LI05CH08_Elfner ARI 29 November 2018 14:54

Annual Review of Linguistics

The Syntax–Prosody Interface

Ryan Bennett1 and Emily Elfner2
1
Department of Linguistics, University of California, Santa Cruz, California 95064, USA;
email: [email protected]
2
Department of Languages, Literatures and Linguistics, York University, Toronto,
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Ontario M3J 1P3, Canada; email: [email protected]

Access provided by 36.227.193.71 on 03/15/24. For personal use only.

Annu. Rev. Linguist. 2019. 5:151–71 Keywords

First published as a Review in Advance on syntax–prosody interface, prosody, phrasal phonology, recursion, prosodic
October 4, 2018
hierarchy
The Annual Review of Linguistics is online at
linguist.annualreviews.org Abstract
https://doi.org/10.1146/annurev-linguistics- This article provides an overview of current and historically important is-
011718-012503
sues in the study of the syntax–prosody interface, the point of interaction
Copyright  c 2019 by Annual Reviews. between syntactic structure and phrase-level phonology. We take a broad
All rights reserved
view of the syntax–prosody interface, surveying both direct and indirect ref-
erence theories, with a focus on evaluating the continuing prominent role
of prosodic hierarchy theory in shaping our understanding of this area of
linguistics. Specific topics discussed in detail include the identification of
prosodic domains, the universality of prosodic categories, the recent resur-
gence of interest in the role of recursion in prosodic structure, crosslinguistic
variation in syntax–prosody mapping, prosodic influences on syntax and word
order, and the influence of sentence processing in the planning and shaping
of prosodic domains. We consider criticisms of prosodic hierarchy theory
in particular, and provide an assessment of the future of prosodic hierarchy
theory in research on the syntax–prosody interface.

151
 LI05CH08_Elfner ARI 29 November 2018 14:54

1. INTRODUCTION: DOMAINS IN PHRASAL PHONOLOGY

The phonetic form of a word often depends on its position within a larger, containing phrase.
To illustrate, consider a process of /r/-assimilation found in Bengali; in some circumstances,
the approximant /r/ may be realized as identical with any following coronal consonant, as in
example 1 (Hayes & Lahiri 1991, Fitzpatrick-Cole 1996, Truckenbrodt 2002). This process applies
optionally, both within and across words, as in examples 2a–c:
(1) /r/ → CX / CX, [+cor]
(2a) [bOrSa] ∼ [bOSSa] ‘rainy season’
 
(2b) [kor-tSh e] ∼ [kotS -tSh e] ‘(s)he does’
   
(2c) [ram-er SoSur-er dZ onno] ∼ [ram-eS SoSur-er dZ onno] ∼ [ram-eS SoSur-edZ dZ onno]
‘for Ram’s father-in-law’
The pronunciation of any given word may thus vary dramatically within the wider context in which
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

it is embedded. Processes like example 1, which apply at the junctures of words and morphemes,
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

are sometimes known as sandhi processes (from Sanskrit sam- . ‘together’ + dhi ‘putting’; Ruppel
2017, p. 109).
It has long been known that sandhi processes do not apply indiscriminately between any adjacent
pair of words (see Selkirk 1980a, Nespor & Vogel 1986, and references therein). More typically,
sandhi rules apply between words standing in a particular syntactic relationship. For example,
consider the syntactic structure of example 2c (‘for Ram’s father-in-law’), adapted from Fitzpatrick-
Cole (1996):

(3) PP [NP [ram-er SoSur-er] dZ onno]
Ram-GEN father.in.law-GEN for
This syntactic structure is reflected in the distribution of /r/-assimilation. Assimilation may ap-

ply between [ram-er] and [SoSur-er], between [SoSur-er] and [dZ onno], or both, as illustrated in
∗  
example 2c. Missing from this list is the form [ram-er SoSur-edZ dZ onno], in which assimila-

tion has applied between [SoSur-er] and [dZ onno] but not between [ram-er] and [SoSur-er]. It thus
appears that the domain of /r/-assimilation is conditioned by syntactic structure: It may apply
between words contained in the same noun phrase (NP) or the same prepositional phrase (PP),

but it may not apply solely between [SoSur-er] and [dZ onno], because these two words do not form
a syntactic unit to the exclusion of [ram-er] in example 3.
Nonetheless, there is reason to believe that the domain of /r/-assimilation is not syntactically de-
 
fined. Hayes & Lahiri (1991) point out that in a sentence such as [OmOr tS ador tara-ke dietShe] ‘Amor
gave a scarf to Tara,’ assimilation is impossible between the first two words ([OmOr] ‘Amor’ and
 
[tS ador] ‘scarf ’) in careful speech. However, assimilation to [OmOtS tS ador. . .] is possible in faster,

less guarded renditions of the same sentence, or in a discourse context in which the word [tS ador]
‘scarf ’ has previously been mentioned (e.g., ‘Shamoli gave a scarf to Ram, and Amor gave a scarf to
Tara’). It seems unlikely that the syntactic structure of this sentence is conditioned by speech rate
or by the larger discourse context, since the semantic interpretation of the sentence (derived, by
hypothesis, from the syntax) is fixed across all of these contexts. Hayes & Lahiri (1991) therefore
conclude that the domain of /r/-assimilation cannot be defined in terms of the syntax itself.
Here we appear to be at an impasse. The domain of /r/-insertion in Bengali seems closely
tied to syntactic structure, and yet it cannot be identical to that syntactic structure. Such partial
(mis)matches between syntax and the domains in which sandhi, intonation, and other aspects of
phrasal phonology apply are at the heart of a research area known as the syntax–prosody interface.
A central problem in research on phrasal phonology is “chunk definition” (Scheer 2012a,b): What

152 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

kinds of structures or domains condition phonology at the phrase level and above? What are the
algorithms which produce those domains? How do we identify phonologically relevant phrasal
domains on the basis of phonetic or phonotactic evidence? These issues were raised in a modern
context at least as early as Chomsky & Halle’s (1968) foundational publication on generative
phonology. Fifty years of subsequent research on the syntax–prosody interface have broadened
the empirical basis for these questions, and have given rise to a range of different perspectives on
what an adequate theory of phrase-level phonology should look like.
Prosodic phrasing is often invoked to account for sandhi processes, such as Bengali /r/-
assimilation, which affect the segmental or tonal content of individual words. Prosodic domains
also play a key part in conditioning phrase-level intonation (see, e.g., Jun 2005, 2014), as well as
“lower-level” phonetic properties such as duration, voice quality, and so on (see, e.g., Fougeron
& Keating 1997, Keating et al. 2003, and discussion in Sections 3.1 and 5). We thus adopt a broad
view of phrasal phonology, in which segmental phonotactics, stress, tone, intonation, morphology,
and subsegmental phonetic patterning all count as potential evidence for theories of phrase-level
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

domains.
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

In this review, we survey some current and historically important issues in the study of the
syntax–prosody interface. Our intent is to provide some basic background on this research tradition
while highlighting debates and empirical findings which strike us as particularly important for the
ongoing development of the field. In an article of this scope we cannot hope to do justice to all
of the phenomena and analytical issues which are important for understanding the relationship
between syntax and prosody. To supplement the material presented here, we refer readers to
various overviews (e.g., Inkelas & Zec 1995, Shattuck-Hufnagel & Turk 1996, Turk et al. 2006,
Truckenbrodt 2007, Elordieta 2008, Wagner & Watson 2010, Selkirk 2011, Frota 2012, Cole
2015, Ishihara 2015, Wagner 2015, Elfner 2018), collected volumes (e.g., Kaisse & Zwicky 1987;
Inkelas & Zec 1990; Jun 2005, 2014; Selkirk & Lee 2015), and books (e.g., Selkirk 1984, Nespor
& Vogel 1986, Pierrehumbert & Beckman 1988, Gussenhoven 2004, Ladd 2008, Féry 2016).

2. DIRECT AND INDIRECT REFERENCE

Theories of chunk definition in phrasal phonology fall into one of two broad camps. Indirect
reference theories assume that syntactic structure is first mapped to a separate representation—
typically called prosodic structure—which provides the groupings of words to which phrase-level
phonological processes are sensitive. To illustrate with Bengali, Hayes & Lahiri (1991) argue that
syntactic phrases are mapped to a set of nonsyntactic units known as phonological phrases (ϕ).
These ϕ condition /r/-assimilation and other phrasal phenomena. The mapping from syntax to
ϕ-structure yields groupings like those in examples 4a–d, which reflect syntax to the extent that
each ϕ corresponds to a syntactic constituent of some kind (see the ungrammatical example 4d,
which violates this requirement; Fitzpatrick-Cole 1996, Truckenbrodt 2002):

Possible ϕ groupings for PP [ NP [ram-er SoSur-er] dZ onno]
 
