This work deals with five phenomena of Latin phonology, all concerning sound change, that so far have not been tackled in a comprehensive way. They are: vowel colouring before clear and dark /l/ (Ch. 2); the littera-rule (Ch. 3); syllabification and vowel reduction before stop plus liquid (Ch. 4); vocalic epenthesis in stop plus /l/ (Ch. 5); and, finally, assimilations: syllable structure and segment sequence (Ch. 6). Previous studies focused on syllable structure and its influence on the phonological change in the last decades (see p. 3). In our opinion this book stands as an innovative, well documented and important contribution to the solution of some problematic sound changes in Latin on the basis of the most recent discussion in phonological theory. The author intends to isolate “the precise phonological conditions” (p. 1) of these phenomena and motivate “why those conditions existed and were instigators of change in the language” (p. 1). Synchronic phonological structure is contrasted to phonetic pressure alone in order to find a solution. This is certainly one of the major points of interest in the book.
Chapter 1 briefly introduces the debate between “reductionists” and “non-reductionists”: the former argue that phonetic pressures alone motivate sound change; the latter seek an innate or universal mental linguistic structure that can guide phonological changes. As the author argues (p. 5) “the present volume does not focus upon addressing this debate”, but “the analyses offered are most in harmony with a reductionist account”, and this is a second major point of interest, in that the author embraces both synchronic and diachronic perspectives. In particular the question is focused on how “synchronic phonological structure, in the shape of syllable structure” (p. 6) could influence diachronic sound change. Syllable structure can influence sound change in two ways, direct, and indirect: in the second case, it conditions surface variants that will drive sound change over time. The majority of the phenomena under scrutiny are indirectly influenced by syllable structure. Vocalic epenthesis, treated in Ch. 4, however, is somehow considered an exception and motivated as an analogically driven process: it is still a case of indirect influence, but highly sensitive to morphological structures. The author considers also the role of synchronic morphological structure, crucial in vowel reduction (Ch. 4) and assimilation (Ch. 6).
Chapter 2 deals with the different realizations of /l/: clear and dark variants depend respectively on the onset or coda position of the segment in the syllable structure, resulting into the reconstructed non-contrastive categories of archaic Latin /l/: dark /l/ in codas, clear /l/ in geminates and /l/ underspecified in onset. The last condition is a novelty in Sen’s explanation: the “darkness” of /l/ before vowels, hence in onset, was gradient and conditioned by the environment, i.e. its realization was darker before /a, o, u/ than before /e/. Sen analyses direct and indirect sources to highlight this evidence, from the attestations in Pliny the Elder and the quotations in Priscian G.L. 2.29 and Consentius, who use the metalinguistic terms plenus, ‘full’ or pinguis, ‘fat’, for dark /l/ in word and syllable final position, medius, ‘middle’, for ambiguous realizations in other positions, and exilis, ‘thin’, for clear geminate /ll/. The analysis of the Latin grammarians is described in Table 2.2, p. 33. Sen presents the phonological analysis of /l/ with its three surface variations as evidence for contrast in backness: dark /l/ in coda is [+back], clear geminate /ll/ is [-back], onset /l/ is underspecified, [Ø back]. The author resumes in very clear Tables (2.3, pp. 38-40) the sound change with vowel colouring in five environments: (a) coda /l/; (b) onset /l/ before /a o u/; (c) before /e/; (d) onset /l/ before /i/; and, finally, (e) geminate /ll/. The back – front continuum of the tongue position accounts for each variant realization as darkest, darker, dark or clear resonance of /l/. Some examples are: (a) sepeltos : sepultus; (b) Grk. epistolá = Lat. epistola (and epistula) ; (c) silentus; (d) agilis; (e) fefelli. Underspecified /l/ is darkened by a following back vowel, /a, o, u/, but is not as dark as in coda position: compare */wél.tes/ > /vul.tis/ with */wé.lo:/ > /vo.lo:/; the latter is backening the preceding vowel only to /o/. Similarly onset /l/ before /e/ was dark, but less dark than all others. We wonder why the counter-evidence to colouring before /le/ in cases such as (*/akwalentus/ >) /aquilentus/ or /pestilentus/, which are compounds, were not explained by the author as conditioned by morphology: the final /i/ in the first member of a compound does not change.
In Chapter 3, under the label “littera-rule”, Sen deals with a sporadic Latin sound change whereby a string composed by a long vowel and a simple consonant changes into a short vowel plus a double consonant (/V:C/ > /VCC/, /li:tera/ > /littera/). The first important step is to isolate the correct phenomena in term of diachronic change. As a matter of fact there are very similar phenomena involving this alternation that need not be examples of the “littera-rule”, but are most probably assimilations (pp. 53-54); analogical processes (pp. 54-55), spontaneous geminations (pp. 55-56, 60-62) or synchronic variation (pp. 72-73). Sen gives an accurate philological examination of the phenomena and is able to reduce to 35 the number of possible examples of diachronic change. Within these we have to eliminate some more ten examples, which for phonetic reasons are better interpreted as synchronic variations even if this alternation is philologically not recoverable (pp. 72-73). Then the author (p. 65) distributes the examples in three phonological patterns: (a) high vowel + voiceless stop; (b) /a/ + sonorant; (c) front vowel + /l/. Only (a) is the relevant sequence for a diachronic sound change, while (b) more likely results from synchronic variation; (c) is also to be interpreted as synchronic variation, but under different conditions, where morphology plays a significant role. The “littera-rule” is sometimes listed as an example of “inverse compensatory lengthening”, because the long feature of the vowel is transferred to the following consonant that works then simultaneously as coda and onset in contiguous syllables resulting in syllable length preservation. This can be described, in Latin, in terms of syllable weight as well: /CV:/ syllables have the same mora counting as /CVC/ ones. Still, the preservation of a phonological structure that has two alternative surface realizations cannot motivate the reasons for a sound change over time but can provide the environment for it. Sen makes some important observations in cross-linguistic phonology. High vowels are phonetically shorter than others, and any vowel is phonetically shorter in front of voiceless stops. So our pattern (a), i.e. phonologically long high vowel plus voiceless stop, will tend to be realized with vowels that are phonetically shorter than expected. Moreover, a voiceless stop can easily be prolonged as this gesture does not present aerodynamic problems. So, even if Sen tends to prefer a “reductionist” account for sound change, we still see phonological structures reducing the possible outcomes of changes motivated firstly by phonetic reasons, as some sort of structure preservation over time is evident.
