Observations on the Roles of Prosody and Syntax in the Phonological Phrasing of Barcelona Spanish.

| January 7, 2014

December 2008. Volume 3 Issue 3
| December home | PDF version | SWF version |

Title
Observations on the Roles of Prosody and Syntax
in the Phonological Phrasing of Barcelona Spanish

Author
Rajiv Rao
Northern Illinois University, USA

Bio-Data
Rajiv Rao is an Assistant Professor of Spanish Linguistics at Northern Illinois University. He completed his Ph.D. in Spanish Linguistics at the University of California at Davis with a focus on the suprasegmental phonetics and phonology of various dialects of Spanish. His current research interests surrounding Spanish include prosodic realizations of focus, phrasing in Optimality Theory, Afro-Hispanic phonetics and phonology, and second language phonology.

Abstract

This paper aims to further enhance our knowledge of phrasing (i.e. the prosodic chunking of discourse) in the elicited, lab speech of Spanish. Specifically, Barcelona Spanish represents an intriguing object of study because very little previous work has addressed its phrasing, intonation, or its phonetics and phonology in general. Inspired by Prieto’s (2006) work on phonological phrases (PPHs) and their organization within intonational phrases (IPs), 18 speakers produced lab speech data by reading 90 sentences at a normal speech rate. The sentences consist of 18 sets of five sentences each. Four of the sets are comprised of simple Subject-Verb-Object utterances. The remaining 14 sets begin with a shorter, less complex utterance containing three or four stressed words, followed by a gradual increase in syntactic branching of the subject Noun Phrase (NP), the direct object NP, or both of these NPs. The data analysis first involved locating PPH boundaries, which was done by searching for phonetic cues previously documented as indicators of phrase breaks. The most common cues observed in the present data are pitch rises, preboundary lengthening of stressed syllables, pitch reset, and pauses. For each type of syntactic structure, the observed frequencies of different phrasing patterns produced were calculated. Patterns are described in terms of how many prosodic words (PWs) are contained within each phrase and whether prosodic boundaries coincide with syntactic maximal projections. The empirical results reveal that the ideal PPH length is two PWs. Furthermore, balanced length, symmetry and a rightward increase in length are crucial in dictating phrasing preferences, especially in sentences with complex syntactic structure. These findings run contrary to earlier proposals that place more emphasis on phrase boundaries being determined mainly by syntactic boundaries.

Keywords: prosody, intonation, phonological phrasing, Spanish, accent, stress

1. Introduction
Spanish intonation is a scope of investigation that has shown continual growth over the last ten to fifteen years. Studies produced during this period have described patterns in pitch contours in terms of where peaks and valleys occur in different types of utterances as well as how these tonal highs and lows are phonologically represented. Within this larger domain of intonation lies phrasing, or the grouping of discourse into smaller sized units, which did not receive considerable attention in Spanish until the mid-to-late 1990s.
This paper considers how speakers of the Barcelona dialect prosodically chunk information contained in utterances produced in a lab setting. The goal of the present study is to significantly contribute to the developing body of work on Spanish phrasing by discussing the ongoing debate as to whether phrasing is more influenced by syntactic factors, such as alignment with maximal projections (XPs), or prosodic conditions such as weight balance and symmetry within an utterance. This is done by utilizing recordings and pitch contours of Subject-Verb-Object (SVO) sentences with extensive syntactic branching of the subject, the direct object, and both the subject and direct object.
Additionally, focusing on data from the Barcelona dialect is noteworthy because this dialect has received little to no attention with regard to intonation. The main findings reflect the proposals of those such as Ghini (1993) for Italian, who claims that prosodic factors are much more powerful than syntactic ones in explaining phrasing patterns. In particular, weight balance (i.e. equal number of stressed words in phrases), symmetrical distributions, and increasing phrase length across an utterance are discovered as crucial to the production of speakers. On the whole, such prosodic influences outweigh those related to alignment of prosodic and major syntactic boundaries and cohesion of syntactic constituents within phrases. Before reviewing previous literature on phrasing in Spanish and the methodology and results of the present study, it is crucial to introduce certain theoretical concepts that serve as a foundation for the discussion of phrasing in lab speech.

1.1 Prosodic Phonology and Phrasing

Prosody in general is used to divide information into ‘chunks’ demonstrating definite size and internal structure (Steedman 1991; Zubizarreta 1998; D’Imperio et al. 2005). Prosodic Phonology (Selkirk 1984, 1986; Nespor and Vogel 1986), which considers the relationship between syntax and prosody, hierarchically organizes prosodic constituents in the fashion shown in (1). Each abstract constituent of this arrangement corresponds with concrete phonetic evidence found in pitch contours.

(1) Prosodic Hierarchy2
IP Intonational Phrase (Major Phrase)
PPH Phonological Phrase (Minor Phrase)
PW Prosodic Word
F Foot
σ Syllable
The various levels in (1) are defined in detail by Selkirk (1984), with the top three being most pertinent to this paper. An IP is a unit that corresponds to a portion of a sentence associated with a characteristic intonational contour or melody. In general, IPs have easily perceived pauses before and after the left and right edges, respectively. They also have boundary tones, are non-isomorphic in relation to syntactic structure, and are units that carry meaning (van Heusinger 2007). In Spanish, the conclusion of an IP is mainly signaled by a final high (H) or low (L) boundary tone (%) or by a clear pause.

A PPH provides additional depth of phrasing. It is similar to the ‘intermediate phrase’ that is commonly referred to in Intonational Phonology. The common thread between phonological and intermediate phrases is that they are both seen as smaller ‘minor phrases’ contained within IPs. Those such as Sosa (1999), Beckman et al. (2002), and Jun (2005) argue that a second level of phrasing is unnecessary in Spanish. However, Nibert (1999, 2000) perceptually justifies two levels of phrasing by showing that the second level is used to disambiguate meaning in syntactically identical utterances. Hualde (2002) and Toledo (2007), among many others, also recognize the importance of a second tier of phrasing constituent. This perspective is assumed in the present study as well. Furthermore, a major difference between PPHs and IPs is that the disjuncture at PPH boundaries is less clearly defined. The boundaries of PPHs can be located in Spanish by using phonetic cues such as F0 continuation rises ending in the final syllable of a word, longer duration of stressed syllables, large pitch range increases or decreases, and pauses (Elordieta et al. 2003; Hualde 2003; D’Imperio et al. 2005; Prieto 2006; Toledo in press; among others). Syntactically speaking, a PPH denotes any level of prosodic constituent structure that may include one or more major category words (i.e. lexical categories of Noun, Verb, Adjective, and Adverb, from Chomsky 1965). According to Truckenbrodt (1999, in press), the PPH and IP differ in that the former refers specifically to syntactic phrases (XPs), such as Noun Phrases (NPs), Verb Phrases (VPs), and Adjective Phrases (APs), while the latter deals with larger syntactic clauses.
A PW is a phonologically relevant idea that plays a metrical role in describing main word stress. According to Ladd (1996), ‘stress’ concerns perceived prominence of lexical items in an utterance, where as ‘accent’ refers specifically to intonational F0 movement, which serves as one possible phonetic cue to the location of perceived prominence. It also merits mention that earlier work using this hierarchy, such as that of Nespor and Vogel (1986), included a Clitic Group level between the PW and PPH. However, this level has been excluded from the hierarchy in more recent studies. With regard to PWs, Contreras (1964) and Navarro Tomás (1964) argue that intensity is the most salient cue to stress. Recently, Ortega-Llebaria and Prieto (2007) found that contexts in which accent is not an indicator of stress tend to rely on differences in duration and spectral tilt (and vowel quality, but to a much less degree). Furthermore, other investigations in the past two decades such as Garrido et al. (1993), Quilis (1993), Garrido (1996), and Face (2003) note that a change in F0 through the stressed syllable is actually the strongest cue to stress in Spanish. This is the assumption made in the current paper as well. Therefore, in many cases, a lexical item is a PW if it is prosodically accented, meaning it contains F0 movement through the stressed syllable. However, it should also be noted that in nuclear position of some phrases, where F0 is often reduced, prominence is still achieved through sources such as final lengthening. In fact, a Nuclear Stress Rule was formalized as a generally occurring phenomenon by Chomsky and Halle (1968).

This rule is commonly applied to constituents of Romance languages (see Prieto 2005; Hualde in press; among others). Therefore, while F0 movement is assumed to indicate PW status in prenuclear position, other factors must be examined in nuclear position. Concerning F0 movement, the minimum pitch excursion to attain PW status is 7 hertz (Hz), following O’Rourke (2006). Quilis (1993) provides a very extensive and useful list of types of palabras acentuadas e inacentuadas (‘stressed and unstressed words’). The stressed words (e.g. mainly content words) are those that are expected to be accented due to the presence of F0 movement in the stressed syllable. Unstressed words (e.g. namely function words) are those in which such tonal movement is not normally anticipated. However, factors such as speech rate and emphasis can lead to opposing trends. Overall, while F0 is used as the most prominent correlate to stress and PW status in this study, factors such as duration, intensity, and spectral tilt remain deserving of further notice in investigations of accent and stress.
In terms of the two lowest levels of the hierarchy, F refers to a suprasyllabic unit smaller than the word that helps describe stress patterns. However, Selkirk (1984) also notes that there is little evidence that this unit is a relevant domain for phonological rules. On the other hand, F is shown to be crucial to flapping in American English in Carr and Durand (2004). Finally, although one of Selkirk’s (1980) earlier versions of the prosodic hierarchy places the mora at the lowest level, the syllable is assumed to be the smallest prosodic constituent here.
The prosodic hierarchy abides by a certain set of rules and restrictions (Selkirk 1984; Nespor and Vogel 1986). Two fundamental rules pertaining to the prosodic hierarchy that have been formalized in frameworks such as Optimality Theory (OT) (Prince and Smolensky 1993) are outlined in (2) (outlined in Truckenbrodt in press, based on Selkirk 2000).

(2) a. Exhaustivity
Every constituent of level l is contained in a constituent of level l+1. For example,
a PW is contained in a PPH.
b. Nonrecursivity
No constituent of level l is contained in another constituent of level l. For example, a PPH is not contained in another PPH.

Previous studies show that these rules apply to different levels of prosodic representation, such as Ladd (1986) and Frota (2000) for IPs, Truckenbrodt (1999) and Gussenhoven (2004) for PPHs, and Vigário (2003) for PWs.
Phonological rules are applied to the prosodic constituents of the hierarchy. Previously, those such as Nespor and Vogel (1986) claimed that syntactic structure is that which dominates the distribution and division of prosodic constituents. This idea especially pertains to the top two levels of the hierarchy, the IP and PPH. Although it was mentioned in such older studies that speech rate, style, and emotion can lead to restructuring of IPs into shorter IPs, D’Imperio et al. (2005) clearly emphasize that more recent studies have shown that prosodic boundary placement in different languages is determined by factors other than merely syntax. Namely, these factors deal with constituent weight, symmetrical distribution of constituents, and information structure (see Steedman 1991; Ghini 1993; Truckenbrodt 1995, 1999; Ladd 1996; Selkirk 2000; among others).
The remainder of this paper is organized as follows: Section 2 provides an overview of previous work on phrasing in lab speech in Spanish, Section 3 describes the methods and materials used to collect and analyze lab speech data in the present study, Section 4 presents the empirical results of phrasing patterns in a series of tables, and Section 5 concludes by summarizing the main findings, mentioning some limitations of the study, and suggesting future research paths.

