John T. Jensen (2022). The Lexical and Metrical Phonology of English: The Legacy of The Sound Pattern of English. Cambridge: Cambridge University Press. Pp. xv + 379.

1 Overview of the volume

While there have been several books on the topic of English phonology, such as Burzio (1994), Giegerich (1992) or Hammond (1999), none of them covers all types of phonological processes, segmental and suprasegmental, and tries to unite them within a single theory, as Chomsky & Halle (1968) had done in their foundational The Sound Pattern of English (henceforth SPE). The last one being over 20 years old, it is more than welcome to have a new general book on the phonology of English.

In this review, I will first give a detailed overview of the book in §2 before turning to a more critical discussion in §3, in which I will argue that this book, as rich as it is in details and analyses, is disconnected from the empirical methods and theoretical discussions that have animated the field in the past 20 or 30 years.

2 Summary of the volume

Jensen’s volume gives a rich description of various facts regarding English phonology. The book is well-written and is quite easy to read. Many concepts are introduced using data from other languages, which broadens the scope of the book and shows how theoretical tools that are useful for English may also be useful for other languages.

The book begins with a brief preface giving the rationale for the book and how it relates to existing book-length works on English phonology, along with a word on the transcription system used throughout the book. The goal is ‘to give an internally consistent, coherent and complete analysis of the phonology (and morphology) of English’ (p. xi), and it is a considerably revised and expanded version of Jensen (1993).

Then, the book carries on to Chapter 1 on ‘Theories of Phonology’, which starts with a discussion of generative phonology and SPE, introducing some of the key concepts of that framework, such as the use of underlying representations, of sequential rules and how those can be useful to deal with cases of phonological opacity. Five ‘principles of generative phonology’ are also presented (morphological uniqueness, predictability, naturalness, simplicity and preference for phonological solutions). The chapter then covers Metrical Phonology and grid theory, and Jensen argues in favour of adopting arboreal representations. The Prosodic Hierarchy is briefly introduced and will be the focus of Chapter 5. Then, Jensen discusses some of the key concepts of Lexical Phonology and Morphology, such as stratification, Structure Preservation and Strict Cyclicity. Those theories will be those on which Jensen bases his analysis. The chapter concludes with a brief presentation of ‘other post-SPE developments’, such as Autosegmental Phonology (tonal phenomena and Feature Geometry are discussed), Underspecification Theory and Optimality Theory.

Chapter 2 introduces the segmental system of English. The chapter begins with a discussion of the levels of representations used in Jensen’s analysis: the underlying, lexical, systematic phonetic and physical levels. Then, he proceeds with a description of English consonants and vowels using distinctive features (the classification of vowels at the underlying and lexical levels is nearly identical to that of Halle & Mohanan 1985). The notable aspects of the analysis are that Jensen uses [±ATR] instead of SPE’s [±tense] (slightly differently though, as he assumes that certain [−ATR] vowels may be long), and the assumption that there are three possible unstressed vowels [ɪ], [ʊ] and [ə], but that all three may be stressed. Jensen also uses quite abstract underlying representations that are reminiscent of SPE (e.g., ‘lexical’ [eɪ̯] is assumed to derive from underlying //), although those representations are only posited to be different from surface representations if there are alternations or rules to derive the correct surface representations. For example, reduce is assumed to be /re + dʌk + ɛ/, where /e/ undergoes Vowel Shift to [iɪ̯] and the final /ɛ/ is needed to derive the surface [s] through Velar Softening (cf. redu[k]tion). Thus, underlying representations contain what cannot be predicted through lexical rules. The chapter concludes with a small section on orthography and phonology, in which Jensen seems to embrace SPE’s view that English orthography is close to optimal for the underlying level.

