This monograph focuses on sentential stress in both informationally neutral and informationally non-neutral sentences within a phase-based framework. The resulting system relies solely on surface structure and is consistent with the Y-model assumed in generative grammar, contra related work by Zubizarreta (Reference Zubizarreta1998) and Selkirk (Reference Selkirk and Goldsmith1995) . The following issues figure prominently in the discussion: the distribution of primary and secondary stress, the observation that in both S(ubject)V(erb)O(bject) and SOV languages stress falls on the object in informationally neutral sentences, and the peculiar behavior of sentence stress in sentences involving unaccusatives and passives. A major achievement of this book is that it provides a unified approach to such a heterogeneous set of issues.
Chapter 1, ‘Setting the stage’, introduces the framework and summarizes the main proposal. Chapter 2, ‘Sentential stress: Phonological accounts’, claims that phonological accounts of sentential stress suffer from an overgeneration problem – they allow stress patterns that have not been attested across languages. In particular, it is shown that Phrasal Phonology and Metrical Grid Theory allow SOV languages where the default stress is on the subject, in spite of the fact that this pattern is not attested cross-linguistically. Furthermore, these proposals run into trouble because in passive and unaccusative sentences uttered in out-of-the-blue contexts, primary stress falls on the subject, as is shown in (1), in contrast to transitive sentences, as in (2) (stress is marked by underlining).
(1) A letter arrived for you today.
(2) John read a book.
To capture these facts in a phonological account would require construction-dependent mechanisms, which is an unnatural explanation at best.
In Chapter 3, ‘Sentential stress: Syntactic accounts’, it is shown that the stress facts involving passives and unaccusatives pose a problem also for Cinque (Reference Cinque1993), whereas Legate (Reference Legate2003), who specifically addresses this issue, faces empirical problems of a different order. Kahnemuyipour further discusses a number of problematic aspects of Zubizarreta's (Reference Zubizarreta1998) proposal in this chapter.
With regard to Cinque's account, the author shows that it is reliant on the head parameter – a fact that undermines this approach once Kayne's (Reference Kayne1994) Antisymmetry proposal is adopted. Furthermore, the unaccusative and passive facts as well as new data from German and Persian cast doubt on the validity of Cinque's proposal. More specifically, within his system stress is assigned to the most deeply embedded constituent, a fact at odds with data such as (1). Similarly, when adding manner or measure adverbs (left edge markers) in Persian SOV sentences (which results in SAdvOV order), the adverb is stressed, whereas in ditransitives, stress falls on the direct object and not the Prepositional Phrase, contrary to expectation.
Legate (Reference Legate2003) attempts to deal with the unaccusative and passive stress facts by arguing that stress is sensitive to the base-generated position of a constituent and that a moved constituent inherits stress from its lower copy. However, this does not make the right predictions for the behavior of bare wh-objects and topicalized objects, which are stressed in their base-generated position but unstressed in their surface position, as is shown in (3) and (4).
(3) John bought books.
- (4)
(a) What did John buy?
(b) *What did John buy?
As concerns Zubizarreta's (Reference Zubizarreta1998) approach to nuclear stress in focus-neutral contexts, Kahnemuyipour claims that it fails to account for secondary stress in the sentence, includes unwanted redundancy and cannot derive sentential stress assignment in German ditransitives and Persian.
Chapter 4, ‘Sentential stress: A phase-based account’, develops Kahnemuyipour's own proposal. His Sentential Stress Rule posits that stress is assigned at the (strong) phase to the highest phonologically non-null element of the spelled-out constituent or ‘spellee’, i.e. the complement of a phase-head. An element X is considered to be higher than an element Y if X asymmetrically c-commands or dominates Y (68). Under the assumption that in both SVO and SOV languages the object moves to the specifier of an Aspect Phrase (AspP), which is located between the lexical verb and the functional head v, the word order difference arises from the presence/absence of the movement of the verb to a position higher than the object. Furthermore, within the phase-based framework, vP is a phase and consequently, AspP is a spellee, that is to say, a stress domain in Kahnemuyipour's terms. This results in the object being stressed in both SOV and SVO sentences, as it will be the highest non-null element in the stress domain; see (5). If, as in (6), there is no object, it is the verb that is stressed as the highest element in the stress domain.