2. Previous Studies
The majority of studies on phonological phrasing decisions in Spanish and in Romance in general have been carried out only in recent years.3 For example, Garrido et al. (1995) investigate the link between prosody and syntactic boundaries by examining the phonetic variables that can serve as indicators of such boundaries and by comparing the prosodic features of four different boundary types. It is found that vowel duration tends to increase at a syntactic boundary (non-prepausal, in order to avoid pre-pausal lengthening effects) when compared to word-internal vowels. These results are obtained at a statistically insignificant level. Furthermore, F0 peaks normally occur just before syntactic boundaries, regardless of type of boundary. These peaks tend to be located in the syllable immediately prior to the boundary, regardless of its status as stressed or unstressed. F0 reset appears to occur more in valleys and is more common at higher level syntactic boundaries, such as NP-VP. Therefore, the results suggest that reset could be a cue to hierarchical differences between syntactic boundaries.
Elordieta et al. (2003) study the effects of syntactic branching and constituent length on the placement of prosodic boundaries in Spanish, European Portuguese, and Catalan. This study uses a database containing broad focus SVO sentences with combinations of two constituent length conditions and seven syntactic branching conditions, with each of these conditions applying to the subject and object. The most common phrasing pattern found for Spanish is (S)(VO). The dividing point between the subject and verb is manifested as a F0 rise, often accompanied by final lengthening, and sometimes including pauses as well. There is also reset and a drop in F0 at the beginning of the verb. In examples of non-branching subjects or objects, a few cases of (SVO) and (S)(V)(O) phrasings are found. However, when either the subject or object is branching (short or long), (S)(VO) groupings are completely dominant. Spanish and Catalan prefer (S)(VO) divisions, although the latter favors balancing, meaning it divides utterances according to number of PWs per phrase. European Portuguese favors (SVO) in the majority of examples. Overall, the study demonstrates that these three languages share the fact that length is a stronger determiner of phrasing decisions than is syntactic complexity.
D’Imperio et al.(2005), which expands the previous study to include five Romance languages, mentions the same results for Spanish while also providing further comments. In this study, a database is used containing complex objects of two PWs composing one noun (i.e. a proper name), in order to have objects that are prosodically branching but syntactically non-branching. The investigators utilize this as a method to discover whether syntactic or prosodic branchingness more highly influences intonational phrasing. For Spanish, it is found that objects displaying prosodic branching, such as proper names, display a decreased percentage of (S)(VO) phrasings, with other types of phrasing divisions increasing such as (SVO), (S)(V)(O), and (SV)(O). These results differ from those for syntactically branching common nouns in object position, which, as shown in Elordieta et al. (2003), occur almost always in (S)(VO) groupings. D’Imperio et al. (2005) claim that this difference may be attributed to employing phrasing or accentuation (i.e. separating the object by phrase boundaries) in order to make an object intonationally salient. This strategy could be initiated by retrieval difficulties of proper names in comparison to common nouns and by the salience of proper names both cognitively and pragmatically (see Burgess and Conley 1999).
A study by Frota et al. (in press) addresses the phonetics and phonology of intonational phrasing in Romance. The results indicate that in Spanish, H boundary tones usually serve as markers of prosodic breaks. Furthermore, the F0 stretch before the boundary tends to appear as a continuation rise. This type of rise takes place in the stressed syllable in nuclear position, often extending to the boundary syllable. It is also manifested as a rise followed by sustained pitch, leading to a plateau up to the boundary syllable. F0 reset and preboundary lengthening of a word or stressed syllable are shown to be used by the languages with different frequencies. For Spanish, pitch reset is found to be a common indicator of a phrase boundary. Overall, the study illustrates that variation may occur in the phonology (choice of boundary tone) and phonetics (scaling of boundary tone, scaling of pre- and post-boundary peaks) of boundaries, even in languages that are related.
Prieto (2006) examines patterns of phonological phrasing in Peninsular Spanish at slow, normal, and rapid speech rates by having participants read SVO sentences with various degrees of syntactic branching of subject NPs and direct objects contained within following VPs. In terms of accounting for PPHs when looking at F0 contours, Prieto states that Spanish speakers prominently stress or accent the last tonic syllable of a PPH. Optionally, there is a continuation rise at the right boundary of a PPH, even though a phrasing break is often perceptible without this rise. The remainder of Prieto’s study, which analyzes the phrasing patterns observed using OT, shows that a full explanation of the trends involves the interaction of both constraints on eurhythmicity (i.e. balanced stressed periods) as well as syntactic alignment.
Inspired by the aforementioned study by Prieto, Rao (2007) investigates the phrasing patterns produced by speakers from Lima, Perú. Through a detailed analysis of pitch contours, he discovers that final lengthening of stressed syllables is clearly the most salient marker of PPH boundaries in this variety. This finding suggests that phonetic cues to phrase divisions are subject to dialectal variation. Final lengthening is at times accompanied by pitch rises, pitch reset, and pauses. Additionally, in Rao’s OT analysis, prosodic constraints are dominant in accounting for the data. The overall distribution of PWs across PPHs motivates the proposal of a new constraint prohibiting the increase of PPH length across IPs.
Overall, previous accounts of prosodic phrasing patterns in lab speech in Spanish from a more descriptive stance have proved to be fruitful because they have revealed various cues that are indicative of phrase boundaries. They also show that different phrasing patterns are possible in SVO utterances. The phrasing of such utterances seems to be determined by syntactic boundaries and branching, as well as other prosodic conditions. For the purposes of the present study, the overview of literature motivates the following research questions for Barcelona Spanish, which is not a variety appearing in previous work: 1) What are the observed phrasing patterns in SVO lab speech utterances in this dialect?; 2) How does increased syntactic branching of NPs and VPs influence phrasing patterns? The subsequent portions of this paper investigate these questions from an empirical perspective.

3. Methods

3.1 Data Elicitation

Inspired by the methodology of Prieto (2006), the 18 participants in the present study were each asked to read 90 sentences in broad focus at a normal speech rate. This rate is what they consider to be appropriate, for example, when reading a text aloud to a friend. Due to the fact that Barcelona is a city in which Spanish and Catalan are in constant contact, it was important to carefully select participants. Since the focus is on the dialect of Spanish spoken in this region, only speakers who are dominant in Spanish or who consider themselves as balanced bilinguals were eligible to participate. In this context, the term ‘balanced’ refers to speakers that judge themselves as confident and proficient communicating in both Spanish and Catalan. A language history questionnaire helped screen for speakers who are Spanish dominant or balanced. This questionnaire can be found in Appendix A. Upon completing the language history questionnaire, participants carried out the reading task in a phonetics laboratory at the Universitat Autònoma de Barcelona.4 They were recorded using the PitchWorks software package, a head-mounted microphone, and a laptop computer.

The 12 female and 6 male participants selected were all between the ages of 18-35 at the time of the recordings. All but three of the speakers were students at the Universitat Autònoma de Barcelona. Although this was the case, the other three speakers had all earned their undergraduate college degree and were working in Barcelona. Therefore, we can generalize that all participants had either received or were in pursuit of their undergraduate degree. Furthermore, the speakers were all born and raised in or near Barcelona and use Spanish more in their university coursework, as well as at home and when talking to friends. However, they all admitted that Catalan plays a role in their lives, especially when they are in public settings such as the bank, the post office, and restaurants, where most locals are prideful in favoring Catalan. Since Catalonians in general perceive their regional language as a symbol of identity, many treat Spanish negatively and avoid using it as much as possible, even if they have the ability to do so. In a few cases, potential participants were dismissed because of their attitudes toward Spanish, which the investigator felt might impede successful recordings. Attitudes toward each language were addressed with each potential participant orally, after completion of the questionnaire, because the investigator discovered that attitude was a potential confounding variable after creation of the questionnaire.
The 90 sentences used for data elicitation consist of 18 sets of five sentences each. Each set contains sentences with different types of syntactic branching. Although sentences group themselves according to structural similarities, they were randomized when presented to the participants. Examples of stimuli sets are shown in Appendix B. Four of the sets are constructed of simple SVO utterances. The remaining 14 sets begin with shorter, less complex utterances containing three or four PWs, followed by a gradual increase in syntactic branching. Increasing syntactic complexity corresponds with increasing the number of PWs. Increased branching takes place on both the subject NP and the direct object belonging to a VP maximal projection, and is done by adding APs, PPs, and in some cases, an NP functioning as an appositive that is considered to be part of the branching of the subject NP.5 Of the 14 sets with increased syntactic complexity, six demonstrate branching of the subject NP, six contain direct object NP branching via APs and PPs, and two exemplify branching of both the subject NP and the direct object NP. Unfortunately, due to technical difficulties during the recording process, there are fewer tokens of subject NP branching via APs and PPs. Utterances with branching of both the subject and direct object NPs are the most complex structures used in this study, with the longest sentences containing up to 11 PWs.
Another factor considered is whether the presence of a stressed determiner, or an extra PW, affects the parsing of phrases in examples of direct object NP branching. Quilis (1993) served as an invaluable resource in distinguishing those determiners that normally carry stress from those that do not. The presence of both types of determiners is equally divided among the four sets of simple SVO utterances as well as those containing direct object NP branching. An example of a set of sentences with unstressed determiners in the direct object NP is given in set 6 in Appendix B. The determiners are underlined to highlight the contrast between sentences in sets 6 and 7 of this appendix.It should be noted that those such as Quilis (1993) state that indefinite articles such as una (‘a,’ feminine) are generally stressed in Spanish. However, the data here rarely display F0 movement on such items. For this reason, they are treated as unstressed function words located within the same prosodic domain as the following content word. For example, una is within the same word-level domain as carta (‘letter’), and the only stress present is on the latter, content word (see Liberman and Sproat 1992; Hualde in press).
3.2 Data Analysis

The data analysis first involved confirmation that normally stressed words are indeed considered PWs, meaning they contain F0 rises through lexically stressed syllables. Since this portion of the study concerns laboratory speech, we expect tonal movement through almost all stressed syllables, based on the findings of Face (2003), in which deaccenting is rarely found in lab speech. Furthermore, phonetic cues documented in previous studies assisted in locating PPH boundaries. The most common cues observed in the present data are continuation rises, large decreases in pitch range, pre-boundary lengthening of stressed syllables, and pauses. The following series of figures shows examples of these signals to phrase boundaries found in F0 contours. The vertical lines in each figure show where PPH boundaries occur and the symbol ‘Ф’ serves to demarcate PPHs.
In Figure 1, there is a continuation rise ending on the final syllable of Bárbara, followed by a large decrease in pitch range, both of which are indicators of a PPH boundary. The F0 measurement of the first peak realized in the first PPH, Bárbara, is 292.2 Hertz (Hz), while the second peak drops to 209.5 Hz. Although peak downstep is often seen within PPHs in Spanish declaratives, Prieto et al. (1995, 1996), Prieto (1998), Face (2001, 2003) and Hualde (2003), among others, have described this pattern as a gradual decrease in F0 peak height rather than the drastic decrease seen here. For example a downstepped contour of a declarative shown in Face (2003: 118) illustrates a decay of just 18 Hz between the first two peaks of the utterance. Further evidence of a phrase boundary after the subject of the utterance in Figure 1 is that the stressed syllable, Bar., is noticeably longer here than it is in a case in which the word Bárbara is not directly followed by a phrase boundary.6 In this example, the duration of the stressed syllable is 193.2 milliseconds (ms), which is much longer than the 138.3 ms duration of this same syllable found in the sentence La Bárbara rubia lleva el bolígrafo (‘The blond Barbara carries the pen’), in which the phrase boundary is located after the AP complement rubia (‘blond’), and not after Bárbara, the NP head. This contrast can be seen when comparing Figures 1 and 2.

Figure 1: The phrasing of the simple SVO sentence Bárbara lleva el bolígraf (‘Barbara carries the pen’). The phrasing pattern observed is (Bárbara)Ф(lleva el bolígrafo)Ф. The highlighted portion represents the duration of the stressed syllable of the subject Bárbara.

In addition to the shorter duration of the stressed syllable of the word Bárbara in comparison to Figure 1, Figure 2 presents further evidence for a phrase boundary occurring after rubia, the second PW of the utterance. The first two peaks are manifested in a similar pitch range, with the first peak being at 260 Hz and the second at 250.4 Hz. The second peak, associated with the word rubia, is followed by an increased drop in F0 to 205.7 Hz, which is the height of the third peak, associated with the word lleva (‘carries,’ 3rd person singular). The gradual decrease in F0 peak height between the first two peaks, which are within the same PPH, is expected due to downstep. On the other hand, the exaggerated drop to a much lower pitch range after the F0 rise to the second peak is indicative of a PPH boundary. Additionally, the stressed syllable of rubia, ru., is 188.7 ms in length. This duration is significantly longer than that found in other sentences in the data in which rubia is not before a PPH boundary.

