1 Overview of the volume
In Phonetic causes of sound change: the palatalization and assibilation of obstruents, Daniel Recasens investigates the phonetic motivation for the ‘softening’ of velar and labial stops to (alveolo-)palatal affricates or fricatives (traditionally called ‘palatalisation’), arguing for the articulation-based nature of this sound change and for the necessity of an intermediate stage of an (alveolo-)palatal stop.
The book is interesting, timely and necessary. The topic of the phonetic causes of sound change has been discussed for a long time, but it is far from being resolved. Palatalisation provides an excellent test case for the phonetic scenarios that drive the change, since it is cross-linguistically common and has a wide range of triggers and targets. In fact, it ranges so widely that there is plenty of literature covering different aspects of palatalisation in different language families. As to the bigger picture, there are several typological surveys of palatalisation patterns in the languages of the world, but not necessarily from the point of view of sound change, and definitely not exhaustive. As a typological study, the book is quite remarkable, as it covers a wide range of processes, concentrating mostly on Romance languages.
Even though a wide range of typological data is covered in the book, the subtitle The palatalization and assibilation of obstruents appears to be somewhat misleading, since the book discusses only velars and labials. There is a good explanation for this choice (there is a vast amount of literature on the palatalisation of coronals), but one cannot help but dream about seeing a discussion in a single volume of all the places of articulation affected by palatalisation. Also, although it is very difficult to cite all the relevant literature in such an ambitious project, one important missing reference on the typology of the relevant sound changes is Kümmel (Reference Kümmel2007).
The book emphasises the importance of distinguishing one-stage vs. multiple-stage sound changes, and thus the multiple possible pathways of diachronic development. As Recasens puts it, the book ‘is especially interested in ascertaining the intermediate stages which may occur during these sound changes and also to make sense of the articulatory and perceptual factors which render them possible’ (p. 4). Another important aim of the book is to put forward an articulatory account of change which addresses both articulatory and acoustic data, as well as offering both articulatorily and acoustically based explanations.
2 Summary of the book
The book contains an introduction, chapters on velar softening (Chapter 2), velar palatalisation (Chapter 3), velar assibilation (Chapter 4) and labial softening (Chapter 5), and a conclusion. The term ‘velar softening’ is introduced as a cover term for two changes, palatalisation and assibilation. Velar softening is defined as a shift of a velar to an affricate or a fricative (palatal, alveolo-palatal, palato-alveolar or alveolar), mostly preceding a front vocalic segment. Recasens’ assumption, following scholars since Rousselot (Reference Rousselot1924–25), is that velar softening involves the application of two distinct processes: first velar palatalisation, i.e. the shift of a velar to a palatal stop (e.g. [k] > [c]), then assibilation, i.e. the affrication of the result of the first change (e.g. [c] > [ʧ]). Recasens notes that ‘velar softening’ is just a convenient cover term for what is termed ‘velar palatalisation’, for instance in Slavic (for a recent discussion of Slavic palatalisation terminology, see Iskarous & Kavitskaya Reference Iskarous and Kavitskaya2018). Indeed, one of the problems in the discussion of palatalisation in the literature is that the terms for the palatalisation-related processes and sound changes are different in each language family tradition. It would have been informative and useful to make these terminological differences more transparent.
The production of velar stops explains the fact that the most common triggers of velar softening are front vowels and glides (Chapter 2). A (traditional) articulation-based account of softening is argued to be preferable to an acoustics-based account, and positing an intermediate stage of the change is crucial for the argument.
Chapters 3 and 4 of the book explore the typology of and the conditions on velar palatalisation and velar assibilation respectively. While the chapters closely analyse a considerable amount of data from various Romance languages, the data from many other language families is mentioned in a somewhat haphazard way, making the book seem skewed towards Romance. The division of languages into ‘Non-Romance’ (e.g. §3.1, §3.1.3.1, §3.1.4.1) and Romance (e.g. §3.1, §3.1.3.2, §3.1.4.2) throughout the book is somewhat puzzling, and sometimes seems unwarranted.
Chapter 3 considers the typology of velar stop fronting before a front vowel or glide (/k/ > [c]). It looks at the languages and dialects where [c] either has phonemic status or is an allophone of the velar stop, amassing data showing that both phonemic and allophonic (alveolo-)palatal stops differ considerably with regard to their place of closure, exhibiting different degrees of fronting of the closure, which is distributed over the alveolar and palatal zones even amongst speakers of the same dialect (§3.2). Variability in the degree of closure fronting is argued to be crucial for an articulatory account of velar softening. It is conjectured that one of the factors responsible for the variability of palatal stops is the phoneme inventory size, i.e. the extent of crowding in the palatal region, which may result in an alveolar rather than a palatal realisation of /c/. Other factors discussed are speaker, surrounding vowels, word position and palate shape. The chapter goes on to show that [c] is indeed perceptually ambiguous between [t] and [k] as an intermediate step in velar palatalisation. This variability and ambiguity lead to /k/ and /t/ being frequently exchanged in those languages and dialects where [c] is a variant of /k/. An interesting point the chapter makes in this respect is that /t/ > /k/ and /k/ > /t/ sound changes are both plausible and attested. The changes happen not because /k/ can be misperceived as /t/, but because [c] can be perceived as either /k/ or /t/, contra Ohala (Reference Ohala, Breivik and Jahr1989). The categorical /t/ > /k/ change, where [tʲ c] are the allophones of /t/, is still argued to be more likely than /k/ > /t/, where [kʲ c] are the allophones of /k/, for articulatory reasons (§3.3).