Chapter 3 deals with syllables and moras, starting with a definition of those two units and giving some evidence for the mora based on Japanese orthography. Then, Jensen evokes the absence of the syllable in SPE, which uses strong and weak clusters to capture generalisations about stress placement, and some early approaches to the syllable: the syllable boundary approach, the autosegmental approach, the constituent structure approach and the moraic approach. Jensen adopts the latter, and one originality is that he assumes that onsets are connected to the first mora of the syllable. Moreover, in his approach, a segment that is connected to two moras is strengthened, not lenited, and so he rejects ambisyllabicity. He also introduces several possible constraints on syllable structure taken from the existing literature, such as the Sonority Sequencing Generalization, the Linking Constraint and conditions on codas. The chapter continues with a description of the phonotactic patterns of onsets and codas in English. The last comparable description is Hammond (1999), and so this is a welcome section, especially considering the distinction of ‘marginal’ clusters (although we are not told how this is determined).

Chapter 4 covers stress placement. Jensen uses a metrical approach using parameters, following Hayes (1980, 1982, 1995). As opposed to most more recent analyses (see Bermúdez-Otero 2018a), Jensen does not consider the moraic trochee to be the basic foot structure in English. The stress generalisations that are described are close to identical to those developed in SPE, enriched with later developments such as extrametricality. The chapter carries on with destressing rules, which account for the distribution of reduced vowels and ‘trim back excess stresses’ assigned by the previous rules. Thus, a word like abracadabra is initially footed as (a)(braca)(dabra), and destressing and adjunction rules remove the middle foot and adjoin the two stray syllables to surrounding feet, yielding ((a)bra)(ca(dabra)). The chapter carries on with a discussion of the effects of cyclicity on stress. The absence of destressing in còndènsátion, elàstícity or the relative prominence in the first two syllables of sènsàtionálity are attributed to cyclic preservation from the bases condénse, elástic and sènsátional. Additionally, there is an original analysis assuming that suffixes that may be metrified on their own (e.g., -ate, -ize and -ation) do not necessarily cause remetrification in derived words. Thus, extrafenestral stress in words such as óxygenàte is assumed to be regular cyclic preservation from óxygen, while hỳdrógenàte is assumed to be derived from the root hydrogen-. This is different from approaches that assume a dual-affiliation of certain affixes to Strata 1 and 2 (e.g., Giegerich 1999; Bermúdez-Otero 2018b), and here no criteria are given to predict when a word will be derived from a stem or from a root. Such an approach is also at odds with traditional approaches to cyclicity, in which stress rules apply at each Stratum 1 cycle, possibly resulting in stress shifts. In those approaches, the preservation of primary stress (which Burzio 1994 calls ‘Strong Preservation’) is not the default case under Stratum 1 affixation, while ‘Weak Preservation’ (i.e., preservation of stress in the form of subsidiary stresses) may occur, but not systematically (see §3.1).

Chapter 5 deals with the Prosodic Hierarchy. Jensen starts by exposing the main assumptions of the theory: non-isomorphism between prosodic and morphosyntactic boundaries, the Strict Layer Hypothesis, the different types of branching (binary vs. n-ary) and the different types of rules (domain span, domain limit and domain juncture). He assumes eight prosodic categories: the mora, the syllable, the foot, the phonological word, the clitic group (although most prosodic phonologists have rejected it; see Scheer 2011: fn. 36), the phonological phrase, the intonation phrase and the phonological utterance. The rest of the chapter is an exemplification of prosodic categories using processes that refer to them, not all from English. A substantial amount of this chapter is taken from Nespor & Vogel (1986).

