Articulation and sound change in Romance Daniel Recasens† † Universitat Autònoma de Barcelona and Institut d’Estudis Catalans, Spain E-mail: [email protected] on-glide, i.e., [j j] > [jej] > [i] and [j w]> [jew]> [iw], as ABSTRACT exemplified by Haute Loire [sjejs] (SEX, “six”), [seizo] Research carried out by Romance dialectologists, (CERESIA, “cherry”), [djew], [diw] (DEU, “God”) [4: 82, historical linguists and phoneticians since the end of the 96-97]. On the other hand, vowel lowering has applied to XIX century has contributed to clarify the role of [jj] and [jw] but not so to [jej] and [jew], e.g., [mjj]> articulatory factors in sound change. By focusing on data on Romance dialects and on theoretical approaches [mjaj] (MEDIU, “half”) in Vaux and [djw] > [djaw] > advocated by Romance linguists in the past, the present [djçw] (DEU) in Diois [5: 177; 6, vol. I: 373]. paper attempts to show the relevance of articulatory detail The fact that dissimilatory lowering is not prone to to the understanding of sound change processes and apply to rising diphthongs with (mid) high vowels may be phonetic production mechanisms. associated with the strict articulatory requirements 1. INTRODUCTION involved in the production of these vocalic sequences. Regarding sequences with the on-glide [j], high Analyses of available data on Romance dialects in articulatory demands appear to be related to the monographs and linguistic atlas es allow us to formulate prominence of the carryover effects associated with credible hypotheses about the intermediate stages in tongue dorsum raising and fronting and with jaw raising phonetic derivations connecting Latin input forms with both for the glide and for the vowel nucleus ([7]). A high their output correlates in Romance. This reconstruction constriction degree at the lips and at the tongue dorsum for task puts us in an invaluable position for gaining a better the on-glide and the vowel in the case of [wo] may also understanding of those articulatory mechanisms involved result in salient C-to-V carryover effects and a heavily in the implementation of phonetic evolution. Within this constrained CV sequence. Evidence for salient carryover framework, I will focus on data collected by Romance effects associated with lip rounding accounts for specific linguists on several sound change topics, namely, vowel changes occurring after a labial consonant or a labial shifts in diphthongs and triphthongs (section 2), velar vowel: glide insertion, as for Judeo-Spanish [pwade] palatalization and velar softening (section 3), segmental (PATRE, “father”) and dialectal Catalan [kow ] (CODA, categorization of the CV and VC transitions (section 4), ´ “tail”) [8: 357; 9: 302]; vowel assimilation, as for Tuscan and consonant weakening and strengthening (section 5). fucina (OFFICINA, “office”) [10: 169]; consonantal Section 6 deals with sound changes which have been explained in articulatory terms by Romance linguists but absorption, as for [awi] for [awi] in dialectal are better accounted for on a perceptual basis. Catalan (OPERIRE, “to open”) [9: 201]. (b) Another relevant trend is the reluctance for 2. VOCALIC SEQUENCES heterorganic rising and falling vocalic sequences to Production factors appear to play a relevant role in undergo assimilatory vowel raising and dissimilatory the development of diphthongs and triphthongs arisen vowel lowering. Indeed, vowel lowering is not prone to through contextual or spontaneous vowel affect the rising diphthongs [jç], [jo], [ju], [w], [we] and diphthongization in Romance. This view is in contrast [wi] and the falling diphthongs [çj], [oj], [uj], [w], [ew] with perception- based accounts of segmental assimilation and [iw], or the triphthongs [j j], [joj], [w w] and [wew]. and dissimilation according to which those processes ç result in an increase in perceptual salience and Vowel raising may also be blocked in the tripthongs just distinctiveness [1, 2] or else are motivated by mentioned. hypocorrection or hypercorrection [3]. A possible rationale for the absence of changes in (a) Data from Romance dialects show that (mid) vowel height in this case could be associated with the fact high vowel nuclei may undergo assimilatory raising but that the production of sequences such as [we] and [jo] are reluctant to undergo dissimilatory lowering when involves antagonistic gestures, i.e., tongue occurring in rising diphthongs and triphthongs with a fronting/raising and lip unrounding after [w] ([we]) and homorganic on-glide (i.e., [ji], [je], [wu], [wo], [woj], tongue backing and lip rounding after [j] ([jo]). Rather than being subject to vowel raising and lowering, those [wow], [jej], [jew]). Thus, Latin [] has undergone a vocalic complexes appear to undergo changes two-step raising