(4a) (ram-eS SoSur-edZ dZ onno)ϕ

(4b) (ram-eS SoSur-er)ϕ (dZ onno)ϕ

(4c) (ram-er)ϕ (SoSur-er)ϕ (dZ onno)ϕ
 
(4d) ∗ (ram-er)ϕ (SoSur-edZ dZ onno)ϕ
The fact that prosodic structure is derived from syntax, but need not be identical to it, thus
explains why processes such as /r/-assimilation are only imperfectly conditioned by syntactic
constituency. Additionally, if factors such as speech rate and discourse status can affect groupings

www.annualreviews.org • The Syntax–Prosody Interface 153

 LI05CH08_Elfner ARI 29 November 2018 14:54

of words at this level of representation, we expect such factors to condition processes such as
/r/-assimilation even when the underlying syntax (and resultant semantics) remains constant.
In contrast, direct reference theories assume that the domains which condition segmental
sandhi and other types of phrasal phonological processes are defined solely with reference to syn-
tactic (and/or morphological) structure. Proponents of direct reference theories include Rotenberg
(1978), Kaisse (1985), Odden (1987), Chen (1990), Cinque (1993), Seidl (2001), Wagner (2005,
2010), Arregi (2006), Newell (2008), Pak (2008), Samuels (2009), Scheer (2010; 2012a,b), and
Newell & Piggott (2014), among others (see Elordieta 2008 for additional references). As Elordi-
eta (2008) emphasizes, existing direct reference theories do not claim that the domains of phrasal
phonology are exactly identical to syntactic units (indeed, such a position is likely untenable, given
the existence of well-known mismatches between syntax and the domains of phrasal phonology;
see Section 2.1 and Pak 2008, section 2.2.1). Instead, direct reference theories claim that phono-
logically relevant groupings of words are determined by the syntactic relationships which hold
between those words. These relationships may be defined in terms of structural properties such as
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

c-command (Kaisse 1985) or in terms of the kinds of nodes which intervene between words in the
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

same syntactic structure (Odden 1987; see also Elfner 2018 on cyclic/phasal models of prosodic
phrasing, and references therein). In a direct reference framework, we might explain the ungram-
maticality of an output such as example 4d by assuming that /r/-assimilation applies more readily
between two words which are structural sisters ([ram-er] and [SoSur-er]) than between two words

which are not ([SoSur-er] and [dZ onno]), such that assimilation in the second pair of words entails
assimilation in the first. Importantly, these are purely syntactic conditions on sandhi—there is no
mediation by a separate level of structure such as ϕ.
Although we adopt the traditional division between direct and indirect reference theories of
phrasal phonology here, we recognize that this dichotomy is probably too simplistic a charac-
terization of the actual theoretical landscape (see also Elordieta 2008). Seidl (2001), for instance,
advocates a Minimal Indirect Reference theory in which some phonological processes apply at a
level of representation (M0 ) which contains only morphosyntactic structure, while other processes
apply at a level of representation (P0 ) which is derived from but not identical to the morphosyntax
(see also Rotenberg 1978). Pak (2008, pp. 33–36) proposes a similar direct reference architecture
but further adopts rules which optionally rebracket groupings of words on the basis of speech rate.
The space of theories here may be better characterized as a continuum, with more or less rigid
adherence to the underlying syntax (see also Wagner 2010, Selkirk 2011).
Indirect reference theories are clearly ascendant in research on the syntax–prosody interface.
In Section 2, we discuss some additional evidence favoring indirect over direct reference as the
source of chunk definition in phrasal phonology and provide an overview of prosodic hierarchy
theory (PHT), the most widely employed indirect reference theory of phrasal phonology and
phonetics. In Section 4, we consider some recent critiques of PHT.

2.1. Arguments in Favor of Indirect Reference

Linguists have offered a number of arguments in support of indirect reference theories of phrasal
phonology, many originating in Nespor & Vogel’s (1986) seminal book on the topic. As discussed
in Sections 3–5, some of these arguments are more convincing than others; we present a few of
them here without commenting on their soundness.

2.1.1. Blindness to syntactic category distinctions. Processes of phrasal phonology do not

generally distinguish between words on the basis of their lexical or syntactic category. For example,
some dialects of American English insert [ô] between lexical words (nouns, verbs, adjectives) ending

154 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

in [A @ O] when followed by a vowel, as in examples 5a and 5b. This applies regardless of the category
of the following word. This indifference to syntactic category distinctions suggests that sandhi
processes apply at a level of representation which no longer encodes such category information:
Intrusive [ô]
(5a) Did Wanda[ô]NOUN eatVERB much at dinner?
(5b) The boat will yaw[ô]VERB [aDET littleNOUN ]NP .
(McCarthy 1993)

2.1.2. Nonisomorphisms (mismatches) between syntax and prosody, and eurhythmic ef-
fects. Some domains for phrasal prosody do not match the groupings of words provided by
morphology or syntax, and they are often shaped by factors which are purely phonological in
nature. In Catalan, for instance, there is a preference for the last ϕ domain in the utterance to
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

contain no more than two words (ω) (Prieto 2005, 2014). This creates a contrast in intonational
grouping between [verb object] clauses with a one-word object, which are phrased (ω1 ω2 ), and
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

those with a two-word object, which are phrased (ω1 )(ω2 ω3 ) (see, e.g., Ghini 1993; Selkirk 2000;
Elordieta 2007; Elfner 2012, 2015; Myrberg 2013). Eurythmic effects (i.e., preferences relating
to the size and rhythmic patterning of prosodic constituents) are a common source of noniso-
morphism, though nonisomorphisms can also emerge from other factors, such as the particular
mapping algorithm used to derive prosodic structure from syntactic structure (see, e.g., Selkirk &
Shen 1990 on Shanghai Chinese, and the discussion of Kwak’wala determiners in Section 4).

2.1.3. Insensitivity to phonetically null elements. Syntacticians frequently assume that pho-
netically null elements may still be present in the syntactic representation (e.g., traces, unpro-
nounced copies, PRO). These elements seem to have no effect on sandhi rules or other processes
of phrasal phonology (see, e.g., Kaisse 1985; Nespor & Vogel 1986; Truckenbrodt 1995, 1999;
Elfner 2012, 2015).

2.1.4. Variability and optionality. Many processes of phrasal phonology apply variably and/or
optionally (e.g., Bengali /r/-assimilation, which can be affected by factors such as speech rate).
Such optionality is not characteristic of syntactic structure (but see Section 5 for explanations of
such effects which make reference to speech processing and production planning).

2.2. Prosodic Hierarchy Theory

PHT is the dominant indirect reference theory [Selkirk 1981 (1978), 1980a, 1984; Nespor & Vogel
1986; Pierrehumbert & Beckman 1988]. PHT assumes that phrasal phonology is conditioned not
by syntax but by abstract phonological constituents known as prosodic categories. These prosodic
categories are derived from the syntax and come in different types (or “levels”), depending on what
syntactic units they characteristically correspond to. As these prosodic categories are each derived
from syntactic units of different sizes—at least clauses (CP), maximal projections (XP), and words
(X0 )—they too can be arranged into a “hierarchy” reflecting their relative sizes (Figure 1).1
Evidence for the prosodic hierarchy primarily comes from the observation that phonological
domains of different sizes may be associated with categorically distinct phonological processes (e.g.,

1
Various other views exist on the composition of the prosodic hierarchy, in addition to that shown in Figure 1 (see, e.g.,
Nespor & Vogel 1986; Jun 2005, 2014; Ladd 2008 for additional discussion).

www.annualreviews.org • The Syntax–Prosody Interface 155

 LI05CH08_Elfner ARI 29 November 2018 14:54

CP ι Intonational phrase

XP φ Phonological phrase Interface categories

X0 ω Prosodic word

Figure 1
The prosodic hierarchy above the word.

Nespor & Vogel 1986, Vogel 2009, Vigário 2010), suggesting that different types of prosodic
units may be copresent in the same representation, as in PHT. Relatedly, in many languages
multiple phonological processes converge on the same domains, suggesting that a relatively small
number of prosodic categories (as in Figure 1) may be sufficient to account for most patterns of
word- and phrase-level phonology. An example of such domain clustering comes from European
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Portuguese, in which the prosodic word conditions both stress assignment and a diverse set of
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

segmental phenomena (Vigário 2003, chapter 5; see also Peperkamp 1997).

The prosodic categories in Figure 1 can be nested: A sentence such as She loaned her rollerblades
to Robin would have (at least) the structure {([She loaned]ω [her rollerblades]ω )ϕ ([to Robin]ω )ϕ }ι
(Selkirk 2000). Early versions of PHT adopted the strict layer hypothesis, which proposes that
prosodic constituents can only dominate constituents at the next level down on the hierarchy
[Selkirk 1984, p. 26; see also Selkirk 1981 (1978), Beckman & Pierrehumbert 1986, Nespor &
Vogel 1986, Pierrehumbert & Beckman 1988]. It is now clear that the strict layer hypothesis is too
strong: At a minimum, categories may dominate categories which are more than one step lower
on the hierarchy (“level skipping”; Selkirk 1995). Verbal clitics in Standard Italian furnish a clear
example. These clitics are outside the domain of stress assignment—the prosodic word (ω)—
which includes the verb, and must therefore be dominated directly by ϕ [e.g., pórtamelo ([pórta]ω
me lo)ϕ ‘Bring it to me!’; Peperkamp 1997, Anderson 2005, Vogel 2009]. Recent research has also
revived the possibility that prosodic constituents may dominate other constituents of the same
type, a configuration also banned by the strict layer hypothesis (“recursion”; see, e.g., Ladd 2008,
chapter 8, and Section 3.3, below).

3. SOME CURRENT ISSUES IN INDIRECT REFERENCE THEORY

3.1. Identifying Domains: How Are Mapping Algorithms Distinguished?

Broadly understood, the term “prosodic domain” refers to a portion of an utterance which is
identifiable due to its behavior with respect to some phonetic or phonological process. As discussed
above, /r/-assimilation in Bengali applies across domains larger than the word, yet does not apply
indiscriminately across the entire utterance. The domain of application of /r/-assimilation is thus
definable according to those portions of the utterance in which the phonological process may be
observed, whether those domains are defined according to syntactic constituents (under a direct
reference approach) or according to prosodic constituents, such as ϕ (as argued in Hayes & Lahiri
1991). Thus, we might infer that the edges of phrase-level phonological domains in Bengali fall
between those words in which /r/-assimilation is blocked.
As /r/-assimilation in Bengali applies only between words contained in the same domain, it can
be considered a domain-span process (in the terminology of Selkirk 1980a). Other phonological
and phonetic processes target elements at domain edges: typically, either the left or right edge of
some prosodic domain. For example, in Japanese the prosodic domain traditionally referred to as