Chapter 6 deals both with a more detailed presentation of Lexical Phonology and with the ‘cyclic rules’, that is, the rules of Stratum 1. Jensen briefly presents the idea of stratification and begins with arguments taken over from Kiparsky (1982a, 1982b) on affixes that are sensitive to stress and zero derivation. However, we are not told why the former are relevant to the model: this was a key argument to show that some phonological computation must have taken place before certain morphological operations, thus showing the necessity of an interleaving of phonology and morphology. Then, Jensen describes the general properties of each stratum (e.g., cyclicity, Structure Preservation, the Strict Cycle Condition and lexical exceptions) and then proceeds with a discussion of several cyclic rules (Trisyllabic Laxing, Degemination and CiV Tensing) and of the Strict Cycle Condition. Recent approaches unify the fact that -ic Laxing is assumed to be an exception to Syllable Extrametricality (see, e.g., Bermúdez-Otero & McMahon 2006), but this connection is not mentioned here. The chapter concludes with how stress rules interact with cyclic segmental rules.

Chapter 7 addresses the rules posited to apply on Stratum 2. Those include rules that affect vowels (e.g., Vowel Shift, Diphthongisation and j-insertion) or consonants (e.g., Velar Softening, Palatalization and simplification of stem-final clusters involving a nasal). The chapter concludes with a word on the different types of rule ordering: (counter)bleeding and (counter)feeding.

The final chapter of the book covers ‘further issues in phonological theory’, and it opens with a discussion of umlaut and ablaut, which are eventually analysed as morphological rules. Then, following Kiparsky (1983), Jensen rejects apparent violations of affix ordering and analyses bracketing paradoxes by allowing certain forms to be exceptions to Bracket Erasure. He also discusses the suffix -able, arguing that there are actually two suffixes, one on each lexical stratum, in line with Aronoff (1976). The last section is a short discussion of Optimality Theory, which is briefly presented and judged inappropriate to deal with the various generalisations detailed in the book, especially opacity.

3 Critical discussion of the volume

As can be seen from the summary in the previous section, Jensen covers a broad range of processes of English phonology, and some might argue that it is too broad and would assume that certain processes are actually not phonological (see, e.g., Szigetvári 2018 on vowel reduction or Kaye 1995 on Velar Softening). The book is thus extremely rich and insightful, even for specialists of English phonology, although I am sceptical about its use with undergraduate students due to the complexities of the proposed analyses.

However, the book is mainly taking over and arranging examples and analyses from the early generative literature, and does not engage with most of the issues that have animated the discussions surrounding English phonology in recent years. This can be seen straightaway from the cited references: only 20/175 (11%) were published after 2000, and those are not discussed in as much detail as the literature of the 1960s–1980s is. In the following two subsections, I first detail the issues that I see with the data used in the book and then those regarding the theoretical model.

3.1 The empirical data

The main issue that I have with the data used in the book is that we are almost never told where the generalisations and the examples that illustrate them come from, carrying on a way of doing phonology by relying on examples taken from previous work and introspection. This was the case in most of the literature on English stress of the second half of the twentieth century, where the same examples are used repeatedly, including non-existing words such as hamamelidanthemum. Jensen continues with this tradition, and thus no quantification of the size of the classes and the efficiency of the rules is given. Exceptions are sometimes discussed, but we are not told whether a rule applies to few or many items (sometimes resulting in ‘rules’ that apply to as few as two words, in the case of the rule of Nasal Deletion given on p. 327). Variation is almost absent from the discussion, as illustrated, for example, by the cases of cyclic stress preservation in words such as orìginálity (with second-syllable stress arguably preserved from oríginal), ìnf[ə]rmátion vs. ìmp[ɔː]rtátion or elàstícity vs. àdaptátion (all derived from a base with second-syllable stress) which are known to be variable (Collie 2007, 2008; Kraska-Szlenk 2007; Dabouis 2019; Dabouis & Fournier 2019). Certain points are also asserted without any supporting evidence, such as the assumption that there is no syllabification after Stratum 1, although there is evidence pointing to the contrary (Bermúdez-Otero 2007).