(5) Ali [AspPqazaaxord].Persian(71, ex. (6c))Alifoodate‘Ali ate food.’
(6) Ali [AspPxord].Persian(71, ex. (6a))Aliate
If the stress domain is empty, the closest phonologically non-null element will be stressed.
Kahnemuyipour proceeds to analyze the presence/absence of stress placement on adverbials as a matter of parametrization. He claims that manner adverbials are merged inside the stress domain (i.e. inside vP) in Persian and Armenian but outside (i.e. above vP) in English and German, as seen by the fact that only in the former languages does the adverb receive sentential stress; see (7)–(8).
(7) Ali [AspPxubqazaaxord].Persian(71, ex. (6d))Aliwellfoodate‘Ali ate well.’
(8) John slowly ate the cake.
Evidence in favor of Kahnemuyipour's hypothesis comes from the fact that under VP-fronting, the adverb can be left behind in English, but not in Persian (89). In turn, sentence-final adverbs in English are considered to be the result of movement of the lower elements around the adverb, which ends up being stressed by virtue of being the phonologically non-null element closest to the emptied stress domain; see (9).
(9) John ate the cake slowly.
Finally, Kahnemuyipour considers the stress behavior of adjunct and argument PPs in this chapter, deriving stress assignment on the basis of the position where these elements are merged.
As far as the behavior of passives and unaccusatives is concerned, Kahnemuyipour suggests that within the phase-based system, the Complementizer Phrase (CP) is a phase, but unaccusative and passive verb phrases are not. Therefore, in the sentences in (1), the first phase is CP, and it is the subject that is the highest element in the spellee (T(ense)P(hrase)). According to the Sentential Stress Rule, stress is then assigned to the subject in languages as diverse as English and Persian. However, if another constituent, such as a manner adverb, is added to the sentence, primary stress is no longer on the subject, as seen in (10).
(10) A boy mysteriously disappeared.
Kahnemuyipour takes this behavior to indicate that an unaccusative or a passive verb phrase does not constitute a phase in its barest form. Only once modifiers are added does the Verb Phrase become phasal. Legate's (Reference Legate2003) arguments that unaccusative and passive verb phrases are phasal are argued to be based on data illustrating this phenomenon. Under this view, stress assignment in (10) follows the same logic as stress assignment in transitive sentences, such as (8).
With regard to the Romance languages, the rightmost position of stress in unaccusatives is taken to suggest that unaccusative verb phrases are inherently phasal in Romance. On the assumption that there is a low phase in these languages, in Spanish the highest element in the spellee is stressed, viz. the subject, as seen in (11). In French, given that the lower phase is empty due to verb movement, it is the verb as the closest phonologically non-null element that is stressed, cf (12). A detailed exposition of this idea is left for further research.
(11) Llegóelcorreo.Spanish(108, ex. (70b))arrivedthemail‘The mail arrived.’
(12) Lecourierestarrivé.French (109, ex. (71))Themailisarrived‘The mail has arrived.’
In Chapter 4, Kahnemuyipour also puts forward an account of secondary stress. He claims that stress is assigned iteratively. At the CP-phase, TP is the spellee, hence, secondary stress is found on the subject, the highest non-null element within this stress domain (the following examples use numbers to indicate primary and secondary stress):
(13) John sawMary.21
Kahnemuyipour speculates that the reason or mechanism responsible for the fact that primary stress is found on the lower phase might not be built into the system under discussion, but may be due to an invariable cross-linguistic phonological/phonetic principle or, alternatively, may be derived from the notion of ‘first’ in a strictly bottom–up system of syntactic derivation.