Recasens also discusses cases of regressive velar palatalisation before high (or non-low) front vowels and /j/, regressive palatalisation before low vowels and schwa, e.g. /k/ > [c] (and later [ʧ]) before /a/ in Old French, and progressive palatalisation after vocalic segments, the latter two cases being not particularly rare, but commonly overlooked. Depending on the language or dialect, progressive palatalisation can happen after just /i/, after any front vowel, after any front vowel and /a/, after any vowel, or even after a consonant. The proposal is that articulatory strengthening, or reinforcement, contributes at least to velar palatalisation in the context of non-front vowels (§3.4), i.e. /k/ > [c] is an instance of strengthening. The strengthening analysis accounts for the preferential application of velar palatalisation before the low vowel /a/ word-initially and postconsonantally. Why is this treated as an instance of strengthening? Recasens defines strengthening as ‘an increase in tongue contact and duration’ (p. 76). However, the process which is described as strengthening here represents a step on the path to (further) lenition. Moreover, later discussion (§4.5.4) refers to velar assibilation, which is usually interpreted as lenition, as ‘stop strengthening’. This is rather confusing, and more discussion of the relation between phonetic and phonological strengthening/fortition and lenition would strengthen the analysis. The failure of velar palatalisation to occur before back rounded vowels, as opposed to before front rounded vowels, requires special treatment, and is given an acoustically based explanation.
Chapter 4 provides a typological overview of velar assibilation, showing that the (alveolo-)palatal stop [c] gives rise to various affricates that differ in place of articulation, such as [ʧ], [ʦ], [ʨ] and [ʨ̧], which at a later stage may become fricatives. The chapter also distinguishes between cases where assibilated variants are in free distribution with palatal stops or palatalised velars (§4.3.1), where assibilation happens before front vowels and glides (§4.3.2), where it is found before low vowels (§4.3.3) and where it is found after front vowels and word-finally (§4.3.4). It is noted that high front vocoids can be an exclusive context for velar assibilation in some languages, and that velar affrication to [ʧ] before front vocoids is more likely than palatalisation to [c], supporting an articulatory account in which the change of a stop into an affricate is especially likely in the context of a constricted vocalic segment.
Affrication is argued to be an intrinsic property of palatal stops for articulatory reasons. Just as in the case of /k/ > [c], assibilation happens both before front and back segments, the former being arguably more commonly attested. Progressive and word-final assibilation is attested as well.
§4.4 addresses the less common alveolar outcomes ([ʦ] and [ʣ]) of velar softening. These are problematic for both acoustics-based and articulation-based accounts, due respectively to the too high frequency of the frication noise and the extremely anterior articulation of the alveolar affricates. The hypothesis of the necessary intermediate stage, as in [c] > [ʧ] > [ʦ], is rejected in favour of [c] as the intermediate realisation, as in (/k/) > [c] > [ʦ].
§4.5 deals with the phonetic motivation for velar assibilation. The claim is that in order for a palatal stop to be categorised as an affricate, the stop burst must be prominent enough. An articulatorily based account holds that there are several factors that can increase this prominence, such as the intensity level and duration of the stop burst and vocalic transitions. In low and back vowel contexts in particular, the transitions may be reinterpreted as the glide /j/, which in turn is perceived as a fricative, due to its lingual configuration and spectral characteristics. Yet another factor that contributes to velar assibilation is the position of the stop in a word. Affrication is favoured in strong positions, i.e. word- and syllable-initially, which argues against the proposal that velar softening is a lenition process that originates in weak positions (Minkova Reference Minkova2014). Finally, stress is also a factor in velar softening, which is more likely to occur in stressed syllables.
At this point it becomes necessary to compare the affrication of /t/ and /k/, even though the book is focused on velar softening. §4.6 does just that, making the prediction that /t/ > [c] is less likely to occur than /k/ > [c], and that, given that palatalisation is not required for /t/ to become [ʦ], [ʦ] is the preferred outcome of the assibilation of /t/, but not of /k/. A typological overview provides support for this prediction.
In §4.7 it is posited that an important factor in determining of the outcome of velar softening is the base of articulation, or the ‘articulatory setting by which a subset of vowels or consonants exhibit analogous articulatory configurations in a given dialect’ (p. 126). The base of articulation distinguishes, for instance, languages with a laminal setting, such as French, from languages with an apical setting, such as English, and is used to explain the differences between dialects that show a dental or alveolar place of articulation in the original /ki ke ka jV/ sequences (e.g. Franco-Provençal and Friulian) and those where /ki ke/ result in a dental or alveolar and /ka jV/ yield a palato-alveolar or palatal outcome (e.g. West Romance). This analysis is extended to labial softening in §5.4.