Moreover, it is unclear which variety of English is described throughout the book, although there are some references to specific varieties (mainly RP). Recent developments in English phonology are not discussed. For example, the increasing rates of antepenultimate stress observed among words with suffixes that are often classified as ‘stress-neutral’ such as -arily, -orily, -atory (British English) or -\=VCable (Trevian 2007; Arndt-Lappe & Hedia 2019), or the development of palatalisation in foot-initial position (e.g., tune [ˈtʃuːn], dune [ˈdʒuːn] and assume [əˈʃuːm]), which is said to be impossible (p. 295) but has been developing in recent years (Glain 2012, 2014), could have been mentioned.

Finally, in line with my general criticism about the disconnect between this book and the recent literature, many of the empirical facts that have been described and analysed in recent years are absent. The list cannot be exhaustive and so I will only mention four of those. First, Jensen, following SPE, assumes that stress placement in nouns obeys the following rule, sometimes called the Latin Stress Rule (here taken over from Moore-Cantwell 2020) in (1).

• 

While (1a) generally holds true (although that depends on the part of the lexicon considered; see Dabouis & Fournier 2022), (1b) has been shown to be riddled with exceptions, especially at the periphery of the lexicon (Pater 1994). Recent work by Moore-Cantwell (2020) also reports that the nature of the final vowel interacts with the position of stress in nouns. Second, although there are quite extensive discussions of the behaviour of suffixed words, the book hardly mentions the prosodic and segmental characteristics of prefixes, especially productive ones. One footnote on p. 215 mentions that some have ‘phonological word status’, although there is a wide body of research on this (Booij & Rubach 1984; Wennerstrom 1993; Raffelsiefen 1999, 2007). Related to this is the treatment that Jensen makes of the negative prefix in-. As in most works of early generative phonology, he assumes that it is a Stratum 1 prefix because of its assimilation behaviour, but Raffelsiefen (1999) shows that it has the properties of prefixes that form phonological words, which would be Stratum 2 in Jensen’s model. Those properties even include gemination (contrary to what Jensen claims on p. 254), which has been found for in- in all acoustic studies on nasal gemination (Kaye 2005; Oh & Redford 2012; Oh 2013; Ben Hedia & Plag 2017; Yu 2022), and even when it is realised as ir- (Dabouis et al. 2023). Third, there is no discussion of compound stress except through the general principles of Word Tree Construction (p. 172) in the context of the Rhythm Rule (pp. 213–216), even though there is ample recent literature on the issue that seeks to account for its variability (Giegerich 2004, 2009; Plag 2010; Arndt-Lappe 2011; Bell & Plag 2012, 2013). Finally, there is no mention of paradigmatic dependencies between words that do not stand in a relationship of containment. For example, Steriade (1999) attributes the stress-shifting behaviour of -able in remédiable (cf. its local base rémedy) to the existence of remédial, a behaviour that is distinct from párodiable (cf. párody), as there is no *paródial. This issue has been central in the debate between Lexical Conservatism (Steriade 1999; Steriade & Stanton 2020; Breiss 2021), which assumes that bases need not be contained within derivatives, and Stratal Phonology (Bermúdez-Otero 2018b), which assumes that they must.

Thus, the book discusses a wealth of existing data on English phonology, and it certainly cannot deal with everything given the richness of the literature, but it would have been welcome for Jensen to engage more with some of these contemporary issues.

3.2 Theoretical issues

On the theoretical side, I will focus on two points that seem problematic to me: the treatment of the morphosyntax–phonology interface and the theory of Lexical Phonology used in the book.

