DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect Lorenzo Ferrarotti (Università degli Studi di Bergamo) https: / / orcid.org/ 0000-0002-8741-6812 Riassunto: L’articolo si propone di ricostruire alcuni aspetti del sistema fonologico del dialetto piemontese astigiano del XVI secolo sulla base del corpus testuale fornito dall’ Opera Jocunda di Giovan Giorgio Alione (1521). Attraverso la ricostruzione interna e quella comparativa, ne sono esaminate le relazioni con i sistemi fonologici dei dialetti contemporanei. Se, in molti casi, sono state riscontrate alcune importanti continuità, sono presenti anche molte discontinuità: all’epoca, questo dialetto aveva un inventario fonematico più ricco rispetto ai principali dialetti piemontesi contemporanei, almeno per quanto riguarda i fonemi consonantici. In termini di evoluzione diacronica generale, la ricostruzione mostra che il piemontese è del tutto assimilabile a una lingua romanza occidentale perché ha condiviso diversi sviluppi fonologici con il galloromanzo e l’iberoromanzo, ma non con l’italiano. In questo contesto, hanno avuto luogo diverse fusioni di fonemi comuni anche ad altri dialetti galloitalici (ad es. la deaffricazione delle affricate alveolari). Dal punto di vista del contatto linguistico, la fonologia dell’antico dialetto di Asti mostra interferenze significative con il francese medio (prevalentemente tramite prestiti lessicali), ma anche con alcuni volgari «alti» dell’Italia settentrionale, in una dinamica di contatto che era comune ai dialetti lombardi dell’epoca. Keywords: Piedmontese, Asti, Phonology, Reconstruction, Gallo-Italian, Western Romance 1. Object of analysis and methodology Piedmontese, one of the so-called dialects ( dialetti ) of Italy, is, from a structural point of view, a language separate from Italian. It is commonly deemed to belong to the socalled «Gallo-Italian» group of the Romance languages, which lies typologically between the Gallo-Romance and the Italo-Romance groups (see the classification proposals in Bossong 2016, Regis 2020, cf. Benincà, Parry, Pescarini 2016). The most well-known and well-described variety of Piedmontese is the dialect of the city of Turin, which has been called outright «Piedmontese» since the 18 th century. After the city became the seat of the government of the Duchy of Savoy in the second half of the 16 th century, it grew into a demographically relevant urban center in the 17 th -18 th . Only at that time its dialect developed an extensive literature and underwent a codification process (see Regis 2013). In the following centuries, this dialect significantly 104 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 influenced the neighboring varieties and somehow modified the pre-existing linguistic setting, as it was adopted as a koine variety in Western Piedmont and spread its linguistic features in several parts of the region through dialect contact. In any case, the formation of the Piedmontese linguistic area was not entirely created by the leveling action of Turin, as historical texts point to the preexistence of several characteristic Piedmontese features (see the reconstruction in Ferrarotti 2022). In this respect, Piedmontese phonology has never been thoroughly examined from a diachronic point of view. Even if many changes can be assumed just by reconstructive hypotheses, it is uncertain when major changes in phonology occurred. It is also unclear how much local varieties differed from the dominant Turin variety in the past as well due to late or scarce historical records. This study proposes a phonological reconstruction of a historical Piedmontese dialect based on the longest text written before the 18 th century, namely the 1521 Opera Jocunda ( OJ from now on) by Giovan Giorgio Alione. It contains texts in macaronic Latin, French (a language the author was very proficient in 1 ), Flemish, and, above all, ten farces and six poems (totaling more than 6000 lines) in the native dialect of the author 2 , that I will refer to here as the old dialect of Asti (OAD from now on). Its language remains relatively understudied, with some exceptions (Villata 2008, Parry 2017, Ferrarotti 2021, Parry 2023). However, the largest study is still Giacomino (1901), a pre-structuralist analysis of the historical phonetics and morphology of OAD as they emerge from the text. It offers several good insights into its linguistic features but lacks a systemic view on phonology (as is typical for such an older study) and misses some essential points, according to Salvioni’s (1905) review. From a methodological point of view, a great deal of philological caution is needed in exploiting this text for linguistic purposes because its editions have some severe problems in terms of reliability 3 . Therefore, the present analysis will be carried 1 Note that Asti was under direct French rule from 1389 to 1526 and that Alione was a politician who had direct links with France, as he was rewarded with a post by king Francis I of France. 