Figure 2: The phrasing of the sentence La Bárbara rubia lleva el bolígrafo (‘The blond Barbara carries the pen’). The phrasing pattern observed is (La Bárbara rubia)Ф(lleva el bolígrafo)Ф. The shorter duration of the stressed syllable of Bárbara is noteworthy because it helps prove that the phrase boundary is not after this subject NP,
but rather after the adjective rubia.

Figure 3 presents the division of PPHs in an utterance characterizing syntactic branching of a direct object NP belonging to a higher VP. The continuation rises leading up to the ends of the words Marcelo, rica (‘rich/delicious’), and supermercado (‘supermarket’) are phonetic cues to PPH boundaries. Such rises in F0 have been documented as far back as Navarro Tomás (1944) as serving a pragmatic function for speakers. That is, they are often used to signal incomplete thoughts for which additional information is to follow.

Figure 3: The phrasing of the VP branching sentence Marcelo compró la comida rica del nuevo supermercado en la esquina (‘Marcelo bought the delicious food at the new supermarket on the corner’).The phrasing division observed is (Marcelo)Ф(compró la
comida rica)Ф(del nuevo supermercado)Ф(en la esquina)Ф.

Figure 4 illustrates the cues to PPH boundaries found in an utterance with branching of both the subject NP and the direct object NP. Due to the presence of the conjunction y (‘and’), which creates a slightly different syntactic structure than that of other double branching sentences, this specific utterance was excluded from the analysis of phrasing. However, the F0 contour in Figure 4 effectively illustrates cues to PPH boundaries. The continuation rises on the words Javier, Barcelona, and libro (‘book’), reveal the presence of PPH boundaries. The rises in the second and third PPHs actually create a situation in which the final pre-boundary F0 height measurement is much higher than that of the preceding peak belonging to the same PPH. The evidence for a phrase boundary after Barcelona is the strongest because the continuation rise is also accompanied by a short pause of 116.7 ms before the beginning of the VP, me dona el libro verde (‘he/she donates the book (to me)’). This suggests that cues to PPH phrase boundaries may be stronger at higher level syntactic boundaries, which complements the ideas of Garrido et al. (1995).

Figure 4: The phrasing of El Javier inteligente y gordo de Barcelona me dona el libro verde (‘The intelligent and fat Javier from Barcelona donates the green book to me’), parsed into (El Javier)Ф(inteligente y gordo de Barcelona)Ф(me dona el libro)Ф(verde)Ф.

4. Results
This section provides tables with the observed frequencies of phrasing patterns. The tables are organized according to number of PWs per utterance as well as type of syntactic branching. Following Prieto (2006), syntactic structures of utterances are noted in bracket format. In order to slightly simplify these representations, an exhaustive account of all constituents, especially those that are embedded, may not be given. One example of such simplification is assuming that determiners are included with the NPs with which they are associated. Reference is made to X’ levels in order to distinguish between constituents such as complements and adjuncts. However, all X’ levels do not accompany brackets (especially due to the inclusion of NPs rather than DPs, after Prieto 2006). Overall, a debate surrounding theories of syntactic representation is outside the scope of the current investigation.
The discussion of results from each table will particularly emphasize the locations of prosodic and syntactic boundaries as well as the distribution of PWs across PPHs of utterances. When considering a match between the two boundary types, the main syntactic cases are described within the context of ‘major’ or ‘high-level’ boundaries. In the horizontal organization used here to represent hierarchical constituents, these labels mainly denote extreme right edges of both the subject NP and the VP maximal projections. In terms of distribution of PWs, it should be stated that ‘prosodic weight’ of PPHs is used synonymously with ‘length’ of PPHs. Balanced weight corresponds with an increase in eurhythmicity. Finally, phrasing patterns that show a frequency of 5% or less are collapsed into the ‘others’ row in all tables of this section.

4.1 Simple SVO Utterances

To begin, Table 1 deals with simple SVO utterances with unstressed determiners in the direct object NP. All sentences belonging to this classification have a total of three PWs. In this case, there are three PWs since the determiner una (‘a’, feminine) is unstressed. The letter ‘W’ denotes any PW in all tables.
Table 1 shows that speakers prefer isolating the subject NP in its own PPH, followed by a PPH boundary aligning with the NP-VP syntactic boundary, and finally another PPH containing the two PWs belonging to the VP. Other patterns were articulated, but at very low frequencies when compared to the dominant trend seen in (a).

Syntactic structure: [[N]np[V[N]np]vp]cp/ip
(3 PWs)

Example: Javier escribe una carta (‘Javier writes a letter’)

Phrasing Pattern

Frequency (%)

a. (W)Ф(WW)Ф

75%

b. (WW)Ф(W)Ф

17%

c. (WWW)Ф

4%

d. (W)Ф(W)Ф(W)Ф

4%

Table 1: Phrasing patterns of simple SVO utterances with unstressed determiners in the direct object NP.

When the number of PWs of SVO utterances is altered by adding a stressed determiner to the direct object NP, such as muchas (‘many,’ feminine), a different trend of chunking constituents occurs. Table 2 shows the results of utterances such as Javier escribe muchas cartas (‘Javier writes many letters’). It is worth noting that although the number of PWs increases in Table 2, the overall syntactic structure of utterances considered in Tables 1 and 2 remains unchanged. The results in Table 2 reveal the emergence of two preferred phrasing patterns, which are present at almost the same frequencies. In some cases, speakers prefer to place a PPH boundary at the NP-VP syntactic boundary, which yields (b). However, speakers also produce two phrases containing equal numbers of PWs, as seen in (a). When (a) is favored speakers show that equal distribution of PPH weight is a top priority. This sacrifices placing a PPH boundary after the NP maximal projection, Javier. On the other hand, a propensity toward (b) values aligning PPH boundaries with the right edges of both the subject NP and VP. The right edges of boundaries are the only relevant edges here, since Spanish is a right branching language.

Syntactic structure: [[N]np[V[N]np]vp]cp/ip
(4 PWs)

Example: Javier escribe muchas cartas (‘Javier writes many letters’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WW)Ф

37%

b. (W)Ф(WWW)Ф

36%

c. (W)Ф(W)Ф(WW)Ф

15%

d. (W)Ф(WW)Ф(W)Ф

6%

e. Others

6%

Table 2: Phrasing patterns of simple SVO utterances with stressed determiners in the direct object NP.

4.2 Syntactic Branching of the Subject NP
This section gives the observed frequencies of the phrasing patterns of utterances with syntactic branching of the subject NP. All such utterances contain between four and eight PWs. The results include examples of branching via an adjective, such as La Bárbara rubia lleva el bolígrafo (‘The blond Barbara carries the pen’), as well as cases of branching via appositives, such as Carmen la maestra lava la bicicleta (‘Carmen the teacher washes the bike’). In terms of syntax, it is assumed that adjectives and appositives are syntactically analyzed in the same manner. Both cases are classified as modifiers of the subject head, as opposed to complements. These modifiers are adjoined to a higher N’ node than that of complements. The representation of this similarity using a portion of sentences produced by speakers is given in (3). This schema is partial because only the subject portion of the sentence is relevant to the present discussion.
(3) Syntactic structure of La Bárbara rubia (‘The blond Barbara’) and Carmen la maestra (‘Carmen the teacher’)

Prosodically speaking, Ladd (1986) proposes that NPs that are syntactically embedded in a sentence, which include appositives, must be treated as a special case of prosodic branching. He argues that embedded NPs form a recursive IP structure belonging to a higher node, PPH’, that dominates the noun head and the appositive. The PPH’ is a constituent of the highest IP node. Ladd states that experimental evidence supporting this idea comes from studies on declination and declination reset. This type of structure involving a recursive IP violates Selkirk’s (2000) proposal of Nonrecursivity (see (2b)). In Ladd’s proposal, an IP contains another IP, which (2b) does not allow. A specific example of the prosodic structure in question is displayed in (4), using an utterance produced by speakers in this study.

(4) Ladd’s (1986) prosodic representation of appositives applied to Spanish.

Ladd (1986) uses the terms Major Phrase (MP) and Tone Group (TG) to refer to Selkirk’s (1984, 1986) IP and PPH, respectively. For our purposes, the two sets of terms are considered to be synonymous.
Tables 3 through 12 display the frequencies of phrasing patterns in utterances with subject NP branching via appositives or APs and PPs. The two types of branching are considered in separate tables due to the recursive IP structure that is unique to appositives.
The phrasing trends displayed in Table 3 indicate that speakers prefer to group PWs in two PPHs of equal lengths, while also placing a prosodic boundary at the NP-VP syntactic boundary. The overwhelmingly preferred pattern in (a) proves that speakers desire wrapping constituents belonging to the entire subject NP and VP maximal projections in the same PPH. Along with this preservation of syntactic integrity is an attempt to equally distribute the weight of PPHs. The less frequent patterns in (b-d) divide the constituents belonging to subject NP and VP maximal projections and fail to evenly distribute the number of PWs per PPH.

Syntactic structure: [[N[A]ap]np [V[N]np]vp]cp/ip
(4 PWs)

Example: La Bárbara rubia lleva el bolígrafo (‘The blond Barbara carries the pen’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WW)Ф

80%

b. (W)Ф(W)Ф(WW)Ф

10%

c. (WWW)Ф(W)Ф

5%

d. (WW)Ф(W)Ф(W)Ф

5%

Table 3: Phrasing patterns of subject NP branching with an AP constituent.

The results in Table 4 suggest that speakers interpret the reading of appositives in two possible ways. Those that produced pattern (a) did not produce a prosodic boundary after the subject Carmen, which points to the fact that they do not believe that a comma is necessary between this first PW and the appositive NP modifier. Such a reading is prosodically similar to the dominant pattern found in Table 3 and appears to be a contrastive interpretation in which one needs to specifically state that Carmen is a teacher (as opposed to any another profession). On the other hand, when there are a total of three PPHs, as seen in (b), speakers prosodically divide Carmen and the appositive NP, which evidences a reading in which a comma is needed after the first PW of the utterance. Consultation with a native speaker of Peninsular Spanish helped determine that this interpretation can be described as a type of ‘non-contrastive’ focus, in which certain information about the subject is highlighted without comparing this added information to some other possibility. For example, the reading in (b) simply emphasizes that the subject is a teacher, where as (a) shows that the subject is a teacher, as opposed to another profession. Therefore, (a) distinguishes itself from (b) by exemplifying ‘contrastive’ focus (see Face 2001).

Syntactic structure: [[N[N]np]np [V[N]np]vp]cp/ip 
 (4 PWs)

Example: Carmen la maestra lava la bicicleta (‘Carmen the teacher washes the bicycle’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WW)Ф

57%

b. (W)Ф(W)Ф(WW)Ф

43%

Table 4: Subject NP branching via an appositive NP (la maestra, ‘the teacher’) containing one PW.

As syntactic branching of the subject NP increases via the addition of adjectives and prepositions, more patterns are observed. Upon reviewing the top four patterns in Table 5, it is apparent that maintaining the entire VP intact is a high priority. The two most frequent patterns, (a) and (b), divide the subject NP into two PPHs. Pattern (b) exhibits a rightward increase in length among the two PPHs that form the subject NP, from one PW to two PWs, while pattern (a) shows the opposite distribution. Pattern (a) is also symmetrical, meaning the PPHs surrounding the medial PPH are equal in length, which is not the case in (b). When comparing (a) and (b) to (c) and (d), it becomes clear that in examples of five total PWs, speakers prefer to parse utterances into three total PPHs, as opposed to two or four. Specifically, it is better to divide the overall subject NP into two PPHs and less common to produce one long PPH or three individual PPHs from this high-level constituent.