156 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

the Minor Phrase is marked at its left edge by a rise in pitch (a %LH boundary tone), while the
Major Phrase domain is marked at its left edge by a pitch reset, which undoes the pitch downtrends
of the preceding prosodic phrase (McCawley 1968, Pierrehumbert & Beckman 1988, Selkirk &
Tateishi 1991). In ChiMwiini, the right edge of intonational phrases is marked by lengthening of
the vowel in the penultimate syllable, as well as by the presence of a right-edge phrasal high tone
(Kisseberth & Abasheikh 1974, Selkirk 2011).
As discussed by Selkirk (2011), the diagnostics for prosodic domain edges are often partial
and asymmetric, such that prosodic domains are typically diagnosable only on the left or right
edge, but not both. This observation inspired the “end-based” theory of syntax–prosody mapping,
which claims that languages differ parametrically as to whether the left or right edges of syntactic
constituents are referred to in the mapping to corresponding prosodic domains (Selkirk 1986,
Selkirk & Shen 1990). ChiMwiini, under Selkirk’s (1986) proposal, would map the right edges
of syntactic phrases onto the right edges of phonological phrases, leaving the position of left
edges unspecified. Japanese, alternatively, would be specified with a left-edge setting (Selkirk &
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Tateishi 1991). Under the strict layer hypothesis, the location of the unspecified domain edge
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

from such incomplete mappings would be determined by general, formal constraints imposed on
the prosodic hierarchy, such as a ban on recursive nesting of prosodic domains (see Section 2.2).
Selkirk (2011) advocates for an approach to syntax–prosody mapping that moves away from
reference to phrase edges and permits certain structures banned under strict layering, such as
recursion. Selkirk (2011) instead proposes that both the left and right edges of syntactic constituents
are transparently mapped to their corresponding prosodic constituents. Evidence for such an
approach comes from languages in which it is possible to diagnose both the left and right edges of
prosodic domains simultaneously. Selkirk (2011) discusses the case of Xitsonga, in which a lexical
high tone spreads throughout ϕ-level domains. It is blocked from spreading onto the final syllable
of ϕ (providing a diagnostic for the right edge), and is further blocked from spreading onto the
initial syllable of the following ω (thus providing a diagnostic for the left edge). Connemara Irish
(Elfner 2012, 2015) is another example of a language with diagnostics for both the left and right
edges of ϕ, which are marked with rising and falling phrasal pitch accents, respectively.
In the case of languages such as Japanese and ChiMwiini, the absence of a clear phonological
diagnostic for one edge or the other would not necessarily constitute evidence against the presence
of prosodic boundaries in locations predicted by syntactic structure. Selkirk’s (2011) proposal thus
assumes that prosodic boundaries may be present even when there is no explicit phonological or
phonetic evidence for their existence, and instead places the onus on the theory of syntax–prosody
mapping to predict the locations for prosodic boundaries.
Such an approach contrasts with what is typically assumed in psycholinguistic (see Section 5)
and “intonation-first” approaches to prosodic domain demarcation ( Jun 1998, 2005, 2014), namely
that prosodic boundaries are present only when there is some explicit phonetic or phonological
cue to their existence. Such evidence may come from categorical measures, such as the presence
of a boundary tone, a pause, or final lengthening, or from gradient measures which are equated to
the relative strength of a prosodic boundary, such as the degree of final lengthening, the degree
of pitch reset or pitch scaling, the magnitude of rises or falls in pitch, the duration of pauses, the
degree of initial strengthening, and the frequency that phonological processes apply across word
boundaries (see Section 5 for further discussion).

3.2. Universality of Categories

The prosodic hierarchy is typically assumed to be a component of universal grammar: All languages
have hierarchically ordered prosodic structure, and languages are thought to make use of the

www.annualreviews.org • The Syntax–Prosody Interface 157

 LI05CH08_Elfner ARI 29 November 2018 14:54

same set of prosodic categories in the structuring of utterances. Theories of the syntax–prosody
interface which adopt PHT similarly assume that the principles or constraints governing syntax–
prosody mapping are universal in nature. Thus, although languages may differ in terms of their
surface syntactic structure and in the explicit marking of prosodic domains using phonetic and
phonological processes, the grammatical mechanism underlying both the mapping of prosodic
structure from syntactic structure and the hierarchical organization of prosodic structure into
distinct prosodic categories is thought to remain constant across languages.
To what extent is this an accurate representation of prosodic structure crosslinguistically? In
terms of the number and types of prosodic categories that are present in the prosodic hierarchy,
there are languages which arguably underrepresent or overrepresent the prosodic hierarchy shown
in Figure 1. For example, Japanese and Basque, both lexical pitch accent languages, show evidence
for an additional prosodic domain which is larger than the prosodic word but smaller than the
phonological phrase, traditionally referred to as the Accentual or Minor Phrase (McCawley 1968,
Pierrehumbert & Beckman 1988, Selkirk & Tateishi 1991, Jun & Elordieta 1997). Other lan-
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

guages, alternatively, appear to provide positive evidence for only some of the prosodic domains
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

depicted in Figure 1. For example, the Inuit languages demarcate prosodic domains using tonal
cues at the level of the prosodic word and the level of the intonational phrase but (seemingly)
provide no evidence for an intermediate, ϕ-level prosodic domain (Arnhold 2014, Arnhold et al.
2018; see also Bennett 2015 for discussion).
Such languages call into question the universality of category labels in prosodic typology. For
example, does the presence of an additional intermediate-level domain in Japanese and Basque
necessitate a revision to the prosodic hierarchy, as shown in Figure 1? And conversely, if Inuit
differentiates between only two types of higher-level domains, what determines which two cate-
gories these domains correspond to? Recent research by Ito & Mester (2012, 2013) on Japanese
and by Elordieta (2015) on Basque argues that the additional intermediate-level categories may
be captured under the assumption that ϕ domains may be recursive, such that at least some ap-
parent category distinctions actually derive from the depth of embedding relative to the amount
of structure present in the utterance, and not (necessarily) to distinctions made in the number and
type of prosodic categories (see Section 3.3, as well as Selkirk 2011; Elfner 2015, 2018 for further
discussion). Wagner (2005, 2010) argues in favor of a more radical, “label-free” version of this
approach, such that the hierarchical structure observed in the prosodic organization of sentences
derives not from a universal prosodic hierarchy at all but rather from the hierarchical, recursive
nature of syntactic structure, on which prosodic domains are built (see Sections 3.3 and 5).
Wagner’s label-free approach contrasts with the “syntactic grounding” approach proposed
most explicitly by Selkirk (2011), who derives the universality of prosodic categories from the
universality of syntactic constituent types. More specifically, Selkirk (2011) proposes that there is
a direct correspondence between three syntactic constituents, word (X0 ), phrase (XP), and clause
(CP), each of which maps onto a corresponding prosodic category, ω, ϕ, and ι, respectively.
The hierarchical structure of syntax in this way results naturally in the hierarchy of prosodic
categories, as espoused by PHT, under the assumption that prosodic domains may be recursive, as
discussed above. Language-specific differences in prosodic category distinctions may thus derive
from differences in the syntactic organization of individual utterances. A challenge for this type of
approach is a relative lack of understanding regarding how such mapping constraints are applied in
languages with radically different systems of morphosyntactic organization, such as polysynthetic
languages (see Elfner 2018 for discussion and references).
Another prediction made by PHT is that processes of phrasal phonology should apply within
the same phonological domains, namely those supplied by the prosodic hierarchy in Figure 1. This
prediction, known as domain clustering (see Section 2.2; Inkelas 1990), has been the target of some

158 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

criticism. Bickel et al. (2009) and Schiering et al. (2010) claim that word-level prosody in Limbu
violates domain clustering (as well as nesting; see Section 4) because domain-bounded phonological
processes diagnose four distinct groupings of morphemes in [prefix–stem–suffix = clitic] strings.
These four domains do not correspond neatly to ω or ϕ. Padgett (2014) notes that stress, vowel
reduction, and final devoicing in Russian—which generally converge on a single definition of
the prosodic word—come apart in compounds, where they appear to have nonidentical domains
of application. Seidl (2001) discusses similar facts in Mende. All of these patterns seem to point
toward a richer prosodic hierarchy than that in Figure 1, perhaps making use of language-specific
or even process-specific domains. Seidl (2001) and Schiering et al. (2010) consider (but dismiss)
the possibility that these apparent “extra” domains actually reflect recursion of ω and ϕ (see
Section 3.3); it would be worthwhile to revisit these arguments in light of recent developments in
the theory of prosodic recursion (particularly the recursive prosodic subcategories of Ito & Mester
2009, 2010, 2012, 2013). See Wagner (2010, 2015) for additional critical commentary.
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

3.3. Recursion in Prosodic Structure

Access provided by 36.227.193.71 on 03/15/24. For personal use only.

The strict layer hypothesis (described in Section 2.2) sharply restricts the nesting of phonological
domains, prohibiting recursion in prosodic structure: A node of type κ cannot dominate another
node of the same type. This ban on prosodic recursion was questioned early on in the development
of PHT, most notably by Ladd (1986, 1988). Ladd investigated the strength of intonational
boundaries in coordinate structures such as [[A and B] but C] versus [A [but B and C]], where {A,
B, C} are all full clauses. He found that intonational boundaries were stronger before ‘but’ than
before ‘and.’ Under the assumption that all clauses correspond to intonational phrases (ι), the only
way to represent this distinction is through recursion: [[A and B]ι but C]ι versus [A [but B and C]ι ]ι
(see also Dresher 1994; Kubozono 1989; Féry & Truckenbrodt 2005; Wagner 2005, 2010; Ladd
2008).
Throughout the 1990s, recursion was commonly invoked for the prosodic word (ω), chiefly as
a means of understanding how unstressed affixes, clitics, and function words are phonologically
incorporated into their hosts (e.g., Inkelas 1990, Selkirk 1995, Booij 1996, Peperkamp 1997,
Vigário 1999). More sporadically, recursion was also invoked for the phonological phrase (ϕ)
(e.g., Truckenbrodt 1995, 1999). The last 10 years have seen a resurgence of interest in the
possibility of prosodic recursion at both higher levels (e.g., ϕ: Ito & Mester 2007, 2012, 2013;
Selkirk 2011; Elfner 2012, 2015; Elordieta 2015; ι: Selkirk 2009, Myrberg 2013) and lower levels
(e.g., the metrical foot; see Bennett 2013, Martı́nez-Paricio 2013, Martı́nez-Paricio & Kager 2015,
and references therein).
Some researchers remain skeptical about the possibility of prosodic recursion, either rejecting
it outright (e.g., Vogel 2009, Schiering et al. 2010) or arguing that it has a limited role to play
in prosodic systems (e.g., Vigário 2010, Frota & Vigário 2013). This debate largely concerns the
kinds of diagnostics which are taken to be valid indicators of recursion. Vogel (2009) and Vigário
(2010) argue that recursion of a category κ should only increase the strength of the phonetic
cues associated with that category, as in Ladd’s (1986) study of coordinate structures in English.
In contrast, Ito & Mester (2007, 2009, 2010, 2013), Martı́nez-Paricio (2013), Elfner (2015), and
others argue that different levels of recursive structure can show not only gradient differences in
boundary strength but also categorical differences in the kinds of phonological phenomena which
occur at each level (e.g., the topmost ω in a recursive prosodic word structure can show different
behavior than the bottommost ω; see Bennett 2018, Elfner 2018 for discussion). The question of
how recursive prosodic structures are interpreted by the phonetics and phonology clearly merits
further investigation.