2 One Farce ( Farsa del franzoso alogiato a l’hostaria del lombardo ) is written in a super-regional Northern Italian koine that blends some features of Lombard dialects with Tuscan and local ones. 3 Tosi (1865) and Villata (2007) offer just a basic unedited (or barely edited) text, and only the latter features a translation (which is not always reliable and often incomplete). Bottasso’s (1953) text is still the most widespread today. It features a final glossary but no translation, though the main issue with it is the heavy writing normalization (based on the norms of contemporary Turin Piedmontese) that alters and sometimes obliterates many graphemic contrasts of the original text: as we will see, they are of paramount importance in a historical phonological reconstruction. With some minor differences, the same goes with Clivio’s (2003) edition of two farces. This is a significant methodological flaw: Alione’s work is, in fact, a unicum , because it originates from a relatively marginal context (Asti) in a time when neither a standard form nor any literary tradition in Piedmontese existed (and even the printing press was a novelty in that area). At that time, the only literary or learned models for that area were French and Northern Italian (or «Lombard») vernaculars. The latter were still out from the Tuscan mainstream, which would impose itself in a more definite way only in the second half of the 16 th century (after the publication of the landmark Le prose della volgar lingua by Pietro Bembo dates 1525, four years after the OJ ). 105 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect out on the original 1521 text with the addition of the necessary diacritics and some minimal adjustments 4 . The richness of Alione’s texts makes possible an extensive reconstruction of several aspects of the phonology of OAD. As a matter of fact, the OJ is laid out in a writing system that tends to reproduce the spoken language phonemically, in contrast to many texts of the previous centuries that are written in a supra-regional scripta that in several cases obscures the features of local dialects (Sanga 1990). Earlier attempts at understanding Alione’s phonology consist predominantly of grapheme-to-sound guesswork with no phonological awareness. Besides Giacomino’s (1901) extensive account, some suggestions are present in Bottasso’s (1953: XI-XIV), Clivio’s (2003: 137-39) and Villata’s (2007: VIII-XVII) prefaces to their editions. However, their observations are based mainly on generic similarities with the contemporary dialect of Turin and only in a limited way on other criteria such as the ones that will be adopted here. Firstly, an internal reconstruction will focus on the identification of the correct phonemic values of graphemes, primarily through the individuation of contrasts (a crucial feature of phonological reconstruction, according to Minkova 2015). Also, the analysis of the diachronic evolution of Latin phonemes or consonant clusters 5 and the comparison with the etymon of the lexeme they appear in (see, e.g., Rohlfs 1966 for Italo-Romance dialects) will be added. This process, well-established in Romance linguistics, gives quite solid results in reconstructing dialect kinships, as irregular outcomes allow to identify borrowings from other languages, other dialects, or learned «high» varieties (see Campbell 2013: 187-97). Furthermore, Alione’s texts are written in rhyme, predominantly in the octosyllable verse, in which the eighth syllable is always stressed, and the verse can be composed of 8 or 9 syllables. The rhyme is usually perfect («syllabic») or imperfect (with identity between the nucleus and the coda of the last syllable of the verse), with simple rhyming couplets 6 (AABB). This allows to see phonemic identity even when there is no graphemic identity due to fluctuations in the usage of the writing system 7 . In some cases, a few 4 As the normalization of v/ u and j/ i , the addition of acute/ grave accents on paroxytones and the separation of the words according to the contemporary usage. 