Chapter 4 deals with the syllabification of the cluster stop (T) plus liquid (R) and vowel reduction before the cluster. Heterosyllabic coda plus onset or tautosyllabic complex onset conditioned surface variants of the preceding vowel, resulting in diachronic vowel reduction. Vowel reduction is not always straightforward: sometimes the vowel rises to /i/, as if it were in open syllable, sometimes to /e/, as if it were in closed syllable. Therefore analysis of the syllabification process of the TR cluster is decisive. The author describes Old Latin evidence such as *kekadai > cecidi, with open syllable, and *perfaktos > perfectus, with closed syllable. Some developments are unconditioned: all vowels in internal open syllables were neutralized to /i/; others are conditioned: foros > -feros showing /r/ conditioning or *opitemus > optimus / optumus showing labial conditioning. Tables 4.1-4.3 summarize the phenomena. Sen challenges some explanations for vowel reduction before TR under the premise that TR clusters were heterosyllabic in Old Latin. Morphologically governed syllabification of TR as unambiguous feature in literary Latin seems to be a plausible support to Hoenigswald’s theory, which states that the quality of the vowel before TR depends on syllable structure (open or close) and the “syllable boundaries in TR were aligned with morphological boundaries, rather than uniformly preceding or bisecting the sequence” (p. 92). The analysis of extensive material on pp. 95-111 and the discussion of forms examined therein (derived with –bra, -trum/-tra, -bulum, -culum/-cula, -crum, -trix, -bilis, -bris, -cris, -brum/-bra, -bris; pp. 112-118) lead to the conclusion of a progressive development of the syllabification of TR, based on the hypothesis that TR was heterosyllabic in archaic Latin, but, due to the alignment of morpheme and syllable boundaries, open or closed-syllable vowel reduction were realized on the basis of the syllabification; open-syllable reduction before TR was influenced by three different environments: r-conditioning, labial conditioning and back-conditioning.
The analysis of vocalic epenthesis in Tl (Chapter 5) has been undertaken according to different parameters: syllable structure, metrical structure, morphological structure, diachronic development of the single clusters (/bl/, /kl/, /pl/ and /gl/). Chronologically, vocalic epenthesis took regularly place in the fourth century in word-internal onset /bl/ and in the late fourth to the second century in onset /kl/; evidence for epenthesis in onset /gl/ is practically non-existent. Syllabification in /pl/ was operative over a shorter period of time, from the mid-third to the second century and epenthesis failed in heterosyllabic /bl/ and /kl/ as the following examples show: *pe.ri:klom (5th cent.) > /pe.ri.cu.lum/ (2nd); *póp.li.kos (5th cent.) > /pú-bli.cus/ (2nd ) and *sta-blis (5th cent.) > /sta.bi.lis/ (3rd) and *sta-blom (5th cent.) > /sta.bu.lum/ (3rd).
Chapter 6, on assimilations in consonantal clusters, is again an impressive piece of scientific, philological and linguistic analysis. Again the author gives a comprehensive account of diverse phenomena concerning syllable structures and segmental phonemic lines. Sen’s hypotheses heavily relies on typology: assimilations in Latin are described in terms of known phonological hierarchies. Latin assimilations conform to the “Place Hierarchy” (p. 179) and to the “Manner Hierarchy”. It is important to stress that the mentioned hierarchies are better understood in a linear approach, and “there is no motivation for reference to syllable position” (p. 179). Assimilation in nasality is treated along similar lines and with similar conclusions: “a parsimonious approach would do away with reference to syllable position, referring only to linear sequence” (p. 183). Things, however, get a little more complicated when dealing with voice assimilation in TR clusters (pp. 186-194). Latin evidence may be distributed in this way: sonorants are underspecified for voice [∅voice], but get a [+voice] feature in syllable onsets. In this case, there is voice assimilation of the preceding obstruent: *sek.men.tom > segmentum. In all other cases assimilation is not active, so we can find contrast in voice: e.g. ver.pa vs. ver.ba. It is however quite important to notice that a complex onset does not allow voicing of the sonorant: e.g. planta vs. blanda. In this case syllable structure is decisive. Sen concludes the chapter treating some counterexamples, which can all be convincingly explained.
In chapter 7 the author offers a set of interesting conclusions about the methods for reconstructing phonological change. The Appendix gives a philological examination of the “littera-rule” forms. The book is conveniently furnished with indexes of Latin words, subjects, languages, authors, and selected sources.
While not all solutions may be considered conclusive, the book will certainly stimulate specific studies of Latin sound change as well as more general considerations about theoretical and historical phonology.