Syntactic structure: [[[N AP]n’[P NP]pp]np [V[N]np]vp]cp/ip
(5 PWs)

Example: La Bárbara morena del valle lleva el bolígrafo (‘The brunette Barbara from the valley carries the pen’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(W)Ф(WW)Ф

36%

b. (W)Ф(WW)Ф(WW)Ф

23%

c. (W)Ф(W)Ф(W)Ф(WW)Ф

14%

d. (WWW)Ф(WW)Ф

14%

e. Others

13%

Table 5: Subject NP branching via an AP and a PP.

When the length of the appositive NP is increased to two PWs, as shown in Table 6, three different phrasing patterns are articulated. Similar to what is shown in Table 4, a prosodic boundary after the subject, as seen in (a), cues the interpretation of the appositive NP, la maestra nueva (‘the new teacher’), as a recursive IP. Such a reading led to the most common phrasing pattern, in which the noun, Carmen, the appositive NP, and the entire VP, lava la bicicleta (‘washes the bicycle’), each form their own PPHs. In this distribution, the PPH with the least number of PWs is at the beginning and is followed by two longer PPHs of equal length. Pattern (b) does not contain a recursive IP and is the second most frequent pattern. Here, the entire NP subject fits into one PPH of three PWs and is followed by another PPH of two PWs containing the VP’s constituents. In both (a) and (b) a prosodic boundary is placed after nueva and bicicleta, which both represent the right edge of syntactic maximal projections. Pattern (c) is a symmetrical production, but it separates an adjective from the noun that it complements, which is not highly accepted.

Syntactic structure: [[[N]n’[N AP]np]np [V[N]np]vp]cp/ip
(5 PWs)

Example: Carmen la maestra nueva lava la bicicleta (‘Carmen the new teacher washes the bicycle’)

Phrasing Pattern

Frequency (%)

a. (W)Ф(WW)Ф(WW)Ф

48%

b. (WWW)Ф(WW)Ф

37%

c. (WW)Ф(W)Ф(WW)Ф

15%

Table 6: Subject NP branching with an appositive NP (la maestra nueva, ‘the new teacher’) containing two PWs.

When syntactic branching of the subject NP using APs and PPs increases to create utterances of six total PWs, several phrasing patterns emerge. Such patterns are illustrated in Table 7.7 The most frequent, (a), equally balances the distribution of PWs, forming three PPHs of two PWs each. In this example, there are prosodic boundaries after central (‘central’) and bolígrafo (‘pen’), which respectively represent the end the subject NP and VP syntactic maximal projections. In fact, in all examples of this type of utterance, prosodic boundaries are present at the two aforementioned syntactic boundaries. Patterns (b), (c), and (d) parse the constituents of the entire subject NP projection in different unbalanced combinations of PPHs. The second most frequent pattern, (b), equally balances the first two PPHs and then increases the number of PWs in the third PPH, while (c) and (d) lack any balance of PPHs within the subject NP’s constituents. Pattern (c) prefers an increase in PWs per PPH across the subject NP, while pattern (d) displays a decrease in PPH length across the same domain.

Syntactic structure:  [[[N AP]n’[P[N AP]np]pp]np [V[N]np]vp]cp/ip
(6 PWs)

Example: La Bárbara testaruda del valle central lleva el bolígrafo(‘The stubborn Barbara from the valley carries the pen’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WW)Ф(WW)Ф

38%

b. (W)Ф(W)Ф(WW)Ф(WW)Ф

19%

c. (W)Ф(WWW)Ф(WW)Ф

14%

d. (WWW)Ф(W)Ф(WW)Ф

10%

e. Others

19%

Table 7: Increased branching of the subject NP through APs and PPs.

In Table 8, pattern (a), which is the most common, has a recursive IP, la maestra del segundo grado (‘the second grade teacher’), that is phrased in one PPH containing three PWs. Similar to previous examples, in the two most frequent patterns, prosodic boundaries occur after the last element of the entire subject NP and VP maximal projections (grado and bicicleta, respectively). Rather than separating the appositive from Carmen, the non-recursive IP structure in (b) favors equal weight balance of PPHs across the entire utterance. This contrasts with (a), in which there is an initial increase in PPH length, followed by a decrease.

Syntactic structure: [[[N]n’[N[P[N AP]np]pp]np]np [V[N]np]vp]cp/ip 
(6 PWs)

Example: Carmen la maestra del segundo grado lava la bicicleta(‘Carmen the second grade teacher washes the bicycle’)

Phrasing Pattern

Frequency (%)

a. (W)Ф(WWW)Ф(WW)Ф

50%

b. (WW)Ф(WW)Ф(WW)Ф

36%

c. Others

14%

Table 8: Increased branching of the subject NP through a longer appositive (la maestra del segundo grado, ‘the second grade teacher’) of three PWs.

Table 9 reveals that non-appositive subject NP branching resulting in seven total PWs has two trends, (a) and (b), tied for highest frequency. Once again, there are still PPH boundaries at the end of the whole subject NP and the VP. The difference between the most frequent parsing of PPHs lies in the division of the phrases belonging to the overall subject NP. Pattern (a) has two total PPHs in the subject NP, with the first being shorter in length than the second. Conversely, three PPHs compose this same structure in (b), with the first two being equal in weight and the third decreasing by one PW. The third most common phrasing pattern, (c), is similar to (b) in that the syntactically branching subject NP is comprised of three PPHs. Although the first two PPHs in (c) are balanced, they each only contain one PW, and are followed by a PPH with three PWs. Therefore, the increase in PWs per PPH on the recursive side of the subject NP, along with the number of PWs in the first two balanced PPHs, are that which distinguish (b) from (c).

Syntactic structure: [[[N AP]n’[P[[N AP]n’[P NP]pp]np]pp]np [V[N]np]vp]cp/ip
(7 PWs)

Example: La Bárbara preciosa de la ciudad húmeda en la costa lleva el bolígrafo (‘The precious Barbara from the humid city on the coast carries the pen’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WWW)Ф(WW)Ф

31%

b. (WW)Ф(WW)Ф(W)Ф(WW)Ф

31%

c. (W)Ф(W)Ф(WWW)Ф(WW)Ф

19%

d. Others

19%

Table 9: Increased syntactic branching of the subject NP via APs and PPs, leading to an utterance with seven total PWs.

The frequency results in Table 10 indicate that in three of the four most commonly observed phrasing patterns (b-d), speakers produce this type of sentence as if there were a comma separating Carmen, from the appositive NP. The differences between these three patterns deal with the divisions of phrases across the appositive NP. Pattern (b) prefers chunking all four PWs in the appositive together, while (c) parses the two PPHs that form the appositive into equal groups of two PWs. In (d), the PPHs in the appositive show a decrease in length from three PWs to one PW. Furthermore, pattern (a), which does not treat the appositive NP as a recursive IP, points to the preference of grouping Carmen with the noun head and AP complement of the appositive. This creates a division before the PP of the appositive and splits this modifier NP across two PPHs. In this trend, the first PPH of the utterance is the longest, and is followed by two more with equal weights of two PWs each. Finally, consistent with the previous results, there is a prosodic boundary marking the end of the entire subject NP (secundaria, literally ‘second’) and the VP (bicicleta, ‘bicycle’) projections in all phrasing patterns produced.

Syntactic structure: [[[N]n’[[AP N]n’[P[N AP]np]pp]np]np [V[N]np]vp]cp/ip
(7 PWs)

Example: Carmen la única maestra de la escuela secundaria lava la bicicleta (‘Carmen the only teacher at the high school washes the bicycle’)

Phrasing Pattern

Frequency (%)

a. (WWW)Ф(WW)Ф(WW)Ф

29%

b. (W)Ф(WWWW)Ф(WW)Ф

25%

c. (W)Ф(WW)Ф(WW)Ф(WW)Ф

25%

d. (W)Ф(WWW)Ф(W)Ф(WW)Ф

14%

e. Others

7%

Table 10: Subject NP branching with an appositive containing four PWs (la única maestra de la escuela secundaria, ‘the only high school teacher’).

Table 11 presents the results for cases of the longest syntactic branching of subject NPs containing APs and PPs in the data. The dominant pattern that emerges can be seen in (a). This pattern groups the eight total PWs into four PPHs that each house two PWs. Therefore, there is equal weight distribution across all PPHs of the utterance. Prosodic boundaries also coincide with major syntactic boundaries, since there is a PPH boundary after oeste (‘west’) and bolígrafo (‘pen’). Pattern (b), which is not nearly as common as (a), also demonstrates such a match between prosody and syntax. In (b), the final three PPHs of the utterance are equally distributed for weight, while the first two PWs are individually phrased. By focusing on (a) and (b), it becomes apparent that balance of PPH weight across the entire utterance is desirable.

Syntactic structure:
[[[N AP]n’[P[[N AP]n’[P[N AP]np]pp]np]pp]np [V[N]np]vp]cp/ip
(8 PWs)

Example: La Bárbara dormilona de la ciudad húmeda en la costa oeste lleva el bolígrafo (‘The sleepy Barbara from the humid city on the west coast carries the pen’)

Phrasing Pattern

Frequency (%)

a. (WW)Ф(WW)Ф(WW)Ф(WW)Ф

56%

b. (W)Ф(W)Ф(WW)Ф(WW)Ф(WW)Ф

25%

c. Others

19%

Table 11: The longest examples of subject NP branching via APs and PPs.

When syntactic branching of appositive NPs is carried to an extreme, numerous phrasing patterns emerge, as shown in Table 12. Patterns (b-e) contain appositives that are recursive IPs due to the isolation of Carmen in its own PPH. Once again, prosodic boundaries and high-level syntactic boundaries coincide in all cases, as there are PPH boundaries after Bilbao (‘Bilbao’) and bicicleta (‘bicycle’). The PWs belonging to the recursive IP in cases (b-d) are parsed into two or three PPHs. When there are two PPHs, as in (b) and (c), one contains two PWs and the other has three PWs, with the difference being their order. Patterns (d) and (e) both have recursive IPs of three PPHs that are distinguished by weight balance between the first and second PPHs or between the second and third PPHs. Pattern (e) shows a more drastic decrease in prosodic weight, since there is a decrease of two PWs between the first and second PPHs of the recursive IP. Patterns (a) and (f) do not treat the appositive NP as a recursive IP. In (a), this leads to a fewer number of PPHs across the utterance. The entire subject NP is composed of two PPHs that are balanced for prosodic weight, followed by the VP in another PPH. Comparing the frequencies of (a) and (f) uncovers that speakers find it better to maintain more weight balance (especially in adjacent PPHs). This balance is favored over gradually decreasing PPH length, for example, by placing PPs in their own PPHs.

Syntactic structure: 
[[[N]n’[[AP N]n’[P[[N AP]n’[P NP]pp]np]pp]np]np [V[N]np]vp]cp/ip 
(8 PWs)

Example: Carmen la única maestra de la escuela secundaria en Bilbao lava la bicicleta (‘Carmen the only teacher at the high school in Bilbao washes the bicycle’)

Phrasing Pattern

Frequency (%)

a. (WWW)Ф(WWW)Ф(WW)Ф

22%

b. (W)Ф(WW)Ф(WWW)Ф(WW)Ф

22%

c. (W)Ф(WWW)Ф(WW)Ф(WW)Ф

14%

d. (W)Ф(WW)Ф(WW)Ф(W)Ф(WW)Ф

11%

e. (W)Ф(WWW)Ф(W)Ф(W)Ф(WW)Ф

11%

f. (WWW)Ф(WW)Ф(W)Ф(WW)Ф

8%

g. Others

12%

Table 12: Longest examples of subject NP branching via an appositive NP, which contains five PWs (la única maestra de la escuela secundaria en Bilbao, ‘the only high school teacher in Bilbao’).