5 I will employ here attested or reconstructed Latin etyma in small capital letters (e.g., pratuM ). At the same time, intermediate proto-Romance or late Latin forms that have to be posited to explain some phonological phenomena will be written in italics (e.g., *arjola < areoLaM ‘flowerbed’). 6 In a few exceptions, some other rhyming schemes are employed. Apparently, no assonances are attested. In a small number of recognizable cases, rhyming words are blatantly distorted for a comical effect. 7 A clear model in this type of dialect reconstruction is Salvioni (1911), who could reconstruct several aspects of the vowel system (and indirectly some facts about the consonant system) in the medieval dialect of Milan on the basis of rhymes. Other important cases are Shakespearean English (Viëtor 1906, Crystal 2016) and Old and Middle French (Fouché 1966), for which similar methods (graphemic variations, rhyme etc.) have been employed for the reconstruction of their phonological system. 106 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 puns in the text can also be helpful in the reconstruction process because their interpretation relies on the identity or non-identity of sound. Secondly, a comparative reconstruction can be attempted, comparing the «internal» evidence with other languages and dialects that can be found in the OJ . In fact, in Alione’s work, several languages and dialects whose linguistic history is better attested are imitated, and this can be used to understand the phonetic value of some graphemes employed by the author. In this context, there are several Piedmontese lines rhyming with French lines that can be used to ascertain phonetic values. Also, some more general «external» evidence can be considered, such as the graphemic usages and writing traditions 8 of Northern Italy and France and the linguistic history of other languages and dialects. Finally, the diachronic comparison with contemporary dialect geography evidence is of paramount importance in the reconstructive process because OAD exhibits archaisms that are often preserved in present-day isolated or conservative dialects but not in the contemporary dialect of Asti and its surroundings. The reconstruction will be divided into sound classes, and only the most problematic cases will be analyzed. The sources of all examples quoted from the text will be shown in the text before every one of them with a letter and a number, indicating respectively the single work (farce or poem) and the verse they appear in 9 . The matter will be organized as follows: in §2 consonantal phonemes will be analyzed, in §3 vocalic phonemes, in §4 other features. In §5, there will be a general evaluation of the reconstructed sound system. 2. Consonants 2.1 Postalveolar affricates One of the most complex and puzzling features of the OJ writing system is the graphemic representation of postalveolar affricates, which is very different from Italian and Piedmontese contemporary and modern usages (see Genre 1978, Miola 2015). 8 Alione’s spelling choices for the representation of his native dialect must be put in a context where no clear pre-existing literary tradition in this language was present. For this reason, the author draws on French and Northern Italian (or «Lombard») written usages. It is useful to remember that the text comes from a printed edition from the 16 th century. For this reason, spelling fluctuations are expected, as there were no fixed norms, even in more codified languages. Also, the experimental character of Alione’s work influences the uncertainties of rendering in writing a previously unrepresented dialect. 9 Pr. Prologo de l’auctore , A. Comedia de l’homo; B. Farsa de Zohan Zavatino ; C. Farsa de Gina et de Relucha ; D. Farsa de la dona ; E. Farsa de Nicolao Spranga ; F. Farsa de Pèron e Cheyrina ; G. Farsa del lanternero ; H. Farsa de Nicora e de Sibrina ; I. Farsa del bracho e del milaneyso ; L. Farsa del franzoso ; M. Conseglo in favore de doe sorelle ; N. Frotula de le done ; O. Cantione de li disciplinati de Ast ; P. Altra cantione ; Q. R. Uno benedicite Dominus e uno Reficiat ; S. Le dit du singe . 