www.annualreviews.org • The Syntax–Prosody Interface 159

 LI05CH08_Elfner ARI 29 November 2018 14:54

3.4. Crosslinguistic Variation in Syntax–Prosody Mapping

Prosodic structure is derived, at least in part, through reference to syntactic structure. While
the basic building blocks of syntactic structure may be universal in nature, languages differ in
terms of surface structure, after syntactic operations such as movement have taken place. Under
most current theories of the syntax–prosody interface, at least those deriving from generative
syntax and, more specifically, Minimalism (Chomsky 1995), prosodic domains are created on the
basis of syntactic constituent structure, and not vice versa (see also Section 3.5). This view of the
architecture of the grammar is commonly referred to as the Y model, which holds that syntactic
structure feeds into the phonological component of the grammar. This assumption holds even in
views assuming that prosodic domains are created via direct reference, where there is no mediating
component of prosodic structure.
To what extent, therefore, can crosslinguistic variation in syntactic structure account for the
typology of patterns found in prosodic structure? Syntactic structure provides, in a sense, the
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

blueprints for prosodic domains; the particular phonological properties of sentences, including
the number and type of words, their order, and their organization in terms of constituents, will
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

be determined by the syntactic structure. Different theories of syntax–prosody mapping will make
different predictions regarding the particulars of how these constituents are mapped to prosodic
domains, but language-specific syntactic structure will determine the basis of prosodic constituents
in any given utterance.
Given the basic mapping of syntactic constituents to prosodic constituents shown in Figure 1,
we assume that all languages contain these basic syntactic building blocks: words (X0 ), phrases
(XP), and clauses (CP). Selkirk’s (2011) Match Theory predicts that languages will map these basic
syntactic elements onto corresponding prosodic domains, ω, ϕ, and ι, making the prediction that
all languages will distinguish at least three types of prosodic domains (see Section 3.2 for further
discussion).
Alternatively, it may be that prosodic domains are relative and do not require specific category
labels themselves, as would arise under the label-free theory discussed in Section 3.2 (Wagner
2005, 2010). Owing to the relatively recent resurgence of interest in the role of recursion in
prosodic structure (Section 3.3), such views of prosodic structure are becoming more common
in the literature. At one extreme, Wagner (2005, 2010) proposes that the prosodic hierarchy be
re-envisioned as recursive prosodic domains, where relative boundary strength and the depth
of embedding, rather than category labels or correspondence with particular syntactic elements,
are responsible for the effects of the prosodic hierarchy (see also Ladd 1986, 1988); as such,
crosslinguistic variation in syntax–prosody mapping may be derived exclusively from differences
in surface syntactic structure, leaving only a minor or no role for differences in terms of prosodic
structure and the prosodic hierarchy. However, even within PHT, the presence of recursion has
been used to argue for a simplification of the number of distinctions necessary in the prosodic
hierarchy (see Section 3.2, as well as Selkirk 2009; Elfner 2012, 2015; Ito & Mester 2012, 2013;
Myrberg 2013; Elordieta 2015).
In summary, while languages differ in terms of the details of their surface syntactic configura-
tions, the mechanisms governing syntax–prosody mapping are sensitive only to the larger patterns
of constituent structure. This means that if languages systematically map syntactic constituents
onto prosodic domains in universally consistent ways, we can expect to see commonalities in terms
of how prosodic domains are constructed across languages.

3.5. Prosodic Influence on Syntax and Word Order

Various phenomena suggest that the phonological size of a syntactic constituent can affect its
position within the sentence. For example, Zec & Inkelas (1990) observe that topicalization in
160 Bennett · Elfner
 LI05CH08_Elfner ARI 29 November 2018 14:54

Serbo-Croatian is subject to a phonological size condition: Prosodically large units such as u Rio
de Žaneiru ‘in Rio de Janeiro’ can be moved to a sentence-initial topic position, but prosodically
small units like u Riju ‘in Rio’ cannot (see also Ryan 2018 on “end weight” effects such as Heavy XP
shift). The distribution of “small” words, such as clitics and pronouns, can also be conditioned by
constraints governing the position of such words within phonological domains such as ϕ or ι (Zec
& Inkelas 1990, Halpern 1995, Anderson 2005, Werle 2009, Bennett et al. 2016). A growing body
of research suggests that even phenomena traditionally (and uncontroversially) taken to belong
to the syntax proper, such as wh-movement and argument incorporation, may be influenced by
fundamentally prosodic factors (e.g., Aissen 2000; Kandybowicz 2009, 2015, 2017; Agbayani &
Golston 2010, 2016; Richards 2010, 2016; Sabbagh 2013; Clemens 2014).
Anttila (2016) and Shih (2017) offer surveys of recent research in this area, to which we refer
the reader for further discussion and references. Almost all work dealing with phonological effects
on syntax and word order assumes some version of PHT; we are unaware of any research on
prosodically motivated syntactic variation which adopts an explicit direct reference framework. A
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

possible exception is research by Wagner (2005, 2010), who proposes that prosody may be a de-
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

ciding factor when there is more than one syntactic parse available, as in the optional extraposition
of relative and complement clauses.

4. A RETURN TO DIRECT REFERENCE?

4.1. Critiques of Prosodic Hierarchy Theory

The last 20 years have witnessed a revived (although somewhat muted) debate over chunk definition
in phrasal phonology. In particular, there has been renewed skepticism over the abstract categories
assumed by PHT (Section 2.2), with some authors advocating the rejection of indirect reference
theories as a whole [recalling and, to some extent, updating arguments made in the 1980s by Kaisse
(1985), Odden (1987), and others].
As noted above, critics of PHT have sometimes seized on the apparent nonuniversality of
prosodic categories as evidence that PHT is fundamentally misguided as a theory of phrase-level
phonological domains (Section 3.2). A second, related argument against PHT concerns patterns
of overlap between phonological domains. PHT makes at least the following claims about the
ways in which phrasal domains can be related to each other:
 Nesting. If the domains of two processes overlap, they must be nested, with one domain
containing the other entirely (also known as proper bracketing; see Ito & Mester 2009).
 Layering. If the domains of two processes overlap, one of them should be consistently larger
than (i.e., contain) the other, in all contexts.
It has been suggested that both the nesting and the layering requirements of PHT are too strong.
Seidl (2001), drawing on Akinlabi & Liberman (2000), argues that Yoruba falsifies the nesting
provision of PHT. Yoruba has a ban on two adjacent identical tones within the same word. This
restriction also applies between a verb and following enclitic, but not between verbs and preceding
proclitics, which suggests the domain structure [proclitic–[verb–enclitic]]. Some dialects of Yoruba
have another process of vowel harmony which takes the proclitic–verb sequence as its domain,
suggesting the structure [[proclitic–verb]–enclitic]. This appears to be a violation of nesting, in that
the verb must belong to two different domains but these domains are only partially overlapping—
neither one is contained within the other (see also Chen 1987; Pak 2005, 2008, section 6.4.1;
Samuels 2009, section 5.4.4).
Seidl (2001) and Pak (2008) argue against layering on the basis of phrasal processes in Luganda
(an argument originally due to Hyman et al. 1987). Luganda has a process which spreads high

www.annualreviews.org • The Syntax–Prosody Interface 161

 LI05CH08_Elfner ARI 29 November 2018 14:54

tone between certain words and a second process shortening word-final long vowels in particular
contexts. The domains of high-tone spreading and shortening do not stand in a consistent contain-
ment relation: Both {. . .[. . . .]SHORTEN . . .}SPREAD and [. . .{. . .}SPREAD . . .]SHORTEN are possible nestings of
these domains, in violation of layering.2
The apparent empirical advantages of PHT (Section 2.1) have been called into question as well.
For one, some phrasal sandhi processes seem to be sensitive to syntactic category distinctions. For
example, Puerto Rican Spanish deletes word-final stressed ["a] before mid vowels, but only in verbs
(Kaisse 1985, p. 128). Such phenomena are beyond the reach of basic indirect reference theories,
given the assumption that phrasal phonology cannot “see” syntactic category distinctions (see also
Nespor & Vogel 1986, chapter 2; Hayes 1990).
A pillar of indirect reference theories is the existence of mismatches (“nonisomorphisms”)
between syntax and the groupings of words which are relevant for phrasal phonology. Such mis-
matches demonstrate that phonological constituents cannot be reduced to syntactic constituents.
As another example, determiners in Kwak’wala form a syntactic unit with the following word(s),
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

but a phonological unit with the preceding word, as diagnosed by stress and segmental patterning
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

(Anderson 2005).
Such mismatches seem to support indirect reference theories, to the extent that they show that
syntactic constituency fails to determine the domains of phonological processes. Although sug-
gestive, such arguments must be evaluated on a case-by-case basis. Pak (2008, section 2.2.1) points
out that most direct reference theories do predict nonisomorphisms between syntactic domains
and phonological domains. This is because direct reference models typically apply phonological
processes on the basis of syntactic relations, rather than syntactic constituency as such. Other au-
thors have argued that some apparent nonisomorphisms actually reflect a misunderstanding of the
underlying syntax rather than nonisomorphism per se (e.g., Seidl 2001, Wagner 2010).3 Lastly,
some direct reference theories (e.g., Seidl 2001, Pak 2008) make use of rebracketing operations
which produce groupings of words that deviate from the syntax itself (see Section 2). The question,
then, is whether there exists a residue of nonisomorphism which truly cannot be accommodated
by principled direct reference theories of phrasal phonology.
Apart from these critiques of PHT, conceptual arguments against the prosodic hierarchy
(mostly drawing on considerations of parsimony) have been set forth by Samuels (2009) and
Scheer (2012a,b), among others. Readers are referred to those works for details.