Chapter 5, ‘Sentential stress and information structure’, deals with non-neutral contexts. Kahnemuyipour proposes the Focus Stress Rule in (14) to handle the interaction between sentential structure and information structure.
(14) Focus Stress Rule: At the phase HP, mark a focused subconstituent C to receive focus stress. (129)
The Focus Stress Rule ensures that the focused constituent receives the highest clausal prominence in languages which mark focus prosodically. The two rules (Focus Stress Rule and Sentential Stress rule) apply independently of each other, interacting only in the final phonetic realization of the stress marking. Kahnemuyipour illustrates this with the example in (15), where the subject is focused.
(15) [F John] kissed Mary.
At the vP phase, the Sentential Stress Rule assigns primary stress to the object (SS1). At the CP phase the subject is marked for focus stress (FS) by the Focus Stress Rule as well as secondary stress (SS2) by the Sentential Stress Rule. An element marked for focus stress receives more prominence than an element marked for sentential stress. This results in the stress pattern in (16).
(16) John kissedMary →John kissedMary.(130, ex. (8))FS, SS2SS112
With regard to out-of-the-blue contexts, where the whole sentence is focused, Kahnemuyipour assumes that the Focus Stress Rule fails to apply due to the fact that there is no single subconstituent marked for focus stress, so that the Sentential Stress Rule applies as described above.
The chapter discusses further data concerning bare wh-words and D-linked wh-phrases, and Kahnemuyipour proposes an account for the fact that D-linked wh-phrases are stressed but bare wh-elements are not (see (4) and the related criticism of Legate's Reference Legate2003 approach).
The author further reviews the focus projection framework, as exemplified by Selkirk (Reference Selkirk and Goldsmith1995) . In this framework, sentence stress is given, and focus structure is derived following an elaborate algorithm – a view which is incompatible with the Y-model standardly adopted in generative grammar. The focus projection framework is shown to make predictions different from Kahnemuyipour's proposal on a number of issues. Specifically, the following data are argued to favor Kahnemuyipour's approach: stress on adjuncts, asymmetries in the stress pattern of bare wh-words and D-linked wh-phrases, phonetic differences in some languages between the accent on narrowly and broadly focused elements, and the contrast in the stress patterns of individual level and stage level predicates. Other topics covered in this chapter include the treatment of disjoint focused constituents and the adequacy of the wh-question/answer test to determine the focus of a sentence.
Given that the proposal is phrased in terms of phases, criticisms of the phase-based framework (e.g. Boeckx & Grohmann Reference Boeckx and Grohmann2007) can be at least partially extended to this approach. Furthermore, as the author acknowledges, the view concerning the syntax of unaccusatives and its analysis in terms of phases in Romance is a little bit sketchy and further research is needed (e.g. see Gallego 2007 for an alternative view regarding phases in Romance, which a priori can be reconciled with Kahnemuyipour's proposal). The debate on another aspect of Romance syntax, namely whether preverbal subjects are in TP or CP is also worth further consideration (see Ortega-Santos Reference Ortega-Santos2008 and references therein). If preverbal subjects are situated in TP, the secondary stress on the subject rather than the verb in an SVO sentence would follow straightforwardly. If, however, preverbal subjects occupy CP, the verb and the subject would be in two different stress domains (spellees) and both would qualify to bear secondary stress. An existing criticism of Kahnemuyipour's proposal by Kratzer & Selkirk (Reference Kratzer and Selkirk2007) is that it needs to rely on object movement to AspP. They question whether this movement obligatorily applies to all objects and PPs, pointing out undesirable consequences and predictions that Kahnemuyipour's system makes, and suggest a modification of Kahnemuyipour's framework in order to avoid these shortcomings. All this said, Kahnemuyipour's book is a remarkable contribution because of the breadth of its empirical coverage and because it links recent syntactic developments with longstanding issues in phonology.