107 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect The spelling of these sounds is largely position-dependent, with several overlaps between the positional variants, and is conditioned by some phoneme neutralization processes. As regards the voiceless alveolar affricate / t͡ʃ/ , different spellings are found on the basis of their position in the word: 1) Word-initially, mainly ‹ chi › 10 ; 2) Word-finally, always ‹ g ›; 3) Word-internally, both ‹ chi › and ‹ gi › + ‹ a o u › / ‹ g › + ‹ e i › . The use of ‹ chi › is related to the old French usage 11 , while ‹ g › is quite common in old Lombard texts 12 , but unusual in modern and contemporary languages. This graphemic system is attested, with variable features, in Northwestern Italian texts before the 15 th century (Ghinassi 1976: 90-92). The exact value of chi as / t͡ʃ/ can be reconstructed in the first place on the basis of the comparison with other Italian dialects quoted in the OJ , namely two parodic citations by Alione of the Genoese (1) and the Florentine dialect (2): 1) o zeneyse da cima in fondo / chiù regulau homo dro mondo ‘o Genoese man, from top to bottom the most regulated man of the world’ (E72-73) where it is clear that chiù renders [t͡ʃy] ‘more’ (< pLuS ), in which the outcome / t͡ʃ/ of pL is visibile, one of the most characteristic features of Ligurian dialects (Petracco Sicardi 1992: 19). 2) reca quane / vintechinque onchiuchie de carne ‘bring here twenty-five little ounces of meat’ (E83-84), where vintechinque onchiuchie ‘twenty-five little ounces’ represents [vinteˈt͡ʃiŋkwe onˈt͡ʃut͡ːʃe], cf. Italian venticinque onciuccie . Another proof of its exact phonemic value is internal, as it is found in words that bear several different outcomes. 1) Latin cL (word-initially or word-internally after a consonant), for which / t͡ʃ/ is still the common Gallo-Italian outcome (through * kj ). A490 chiaaf ‘key’ < cLaVeM , C11 chioenda ‘bush’ < cLaudendaM , i580 chiap ‘piece’ < *clap, B428 schiapper ‘to break’ ex -+ *clap + are (possibly from a pre-Latin root whose outcomes merge with Latin cL -, FeW 2: 735), c176 cerg ‘circle’ < cercLuM , g89 mesgia ‘mix’ < * MeScLaM (< MeS cuLaM ), a468 schiayr/ B552 sgeyr ‘(I) see’ < ex + cLaro . 10 Rossebastiano/ Papa’s (2011) account does not notice the usage of ‹ chi › for the postalveolar affricate in old Piedmontese text. 11 Giacomino (1901) misunderstands the value of chi believing that it represents / kj/ , as noted by Salvioni (1905: I 157). 12 Salvioni (1905: I 155) remarks that Giacomino’s (1901: 405) account ignores the exact value of final ‹ g ›, as he believed that it represented / d͡ʒ/ , which is impossible if the correct sound evolution is taken into account. Sanga (1984: 154-55) shows that this writing usage was common until the end of the 16 th century (in Lancino Curti’s works). 108 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 2) ct -, for which / t͡ʃ/ is the typical outcome in southern Piedmontese dialects (Ferrarotti 2022: 102-07, 256-59, Rohlfs 1966: 243-45, 349-52). A351 fag ‘done’ < FactuM , A728 fagia ‘done’ (f. sg.) < FactaM , E268 fagiure ‘tailorings’ < Factu raS etc.; also, B355 aguachiant ‘staring’ (< Germ. *wahtan, in which the outcome of the Germanic cluster -htmerges with Latin ct -, at least in Gallo-Italian dialect, cf. FeW 17: 451-57, cf. rep 1563 vaité ). 3) instances of coalescence of [i̯t] or ti . a34 tug ‘all m.pl.’ < *tuit 13 < * totti , A427 eyg ‘other m.pl.’, < *aiti < aLt ( e ) ri , a882 nosg ‘our m.pl. < noSt ( r ) i . 4) outcomes of ca in French loanwords 14 and hybrid forms. A414 chiera ‘face, good mood, cheer’ < chière < caraM , B79 chiaglia ‘(it) matters’ < chaille (chaloir) < caLere , d106 bouchia ‘mouth’ < bouche < BuccaM (hybrid form uttered by a woman from Asti mocking Frenchmen). Cf. also B81 chianchieme ‘chat to me’ < fr. changez moi crossed with OAD chianchier , a pun in an excerpt in which a Frenchman tries to speak OAD. 5) in words of various origins in which its presence is expected. A32 schiater ‘to burst’, D422 schiancher ‘to rip’, I473 schias ‘thick’(< Germ. *slaitan? FeW 17: 141-44 cf. fr. éclater , It. schiattare , cont. Piedmontese [st͡ʃaˈte], [st͡ʃaŋˈke], [st͡ʃas]) , B128 archichioch ‘artichoke’ (Arabic via Spanish alcarchofa with assimilation), G286 anchioa ‘anchovy’ (< Genoese ancioa ), A471 chianchia ‘(he) chats’, I459 chiangia ‘chat (imperative)’, F276 giangiant ‘chatting’ (cf. It. cianciare ) etc. As can be seen, this kind of spelling is fairly consistent word-initially and word-finally. Still, it is inconsistent word-internally after a consonant, sometimes because derivatives tend to keep the original base form (see fag/ fagia/ fagiura in 2). Still, this alternation is probably due to a context-dependent phonological neutralization (see below) and, in any case, there is some tendency to use ‹ g › for / t͡ʃ/ particularly after a consonant (see above in 1. cerg ‘circle’ , mesgia ‘mix’ , and in 5. chiangia ‘to chat’ etc.). Note that, in many cases, word-final (and sometimes word-internal and word-initial) ‹ ch › stands for the velar stop / k/ , as is common in medieval Latin scripta , and as can be seen in the following subminimal pair: F87 descharrier ‘to unload’ < de ex - 13 [tyi̯t] is possibly a metaphonic plural in origin, but another possible origin is a fusion between Latin * totti x cuncti , yielding * tucti ( DEI 3937). The singular form [tyt] should have had, in any case, back-formed on the plural, as *[tut] is not attested in these varieties. 14 This feature, i.e. the / t͡ʃ/ outcome of ca , seems to be archaic if it is compared with the historical development of French phonology. Both Nyrop (1899: 319) and Fouché (1966: 553 and GGHF 354 as well) state that that phoneme became in / ʃ/ in the 13 th century. Anyhow, Alione’s French texts usually show the «normal» usage of Old French ‹ ch › = / t͡ʃ/ with no ‹ i › added. 109 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect * carricare ~ A866 deschyairer ‘declare’ < de ex cLarare , in which ‹ cha › 15 = / ka/ and ‹ chya › = / t͡ʃa/ . Some problems arise with ‹ chie ›, which is sometimes spelled ‹ che › when representing / t͡ʃ/ : e.g., C240 scheyr instead of * schieyr (not attested), Cheyrina instead of Chieyrina (F passim ). But the biggest issue, in this case, is that ‹ che › and ‹ chi › are used also for representing the velar stop / k/ before / e/ and / i/ , overlapping with the spelling for / t͡ʃ/ , as the following examples show: Pr. 32 cercher ‘to search’ < * circare , C30 chenna ‘chain’ < catenaM , N55 masche ‘witches’ < MaSca ? ( REP 933), C237 cheyt ‘fallen’ < * cadituM , chi/ che ( passim ) i.e., the relative pronoun in sujet and régime case. Cf. some words of Germanic origin as B207 schiffy ‘disgust’ < Germ. *skiuhjan ( FEW 17: 124-26), B452 schina ‘back’ < Germ. *skina ( FEW 17: 112), D205 marchisa ‘marquise’ < Germ. *marka, and other words of various origin as H506 meschina ‘poor’ (< fr. mesquin ? < ar. miskīn, FEW 19: 127), C204 miche ‘loafs’ < * MiccaS , F259 pochin ‘little bit’ (diminutive of poch ) < paucuM . It is also employed in the Genoese word E57 chigomari ‘cucumber’, which has to be linked to the modern [kiˈgømai̯] (sg. [kiˈgømau̯] < cucuMeruM ? < cucuMereM , see Casaccia 1876: 216). In these cases, the distinction between / t͡ʃ/ and / k/ is mainly lexeme-based and cannot be predicted from the writing system. From a reconstructive point of view, this oscillation yields some very troublesome cases as the 3sg personal pronoun Pr.36 chiel / B18 chiella ‘he/ she’. In contemporary dialects (incl. Turin) forms with a velar stop [kjɛl] and [ˈkila] are well attested, but many eastern and some southern Piedmontese dialects have [t͡ʃəl]/ [ˈt͡ʃəlːa] (Ferrarotti 2022: 147-52) with the postalveolar affricate 16 . This form could then actually reflect the realization of the 16 th -century dialect of Asti. Finally, it is noteworthy that initial ‹ chi › is in unequivocal opposition with initial ‹ ci › that bears a different 17 phonemic value (see §2.2 for some minimal pairs). As for the voiced counterpart / d͡ʒ/ , its graphemic representation alternates between an Italian type ‹ g › + ‹ e, i › / ‹ gi › + ‹ a, o, u › both word-initial and word-internal, and the French-like ‹ j › 18 mainly word-initial. In the few contexts where it can appear as word-final, it overlaps with the representation of / t͡ʃ/ , i.e., ‹ g ›. This is due to the fact 15 Cf. also poch ‘less’ < paucuM etc. The use of ‹ h › in these cases does not contribute to representing another phoneme (as would be the case for cha vs. ca in French). For instance, choy (16 tokens) is interchangeable with coy (13) ‘those’. This usage is quite widespread in Alione’s writing. In some cases, there is no fluctuation, as in bocha ‘mouth’ (*< Bucca ), that is always spelled with an added h . Still, the phonemic value of the digraph is clearly [k], because it rhymes with tocha ‘he touches’ (A225-26), which in another place (B359-60) is written toca and rhymes with rocha ‘distaff, spinning tool’ (cf. It. rócca ) . 