4.2. Why Does Prosodic Hierarchy Theory Endure?

Despite these critiques, PHT remains the most widely accepted and widely practiced framework
for understanding the syntax–prosody interface. It is worth asking why. One reason is that PHT
(and the closely related “autosegmental-metrical” theory; Ladd 2008) has proven to be a useful
framework for analyzing phrasal phonology in a wide range of typologically diverse languages (for
examples, see, e.g., Gussenhoven 2004; Jun 2005, 2014; Ladd 2008). The indisputable empirical

2
Hyman et al. (1987) suggest that there is a principled reason why apparent violations of domain nesting in Luganda and
elsewhere often involve tonal processes: Such processes may be subject to their own groupings, independent of the prosodic
hierarchy as it determines segmental patterning (i.e., prosodic groupings are tier specific in an autosegmental sense).
3
Determining the correct underlying syntax is a thorny problem for all approaches to the syntax–phonology interface, not only
for indirect reference theories. For example, certain syntactic structures in both Kaisse (1985) (direct reference) and Nespor
& Vogel (1986) (indirect reference) are manifestly not the structures that most syntacticians would assume today (e.g., Seidl
2001, chapter 3). Similar problems persist in modern theories which give the syntactic “phase” (e.g., Chomsky 2001, 2008)
a central role in the syntax–phonology interface, as there exists no current consensus as to which syntactic units constitute
phases and which do not (see also Newell 2008, Scheer 2012a).

162 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

success of PHT thus provides indirect support for the validity of such a theory, at least in its broad
contours.
Another reason for the persistence of PHT may have to do with its treatment of eurythmic
constraints on phonological domains (Section 2.1). Phrase-level domains are sometimes condi-
tioned by factors which seem purely phonological in nature, often having to do with domain size
and rhythmic balancing. Ito & Mester (2007) analyze such effects as they arise in compounding in
Japanese. Two-member compounds can be parsed together in the same accent domain, (ω1 ω2 ),
but only when ω2 is short. When ω2 is longer (more than four moras), it must be parsed into
its own accent domain, [ω1 (ω2 )]. Similar effects can be observed for other levels of the prosodic
hierarchy, including ϕ (≈XP; e.g., Prieto 2005, Elfner 2015), ι (≈CP; Myrberg 2013), and lower-
level units such as the metrical foot (e.g., Selkirk 1980b, Prince 1991, Hayes 1995). Such effects
are naturally accommodated in indirect reference theories, which assume that eurythmic pressures
operate at a level of representation—prosodic structure—that is essentially phonological rather
than syntactic in nature (Section 2).
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Eurythmic constraints are less naturally accommodated in direct reference theories. Rotenberg
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

(1978) and Kaisse (1985) address in syntactic terms a subset of eurythmic effects—those which
distinguish between branching and nonbranching constituents (see also Inkelas & Zec 1995, Pak
2005, and references therein). The full range of eurythmic effects cannot be analyzed in the same
way (e.g., the size effects in Japanese compounding mentioned above are clearly nonsyntactic).
To the extent that direct reference theories can account for such effects, it appears that they must
invoke at least some level of postsyntactic representation (as in Seidl 2001; Pak 2005, 2008) or,
alternatively, delegate such matters to processing and phonetic implementation (Wagner 2005;
see Section 5 for further discussion). The apparent need for extra levels of representation of course
brings direct reference theories closer to indirect reference theories such as PHT (see Section 2),
perhaps lessening their overall appeal.
Lastly, counterexamples notwithstanding, it appears that most phrasal phonological processes
do have the properties enumerated in Sections 2.1, 3.2, and 4: insensitivity to syntactic categories,
domain clustering, domain layering and nesting, and so on. These tendencies—all predicted by
PHT—must be explained in some other way in direct reference theories.4
Direct and indirect theories are typically pitted against each other because they seem to offer
competing explanations for the same set of facts. An alternative view (suggested to us by a reviewer)
is that there are simply two kinds of phrasal phonology: one conditioned by abstract prosodic
structure, as in indirect reference theories, and another conditioned by the syntax itself, as in
direct reference theories (see also Wagner 2012 and Section 5, below). For instance, tone sandhi in
Xiamen and Taiwanese appears to be conditioned by morphosyntactic structure (Chen 1987, Tsay
& Myers 1996) and is correspondingly insensitive to factors like speech rate and the presence of
pause between words. In contrast, third-tone sandhi in Beijing Mandarin appears to be conditioned
by prosodic structure, as it applies in domains which do not match the syntax, and does show
sensitivity to factors such as speech rate (Chen 2000, Chen & Yuan 2007). It may be, then, that the
tension between direct and indirect reference in phrasal phonology has persisted because phrasal
phonology may be sensitive to either syntactic or prosodic domains. This ambiguity may explain
some apparent violations of domain layering and nesting (see Section 4). In French, for example,
the domains in which liaison applies do not seem to align with intonational domains, but liaison

4
Direct reference theories can account for some of these tendencies, at least in principle (Elordieta 2008, Pak 2008). For
example, direct reference theories which assume that sandhi rules are conditioned only by c-command relations predict that
such rules should be insensitive to syntactic category distinctions.

www.annualreviews.org • The Syntax–Prosody Interface 163

 LI05CH08_Elfner ARI 29 November 2018 14:54

is arguably conditioned by syntax rather than prosodic structure (Kaisse 1985, Pak 2008, Wagner
2012).

5. GRADIENCY, PROCESSING, AND PRODUCTION PLANNING

So far, this review has focused primarily on the grammatical interface between syntax and prosody:
how syntactic domains map onto prosodic domains and how the boundaries of these domains may
be demarcated in terms of phonetic and phonological processes. This section considers the role
of sentence processing and production planning, each of which plays some role in the creation
and demarcation of prosodic domains. For reviews and discussion of the interface of prosody
with sentence processing, production planning, and other psycholinguistic factors, see Wagner &
Watson (2010), Cole (2015), and Wagner (2015).
An assumption implicit in PHT is that the distinction between domain levels is categorical
rather than gradient. While many cues to prosodic boundaries are gradient in character, such as
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

duration and pitch scaling, PHT predicts that speakers should be able to associate such distinctions
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

with a limited set of universal prosodic categories, which align with syntactic constituents in
predictable ways.
To what extent is this the case? As discussed in Section 3.3, evidence for the recursion of
prosodic domains suggests that speakers make use of a greater number of prosodic domains than
the prosodic hierarchy allows, at least under the confines of the strict layer hypothesis, and, further,
that the number of phonetically distinct domains is directly correlated with the syntactic complexity
of the utterance (Ladd 1986, 1988; Kubozono 1989; Wagner 2005, 2010). PHT predicts that the
three basic prosodic categories (ω, ϕ, ι) will be present in every utterance, regardless of syntactic
complexity (though see discussion in Sections 3.2–3.4).
Gradient cues to prosodic domains include durational cues involving final lengthening, the
use of prosodic pauses, and initial strengthening, as well as cues involving fundamental frequency
(pitch), such as the scaling of pitch accents and pitch resets. The magnitude of such gradient cues
appears to depend on relative boundary strength: Stronger cues occur at stronger boundaries.
For example, final lengthening, which involves the lengthening of segments and syllables before
prosodic boundaries, is positively correlated with the relative strength of the boundary (Klatt
1975, Price et al. 1991, Wightman et al. 1992, Shattuck-Hufnagel & Turk 1996, Byrd & Saltzman
1998, Turk & Shattuck-Hufnagel 2000). Similarly, prosodic pauses, although not required even
at relatively strong prosodic boundaries, pattern in conjunction with final lengthening such that
pauses both are more likely to appear and have a relatively longer duration at strong prosodic
boundaries as opposed to weak ones (Ferreira 1991, 1993; Watson & Gibson 2004). Finally,
the phonetic properties of segments in domain-initial positions, such as duration and degree of
stricture, also correlate with the relative strength of the prosodic boundary, resulting in what has
been termed domain-initial strengthening (e.g., Fougeron & Keating 1997, Keating et al. 2003).
With respect to fundamental frequency, the scaling of pitch accents and the use of pitch reset
within an utterance are similarly correlated with relative boundary strength and the presence of
recursion in prosodic structure (Ladd 1988, de Pijper & Sanderman 1994, Féry & Truckenbrodt
2005, Elordieta 2015).
In terms of perception, research has shown that listeners are fairly adept at discerning relative
differences in the strength of prosodic boundaries but are not able to reliably categorize boundaries
in terms of their strength (Price et al. 1991, de Pijper & Sanderman 1994). This observation
may suggest that listeners do not interpret gradient cues as a reflection of specific domain types
(i.e., specific categories), be they syntactically or prosodically defined (Sections 2 and 4). Recent
developments exploring the recursion of prosodic categories (Sections 3.2 and 3.3) may help