A discussion of labial softening (a cover term for labial palatalisation and labial assibilation that is meant to mirror velar softening) follows in Chapter 5. Labial softening is not as frequent as the palatalisation and assibilation of velars and dentals, which is explained by the fact that labials have multiple articulators that move independently. The proposal is that most instances of labial palatalisation and assibilation depend on the spatiotemporal mechanisms of the production of palatalised labials and thus are best explained by an articulatory-based approach, contrary to the acoustically based view of labial palatalisation (Ohala Reference Ohala, Breivik and Jahr1989). The presence of the intermediate phonetic forms in labial palatalisation, such as [pc] and [pʧ] as the outcomes of /pj/, is taken to be the argument for the articulatorily based approach, with both /pj/ > [pc] > [c] > [ʧ] and /pj/ > [pc] > [pʧ] > [ʧ] pathways possible, as shown by the data from Romanian dialects. The first pathway is posited for the dialects that show either [pc] and [ʨ] or [c] and [ʨ] phonetic realisations, but not [pʨ], while the second path is argued for in the dialects with either [pc], [pʨ] and [ʨ] or just [pʨ] and [ʨ] realisations. Recasens also allows for the acoustically based account in some cases, notably when a palatalised labial stop is replaced by a dental stop, involving the change of the place of articulation, but not frication (e.g. Standard Czech [ˈmjɛstɔ] ‘town’ vs. East Bohemian [ˈnɛstɔ]).
Recasens claims that the basic mechanism of labial softening is ‘glide hardening’, i.e. a change of the type /fj/ > [fc fç], whereby the vocalic front segment that follows the labial hardens into a stop or a fricative. It can be followed by further changes, such as cluster simplification or changes in place and manner of articulation. It is also claimed that /Cl/ clusters, where C is a labial, often become [Cj]. The chapter goes on to illustrate these possibilities, using examples mostly from Romance and Bantu languages.
3 Theoretical coverage: issues and implications
The book addresses both phonetic and phonological aspects of palatalisation. As to the phonetic side of the change, Recasens takes a firm stand on the choice of explanation of velar and labial softening, arguing against Ohala's (Reference Ohala, Breivik and Jahr1989, Reference Ohala and Jones1993) listener-based model of sound change. Since Ohala's work has become quite influential in the literature on historical linguistics explanation, the reader might wish that the book offered more discussion of the differences between the approaches. (Some of the evidence that could strengthen the author's argument can be found in Garrett Reference Garrett, Bowern and Evans2015; for an account of Slavic palatalisation that relies on both articulatory and acoustic factors, see Iskarous & Kavitskaya Reference Iskarous and Kavitskaya2018.)
On the phonological side, the book discusses the conditions for phonologisation with respect to the examples of velar and labial softening, addressing important and generally unresolved questions of why certain phonetic distinctions become contrastive while others do not, and why the same distinctions become contrastive in some languages but not others. Palatalisation provides a rich field of study here. There are significantly more languages with allophonic palatalisation than languages in which palatalisation as a sound change introduces a phonological contrast. The book provides in-depth discussion of both possibilities, citing examples from, for instance, Romanian, where (alveolo-)palatal stops remain allophonic, and Rhaeto-Romance, where a phonemic contrast is introduced. While the answer to the question itself remains elusive, there is also a question about why (alveolo-)palatal stops are often phonologised as affricates. It is suggested that this is the result of their low-closure definition, the variability of the closure placement and a long duration of the stop burst. Affrication is argued to increase the acoustic salience of stops, even though it does not occur in all dialects. Recasens notes that it is possible that the outcome is predictable to some extent, but the degree of predictability requires further investigation.
The book goes on to ask questions about phonological implications, asking whether the crowdedness of the phonemic space affects the phonetic variability of segments in that articulatory space. This, too, is a hypothesis that needs further investigation. Finally, the important question of whether a sound change can occur in order to avoid a phonemic merger is posited, but not answered.
Yet another question raised is whether sound change is gradual (in the Labovian sense, expected for vowels) or categorical (i.e. Blevins’ Reference Blevins2004 ‘change’, expected for consonants). The book contends that the palatalisation of velars is somewhat anomalous, because it is gradual, manifested in various realisations for articulatory reasons, which results in ambiguity (i.e. Blevins’ Reference Blevins2004 ‘chance’).
4 Overall evaluation
The book investigates the phonetic motivation for the palatalisation and assibilation (‘softening’) of velars and labials, and proposes an articulation-based account of the change. It is a good and thorough typological and historical study of Romance palatalisation, with other languages and languages families discussed throughout the book as well.
The book is of interest to a broad audience of linguistic scholars. Given the breadth of its typological coverage and the depth of the proposed phonetic explanation, it is useful for general and historical linguists and for researchers working on topics in Romance historical linguistics in particular, both graduate students and those who are further in their careers. The book contains a fair amount of the descriptions of problems that may lead to further exploration, which is particularly useful to scholars at the beginning of their career.