16 The etymon is usually connected to * ecce iLLuM , but this does not fully explain the presence of / t͡ʃ/ , because / səl/ would be expected with / s/ as the outcome of strong c + e , i (see §2.2.2). 17 In Bottasso’s 1953 edition ‹ chi › is normalized as ‹ ci ›, which forces him to change the original ‹ ci › of the text to ‹ çi ›. Sometimes even the value of ‹ che › is misunderstood, e.g., chenna ‘chain’ < cate naM is erroneously normalized as cenna (Bottasso 1953: 279). 18 Often just ‹ i › in the text of the 1521 print. 110 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 that this dialect apparently had a strong devoicing of word-final obstruents (§4.2) and thus there was no need to represent a neutralized opposition graphemically. This fact, and a limited functional yield of the / t͡ʃ/ ~ / d͡ʒ/ contrast (not entirely reconstructable from the evidence provided by the OJ ) is what led to the many overlaps of internal ‹ chi › and ‹ g ›. The native distribution of this phoneme of the OAD is, in principle, quite limited, as it can appear in two instances. 1) as the outcome of gL -: H218 giesia ‘church’ < * glesia < eccLeSiaM , e535 giosa ‘clause’< gLoSSaM and maybe H124 giot ‘scoundrel’ and I639 jotta ‘id., f.’ < gLuttuM , cf. Italian ghiotto ; 2) as outcome of gd -: E81 freg ‘cold’ < * Frigdu < FrigiduM and its derivative A235 fregiure ‘frostbite (pl.)’ < * FrigduM + uraM . As a matter of fact, / d͡ʒ/ is reintroduced in OAD through two channels of language contact. 1) Borrowings from French (hence the significant oscillation with initial ‹ j ›): A199 joyoux ‘joyful’ < gaudiuM + oSuM , A15 jarg(h)on ‘jargon, French’, C262 jantil ‘kind’ < gentiLeM , A199 jeloux ‘jealous’ < zeLoSuM , D52 jach ‘jacket’ (also spelled C307 giach ), I618 giouch ‘henhouse’ < Germ. juk ( FEW 16: 287), A154 rage ‘rage’ < raBieM and probably also Pr. 31 gent ‘people’ < genteM . Every word bearing the -age suffix < aticuM is French as well, as A154 language ‘language’, perhaps the nickname Jan ‘John’ < JohanneM (in contrast to the local OAD form Zan ). Note also D213 chiangiant (< French changeant ‘changing’ but also ‘iridescent’), that overlaps with I459 chiangia as a form of the verb F83 chianchié ‘to chat’ (cf. It. cianciare ). 2) Dialect borrowings from «high» or learned varieties influenced by Latin (or Latin itself) or the so-called Lombard vernacular (see §1): Many words with the Latin suffix egiuM as A844 collegi ‘council’; the learned form A719 judex ‘judge’ < JudiceM and the more native 19 form E319 juz , F411 judiché ‘to judge’ < Judicare , I89 adjusté ‘to adjust’ < ad - + JuStare , B324 Jordan ‘Jordan’ (used in a derogatory way; the river Jordan in I232) < JordanuM . It is important to notice that the use of the phoneme / d͡ʒ/ was evidently salient to Alione, as it is used in one Farce in place of the more typical OAD outcome / d͡z/ to portray the speech of a character speaking another Piedmontese dialect (the maid Minetta in I; see Ferrarotti 2021): I573 già ‘jet’ < JaM instead of za , I558 giόvon ‘young’ < JuVeneM instead of zόvon , I694 jura ‘(he) swears’ < Jurat instead of zura (see §2.2). 19 By the fact that it shows the fall of intervocalic / d/ , a typical feature of Piedmontese dialects. 111 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect No issues are given by ‹ gh › instead, which is used for representing / g/ before front vowels and at the end of the word: A421 ghigné ‘to laugh’, C239 borgheyse ‘burgesses’, A270 digh ‘I say’ etc. 2.2 Alveolar and postalveolar fricatives and affricates In order to reconstruct which phonemes of these sound classes are present in OAD, it is useful to summarize the development of Gallo-Italian alveolar fricatives and affricates, called the «system of the sibilants» by Sanga (1984), which is different from Italian and more similar to the developments seen in Gallo-Romance and Ibero-Romance (or Western Romance, see Bonfadini 1995: 34-35, Loporcaro 2010: 143- 47, cf. Barbato 2019). In a strong position 20 the late