwa Ito, p.c.). However, for our expository purposes it suffices to show that in an answer to a sentence such as ‘Who bought potatoes?’ the subject would appear with –ga, whereas if the question were ‘What did John buy?’, the subject would appear with the topic marker –wa. 5 The following abbreviations will be used: abl = ablative, abs = absolutive, all = allative, aux = auxiliary, dat = dative, erg = ergative, gen = genitive, ines = inessive, loc = locative, pl = plural, sg = singular. 6 It is possible for focalized constituents to appear after the verb, but they are usually uttered as separate intermediate or intonational phrases. They are usually preceded by pauses, fillers such as e ‘err…/um…’, or final lengthening of the verb ending in a rising intonation. It appears that copulas can be followed by focalized constituents even without a pause (Hualde et al. 1994). In central and eastern dialects it is possible to have focalized elements postverbally without a pause (cf. Hidalgo 1994, Elordieta 2003), apart from the usual preverbal position, but the speakers I have consulted cannot have postverbal focus as an answer to a wh-word. In that case preverbal focus is the only option. Perhaps only informational, non-contrastive focus (Kiss 1998) presented by the speaker in her own discourse can appear postverbally in these dialects, but more research is needed in this respect before making any generalizations. 7 Compounds are accented as well. 8 Jun and Elordieta (1997) found that in APs up to four syllables long the peak of H- is reached on the second syllable, and in APs more than fours syllables long it was reached on the third syllable. This H- is not phonetically realized when the second syllable is associated to a pitch accent. 9 For some speakers, in sequences of four or more unaccented words certain dips in pitch can be observed between two unaccented words. Jun and Elordieta (1997) and Elordieta (1998) take these to be AP-boundaries, in the absence of H*+L pitch accents. However, the dips were difficult to perceive and were much smaller than regular drops after H*+L pitch accents (see relevant pitch tracks in the mentioned articles). Also, the factors conditioning these breaks were not very well established; desire for heaviness reduction and slower rate of speech were suggested as factors involved in the insertion of these breaks, but

no systematic study was carried to prove these claims. Moreover, these facts were subject to speaker dependence; some speakers always produce plateaus in sequences of four or more unaccented words, without breaks. This issue deserves a more systematic study, which we plan to undertake in future research. 10 The delayed peaks at the end of prefocal words were already observed for some speakers of LB by Ito, Elordieta and Hualde (in press). However, their data involved cases of corrective focus, which we also discuss below. The patterns presented in this paper show that it is possible to find such delayed peaks in non-corrective narrow focus as well. Other strategies of main prominence that can be observed in these contexts and which are not intonational in nature are a higher intensity and duration on the focalized word. 11 Indeed, the speakers of LB that Elordieta (2003) gathered his data on did not produce utterances in which the second word was made the most prominent word intonationally, and this lead to positing the absence of such a possibility. That conclusion must now be corrected to capture the facts presented in this article. 12 Although the results in Elordieta and Hualde (2001, in press) showed that lengthening applied to words in corrective focus, it must be pointed out that in those utterances speakers were instructed to put special emphasis on those words. In other recordings in which speakers were not told to put emphasis on the correction, I have observed that lengthening did not occur significantly. It seems that a specific experiment is needed to clarify the role of lengthening as a cue of corrective focus, which we leave for future research. 13 Thus, highlighting words following an unaccented word without an accent is possible, but not frequent in LB. Its frequency is speaker dependent, but as stated in note 10, the possibility of finding such patterns has to be incorporated into the intonational grammar of LB, contra what was assumed in Elordieta (2003). 14 Interestingly, the two speakers that patterned differently from the other three speakers in contexts of narrow non-corrective contexts by being able to highlight a word following an unaccented word also patterned differently in a few instances. For contexts in which the first unaccented word was correctively focalized, they produced contours in which this word was prosodically set apart, by having a higher pitch level followed by a fall in pitch for the following word, or by being pronounced with bigger intensity and duration. However, such cases were few in number, compared to the majority of cases in which the unaccented word did not surface with main prominence, thus patterning with the other three speakers. At this point we consider it premature to conclude that highlighting the unaccented word in these contexts is a solid possibility in LB, and leave the issue open for further research with more speakers and more tokens of each type of context.