It is well-established that, within modular frameworks, morphosyntactic information may be translated into the phonology in two ways: procedural and representational (Bermúdez-Otero 2011; Scheer 2011). In Jensen’s analysis, procedural morphosyntactic conditioning is achieved through strata and stratum-internal cyclicity, and representational morphosyntactic conditioning is achieved through the Prosodic Hierarchy. As Bermúdez-Otero (2011) observes, ‘the uncertainty whether a particular instance of morphosyntactic conditioning in phonology should be analysed procedurally or representationally is in fact one of the most serious and recurrent obstacles faced by empirical research into the morphosyntax–phonology interface’, and there is disagreement on the criteria to do so (Raffelsiefen 2005; Bermúdez-Otero & Luís 2009). Here, Jensen does not say how strata and prosodic constituents interact. To take but one example, in Lexical Phonology, certain suffixes are attached on Stratum 2, but Jensen does not discuss how they should be prosodified, although there is debate on the issue (Bermúdez-Otero 2011). For example, should helpless be prosodified as in (2a) or as in (2b) (where ω° represents a minimal phonological word, ω $'$ represents a maximal phonological word and F represents a foot)? In a personal communication, Jensen instead assumes that the suffix would be adjoined to a higher foot projection, following the Strict Layer Hypothesis, as shown in (2c).

• 

Moreover, the author uses two theoretical models which have been designed to make indirect reference to morphosyntactic information, but rules (20), (60), (63), (66) and (75) of Chapter 7 refer directly to word boundaries, which could be considered violations of modularity. Some of these cases could probably have been avoided by referring to the right edge of the phonological word, although some would require an analysis based on cyclic domains (see the discussion of stem-final clusters involving a nasal below).

Then, the version of Lexical Phonology that is used is mostly taken over from Kiparsky (1982a, 1982b) but, in 40 years, many limitations of that model have been identified, and there are still models that preserve a stratal architecture but with significant differences. More recent models typically assign names to strata that clearly indicate which morphosyntactic constituents are assigned to each stratum: the stem level, the word level and the phrase level. Here, Jensen does not state explicitly which units are assigned to each level and concepts like stem, root and word are not defined. Moreover, the assignation of an affix to a given stratum seems to be diacritic, and Giegerich’s (1999) claim that most affixes are dual-level and so stratal affiliation must be (at least partly) driven by the nature of bases, is rejected. Like most lexical phonologists, Jensen assigns a number of processes to Stratum 2 although their domain excludes Stratum 2 affixes. This is the case for the simplification of stem-final clusters involving a nasal (e.g., damning, signing and singer), which Jensen accounts for by assuming that the underlying clusters /mn/, /gn/ and /ng/ are simplified on Stratum 2 before a morphological bracket, in violation of modularity, as pointed out in the previous paragraph. Because of Bracket Erasure at the end of Stratum 1, damnation is assumed to have no internal bracket on Stratum 2 and does not undergo cluster simplification, while damning still has a bracket and therefore undergoes simplification. To avoid the modularity violation, one option would be to assume that the simplification of such clusters occurs at the end of a prosodic constituent, but, as no resyllabification is assumed to take place on Stratum 2 and as Jensen adopts a strict version of the Strict Layer Hypothesis, the difference would have to do with syllabification, an option that Jensen explicitly rejects. An alternative is then to use cyclic domains, and assume that this simplification occurs at the end of a phonological word at the stem level (Bermúdez-Otero & McMahon 2006; Bermúdez-Otero 2011). Incidentally, Jensen’s model cannot account for varieties of English in which such cluster simplification occurs word-finally but not stem-finally (e.g., singing [ˈsɪŋgɪŋ]). Finally, as mentioned above, more recent models such as Stratal Phonology have a far broader empirical coverage which includes up-to-date empirical observations such as what Bermúdez-Otero (2013) calls ‘the stem-level syndrome’, the Russian Doll Theorem (Bermúdez-Otero 2011) or the life cycle of phonological processes (Bermúdez-Otero 2015).

To conclude, Jensen’s book constitutes a rich collection of facts and analyses on English phonology, which makes it a useful read for advanced students of English phonology. It is a welcome contribution by the mere scope of the book, but it does not engage with the issues that have animated the field in recent years. One may get the impression that the author rejects or is unaware of most of the research that has been produced since the 1990s. For those reasons, I believe that this book is very rich, valuable and interesting but cannot be treated as current.