Latin phoneme / t͡ʃ/ (outcome of * cj < c + e , i ), kept as such by Italian, is not usually maintained in Gallo-Italian dialects 21 (see Rohlfs 1966: 201, Sanga 1984: 152 -167, Ferrarotti 2022: 116-21), except for the conservative varieties of Valsesia (and also in Romansh dialects): cenaM ‘dinner’ > [ˈt͡ʃena]. In fact, it usually merges with / t͡s/ (< * tj- , * -ttj- < t + e , i ) as in Old French and Old Ibero-Romance. This merger was common in most Gallo-Italian dialects («Northern Italian system» in Barbato 2019: 962), as it was present in the Medieval dialects of Venice and Bergamo (Sanga 1984: 47-48), where cenaM > [ˈt͡sena]. In many of them, however, the merged phoneme / t͡s/ has then been lost because it was subsequently merged with / s/ (as in French) cenaM > [ˈsena] 22 . This process is well attested for the dialect of Milan at the beginning of 19 th century (Sanga 1985: 16-17), and many Piedmontese dialects underwent this process too, probably at an earlier age (for instance, the 18 th -century Turin dialect already showed no trace of / t͡s/ ). Still, the phoneme / t͡s/ is preserved in a sizable linguistic area 23 between southern Piedmont and northern Liguria, while in Valsesia and in the Biella area (Northeastern Piedmont) its presence is quite reduced and limited to conservative dialects (Ferrarotti 2022: 122-23). 20 Following the traditional usage in Romance linguistics (see, e.g., GGHF §19), this means when the phoneme is either word-initial or after a consonant which is the coda of the previous syllable, i.e. in onset position. As it will be possible to see, an originally geminated/ long consonant can be considered strong, as in Sanga (1984), (1985). 21 This phoneme in these contexts is often reintroduced through borrowings from Standard Italian in urban varieties: for Milan, see Sanga (1985); for Turin and Piedmont in general see Clivio (1972) and Ferrarotti (2022: 119-20). 22 Sanga (1984) attributed this change to the fact that the so-called «rules of tension» that conditioned the evolution of the same Latin sound according to its weak or strong position ceased to operate. 23 The extension of this area is not completely clear. The retention of this phoneme is attested in the linguistic atlases in the following survey points: AIS 176 Cortemilia, 177 Sassello, 184 Calizzano; ALI 68 Cortemilia, 74 Murazzano, 75 Rossiglione. It is also attested in the Alta Langa area, south of Alba (Giamello 2007), Cairo Montenotte (Parry 2005: 108-10) and in conservative varieties neighboring Mondovì (Miola 2013). 112 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 In most western Lombard dialects 24 and in the Piedmontese dialects around Biella (in continuity with the Novara area), there is no such merger, as / ʃ/ is generally well attested as a direct development of / t͡ʃ/ through deaffrication: cenaM ‘dinner’ > [ˈʃena]. According to Bonfadini (1995: 34-35), the preservation of this distinction is typical of Eastern Romance (Italian, Romanian) and Romansh. The voiced counterpart / d͡ʒ/ had a similar development. Several Latin sounds were already merged into / d͡ʒ/ at an early stage ( g + e i , * dj- , * -ddj- , J -). Many Gallo-Italian dialects, probably conditioned by the / t͡ʃ/ >/ t͡s/ merger, mirrored this change by shifting the place of articulation of the phoneme / d͡ʒ/ to / d͡z/ , including several Piedmontese dialects. This sound today is kept only in the areas where / t͡s/ is preserved; otherwise, it is merged with / z/ . Those dialects that keep / t͡ʃ/ or deaffricate it to / ʃ/ (and do not merge it with / t͡s/ ) do the same with / d͡ʒ/ , i.e. they keep it as such (Valsesia) or deaffricate it to / ʒ/ (Western Lombard, dialects of the Biella area). A different, asymmetrical, development is found in French and some Gallo-Italian dialects (such as the dialect of Turin 25 ): / d͡ʒ/ never advanced into / d͡z/ , even if the / t͡ʃ/ > / t͡s/ merger had taken place. In the weak intervocalic position, the development is more straightforward. / t͡ʃ/ merged with / d͡ʒ/ due to the general western Romance voicing, and then the phoneme was reduced to / ʒ/ (Ligurian dialects and conservative Lombard dialects preserve this stage) and then to / z/ (Lombard, Piedmontese etc.), nuceM ‘nut’ > * noge > [nuʒ] > [nuz]. Other sources of alveolar sounds are Latin S / s/ , which is kept in a strong position and voiced to / z/ in a weak position, while Latin x -, pS -, - SS -, - Sc usually yield / s/ . In some varieties - Sc can evolve in / ʃ/ (Rohlfs 1966: 314-16). To summarize, Piedmontese dialects exhibit, just like the Lombard ones described by Sanga (1984: 45-59) several different mergers: 1) The traditional dialects of the Biella area keep / ʃ/ / ʒ/ (< late Latin / t͡ʃ/ / d͡ʒ/ ) and / t͡s/ / d͡z/ as the dialect of Milan until the 19 th century, while the more progressive ones merge / t͡s/ and / d͡z/ with / s/ and / z/ , as the contemporary dialect of Milan. 