7. REFERENCES

Arregi, Karlos. “Focus and Word Order in Basque”. Ms., Massachusetts Institute of Technology, 2001. Bickerton, Derek. 1993. “Subject Focus Pronouns”. In Francis Byrne and Donald Winford, eds., Focus and

Grammatical Relations in Creole Languages, 189-212. Amsterdam: John Benjamins. Bolinger, Dwight. “English Prosodic Stress and Spanish Sentence Order”. Hispania 37 (1954): 152-156. Bolinger, Dwight. “Accent is Predictable (If You’re a Mind-reader)”. Language 48 (1972): 633-644. Cinque, Guglielmo. “A Null Theory of Phrase and Compound Stress”. Linguistic Inquiry 24 (1993): 239-

297. Contreras, Heles. El Orden de Palabras en Español. Madrid: Cátedra, 1978. Contreras, Heles. “Sentential Stress, Word Order, and the Notion of Subject in Spanish”. In Linda Waugh

and C.H. van Schooneveld, eds., The Melody of Language, 45-53. Baltimore: University Park Press, 1980.

Culicover, Peter, and Michael Rochemont. “Stress and Focus in English”. Language 59 (1983): 123-165. Elordieta, Arantzazu. 2001. Verb Movement and Constituent Permutation in Basque. Utrecht: LOT. Elordieta, Gorka. “Accent, Tone and Intonation in Lekeitio Basque”. In Fernando Martínez-Gil and

Alfonso Morales-Front, eds., Issues in the Phonology and Morphology of the Iberian Languages, 4-78. Washington, DC: Georgetown University Press, 1997.

Elordieta, Gorka. “Intonation in a Pitch-Accent Dialect of Basque”. International Journal of Basque Linguistics and Philology 32 (1998): 511-569.

Elordieta. Gorka. “Intonation”. In José I. Hualde and Jon Ortiz de Urbina, eds., A Grammar of Basque, 72-113. Berlin: Mouton de Gruyter, 2003.

Elordieta, Gorka, and José I. Hualde. “The Role of Duration as a Correlate of Accent in Lekeitio Basque” In Proceedings of Eurospeech 2001 - Scandinavia, 105-108, 2001.

Elordieta, Gorka, and José I. Hualde. “Tonal and Durational Correlates of Accent in Contexts of Downstep in Northern Bizkaian Basque”. Journal of the International Phonetic Association, in press.

Etxepare, Ricardo and Jon Ortiz de Urbina. “Focalization”. In José I. Hualde and Jon Ortiz de Urbina, eds., A Grammar of Basque”, 459-515. Berlin: Mouton de Gruyter, 2003.

Frota, Sónia. Prosody and Focus in European Portuguese. Doctoral dissertation, University of Lisbon, 1998 [Published by Garland in 2000].

Frota, Sónia. Review of Intonation, Word Order and Focus Projection in Serbo-Croatian (Godjevac (2000). Glot International 6 (2002): 251-256.

Godjevac, Svetlana. Intonation, Word Order and Focus Projection in Serbo-Croatian. Doctoral Dissertation, Ohio State University, 2000.

Haraguchi, Shosuke. A Theory of Stress and Accent. Dordrecht: Foris, 1991. Hidalgo, Bittor. Hitz Ordenaren Estatistikak Euskaraz. Doctoral dissertation, University of the Basque

Country, 1994. Horvath, Julia. Focus in the Theory of Grammar and the Syntax of Hungarian. Dordrecht: Foris, 1986. Hualde, José I. Euskararen Azentuerak. Bilbao: Servicio Editorial de la Universidad del País Vasco, 1997. Hualde, José I. “Basque Accentuation”. In Harry van der Hulst, ed., Word Prosodic Systems in the

Languages of Europe, 947-993. Berlin: Mouton de Gruyter, 1999. Hualde, José I. “On System-Driven Sound Change: Accent Shift in Markina Basque”. Lingua 110 (2000):

99-129. Hualde, José I., Gorka Elordieta and Arantzazu Elordieta. The Basque Dialect of Lekeitio. Bilbao and San

Sebastián: Servicio Editorial de la Universidad del País Vasco, 1994. Hualde, José I., Gorka Elordieta, Iñaki Gaminde and Rajka Smiljanić. “From Pitch-Accent to Stress-Accent

in Basque”. In Carlos Gussenhoven and Natasha Warner, eds., Papers in Laboratory Phonology VII, 557-584. Berlin: Mouton de Gruyter, 2002.