164 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

capture gradiency in the perception of relative boundary strength while retaining the advantages
of PHT and indirect reference more generally (Selkirk 2009; Ito & Mester 2012, 2013; Myrberg
2013; Elordieta 2015).
To some extent, relative boundary strength may be tied to syntactic structure and the relative
complexity of syntactic constituents. Algorithmic models, common in relatively early research on
the syntax–prosody interface, attempt to derive relative boundary strength (particularly durational
cues to prosodic boundaries) directly from the depth of embedding found in syntactic structure
(Cooper & Paccia-Cooper 1980, Gee & Grosjean 1983, Ferreira 1993). More recently, such
algorithmic approaches have been tied to sentence processing and relative boundary strength, and
more specifically to the notion that the relative length and/or complexity of syntactic material
affects processing time (Watson & Gibson 2004, Watson et al. 2006). Prosodic boundaries are
thus more likely to occur following a complex syntactic constituent (allowing time for recovery)
and more likely to occur preceding a complex syntactic constituent (allowing time for planning).
Speech processing and production planning have also been tied to patterns of variability and
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

optionality in sandhi processes. Some authors have argued that variability in the application of
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

sandhi processes reflects variation in how much of an utterance is concurrently planned during
real-time speech production (see Wagner 2012, Kilbourn-Ceron et al. 2017, Kilbourn-Ceron
& Sonderegger 2018, and references therein). Such variability may also be captured by formal
algorithms which build prosodic domains in a variable or stochastic fashion (as in the account of
Bengali in Hayes & Lahiri 1991, section 1). Still, to the extent that appeals to production planning
might offer a more principled account of variability in phrasal phonology, such variability may
no longer provide an argument for indirect over direct reference theories of phrasal domains
(Section 2).
For example, Kilbourn-Ceron et al. (2017) discuss variability in the application of flapping
in English, which variably occurs across word boundaries and syntactic boundaries of varying
sizes, yet is less likely to occur as the relative strength of the prosodic boundary increases. They
propose that the application of flapping depends, at least in part, on the relative likelihood that
upcoming syntactic material has been planned at the time of production, such that flapping will
occur only if the two words exhibiting the conditioning environment are within the same planning
window. This model predicts that the application of sandhi processes, while partially dependent on
syntactic structure (because words that are closely related syntactically are more likely to be planned
together), will also be variable to the extent that speakers may not fully plan their utterances. The
role of factors such as lexical frequency and speech rate can thus be directly tied to production
planning, and fit well with evidence that prosodic phrasing and prominence are affected by the
relative predictability of lexical items within a given utterance (Aylett & Turk 2004, Turk 2010).
Finally, there is evidence to suggest that the use of prosodic cues is not absolute and is condi-
tioned, at least in part, by the discourse and situational context in which speech is uttered, includ-
ing information structural notions such as focus, givenness, and topicality (for details, see Féry &
Ishihara 2016 and references therein). For example, while it is incontrovertible that prosody may
be used to disambiguate otherwise ambiguous utterances (e.g., Lehiste 1973), speakers are more
likely to employ such prosodic cues when the context allows for multiple interpretations, and
are less likely to use prosodic cues when only a single interpretation is possible, that is, when the
utterance is syntactically, but not contextually, ambiguous (Snedeker & Trueswell 2003). The role
of context in prosody is complex, involving the integration of a number of linguistic modules; al-
though the focus of this review has been on the interface between prosody and syntactic structure,
it is important to keep in mind that even this relationship may be affected by a number of factors
external to a straightforward mapping between syntactic structure and prosodic domains. For a
thorough overview of the role of context in prosody, see Cole (2015).

www.annualreviews.org • The Syntax–Prosody Interface 165

 LI05CH08_Elfner ARI 29 November 2018 14:54

6. CONCLUSION
We hope to have shown in this brief overview that the study of phrasal phonology—including
segmental patterning, intonational patterns, and other phenomena at higher domains—can have
wide-ranging consequences for our understanding of grammar. Phrasal phonology is deeply in-
tertwined with syntax, semantics, pragmatics, phonetics, psycholinguistics, and other components
of language behavior beyond phonology itself. Research on phrasal phonology, and its relation
to syntax, has potentially profound consequences for our understanding of the architecture of
grammar, raising issues related to the modularity and independence of different types of linguistic
knowledge, as well as the traditional division between performance and competence (Chomsky
1965). Phrasal phonology has been an area of intense scrutiny for half a century, but in our view,
many of the most interesting and important issues in this area remain to be definitively settled.
We also expect that new puzzles and problems will emerge from the continued development of
theories which integrate quantitative evidence from experiments and corpus studies with more
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

traditional qualitative descriptions of phonological patterning at the phrase level, and as data from
a wider range of languages are brought into the fold.
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings that
might be perceived as affecting the objectivity of this review.

ACKNOWLEDGMENTS
We thank Jennifer Bellik and an anonymous reviewer for helpful comments on an earlier version
of this article.

LITERATURE CITED
Agbayani B, Golston C. 2010. Phonological movement in Classical Greek. Language 86:133–67
Agbayani B, Golston C. 2016. Phonological constituents and their movement in Latin. Phonology 33:1–42
Aissen J. 2000. Prosodic conditions on anaphora and clitics in Jakaltek. In The Syntax of Verb Initial Languages,
ed. A Carnie, E Guilfoyle, pp. 185–200. Oxford, UK: Oxford Univ. Press
Akinlabi A, Liberman M. 2000. The tonal phonology of Yoruba clitics. In Clitics in Phonology, Morphology and
Syntax, ed. B Gerlach, J Grijzenhout, pp. 31–62. Amsterdam: Benjamins
Anderson SR. 2005. Aspects of the Theory of Clitics. Oxford, UK: Oxford Univ. Press
Anttila A. 2016. Phonological effects on syntactic variation. Annu. Rev. Linguist. 2:115–37
Arnhold A. 2014. Prosodic structure and focus realization in West Greenlandic. See Jun 2014, pp. 216–51
Arnhold A, Compton R, Elfner E. 2018. Prosody and wordhood in South Baffin Inuktitut. In Proceedings of
the 21st Workshop on Structure and Constituency in Languages of the Americas, ed. M Keough, N Weber,
A Anghelescu, S Chen, E Guntly, et al., pp. 30–39. Vancouver: Univ. B. C. Work. Pap. Linguist.
Arregi K. 2006. Stress and islands in Northern Bizkaian Basque. In Studies in Historical and Basque Linguistics:
Dedicated to the Memory of R. L. Trask, ed. J Hualde, JA Lakarra, pp. 81–106. Donostia, Spain: Diput.
Foral Gipuzkoa
Aylett M, Turk A. 2004. The smooth signal redundancy hypothesis: a functional explanation for relationships
between redundancy, prosodic prominence, and duration in spontaneous speech. Lang. Speech 47:31–56
Beckman M, Pierrehumbert J. 1986. Intonational structure in English and Japanese. Phonology 3:255–309
Bennett R. 2013. The uniqueness of metrical structure: rhythmic phonotactics in Huariapano. Phonology
30:355–98
Bennett R. 2015. Review of Sun-Ah Jun (ed.) 2014. Prosodic Typology II: The Phonology of Intonation and Phrasing.
Oxford: Oxford Univ. Press. xv + 587. Phonology 32:337–50

166 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

Bennett R. 2018. Recursive prosodic words in Kaqchikel (Mayan). Glossa 3:67

Bennett R, Elfner E, McCloskey J. 2016. Lightest to the right: an apparently anomalous displacement in Irish.
Linguist. Inq. 47:169–234
Bickel B, Hildebrandt K, Schiering R. 2009. The distribution of phonological word domains: a probabilistic
typology. In Phonological Domains: Universals and Deviations, ed. J Grijzenhout, B Kabak, pp. 47–75. Berlin:
Mouton de Gruyter
Booij G. 1996. Cliticization as prosodic integration: the case of Dutch. Linguist. Rev. 13:219–42
Byrd D, Saltzman E. 1998. Intragestural dynamics of multiple prosodic boundaries. J. Phon. 26:173–99
Chen M. 1987. The syntax of Xiamen tone sandhi. Phonology 4:109–50
Chen M. 1990. What must phonology know about syntax? See Inkelas & Zec 1990, pp. 19–46
Chen MY. 2000. Tone Sandhi: Patterns Across Chinese Dialects. Cambridge, UK: Cambridge Univ. Press
Chen Y, Yuan J. 2007. A corpus study of the 3rd tone sandhi in Standard Chinese. In Proceedings of the
8th Annual Conference of the International Speech Communication Association (INTERSPEECH 2007), ed.
H van Hamme, van Son R, pp. 2749–52. Baixas, Fr.: Int. Speech Commun. Assoc.
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Chomsky N. 1965. Aspects of the Theory of Syntax. Cambridge, MA: MIT Press
Chomsky N. 1995. The Minimalist Program. Cambridge, MA: MIT Press
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

Chomsky N. 2001. Derivation by phase. In Ken Hale: A Life in Language, ed. M Kenstowicz, pp. 1–52.
Cambridge, MA: MIT Press
Chomsky N. 2008. On phases. In Foundational Issues in Linguistic Theory: Essays in Honor of Jean-Roger Vergnaud,
ed. R Freidin, C Otero, ML Zubizarreta, pp. 133–66. Cambridge, MA: MIT Press
Chomsky N, Halle M. 1968. The Sound Pattern of English. New York: Harper & Row
Cinque G. 1993. A null theory of phrase and compound stress. Linguist. Inq. 24:239–398
Clemens LE. 2014. Prosodic noun incorporation and verb-initial syntax. PhD thesis, Harvard Univ., Cambridge,
MA
Cole J. 2015. Prosody in context: a review. Lang. Cogn. Neurosci. 30:1–31
Cooper WE, Paccia-Cooper J. 1980. Syntax and Speech. Cambridge, MA: Harvard Univ. Press
de Pijper JR, Sanderman A. 1994. On the perceptual strength of prosodic boundaries and its relation to
suprasegmental cues. J. Acoust. Soc. Am. 96:2037–47
Dresher BE. 1994. The prosodic basis of the Tiberian Hebrew system of accents. Language 70:1–52
Elfner E. 2012. Syntax–prosody interactions in Irish. PhD thesis, Univ. Mass., Amherst
Elfner E. 2015. Recursion in prosodic phrasing: evidence from Connemara Irish. Nat. Lang. Linguist. Theory
33:1169–208
Elfner E. 2018. The syntax–prosody interface: current theoretical approaches and outstanding questions.
Linguist. Vanguard 4:1–14
Elordieta G. 2007. Minimum size constraints on intermediate phrases. In Proceedings of the 16th International
Congress of the Phonetic Sciences (ICPhS XVI), ed. J Trouvain, WJ Barry, pp. 1021–24. London: Int. Phon.
Assoc.
Elordieta G. 2008. An overview of theories of the syntax–phonology interface. Int. J. Basque Linguist. Philol.
42:209–86
Elordieta G. 2015. Recursive phonological phrasing in Basque. Phonology 32:49–78
Ferreira F. 1991. Effects of length and syntactic complexity on initiation times for prepared utterances.
J. Mem. Lang. 30:210–33
Ferreira F. 1993. Creation of prosody during sentence production. Psychol. Rev. 100:233–53
Féry C. 2016. Intonation and Prosodic Structure. Cambridge, UK: Cambridge Univ. Press
Féry C, Ishihara S, ed. 2016. The Oxford Handbook of Information Structure. Oxford, UK: Oxford Univ. Press
Féry C, Truckenbrodt H. 2005. Sisterhood and tonal scaling. Stud. Linguist. 59:223–43
Fitzpatrick-Cole J. 1996. Reduplication meets the phonological phrase in Bengali. Linguist. Rev. 13:305–56
Fougeron C, Keating PA. 1997. Articulatory strengthening at edges of prosodic domains. J. Acoust. Soc. Am.
101:3728–40
Frota S. 2012. Prosodic structure, constituents and their implementation. In The Oxford Handbook of Laboratory
Phonology, ed. A Cohn, C Fougeron, M Huffman, pp. 255–65. Oxford, UK: Oxford Univ. Press