4.3 Syntactic Branching of the Direct Object NP with an Unstressed Determiner

Tables 13 to 16 break down phrasing patterns of sentences showing increased syntactic branching of a direct object NP containing an unstressed determiner. The total number of PWs in these sets ranges from three to seven. Since those utterances with three PWs are also simple SVO utterances, they were included in the results in Table 1. Therefore, Tables 13-16 deal solely with cases of branching beyond the simple SVO (examples b-e in sets 4 and 6 in Appendix B).
Table 13 reveals two major phrasing patterns in cases of four total PWs. The most frequent pattern prefers equal weight distribution of the four PWs across two PPHs. However, this balance leads to a compromise because it does not allow a prosodic boundary to be placed at the right edge of the subject NP maximal projection. In opposing fashion, pattern (b) highly prioritizes the overlap between syntax and prosody that is violated in (a). While doing so, (b) creates an uneven distribution of PWs across two PPHs, instead preferring PPHs showing an increase from one PW to three PWs. Pattern (c) is similar to (b) in that syntax is respected. However, (b) increases PPH length across two PPHs, while (c) adds an additional PPH by individually phrasing the utterance final adjective. As such, (c) creates a symmetrical distribution of PWs.

Syntactic Structure: [[N]np[V[N[A]ap]np]vp]cp/ip
(4 PWs)

Example: Carmen habló un dialecto nuevo (‘Carmen spoke a new dialect’)

Phrasing Pattern

Frequency

a. (WW)Ф(WW)Ф

40%

b. (W)Ф(WWW)Ф

34%

c. (W)Ф(WW)Ф(W)Ф

14%

d. Others

12%

Table 13: Syntactic branching of the direct object NP through an AP. This type of utterance has four total PWs.

When branching of the direct object NP is increased to form sentences of five PWs in length, two main phrasing patterns emerge once again. These two predominant trends are given in Table 14. Pattern (a) places prosodic boundaries at the right edges of both the subject NP and VP projections. By isolating the PP branching from the direct object, (a) symmetrically distributes PWs in its three PPHs. In contrast with the most frequently observed pattern, (b) balances the prosodic weight of its first two PPHs, followed by a decrease in the final PPH to one PW. Due to this distribution, the right edge of the first PPH and the right edge of the subject NP do not coincide.

Syntactic Structure: [[N]np [[V[N[A]ap]np]v’ [P[N]np]pp]vp]cp/ip
(5 PWs)

Example: Carmen habló un dialecto nuevo con sus colegas (‘Carmen spoke a new dialect with her colleagues’)

Phrasing Pattern

Frequency

a.(W)Ф(WWW)Ф(W)Ф

51%

b.(WW)Ф(WW)Ф(W)Ф

37%

d. Others

12%

Table 14: Syntactic branching of the direct object NP via an AP and a PP. These types of utterances have five total PWs.

Two common phrasing patterns are present when branching leads to a direct object NP containing four PWs. Of the results for these utterances, shown in Table 15, the preferred trend is illustrated in (a). This parsing individually phrases the subject and then divides the constituents of the entire VP into two PPHs of three and two PWs. In terms of PPH length, there is an initial increase between the first two PPHs, followed by a decrease from the second to the third. In this case, PPHs are not balanced for weight, but there are prosodic boundaries at the right edge of the subject NP and VP maximal projections. Pattern (b) is in conflict with (a) because the six PWs of the utterance are equally distributed in sets of two across three PPHs, but there is no PPH boundary after the subject NP projection. The trend observed in (c) is similar to (b). However, in (c) the first four PWs are grouped into one phrase rather than splitting up into two equally balanced PPHs. The long initial PPH in (c) results in a lack of cooperation with both syntax and weight balance. Therefore, it appears that (c) is lower in frequency because it combines two potentially problematic issues seen in (a) and (b).

Syntactic Structure: [[N]np [[V[N[A]ap]np]v’ [P[N[A]ap]np]pp]vp]cp/ip
(6 PWs)

Example: Carmen habló un dialecto nuevo con sus colegas españolas (‘Carmen spoke a new dialect with her Spanish colleagues’)

Phrasing Pattern

Frequency

a. (W)Ф(WWW)Ф(WW)Ф

46%

b. (WW)Ф(WW)Ф(WW)Ф

32%

c. (WWWW)Ф(WW)Ф

12%

d. Others

10%

Table 15: Syntactic branching of the direct object NP via APs and a PP. These utterances have six total PWs.

When branching is increased by one more PW from the type of utterance in Table 15 the two dominant patterns in Table 16 surface. The most frequent trend, seen in (a), fails to distribute PWs across PPHs in a balanced fashion. The first and last of the four PPHs are the same length, but no adjacent PPHs are identical in length. There is an initial increase in length between the first two PPHs, followed by two consecutive decreases between the last two sets of adjacent PPHs. In (a), prosodic boundaries are manifested after the first PW of the utterance, the right edge of the subject NP, and after the last PW of the utterance, which is the right edge of the high-level VP projection. The results in Table 16 are similar to those seen in Table 15 in that the two most commonly produced phrasing patterns satisfy different conditions. In Table 16, (b) balances the weight of the first three PPHs, with each one containing two PWs, and then isolates the final PP in its own PPH. Since there are an odd number of total PWs, all PPHs being equal in length is an unlikely scenario (barring an unexpected case in which all PPHs contain just one PW each). By placing the uneven PPH in final position, (b) provides further evidence in favor of creating weight balance between two adjacent PPHs, as we see when evaluating the first and second PPHs, and the second and third PPHs. However, when such balance is present, the ability to have a prosodic boundary in the same location as the right edge of the subject NP is sacrificed. Once again, the pattern with the third highest frequency, (c), differs from (b) in that a longer PPH is present in initial position. By doing this, both weight balance and correspondence with syntax are disrupted.

Syntactic Structure:
[[N]np [[V[N[A]ap]np]v’[P[[N AP]n’[P[N]np]pp]np]pp]vp]cp/ip
(7 PWs)

Example: Carmen habló un dialecto nuevo con sus colegas españolas de la universidad (‘Carmen spoke a new dialect with her Spanish colleagues from the university’)

Phrasing Pattern

Frequency

a.(W)Ф(WWW)Ф(WW)Ф(W)Ф

44%

b.(WW)Ф(WW)Ф(WW)Ф(W)Ф

30%

c.(WWWW)Ф(WW)Ф(W)Ф

10%

d. Others

14%

Table 16: Syntactic branching of the direct object NP via APs and PPs. These utterances have a total of seven PWs.

4.4 Syntactic Branching of the Direct Object NP with a Stressed Determiner

Tables 17 to 20 present frequencies of phrasing patterns found in utterances containing syntactic branching of the direct object NP with a stressed determiner such as varios (‘various’). As was the case in the previous section, branching is achieved through APs and PPs, with utterance length ranging from five to eight total PWs.
The results in Table 17, for utterances that are five PWs in length, show that pattern (a) distinguishes itself from the others due to its higher frequency. By failing to place a prosodic boundary after the first PW of the utterance and disrespecting a major syntactic boundary, (a) creates effective weight balance. This is because there is an odd number of total PWs and only a one PW difference in length between the two PPHs. The organization of PPHs in (a) also shows a preference for early production of a shorter PPH, which is then followed by a longer PPH. The second most frequently observed pattern, (b), favors PPHs that are symmetrically arranged, with a medial PPH of three PWs surrounded by two PPHs each housing only the utterance initial subject and the utterance final adjective. This leads to a better match between prosodic and syntactic boundaries. Furthermore, the consequence of the harmony between syntax and prosody seen in (c) is a very unbalanced distribution of PWs demonstrating a rightward increase in PPH length. Pattern (d) cooperates with major syntactic boundaries, but also includes a PPH boundary after the verb, which is not the right edge of a maximal projection. In this case, the first two PPHs maintain weight balance, followed by an increase of two PWs in the final PPH of the string. Pattern (e) similarly balances the first two PPHs of its three PPH sequence, while placing the shortest PPH in final position. An overall comparison of patterns (a-e) in Table 17 suggests that the preference is to have fewer total PPHs, and to balance them as best possible. These two features take precedence over matching prosodic boundaries to those of high-level syntactic projections.

Syntactic Structure: [[N]np[V[N[A]ap]np]vp]cp/ip
(5 PWs)

Example: Carmen habló varios dialectos nuevos (‘Carmen spoke various new dialects’)

Phrasing Pattern

Frequency

a. (WW)Ф(WWW)Ф

36%

b. (W)Ф(WWW)Ф(W)Ф

17%

c. (W)Ф(WWWW)Ф

12%

d. (W)Ф(W)Ф(WWW)Ф

11%

e. (WW)Ф(WW)Ф(W)Ф

10%

f. Others

14%

Table 17: Syntactic branching of the direct object NP through an AP. Five total PWs are present in such utterances.

When syntactic branching of the direct object NP is increased to form utterances with six total PWs, as displayed in Table 18, two phrasing patterns are observed at much higher rates than others. The dominant trend is (a), in which there are three total PPHs that are prosodically out of balance. There is also no prosodic boundary after the first PW, which represents the right edge of the subject NP maximal projection. The lack of weight balance of PPHs is forced by the adjunct PP that isolates itself in a PPH in utterance final position. This individual phrasing suggests that adjuncts are more prone to prosodic separation from the rest of the utterance. Based on the patterns shown in (b) and (c), it is apparent that such individual phrasing of the PP at the end of the utterance is preferred. Pattern (b) forms a symmetrically distributed sequence of PPHs by placing its three PPH boundaries at the right edges of the subject NP and entire VP, as well as after the AP constituent within the VP. Finally, pattern (c) increases the total number of PPHs by splitting the first two PWs of the sentence into two individual PPHs. The prosodic boundary dividing these two PPHs coincides with the right edge of the subject NP. Even though the first two PPHs are prosodically balanced, this pattern is not very frequently observed. This low rate indicates that preferred patterns seem to have fewer PPHs than those that are deviant.

Syntactic Structure: [[N]np[[V[N[A]ap]np]v’[P[N]np]pp]vp]cp/ip
(6 PWs)

Example: Carmen habló varios dialectos nuevos con sus colegas (‘Carmen spoke various new dialects with her colleagues’)

Phrasing Pattern

Frequency

a.(WW)Ф(WWW)Ф(W)Ф

47%

b.(WW)Ф(WW)Ф(WW)Ф(W)Ф

31%

c.(W)Ф(W)Ф(WWW)Ф(W)Ф

10%

d. Others

14%

Table 18: Syntactic branching of the direct object NP through an AP and a PP. Six total PWs are present in such utterances.

In Table 19, the most frequent phrasing pattern, (a), distributes PWs symmetrically by surrounding a PPH of three PWs with two PPHs each containing two PWs. This trend also maintains the best weight balance possible with an odd number of PWs. This is done at the cost of not attaining correspondence between the first PPH boundary and the right edge of the subject NP. The second most common parsing of phrases respects major syntactic boundaries but has an increase of three PWs between the first and second PPHs and a decrease of two PWs between the medial and final PPHs. Furthermore, the well-balanced pattern (c) is very infrequent. In fact, this pattern is both well balanced and respectful of syntax. Comparing the more frequent (a) and (b) to (c) and (d) proves that speakers prefer phrasing patterns with fewer PPHs, even when structures with more PPHs respect the overlap between syntax and prosody and maintain some degree of weight balance.

Syntactic Structure: [[N]np [[V[N[A]ap]np]v’[P[N[A]ap]np]pp]vp]cp/ip
(7 PWs)

Example: Carmen habló varios dialectos nuevos con sus colegas españolas (‘Carmen spoke various new dialects with her Spanish colleagues’)

Phrasing Pattern

Frequency

a. (WW)Ф(WWW)Ф(WW)Ф

48%

b. (W)Ф(WWWW)Ф(WW)Ф

23%

c. (W)Ф(WW)Ф(WW)Ф(WW)Ф

9%

d. (W)Ф(W)Ф(WWW)Ф(WW)Ф

7%

e. Others

13%

Table 19: Syntactic branching of the direct object NP through APs and a PP. Such utterances have seven total PWs.