shift in Occitan triphthongs with a palatal contributing to a simplification of the articulatory articulatory gestures for the adjacent vowel and movements involved such as on-glide deletion and vowel consonant. rounding in the case of the vocalic sequences with [w]. Regressive tongue dorsum lowering/ retraction in Illustrative examples are [w] for [ww] (OVU, “egg”) VC sequences with dark [l], [w] and the alveolar trill [r] in Gascon, and [pø] from [pwet] (POTET, “he/she can”) (and also anticipatory lip rounding associated with [w]) in French [11: 122; 12: 91-92]. accounts for the insertion of glides with a lower F2 than the preceding vowel. Illustrative examples are Limousin 3. PALATALIZATION [tjalo] (TELA, “cloth”), [tjaro] (TERRA, “earth”) and Data on velar palatalization in Romance suggest that [vjawre] (VIVERE, “to live”), and Romansh [awlt] palatal stop realizations of front velars differing in degree (ALTU, “tall”) [6, vol. I: 135, 157, 376; 19: 55]. Other of fronting may give rise to palatal and alveolar affricates. consonantal gestures may cause the regressive and Indeed, it has been shown that these palatal stop progressive insertion of high glides, i.e., [j] in the context allophones may be articulated at the medio-postpalatal of a palatal consonant and [w] in CV sequences with a zone but also at the alveolopalatal zone [13]. If realized as labial or velar consonant and a back vowel. The former palatalized dentoalveolars, they are frequently heard as /tj/ insertion type has taken place in French chièvre (CAPRA, or /t/ in Occitan phonetic variants such as [tju] (CULU, “goat”) and Western Catalan [kaja] (CAPSA, “box”) “bottom”) in Haute Loire, and [keti] (‘French celui-ci’, [11: 163; 9: 288], and the latter accounts for Occitan “this one”) in Provence [4: 167; 14: 83]. This view runs [kwolp] (COLAPHU, “blow”) and dialectal Portuguese against the claim that velar softening should be accounted [vwaso] (VASU, “vase”) [6, vol. I: 388; 20: 78]. for through the perceptual identification of the release These data are in accordance with our present noises for a front velar stop and for a front affricate [15]. knowledge about the temporal sequencing of articulatory A possible reason for the existence of a large array of events during the implementation of consonant gestures. palatal articulations originated at the front velar or Thus, regressive glide insertion may be associated with postpalatal zone may be sought in the difficulty involved gestural complexity for dark [l] (the production of which in making a full dorsal closure at the hard palate. involves an alveolar closure and a secondary velar or Palatographic data from different language groups suggest pharyngeal constriction) and for [w] (which is produced indeed that palatal stops and nasals are more frequently with a bilabial and a dorsovelar closing gestures), and articulated at the alveolopalatal zone than at the hard with manner of articulation (trilling for [r] and laterality palate [13]. Also, palatalized dentals and palatals derived from dental stops appear to be taken less often for for [l]). These place and manner requirements cause the palatalized velars than viceversa. Examples of the former tongue dorsum to start moving downwards before the substitution are Franc-Comtois [kjjo] from TILIOLU tongue tip starts raising or the lips start closing [21, 22]. On the other hand, mechanico-inertial properties (“lime tree”) and Francoprovençal [gj ] from DEU O associated with tongue body raising for palatals and velars (“God”) [16: 120; 5: 225]. and with the formation of a narrow labial constriction for In summary, closure fronting along the labials may cause glide insertion to occur at the alveoloprepalatal, mediopalatal and postpalatal zone may progressive level. be the key factor causing front velars to be categorized as Other instances of glide insertion provide some front affricates. Palatal stop releases at the alveolopalatal information about the articulatory configuration and or alveolar zone may sound []-like or [s]-like, thus kinematic properties associated with specific consonantal rendering the change of front velars into [t], [ts] possible. gestures. Thus, [j] and [w] may be inserted in the Existing intermediate forms for originary [pj] also adjacency of dental stops and alveolar fricatives suggest that labial palatalization in Romance is not based presumably since these consonants may be produced with on acoustic similarity [17] but on an increase in some tongue dorsum lowering or raising depending on constriction degree for [j]. Thus, anticipation of the tongue syllable positon and other factors. Illustrative examples dorsum raising gesture for the palatal glide appears to are Franc-Comtois [gwt] (GUTTA, “drop”), Northern have caused