www.annualreviews.org • The Syntax–Prosody Interface 167

 LI05CH08_Elfner ARI 29 November 2018 14:54

Frota S, Vigário M. 2013. Toni Borowsky, Shigeto Kawahara, Takahito Shinya and Mariko Sugahara (eds.)
2012. Prosody Matters: Essays in Honor of Elisabeth Selkirk. (Advances in Optimality Theory.) Sheffield &
Bristol, Conn.: Equinox. Pp. xv + 528. Phonology 30:165–72
Gee JP, Grosjean F. 1983. Performance structures: a psycholinguistic and linguistic appraisal. Cogn. Psychol.
15:411–58
Ghini M. 1993. Phi-formation in Italian: a new proposal. Tor. Work. Pap. Linguist. 12:41–78
Gussenhoven C. 2004. The Phonology of Tone and Intonation. Cambridge, UK: Cambridge Univ. Press
Halpern A. 1995. Topics in the displacement and morphology of clitics. PhD thesis, Stanford Univ., Stanford, CA
Hayes B. 1990. Precompiled phrasal phonology. See Inkelas & Zec 1990, pp. 85–108
Hayes B. 1995. Metrical Stress Theory: Principles and Case Studies. Chicago: Univ. Chicago Press
Hayes B, Lahiri A. 1991. Bengali intonational phonology. Nat. Lang. Linguist. Theory 9:47–96
Hyman L, Katamba F, Walusimbi L. 1987. Luganda and the strict layer hypothesis. Phonology 4:87–108
Inkelas S. 1990. Prosodic Constituency in the Lexicon. New York: Garland
Inkelas S, Zec D. 1990. The Phonology–Syntax Connection. Chicago: Univ. Chicago Press
Inkelas S, Zec D. 1995. Syntax–phonology interface. In The Handbook of Phonological Theory, ed. JA Goldsmith,
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

pp. 535–49. Cambridge, MA/Oxford, UK: Blackwell

Access provided by 36.227.193.71 on 03/15/24. For personal use only.

Ishihara S. 2015. Syntax–phonology interface. In The Handbook of Japanese Phonetics and Phonology, ed.
H Kubozono, pp. 569–618. Berlin: Mouton de Gruyter
Ito J, Mester A. 2007. Prosodic adjunction in Japanese compounds. In Formal Approaches to Japanese Linguistics,
ed. Y Miyamoto, M Ochi, pp. 97–111. Cambridge, MA: MIT Work. Pap. Linguist.
Ito J, Mester A. 2009. The extended prosodic word. In Phonological Domains: Universals and Derivations, ed.
J Grijzenhout, B Kabak, pp. 135–94. Berlin/New York: Mouton de Gruyter
Ito J, Mester A. 2010. The onset of the prosodic word. In Phonological Argumentation: Essays on Evidence and
Motivation, ed. S Parker, pp. 227–60. London: Equinox
Ito J, Mester A. 2012. Recursive prosodic phrasing in Japanese. In Prosody Matters, ed. T Borowsky, S Kawahara,
T Shinya, M Sugahara, pp. 280–303. London: Equinox
Ito J, Mester A. 2013. Prosodic subcategories in Japanese. Lingua 124:20–40
Jun S-A. 1998. The accentual phrase in the Korean prosodic hierarchy. Phonology 15:189–226
Jun S-A, ed. 2005. Prosodic Typology: The Phonology of Intonation and Phrasing. Oxford, UK: Oxford Univ. Press
Jun S-A, ed. 2014. Prosodic Typology II: The Phonology of Intonation and Phrasing. Oxford, UK: Oxford Univ.
Press
Jun S-A, Elordieta G. 1997. Intonational structure in Lekeitio Basque. In Intonation: Theory, Models and
Applications, ed. A Botinis, G Kouroupetroglou, G Carayiannis, pp. 193–96. Athens: Int. Speech Commun.
Assoc.
Kaisse EM. 1985. Connected Speech: The Interaction of Syntax and Phonology. San Diego: Academic
Kaisse EM, Zwicky A, ed. 1987. Phonology Yearbook 4: Syntactic Conditions on Phonological Rules. Cambridge,
UK: Cambridge Univ. Press
Kandybowicz J. 2009. Embracing edges: syntactic and phono-syntactic edge sensitivity in Nupe. Nat. Lang.
Linguist. Theory 27:305–44
Kandybowicz J. 2015. On prosodic vacuity and verbal resumption in Asante Twi. Linguist. Inq. 46:243–72
Kandybowicz J. 2017. On prosodic variation and the distribution of wh-in-situ. Linguist. Var. 17:111–48
Keating PA, Cho T, Fougeron C, Hsu C-S. 2003. Domain-initial articulatory strengthening in four lan-
guages. In Phonetic Interpretation, ed. J Local, R Ogden, R Temple, pp. 145–63. Pap. Lab. Phonol. vol. 6.
Cambridge, UK: Cambridge Univ. Press
Kilbourn-Ceron O, Sonderegger M. 2018. Boundary phenomena and variability in Japanese high vowel
devoicing. Nat. Lang. Linguist. Theory 36:175–217
Kilbourn-Ceron O, Wagner M, Clayards M. 2017. The effect of production planning locality on external
sandhi: a study in /t/. In Proceedings of the 52nd Annual Meeting of the Chicago Linguistic Society. Chicago:
Chicago Linguist. Soc. 15 pp. http://ling.auf.net/lingbuzz/003119
Kisseberth C, Abasheikh M. 1974. Vowel length in Chi Mwi:ni: a case study of the role of grammar in
phonology. In CLS 10 Parasession on Natural Phonology, pp. 193–209. Chicago: Chicago Linguist. Soc.
Klatt D. 1975. Vowel lengthening is syntactically determined in a connected discourse. J. Phon. 3:129–40

168 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

Kubozono H. 1989. Syntactic and rhythmic effects on downstep in Japanese. Phonology 6:39–67
Ladd DR. 1986. Intonational phrasing: the case for recursive prosodic structure. Phonology 3:311–40
Ladd DR. 1988. Declination ‘reset’ and the hierarchical organization of utterances. J. Acoust. Soc. Am. 84:530–
44
Ladd DR. 2008. Intonational Phonology. Cambridge, UK: Cambridge Univ. Press. 2nd ed.
Lehiste I. 1973. Phonetic disambiguation of syntactic ambiguity. Glossa 7:107–23
Martı́nez-Paricio V. 2013. An exploration of minimal and maximal metrical feet. PhD thesis, Univ. Tromsø,
Tromsø, Nor.
Martı́nez-Paricio V, Kager R. 2015. The binary-to-ternary rhythmic continuum in stress typology: layered
feet and non-intervention constraints. Phonology 32:459–504
McCarthy JJ. 1993. A case of surface constraint violation. Can. J. Linguist. 38:169–95
McCawley JD. 1968. The Phonological Component of a Grammar of Japanese. The Hague: Mouton
Myrberg S. 2013. Sisterhood in prosodic branching. Phonology 30:73–124
Nespor M, Vogel I. 1986. Prosodic Phonology. Dordrecht, Neth.: Foris
Newell H. 2008. Aspects of the morphology and phonology of phases. PhD thesis, McGill Univ., Montreal, Can.
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Newell H, Piggott G. 2014. Interactions at the syntax–phonology interface: evidence from Ojibwe. Lingua
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

150:332–62
Odden D. 1987. Kimatuumbi phrasal phonology. See Kaisse & Zwicky 1987, pp. 13–36
Padgett J. 2014. On the origins of the prosodic word in Russian. Paper presented at the 7th International Conference
on Speech Prosody, Dublin, May 20–23
Pak M. 2005. Explaining branchingness effects in phrasal phonology. In Proceedings of the 24th West Coast
Conference on Formal Linguistics, ed. J Alderete, C-H Han, A Kochetov, pp. 308–16. Somerville, MA:
Cascadilla
Pak M. 2008. The postsyntactic derivation and its phonological reflexes. PhD thesis, Univ. Pa., Philadelphia
Peperkamp S. 1997. Prosodic Words. The Hague: Holland Acad. Graph.
Pierrehumbert J, Beckman M. 1988. Japanese Tone Structure. Cambridge, MA: MIT Press
Price P, Ostendorf M, Shattuck-Hufnagel S, Fong G. 1991. The use of prosody in syntactic disambiguation.
J. Acoust. Soc. Am. 90:2956–70
Prieto P. 2005. Syntactic and eurhythmic constraints on phrasing decisions in Catalan. Stud. Linguist. 59:194–
222
Prieto P. 2014. The intonational phonology of Catalan. See Jun 2014, pp. 43–80
Prince A. 1991. Quantitative consequences of rhythmic organization. In CLS26(2): Papers from the Parasession
on the Syllable in Phonetics and Phonology, ed. K Deaton, M Noske, M Ziolkowski, pp. 355–98. Chicago:
Chicago Linguist. Soc.
Richards N. 2010. Uttering Trees. Cambridge, MA: MIT Press
Richards N. 2016. Contiguity Theory. Cambridge, MA: MIT Press
Rotenberg J. 1978. The Syntax of Phonology. Cambridge, MA: MIT
Ruppel A. 2017. The Cambridge Introduction to Sanskrit. Cambridge, UK: Cambridge Univ. Press
Ryan K. 2018. Prosodic end-weight reflects phrasal stress. Nat. Lang. Linguist. Theory. In press. https://doi.
org/10.1007/s11049-018-9411-6
Sabbagh J. 2013. Word order and prosodic-structure constraints in Tagalog. Syntax 17:40–89
Samuels B. 2009. The structure of phonological theory. PhD thesis, Harvard Univ., Cambridge, MA
Scheer T. 2010. A Guide to Morphosyntax–Phonology Interface Theories: How Extra-Phonological Information Is
Treated in Phonology Since Trubetzkoy’s Grenzsignale. Berlin: de Gruyter
Scheer T. 2012a. Chunk definition in phonology: prosodic constituency versus phase structure. In Modules
and Interfaces, ed. M Bloch-Trojnar, A Bloch-Rozmej, pp. 221–53. Lublin, Pol.: Cathol. Univ. Lublin
Scheer T. 2012b. Direct Interface and One-Channel Translation. Boston: de Gruyter
Schiering R, Bickel B, Hildebrandt K. 2010. The prosodic word is not universal, but emergent. J. Linguist.
46:657–709
Seidl A. 2001. Minimal Indirect Reference: A Theory of the Syntax–Phonology Interface. London: Routledge
Selkirk EO. 1980a. Prosodic domains in phonology: Sanskrit revisited. In Juncture, ed. M Aronoff, M-L Kean,
pp. 107–29. Saratoga, CA: Anma Libri