As syntactic branching of the direct object NP in utterances with stressed determiners reaches its maximum, the phrasing patterns observed provide further support for isolating PPs in their own PPHs. In this case the PPs in question are adjuncts and modifiers. This is the case in all of the results shown in Table 20, where the final three PWs compose two PPs and thus two PPHs. The differences among patterns concern the phrasing of the first five PWs of the utterance. Both (a) and (c) do not place a prosodic boundary after the first PW. Although this trait is shared, (a) is preferred because of its relative balance of five PWs across two PWs. At first glance, one would think that (c) would be more common than (a) due to its ability to capture the first five PWs in one PPH. However, the fact that (a) is observed at a higher rate suggests that the idea of ‘fewer PPHs’ has a lower limit and that the number of PWs per PPH has an upper limit. Upon comparing (b) and (d), we see that (b) is probably more frequent because it divides the first five PWs into two PPHs instead of three. Even though (d) balances its first two PPHs for weight and (b) has an increase in three PWs between the first and second PPHs, the latter dominates the former. Finally, when comparing (b) and (c), evidence emerges supporting a four PW maximum per PPH. Pattern (b), owning a PPH of four PWs, is the second most common pattern, while (c) contains a PPH of five PWs and is much less frequent.

Syntactic Structure: 
[[N]np [[V[N[A]ap]np]v’[P[[N AP]n’[P[N]np]pp]np]pp]vp]cp/ip
 (8 PWs)

Example: Carmen habló varios dialectos nuevos con sus colegas españolas de la universidad (‘Carmen spoke various new dialects with her Spanish colleagues from the university’)

Phrasing Pattern

Frequency

a. (WW)Ф(WWW)Ф(WW)Ф(W)Ф

39%

b. (W)Ф(WWWW)Ф(WW)Ф(W)Ф

24%

c. (WWWWW)Ф(WW)Ф(W)Ф

11%

d. (W)Ф(W)Ф(WWW)Ф(WW)Ф(W)Ф

10%

e. Others

16%

Table 20: Syntactic branching of the direct object NP through APs and PPs. Such utterances contain eight PWs total.

4.5 Syntactic Branching of both the Subject NP and the Direct Object NP

The final results section discusses cases of syntactic branching of NPs located on both sides of the verb. Branching is realized once again through the use of APs and PPs and leads to utterances of seven, nine and eleven PWs in length. Especially in the latter two lengths, PPH boundaries are often cued by pauses because speakers can only produce so many consecutive items without any type of rest. If such a physical restriction were clearly proven in future work, it would potentially strengthen the hypothesis that there exists a maximum number of PWs that can be produced in a single PPH. Phrasing patterns of double branching scenarios are presented in Tables 21 through 23.
When there are a total of seven PWs, one dominant and two secondary phrasing patterns are articulated. In all three observed patterns in Table 21, prosodic boundaries are located after the fourth and last PWs of the utterance. These junctures represent the right edges of both the subject NP and VP projections. Another similarity between all three patterns is that all three PWs belonging to the VP are housed in a PPH in final position. Therefore, the major differences deal with the parsing of PPHs containing members of the subject NP. The dominant pattern, (a), which is produced three times as much as (b) and (c), equally distributes the first four PWs into two PPHs. This leads to an adjective and a noun being in one PPH, followed by another PPH containing a modifier PP. This separate phrasing of PP elements goes along with results shown in previous tables as well. Pattern (b) employs three PPHs over the first four PWs, with the first two being balanced for weight, followed by an increase by one PW in the third. Based on ideas emerging in the previous section, this phrasing is probably not as frequent due to the fact that it has more total PPHs. Since there are more PPHs in (b) than (a), there is also more possibility for a lack of prosodic weight balance, which is what is seen in the adjacent second and third PPHs of (b). Furthermore, (c) follows the example set by (a) and divides the first four PWs into two PPHs. However, it is less balanced, showing an increase in length between the two PPHs from one PW to three PWs. The placement of a prosodic boundary after the first PW in (c) also suggests that separating an adjective from the noun it complements is not desirable. Finally, comparing the overall more common distribution of the first four PWs into two PPHs, as shown by (a) and (c), to the distribution in (b) indicates that the target average of PWs per PPH is two. Even though (c) has uneven lengths in its first two PPHs, dividing the four total PWs by its two PPHs shows that the average length is two PWs/PPH. Pattern (a) perfectly satisfies this average and therefore is the most frequently produced.

Syntactic Structure: [[[N AP]n’[P[N AP]np]pp]np [V[N[A]ap]np]vp]cp/ip
(7 PWs)

Example: La Bárbara testaruda del valle central lleva el bolígrafo negro (‘The stubborn Barbara from the central valley carries the black pen’)

Phrasing Pattern

Frequency

a. (WW)Ф(WW)Ф(WWW)Ф

60%

b. (W)Ф(W)Ф(WW)Ф(WWW)Ф

20%

c. (W)Ф(WWW)Ф(WWW)Ф

20%

Table 21: Syntactic branching of both subject and direct object NPs through the addition of an AP and a PP on the subject and an AP on the direct object. Such utterances are seven PWs in length.

As Table 22 shows, when syntactic complexity is increased in the constituents of both the subject NP and VP, phrasing patterns (a) and (b) are the most frequently observed. Both are similar in that they are constructed from three PPHs of two PWs in length, and one PPH of three PWs in length. However, they differ in that the heaviest PPH is located earlier in (a), where it is in second position as opposed to third position. Both of these patterns also have one pair of adjacent PPHs balanced for weight, and do not have any increase or decrease in PWs per PPH of more than one. Therefore, they are overall well balanced, for an utterance type with an odd number of PWs. They also comply, for the most part, with the apparent average number of PWs per PPH of two that was mentioned with regard to the preceding table. Concerning alignment of major syntactic and prosodic boundaries, there is a major difference between patterns (a) and (b). The former respects overlap between the two types of boundaries by locating PPH boundaries after the fifth and last PWs. On the other hand, by putting a prosodic boundary after the fourth PW and then after the seventh PW, (b) fails to align a prosodic boundary with the right edges of high-level syntactic projections. The fact that (b) is well balanced makes it just as frequent as (a). In terms of other parsings of phrases, (c) and (e) are less frequent because they have a higher number of total PPHs. The fact that (e) is infrequent, even though it has two adjacent pairs of PPHs that are well balanced, is evidence supporting the idea that having fewer PPHs, while maintaining an average length of around two PWs, takes precedence over local prosodic weight balance. Patterns (d) and (f) both have the same number of PPHs as (a) and (b), but contain a wider range of overall PPH length leading to less weight balance. Evidence of prosodic well-formedness being prioritized higher than overlap between prosodic and syntactic boundaries by these speakers is found when comparing (b) to (c-f).

Syntactic Structure: 
[[[N AP]n’[P[[N AP]n’[P NP]pp]np]pp]np [V[[N AP]n’[P NP]pp]np]vp]cp/ip
                                                       (9 PWs)

Example: La Bárbara preciosa de la ciudad húmeda en la costa lleva el bolígrafo negro de Manuel (‘The precious Barbara from the humid city on the coast carries Manuel’s black pen’)

Phrasing Pattern

Frequency

a. (WW)Ф(WWW)Ф(WW)Ф(WW)Ф

34%

b. (WW)Ф(WW)Ф(WWW)Ф(WW)Ф

34%

c. (WW)Ф(W)Ф(WW)Ф(WWW)Ф(W)Ф

7%

d. (W)Ф(W)Ф(WWW)Ф(WWWW)Ф

7%

e. (W)Ф(W)Ф(WWW)Ф(WWW)Ф(W)Ф

7%

f. (WW)Ф(WWW)Ф(WWW)Ф(W)Ф

7%

g. Others

4%

Table 22: Syntactic branching of the subject and direct object NP using APs and PPs. Such utterances contain a total of nine PWs.

The type of utterance shown in Table 23 is the longest example of syntactic branching produced by speakers in this study. The subject NP projection is composed of six PWs, while the VP contains five PWs. Interestingly, although there are more total PWs in this structure than that of Table 22, there is a clear-cut favored pattern in Table 23. When taking into account the discussion that has developed to this point, we can firmly posit that (a) is superior because of its overall weight balance across PPHs. This pattern achieves such prosodic balance through a string of PPHs that attempt to stay as close to two PWs in length as possible. While seeking a prosodically uniform organization, (a) also locates PPH boundaries at the right edges of major syntactic constituents (as do the other patterns). These PPH boundaries are after the sixth and the last PWs of the utterance. In pattern (a), the one distinct PPH is chosen with a purpose in that it intends to uphold the previously reviewed desire to jointly phrase adjectives adjacent to the nouns they complement. Although pattern (b) divides the first four PPHs into two adjacently balanced sets, the fact that it separately phrases the first two PWs increases the total number of PPHs, thus making it less preferred. Lastly, speakers produce pattern (c) even less frequently than (b) because it adds more PPHs by individually phrasing the first four PWs of the utterance. The first half of (c) is well balanced, but such a condition sacrifices the average PPH length of two PWs.

Syntactic Structure:
[[[N AP]n’[P[[N AP]n’[P[N AP]np]pp]np]pp]np [V[[N AP]n’
[A PP]ap]np]vp]cp/ip
(11 PWs)

Example: La Bárbara dormilona de la ciudad húmeda en la costa oeste lleva el bolígrafo negro hecho en China (‘The sleepy Barbara from the humid city on the west coast carries the black pen made in China’)

Phrasing Pattern

Frequency

a. (WW)Ф(WW)Ф(WW)Ф(WWW)Ф(WW)Ф

56%

b. (W)Ф(W)Ф(WW)Ф(WW)Ф(WWW)Ф(WW)Ф

19%

c. (W)Ф(W)Ф(W)Ф(W)Ф(WW)Ф(WWW)Ф(WW)Ф

13%

d. Others

12%

Table 23: Syntactic branching of both the subject NP and direct object NP via APs and PPs. Such utterances have eleven total PWs.

5. Conclusions
This paper first provided an overview of theoretical concepts and previous literature relevant to the discussion of prosodic phrasing in lab speech in Spanish. Issues addressed in previous work served as a principle source of inspiration for the present study. Next, the selection of participants and data elicitation procedures were detailed. The methods section also included an in-depth account of data analysis procedures, with special attention to phonetic correlates of phrase boundaries. After phrasing divisions were determined in sentences with different types and degrees of syntactic branching, the empirical results of the patterns produced by speakers were presented in a series of tables. Upon taking frequencies into account, proposals for why certain patterns appear to be preferred were outlined based on specific observations relating to the distribution of PWs across PPHs.
On the whole, the study provides phrasing patterns deriving from sentences with increased and more complex syntactic branching than those of most previous work. The discussion of lab phrasing made constant reference to prosodic concepts such as weight balance, increases or decreases in the number of PWs in adjacent PPHs, symmetrical distributions of PPHs, and limits on PPH length. Other ideas that surfaced pertain to the interface between syntax and prosody, specifically, whether PPH boundaries align with those of high-level syntactic projections. Abundant evidence from the patterns illustrated in the tables points to the following generalization: as syntactic branching of utterances increases, prosodic concepts seem to play a more crucial role than syntactic conditions in determining the parsing of phrases. Syntax is respected in certain cases, especially in shorter utterances. In terms of syntactic restrictions, we noted that adjectives consistently phrase themselves with their noun complements, and that PPs are susceptible to surrounding phrase breaks. However, on the whole, a variety of prosodic conditions on length, balance and total PPHs appear to interact in justifying the presence of higher frequency patterns. This claim about the major role of prosody supports and extends on the previously cited contemporary studies on phrasing, and departs from the fundamental focus on syntax-prosody mapping that was developed in the 1980s. In particular, the findings are along the lines of the theoretical analyses of Prieto (2006) and Rao (2007) done in OT, in which prosodic constraints outrank syntactic ones in many contexts in two different dialects of Spanish.
In order to gain an even clearer understanding of the role of prosodic conditions, more phrasing data needs to be examined in an even greater variety of syntactic structures, speech styles, and dialects. Face (2003) proposes the need to understand the intonation of authentic, spontaneous speech by saying, “while lab speech is invaluable in intonational studies, it cannot be assumed that the intonation patterns produced in lab speech are an accurate representation of the intonation patterns of spontaneous speech” because of “intertwining factors in spontaneous speech that can affect the intonation of an utterance” (116). Such factors that are not present in scripted speech may include different levels of emotion based on the topic at hand, sudden changes in speech rate, turn-taking strategies, and pauses of varying lengths while expressing ideas.8 One crucial factor that has been cited as distinguishing lab speech from spontaneous speech is ‘deaccenting,’ which is the absence of F0 movement as a phonetic correlate of stress (Face 2003). However, as noted earlier, the application of the Nuclear Stress Rule in Romance (see Prieto 2005; Hualde in press; among others) allows prominence to be conveyed through final lengthening or some other cue in nuclear phrase position. Therefore, we can only clearly posit that deaccenting in prenuclear position of a phrase can lead to the loss of a PW.