Inkelas, Sharon, and William Leben. “Where Phonology and Phonetics Intersect: The case of Hausa Intonation”. In John Kingston and Mary Beckman, eds., Papers in Laboratory Phonology I, 17-34. Cambridge: Cambridge University Press, 1990.

Ito, Kiwako, Gorka Elordieta, and José I. Hualde. “Peak alignment and intonational change in Basque”. Proceedings of the 15th International Congress of Phonetic Sciences. Barcelona, Spain, in press.

Jun, Sun-Ah, and Gorka Elordieta. “Intonational Structure of Lekeitio Basque”. In Antonis Botinis, Georgios Kouroupetroglou and George Carayiannis, eds., Intonation: Theory, Models and Applications, 193-196. Proceedings of an ESCA Workshop. Athens, Greece, 1997.

Kiss, Katalin É. “Introduction”. In Katalin É. Kiss, ed., Discourse Configurational Languages, 3-27. New York, Oxford: Oxford University Press, 1995.

Kiss, Katalin É. “Identificational Focus Versus Information Focus”. Language 74 (1998): 245-273. Kubozono, Haruo. The Organization of Japanese Prosody. Tokyo: Kurosio, 1993. Ladd, Robert D. The Structure of Intonational Meaning: Evidence from English. Bloomington, Indiana:

Indiana University Linguistics Club, 1980. Ortiz de Urbina, Jon. Parameters in the Grammar of Basque. Dordrecht: Foris, 1989. Ortiz de Urbina, Jon. “Focus in Basque”. In Georges Rebuschi and Laurice Tuller, eds., The Grammar of

Focus, 311-333. Amsterdam and Philadelphia: John Benjamins, 1999. Pierrehumbert, Janet, and Mary Beckman. Japanese Tone Structure. Cambridge, Mass.: MIT Press, 1988. Poser , William. The Phonetics and Phonology of Tone and Intonation in Japanese. Doctoral Dissertation,

MIT, 1984. Reinhart, Tanya. “Interface Strategies”. Ms., Utrecht University, 1995. Reinhart, Tanya, and Ad Neeleman. “Scrambling and the PF Interface”. In W. Gueder and Myriam Butt,

eds., Projecting from the Lexicon. Stanford: CSLI Publications, 1998. Rialland, Annie, and Stéphanie Robert. “The Intonational System of Wolof”. Linguistics 39 (2001): 893-

939. Selkirk, Elisabeth. “Sentence Prosody: Intonation, Stress, and Phrasing”. In John Goldsmith, ed., The

Handbook of Phonological Theory, 550-569. Cambridge: Blackwell Publishers, 1995. Uriagereka, Juan. “An F Position in Western Romance”. In Katalin É. Kiss, ed., Discourse Configurational

Languages, 153-175. Oxford: Oxford University Press, 1995. Vallduví, Enric. The Informational Component. Doctoral dissertation, University of Pennsylvania, 1990. Vogel, Irene, and István Kenesei. “The Interface between Phonology and Other Components of Grammar:

The Case of Hungarian”. Phonology Yearbook 4 (1997): 243-263. Vogel, Irene, and István Kenesei. “Syntax and Semantics in Phonology”. In Sharon Inkelas and Draga Zec,

eds., The Phonology-Syntax Connection, 365-378. Chicago: University of Chicago Press, 1990. Zubizarreta, María Luisa. Prosody, Focus and Word Order. Cambridge, Mass.: MIT Press, 1998.