www.annualreviews.org • The Syntax–Prosody Interface 169

 LI05CH08_Elfner ARI 29 November 2018 14:54

Selkirk EO. 1980b. The role of prosodic categories in English word stress. Linguist. Inq. 11:563–605
Selkirk EO. 1981 (1978). On prosodic structure and its relation to syntactic structure. In Nordic Prosody, ed.
T Fretheim, pp. 111–40. Trondheim, Nor.: Tapir
Selkirk EO. 1984. Phonology and Syntax: The Relation Between Sound and Structure. Cambridge, MA: MIT Press
Selkirk EO. 1986. On derived domains in sentence phonology. In Phonology Yearbook 3, ed. C Ewen, J Anderson,
pp. 371–405. Cambridge, UK: Cambridge Univ. Press
Selkirk EO. 1995. The prosodic structure of function words. In Papers in Optimality Theory, ed. J Beckman,
LW Dickey, S Urbanczyk, pp. 439–70. Amherst, MA: Grad. Linguist. Stud. Assoc.
Selkirk EO. 2000. The interaction of constraints on prosodic phrasing. In Prosody: Theory and Experiment, ed.
M Horne, pp. 231–61. Dordrecht, Neth.: Kluwer
Selkirk EO. 2009. On clause and intonational phrase in Japanese: the syntactic grounding of prosodic con-
stituent structure. Gengo Kenkyu 136:35–73
Selkirk EO. 2011. The syntax–phonology interface. In The Handbook of Phonological Theory, ed. JA Goldsmith,
J Riggle, ACL Yu, pp. 435–84. Oxford, UK: Wiley Blackwell. 2nd ed.
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Selkirk EO, Lee S, ed. 2015. Phonology 32(1)

Selkirk EO, Shen T. 1990. Prosodic domains in Shanghai Chinese. See Inkelas & Zec 1990, pp. 313–37
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

Selkirk EO, Tateishi K. 1991. Syntax and downstep in Japanese. In Interdisciplinary Approaches to Language:
Essays in Honor of S.-Y. Kuroda, ed. C Georgopoulos, R Ishihara, pp. 519–44. Dordrecht, Neth.: Kluwer
Shattuck-Hufnagel S, Turk A. 1996. A prosody tutorial for investigators of auditory sentence processing.
J. Psycholinguist. Res. 25:193–247
Shih S. 2017. Phonological influences in syntactic alternations. In The Morphosyntax–Phonology Connection, ed.
V Gribanova, S Shih, pp. 223–54. Oxford, UK: Oxford Univ. Press
Snedeker J, Trueswell J. 2003. Using prosody to avoid ambiguity: effects of speaker awareness and referential
context. J. Mem. Lang. 48:103–30
Truckenbrodt H. 1995. Phonological phrases: their relation to syntax, focus, and prominence. PhD thesis, MIT,
Cambridge, MA
Truckenbrodt H. 1999. On the relation between syntactic phrases and phonological phrases. Linguist. Inq.
30:219–56
Truckenbrodt H. 2002. Variation in p-phrasing in Bengali. Linguist. Var. Yearb. 2:259–303
Truckenbrodt H. 2007. The syntax–phonology interface. In The Cambridge Handbook of Phonology, ed.
P de Lacy, pp. 435–56. Cambridge, UK: Cambridge Univ. Press
Tsay J, Myers J. 1996. Taiwanese tone sandhi as allomorph selection. In Proceedings of the 22nd Annual Meeting
of the Berkeley Linguistics Society: General Session and Parasession on the Role of Learnability in Grammatical
Theory, pp. 395–405. Berkeley, CA: Berkeley Linguist. Soc.
Turk A. 2010. Does prosodic constituency signal relative predictability? A Smooth Signal Redundancy hy-
pothesis. Lab. Phonol. 1:227–62
Turk A, Nakai S, Sugahara M. 2006. Acoustic segment durations in prosodic research: a practical guide. In
Methods in Empirical Prosody Research, ed. S Sudhoff, D Lenertova, RMS Pappert, P Augurzky, I Mleinek,
et al., pp. 1–28. Berlin: de Gruyter
Turk A, Shattuck-Hufnagel S. 2000. Word-boundary-related duration patterns in English. J. Phon. 28:397–
440
Vigário M. 1999. On the prosodic status of stressless function words in European Portuguese. In Studies on
the Phonological Word, ed. TA Hall, U Kleinhenz, pp. 255–94. Amsterdam: Benjamins
Vigário M. 2003. The Prosodic Word in European Portuguese. Berlin: Mouton de Gruyter
Vigário M. 2010. Prosodic structure between the prosodic word and the phonological phrase: recursive nodes
or an independent domain? Linguist. Rev. 27:485–530
Vogel I. 2009. The status of the Clitic Group. In Phonological Domains: Universals and Derivations, ed.
J Grijzenhout, B Kabak, pp. 15–46. Berlin: Mouton de Gruyter
Wagner M. 2005. Prosody and recursion. PhD thesis, MIT, Cambridge, MA
Wagner M. 2010. Prosody and recursion in coordinate structures and beyond. Nat. Lang. Linguist. Theory
28:183–237

170 Bennett · Elfner

 LI05CH08_Elfner ARI 29 November 2018 14:54

Wagner M. 2012. Locality in phonology and production planning. In Proceedings of Phonology in the 21st
Century: Papers in Honour of Glyne Piggott, ed. J Loughran, A McKillen, pp. 1–18. Montreal, Can.: McGill
Work. Pap.
Wagner M. 2015. Phonological evidence in syntax. In Syntax—Theory and Analysis. An International Handbook,
ed. T Kiss, A Alexiadou, pp. 1154–98. Berlin: Mouton de Gruyter
Wagner M, Watson D. 2010. Experimental and theoretical advances in phonology: z review. Lang. Cogn.
Process. 25:905–45
Watson D, Breen M, Gibson E. 2006. The role of syntactic obligatoriness in the production of intonational
boundaries. J. Exp. Psychol. Learn. Mem. Cogn. 32:1045–56
Watson D, Gibson E. 2004. The relationship between intonational phrasing and syntactic structure in language
production. Lang. Cogn. Process. 19:713–55
Werle A. 2009. Word, phrase, and clitic prosody in Bosnian, Serbian, and Croatian. PhD thesis, Univ. Mass.,
Amherst
Wightman CW, Shattuck-Hufnagel S, Ostendorf M, Price PJ. 1992. Segmental durations in the vicinity of
prosodic phrase boundaries. J. Acoust. Soc. Am. 92:1707–17
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Zec D, Inkelas S. 1990. Prosodically constrained syntax. See Inkelas & Zec 1990, pp. 365–78
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

www.annualreviews.org • The Syntax–Prosody Interface 171

 LI05_TOC ARI 1 December 2018 14:7

Annual Review
of Linguistics

Contents Volume 5, 2019

The Impossibility of Language Acquisition (and How They Do It)

Lila R. Gleitman, Mark Y. Liberman, Cynthia A. McLemore,
and Barbara H. Partee p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1
How Consonants and Vowels Shape Spoken-Language Recognition
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

Thierry Nazzi and Anne Cutler p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p25

Access provided by 36.227.193.71 on 03/15/24. For personal use only.

Cross-Modal Effects in Speech Perception

Megan Keough, Donald Derrick, and Bryan Gick p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p49
Computational Modeling of Phonological Learning
Gaja Jarosz p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p67
Corpus Phonetics
Mark Y. Liberman p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p91
Relations Between Reading and Speech Manifest Universal
Phonological Principle
Donald Shankweiler and Carol A. Fowler p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 109
Individual Differences in Language Processing: Phonology
Alan C.L. Yu and Georgia Zellou p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 131
The Syntax–Prosody Interface
Ryan Bennett and Emily Elfner p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 151
Western Austronesian Voice
Victoria Chen and Bradley McDonnell p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 173
Dependency Grammar
Marie-Catherine de Marneffe and Joakim Nivre p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 197
Closest Conjunct Agreement
Andrew Nevins and Philipp Weisser p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 219
Three Mathematical Foundations for Syntax
Edward P. Stabler p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 243
Response Systems: The Syntax and Semantics of Fragment Answers
and Response Particles
M.Teresa Espinal and Susagna Tubau p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 261
 LI05_TOC ARI 1 December 2018 14:7

Distributivity in Formal Semantics

Lucas Champollion p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 289
The Syntax and Semantics of Nonfinite Forms
John J. Lowe p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 309
Semantic Anomaly, Pragmatic Infelicity, and Ungrammaticality
Márta Abrusán p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 329
Artificial Language Learning in Children
Jennifer Culbertson and Kathryn Schuler p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 353
What Defines Language Dominance in Bilinguals?
Jeanine Treffers-Daller p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 375
Annu. Rev. Linguist. 2019.5:151-171. Downloaded from www.annualreviews.org

The Advantages of Bilingualism Debate

Mark Antoniou p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 395
Access provided by 36.227.193.71 on 03/15/24. For personal use only.

The Austronesian Homeland and Dispersal

Robert Blust p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 417
Language Variation and Change in Rural Communities
Matthew J. Gordon p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 435
Language, Gender, and Sexuality
Miriam Meyerhoff and Susan Ehrlich p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 455

Errata
An online log of corrections to Annual Review of Linguistics articles may be found at
http://www.annualreviews.org/errata/linguistics