Although Romance is relatively resistant to deaccenting when compared to Germanic languages, Face’s (2003) comments on Spanish spontaneous speech suggest that this phenomenon occurs more in high frequency verbs.9 Other differences between lab and spontaneous speech noted by Face (2003) are: less downstepping (i.e. F0 peak decay across utterances) in spontaneous speech, less final lowering and more F0 rises at the end of spontaneous speech declaratives, more erratic patterns of F0 peak alignment in spontaneous speech declaratives. In Rao’s (2008) empirical work on covariates affecting the odds of deaccenting in spontaneous speech in Barcelona Spanish, he finds that the following are characteristics of stressed words prone to deaccenting: i. fewer syllables; ii. globally frequent in Spanish; iii. grammatically adverbs or verbs; iv. recently repeated in discourse; v. located in initial or medial position of the PPH.
Comparing F0 contours and phrasing patterns from Rao (2008) to the data in the current investigation uncovers that there are notable differences between lab and spontaneous speech that should continue to be researched in future work on Spanish intonation. Although the figures provided below of spontaneous speech are not syntactically identical to those employed for data elicitation in this study, they are similar enough to reveal crucial distinctions. For example, when comparing the production of the utterance in Figure 5 (SV + PPs) to Figures 1-4, we observe that the F0 contour in the figure below is much flatter with a reduced pitch range. This type of pitch range in the spontaneous speech sample is attributed to faster, less deliberate speech, as well as a low level of emotion. The increased speech rate leads to fewer cues that signal PPH boundaries, which is why all four stressed items are in one PPH. The one salient evidence of a boundary is after the final word, where a short pause is present. This speaker was comparing and contrasting the uses of Spanish and Catalan in the state of Catalonia, and he did so in a relatively direct and unemotional way. The fact that emotion and rate of speech are being introduced emphasizes the extra variables that enter the discussion when describing the intonation of more authentic speech. Finally, the deaccenting of the words peligro (‘danger’) and Cataluña (‘Catalonia’), occurring at the conclusion of the idea reduces the number of PWs in the PPH. Though the flat F0 is an example of deaccenting, the expected nuclear stress on Cataluña makes its status as a PW subject to debate. However, prominence is absent from the prenuclear peligro. In lab speech, we expect four PWs from four stressed lexical items. However, only three PWs can be extracted from Figure 5, as peligro does not carry perceptual prominence via F0 movement. Upon examining neighboring PPHs, this reduction in PWs could have implications for the prosodic and eurhythmic principles discussed earlier.

Figure 5: The F0 contour of the utterance El castellano está en peligro en Cataluña (‘Spanish is in danger in Catalonia’). The utterance forms one PPH. The contour is relatively flat, with the final two highlighted stressed words lacking F0 movement. This lack of pitch excursion is a feature of spontaneous speech and not lab speech.

Another F0 contour from Rao’s (2008) spontaneous speech data is displayed in Figure 6 (Adverbial Phrase + PP). In this case, the speaker is a female, as evidenced by the higher overall pitch when compared to the male speaker in Figure 5. In Figure 6, the woman speaks in a somewhat animated fashion about the necessity to work for the entire summer rather than spend time with her friends. The increased emotion leads to a broader pitch range than the flatter contour in Figure 5. Furthermore, the cue to a PPH and IP boundary is a drastically reduced F0 in the final word of the phrase, verano (‘summer’). Such pitch reduction often indicates the conclusion of an idea. Though there is no tonal movement on the nuclear word, stress is achieved through lengthening effects, and thus we can still consider it a PW. Once again, when comparing Figure 6 to Figures 1-4, the faster rate of speech results in the absence of intermediate boundary cues as well as a reduced pitch range. Producing more words in a smaller time-frame correlates with reduced opportunities for the realization of prosodic boundaries as well as less time for ascending and descending pitch movement. In sum, this glimpse at data from spontaneous speech confirms the “intertwining factors” discussed by Face (2003). The factors pointed out in natural speech that are absent in lab speech, associated with deaccenting, emotion, faster rates of speech, and prosodic principles, should all be pursued in future work so that we can learn more about how native speakers of Spanish normally speak in daily conversations.

Figure 6: Deaccenting of the highlighted word verano (‘summer’) in the phrase Trabajando todo el verano (‘Working all summer’). The F0 contour is of one PPH that is in final position of a longer IP.

After introducing examples of spontaneous speech phrasing, it is important to recognize the limitations of the present lab speech data with regard to the psychological reality of speakers. The fact that many of the utterances used for data elicitation are not the most ‘natural’ sounding could have possibly impacted the participants’ production patterns. For example, the sentences with the masculine definite article el (‘the’) before a male name, and the feminine definite la (‘the’) before a female name are generally dialectally marked forms. The potential confounding factor is that if these forms have a pejorative connotation in Peninsular Spanish, the speakers may have been slightly flustered when reading, which may have led to less reliable intonation patterns in their speech. In addition, if some of the contrived sentences rarely appear in natural, everyday speech,the participants may not be fully capable of producing native-like F0 trends.
In sum, the results of this study are valuable because they expose the key role of prosodic and eurhythmic conditions in phrasing patterns. However, this body of data collected in a lab setting is not a complete representation of phrasing in Spanish because it only deals with scripted speech. By including a preliminary comparison of spontaneous speech phrasing to that of lab speech, we see that another contribution of this study is that it serves as a segue to further work on more authentic intonation and phrasing patterns in Spanish, as well as other languages.
Overall, since phrasing is an area of Spanish intonation with plenty of space for further exploration, this study hopes to inspire future research on why phrases are parsed in certain patterns in different dialects and speech styles, and how we can theoretically account for trends observed in data.

Notes
1. I would like to thank Travis Bradley, Robert Blake, and two anonymous reviewers for their constructive comments, which resulted in improvements on an earlier version of this paper. Any errors contained within the paper remain my own.

2. Variations of this hierarchy exist. For versions related to the Autosegmental Metrical model, see Ladd (1996), Nibert (2000) and Gussenhoven (2002).

3. Some other investigations of phrasing decisions in Romance include Sandalo and Truckenbrodt (2003) for Brazilian Portuguese, Prieto (2005) for Catalan, Frota (2000) for European Portuguese, Ghini (1993) for Italian, and Hualde (2003) and Toledo (2007) for Spanish.

4. I sincerely thank Pilar Prieto for granting me access to this facility.

5.Thanks to Raúl Aranovich for the fruitful discussion about the syntactic structure of appositives. In this discussion, he explained that appositives can be syntactically analyzed similarly to other examples of subject NP branching.The issue of whether appositive NPs require isolation through the use of commas, which in turn alters prosodic structure, is the reason why the second and third sets of utterances in Appendix B are separated.

6. Thanks to José Ignacio Hualde for the discussion on why he believes that final lengthening is the strongest indicator of PPH boundaries.

7. In longer branching cases, it may be the case that PPH boundaries are inserted to break up discourse into eight syllable chunks, due to historical trends in Spanish versification (Blake p.c.). Table 7 provides some evidence for this idea, as the most frequent pattern places a boundary after testaruda. Table 10 also demonstrates similar parsing. However, other patterns in tables such as 5 seem to contradict this suggestion. In order to fully evaluate the hypothesis, the data would need to be examined with attention on syllables rather than stressability of words. This concept merits future consideration.

8. See Rao (2006) for the pragmatic implications of the intonation of unscripted (but not truly spontaneous) speech in Peninsular Spanish.

9. For information on deaccenting in Romance, see Bard and Aylett (1999) and Avesani and Vayra (2005) for Italian, Gussenhoven (2004) for French, Ladd (1996) for Romanian and Italian, and Cruttenden (1993), García-Lecumberri (1995) and Hualde (in press) for Spanish, among others. For deaccenting in Germanic languages, see Hirschberg (1993) and Terken and Hirschberg (1994), among others.

Appendix A
CUESTIONARIO DE USO DE IDIOMAS
(‘LANGUAGE HISTORY QUESTIONNAIRE’)

Nombre (‘Name’): _______________________
Edad (‘Age’): _________________________
Fecha (‘Date’): ____________
Lugar de nacimiento (‘Birthplace’): ____________________
Años de residencia en Barcelona (o en Cataluña) (‘Years of residence in Barcelona (or in Catalonia)’): ___________
Años de residencia fuera de Barcelona (o Cataluña) (‘Years of residence outside of Barcelona (or Catalonia)’): ___________
¿Dónde y por cuánto tiempo? (‘Where and for how long?’) __________________________________
Uso del español y del catalán (‘Use of Spanish and Catalan’):
¿Qué idioma ha usado con más frecuencia desde su niñez (o ha usado los dos con más o menos la misma frecuencia)? (‘What language have you use more frequently since your childhood (or have you used both with more or less the same frequency)?’)
¿Qué idioma usa con su familia? (‘What language do you use with your family?’)
¿Qué idioma usa más con sus amigos en reuniones sociales, fiestas, etc.? (‘What language do you use more with your friends in social gatherings, parties, etc.?)
¿Qué idioma usa más para saludar a los amigos en la calle? (‘What language do you use to greet friends on the street?’)
¿Qué idioma usa más en la universidad/ el trabajo? (‘What language do you use more at the university/ at work?)
¿Cuál de los dos idiomas usa con más frecuencia en su vida hoy en día? (‘Which of the two languages do you use more frequently in your daily life?’)
¿Habla otros idiomas además del español y el catalán? (‘Do you speak other languages besides Spanish and Catalan?’)

Appendix B
1. Simple SVO utterances
a. Javier lava muchos platos
‘Javier washes many plates’
b. José bebe varios refrescos
‘José drinks various refreshments ’
c. Marcelo cocina mucho pollo
‘Marcelo cooks a lot of chicken’
d. Carmen habla varios dialectos
‘Carmen speaks various dialects’
e. Bárbara fuma doce cigarrillos
‘Barbara smokes twelve cigarettes’
2. Increased syntactic branching of the subject NP via APs and PPs
a. El Marcelo cubano manda el regalo
‘The Cuban Marcelo sends the gift’
b. El Marcelo mexicano inteligente manda el regalo
‘The intelligent Mexican Marcelo sends the gift’
c. El Marcelo nicaragüense de la costa este manda el regalo
‘The Nicaraguan Marcelo from the east coast sends the gift’
d. El Marcelo costarricense de la playa Manuel Antonio manda el regalo
‘The Costa Rican Marcelo from Manuel Antonio beach sends the gift’
e.El Marcelo costarricense de la ciudad de San José en la mesa central manda el regalo
‘The Costa Rican Marcelo from the city of San José in the central region sends the gift’
3. Increased syntactic branching of the subject NP via appositive NPs
a. Carmen la maestra lava la bicicleta
‘Carmen the teacher washes the bicycle’
b. Carmen la maestra nueva lava la bicicleta
‘Carmen the new teacher washes the bicycle’
c. Carmen la maestra del segundo grado lava la bicicleta
‘Carmen the second grade teacher washes the bicycle’
d. Carmen la única maestra de la escuela secundaria lava la bicicleta
‘Carmen the only teacher at the high school washes the bicycle’
e. Carmen la única maestra de la escuela secundaria en Bilbao lava la bicicleta
‘Carmen the only teacher at the high school in Bilbao washes the bicycle’
4. Increased syntactic branching of the direct object NP
a. José bebió un refresco
‘José drank a refreshment’
b. José bebió un refresco frío
‘José drank a cold refreshment’
c. José bebió un refresco frío en la casa
‘José drank a cold refreshment in the house’
d. José bebió un refresco frío en la casa bonita
‘José drank a cold refreshment in the pretty house’
e. José bebió un refresco frío en la casa bonita de su abuela
‘José drank a cold refreshment in his grandmother’s pretty house’
5. Syntactic branching of both the subject and direct object NPs
a. La Bárbara rubia lleva el bolígrafo
‘The blond Barbara carries the pen’
b. La Bárbara morena del valle lleva el bolígrafo
‘The brunette Barbara from the valley carries the pen’
c. La Bárbara testaruda del valle central lleva el bolígrafo negro
‘The stubborn Barbara from the central valley carries the black pen’
d. La Bárbara preciosa de la ciudad húmeda en la costa lleva el bolígrafo negro de
Manuel
‘The precious Barbara from the humid city on the coast carries Manuel’s black pen’
e.La Bárbara dormilona de la ciudad húmeda en la costa oeste lleva el bolígrafo negro hecho
en China
‘The sleepy Barbara from the humid city on the west coast carries the black pen made in
China’
6. Branching of the direct object NP with an unstressed determiner
a. Javier escribió una carta
‘Javier wrote a letter’
b. Javier escribió una carta larga
‘Javier wrote a long letter’
c. Javier escribió una carta larga a sus amigos
‘Javier wrote a long letter to his friends’
d.Javier escribió una carta larga a sus amigos griegos
‘Javier wrote a long letter to his Greek friends’
e.Javier escribió una carta larga a sus amigos griegos de Atenas
‘Javier wrote a long letter to his Greek friends from Athens’
7. Branching of the direct object NP with a stressed determiner
a. Javier escribió muchas cartas
‘Javier wrote many letters’
b.Javier escribió muchas cartas largas
‘Javier wrote many long letters’
c. Javier escribió muchas cartas largas a sus amigos
‘Javier wrote many long letters to his friends’
d. Javier escribió muchas cartas largas a sus amigos griegos
‘Javier wrote many long letters to his Greek friends’
e. Javier escribió muchas cartas largas a sus amigos griegos de Atenas
‘Javier wrote many long letters to his Greek friends from Athens’

References
Avesani C. & Vayra, M. (2005). Accenting deaccenting and information structure in Italian dialogues. Proceedings of the 6th DIGdial Workshop on Discourse and Dialogue(pp. 19-24).

Bard, E. & Aylett, M. (1999). The dissociation of deaccenting, givenness, and syntactic role in spontaneous speech.Proceedings of the XIVth International Congress of Phonetic Sciences(pp. 1753-1756).

Beckman, M., Díaz-Campos, M., McGory, J. & Morgan, T.(2002).Intonation across Spanish, in the Tones and Break Indices framework.Probus,9: special issue on intonation in Romance.

Burgess, C. & Conley, P. (1999). Representing proper names and objects in a common semanticspace: A computational model. Brain and Cognition,40, 67-70.

Carr, P. & Durand, J. (2004). General American and New York City English. La Tribune Internationale des Langues Vivantes, 36, 56-69.

Chomsky, N. (1965).Aspects of the theory of syntax. Cambridge: The MIT Press.

Chomsky, N. & Halle, M. (1968). The sound patterns of English.Cambridge, MA.: MIT Press.

Contreras, H. (1964). ¿Tiene el español un acento de ntensidad?. ”Boletín del Instituto de Filología de la Universidad de Chile,16, 237-239.

Cruttenden, A. (1993). The deaccenting and reaccenting of repeated lexical items. Proceeding of ESCA Workshop on Prosody(pp. 16-19).

D’Imperio, M., Elordieta, G., Frota, S., Prieto, P., & Vigário, M.(2005). Intonational phrasing in Romance: The role of syntactic and prosodic structure. In S. Frota, M. Vigário & M.J. Freitas (Eds.),Prosodies (pp. 59-98).The Hague: Mouton de Gruyter.

Elordieta, G., Frota, S., Prieto, P.,& Vigário, M. (2003). Effects of constituent length and syntactic branching on intonational phrasing in Ibero-Romance. Proceedings of the XVth International Congress of Phonetic Sciences (pp. 487-490).

Face, Timothy. (2001). Intonational marking of contrastive focus in Madrid Spanish. Unpublished doctoral dissertation, The Ohio State University, Colombus.

Face, T. (2003).Intonation in Spanish declaratives: Differences between lab speech and spontaneous speech. Catalan Journal of Linguistics,2, 115-131.

Frota, S. (2000).Prosody and focus in European Portuguese: Phonological phrasing and intonation. New York: Garland.

Frota, S., D’Imperio, M., Elordieta, G., Prieto, P., & Vigário, M. (in press). The phonetics and phonology of intonational phrasing in Romance. In P. Prieto, J. Mascaró & M.J. Solé (Eds.), Segmental and Prosodic Issues in Romance Phonology.Amsterdam /Philadelphia : John Benjamins.

García-Lecumberri, M. L. (1995).Intonational signaling of information structure in Englishand Spanish: A comparative study. London: University of London.

Garrido, J. (1996). Modelling Spanish intonation for text-to-speech applications. Unpublished doctoral dissertation, Universitat Autònoma de Barcelona, Barcelona, Spain.

Garrido, J.M., Llisterri, J., de la Mota, C., & Ríos, A. (1993). Prosodic differences in reading style: Isolated vs. contexualized sentences. Eurospeech’93. 3rd European Conference on
Speech Communication and Technology (pp. 573-576). Berlin, Germany.

Garrido, J.M., Llisterri, J., de la Mota, C., Marín, R., & Ríos, A. (1995). Prosodic markers at syntactic boundaries in Spanish. Proceedings of the XIVth International Congress of Phonetic Sciences (pp. 370-373).San Francisco.

Ghini, M. (1993). Ф-formation in Italian: A new proposal. In Carrie Dyck (Ed.),Torontoworking papers in linguistics (pp. 41-78). Toronto: Department of Linguistics, University of Toronto.

Gussenhoven, C. (2002). Phonology of intonation. Glot International, 6 (9/10), 271-284.

Gussenhoven, C. (2004). The phonology of intonation. Cambridge: Cambridge University Press.

Hirschberg, J. (1993). Pitch accent in context: Predicting intonational prominence from text. Artificial Intelligence, 63(1-2), 305-340.

Hualde, J.I. (2003). El modelo métrico y autosegmental. In Pilar Prieto (Ed.), Teorías de la entonación (pp. 155-184). Barcelona: Ariel.

Hualde, J.I. (in press). Stress removal and stress addition in Spanish. Portuguese Journal of Linguistics.

Jun, S. (2005). Prosodic typology. In Sun-Ah Jun (Ed.), Prosodic typology: The phonology of
intonation and phrasing (pp. 430-458). Oxford: Oxford University Press.
Ladd, R. (1986). Intonational phrasing: The case for recursive prosodic structure. Phonology Yearbook,3, 311-340.
Ladd, R. (1996). Intonational phonology. Cambridge: Cambridge University Press.

Liberman, M., & Sproat, R. (1992) The stress and structure of modified noun phrases in English. In I.A. Sag & A. Szabolcsi (Eds.), Lexical Matters (pp. 131-181). Stanford, CA: Center for the Study of Language and Information.

Navarro Tomás, T. (1944). Manual de entonación española. New York: Hispanic Institute in the United States.

Navarro Tomás, T. (1964). La medida de la intensidad. Boletín del Instituto de Filología de la Universidad de Chile,16, 231-235.
Nespor, M.& Vogel, I. (1986). Prosodic phonology. Dordrecht: Foris.

Nibert, H. (1999). A perception study of intermediate phrasing in Spanish intonation. In J. Gutiérrez-Rexach & F. Martínez-Gil (Eds.), Advances in Hispanic Linguistics (pp. 231-247). Somerville, MA: Cascadilla Press.

Nibert, H. (2000). Phonetic and phonological evidence for intermediate phrasing in Spanish intonation. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign,
Urbana-Champaign, IL.

Prieto, P., Nibert, H., & Shih, C. (1995). The absence or presence of a declination effect on the descent of F0 peaks? Evidence from Mexican Spanish. In K. Zagona (Ed.),Grammatical
theory and Romance languages(pp. 197-207). Philadelphia: John Benjamins.

Prieto, P., Shih, Chilin.,& Nibert, H. (1996). Pitch downtrend in Spanish. Journal of Phonetics, 24, 445-473.

Prieto, P. (1998). The scaling of the L values in Spanish downstepping contours. Journal of Phonetics, 26, 261-282.

Prince, A., & Smolensky, P. (1993). Optimality theory: constraint interaction in generative grammar. Rutgers Optimality Archive #537.Retrieved January 18, 2006 from
http://roa.rutgers.edu/files/537-0802/537-0802-PRINCE-0-0.PDF

Quilis, A. (1993). Tratado de fonología y fonética españolas. Madrid: Editorial Gredos.

Rao, R. (2006). On intonation’s relationship with pragmatic meaning in Spanish. In T. Face & C. Klee (Eds.), Selected proceedings of the 8th Hispanic Linguistics Symposium (pp. 103-115). Somerville, MA: Cascadilla Proceedings Project.

Rao, R. (2007). On the phonological phrasing in the Spanish of Lima, Perú. Southwest Journal of Linguistics, 26(1), 81-111.

Rao. R. (2008). Deaccenting in spontaneous speech in Barcelona Spanish. Manuscript under review by Studies in Hispanic and Lusophone Linguistics.

Sandalo, F., & Truckenbrodt, H. (2003). Some notes on phonological phrasing in Brazilian Portuguese. D.E.L.T.A., 19(1), 1-30.

Selkirk, E. (1980). The role of prosodic categories in English word stress. Linguistic Inquiry,11, 563-605.

Selkirk, E. (1984).Phonology and syntax: The relation between sound and structure. Cambridge: The MIT Press.

Selkirk, E. (1986).On derived domains in sentence phonology. Phonology Yearbook,3, 371-405.

Selkirk, E. (2000).The interaction of constraints on prosodic phrasing.In M. Horne (Ed.),Prosody: theory and experiment. Studies presented to Gosta Bruce (pp. 231-261). Dordecht, Boston and London: Kluwer.

Sosa, J.M.(1999). La entonación del español: su estructura fónica, variabilidad y dialectología. Madrid: Cátedra.
Steedman, M. (1991). Structure and intonation.Language,67, 260-296.

Terken, J. & Hirschberg, J. (1994). Deaccentuation of words representing given information: effects of persistence of grammatical function and surface position. Language and Speech,37(2), 125-145.

Toledo, G. (2007). Fraseo en español peninsular y modelo autosegmental y métrico. Estudios Filológicos, 42, 227-243.

Toledo, G. (in press). Fonología de la frase entonativa: el contraste continuidad y conclusión.Moenia, Revista Lucense de Lingüística & Literatura,13.

Truckenbrodt, H. (1995). Phonological phrases: Their relation to syntax, prominence and focus. Unpublished doctoral dissertation, MIT, Cambridge, MA.

Truckenbrodt, H. (1999). On the relation between syntactic phrases and phonological phrases.Linguistic Inquiry,30, 219-255.

Truckenbrodt, H. (in press). The syntax-phonology interface. In Paul de Lacy (Ed.), The Cambridge handbook of phonology. Cambridge: Cambridge University Press.

Vigário, M. (2003). The prosodic word in European Portuguese. Berlin/New York: Mouton de Gruyter.

Zubizarreta, M. L. (1998). Prosody, focus and word order. Cambridge, MA: MIT Press.

Tags: , , , , ,

Category: 2008