2) The dialects that keep / t͡s/ and / d͡z/ (in which late Latin / t͡ʃ/ / d͡ʒ/ were already merged) in Southern Piedmont exhibit the system of the older dialects of Bergamo and Venice and the so-called medieval Lombard koine. 24 The dialect of Milan shows both / t͡s/ (nowadays / s/ ) and / ʃ/ , reflecting a possible sociolinguistic variation of the late Middle Ages. The usage of / t͡s/ was typical of the upper class and the more native outcome / ʃ/ was typical of the lower classes: see Sanga (1985) for a reconstruction. 25 The dialect of Turin is ambiguous in this regard because it shows different outcomes in a lexemebased fashion (e.g., gingiVaM > [zanˈziva] ‘teeth gum’, but geLare > [d͡ʒeˈle] ‘to freeze’). This can be attributed both to Italianization and the persistence of a Gallo-Romance outcome (Ferrarotti 2022: 121). 113 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 3) A complete merger of all the previously mentioned phonemes in / s/ / z/ (as in the modern urban dialects of Bergamo, Brescia, and Venice) is the most typical situation of contemporary Piedmontese dialects, especially in the central area of the region (including Asti) and, crucially, in the Turin dialect (in this specific case see N 25). As regards OAD, various authors have posited that the latter merger (3.) was already typical of it, as Bottasso (1953: XI-XIII) and Clivio (2003: 139). Giacomino’s (1901) opinion is less clear, as it will be seen. 2.2.1 Alveolar fricatives To evaluate the presence of the various mergers, one can observe in the first place how the phoneme / s/ and its voiced counterpart / z/ are represented and distributed. In the OJ it is clear that in the word-initial position, ‹ s › represents / s/ , i.e., mainly the outcome of Latin S -: B101 Seynt ‘saint’ < SanctuM , son ( passim ) ‘(they) are’ < Sunt etc. No word-initial / z/ seems to be found, as in contemporary Piedmontese, because it developed from the deaffrication of / d͡z/ into / z/ (see above and below). 1) Word-internally, ‹ ss › is / s/ 26 , stemming from Latin - SS -, - Sc - x -, pS -. A22 esser ‘to be’ < * eSSere , A524 nessir ‘to go out’ < * ne+essir < inde + exire , F118 lasser ‘to leave’ < Laxare , D152 casson ‘big chest’ < * captiaM + oneM etc. A relevant case is Pr.23 cossa ‘thing’ < cauSaM , which shows the lack of the Western Romance voicing after the au diphthong, a once typical 27 Gallo-Italian feature (Hull 2017: 325), such in C248 oche ‘geese’ < * aucaS < aVicaS . Also, in morpheme boundary ‹ ss › is employed to represent / s/ , such as in A510 strassuà ‘sweaty’< extra + SudatuM . 2) Word-internal ‹ s › stands for / z/ , stemming from Latin - S - (always voiced between vowels as effect of the western Romance voicing), c - + e , i and tj in an originally weak position: - S - H204 sposa ‘bride’ < * Spo ( n ) SaM ; c - + e , i , F616 asý ‘vinegar’ < acetuM , A261 piasa ‘he like’ (subjunctive) < pLaceat , amisi 28 ‘friends! (excl.)’ < aMici etc.; tj -: B39 presi ‘price’ < pretiuM , B269 servisi (< SerVitiuM ), A112 rason ‘reason’ (< rationeM ), I35 mangiason ‘binge’ ( manger + ationeM ) 29 etc. 26 It is not entirely clear if the phoneme was lengthened in this position as the etymology would suggest (see §3.7.1). 27 In contemporary Piedmontese [ˈkɔza] is prevalent ( REP 473). 28 Amisi is found only three times (C285, E222, I440) as an exclamation, always in rhyme with servisi ‘service’ (thus confirming the merger of c - + e , i and tj -). It is not clear why the -i inflection was kept in this emphatic form, as the usual plural is B286 amis/ F364 amiz , while the singular form is A114 amý (feminine A226 amìa ), as they come respectively from aMici and aMicuM . This mirrors the use of old Lombard forms amig/ amis , because in many contemporary Piedmontese and Lombard varieties the plural amis has been extended to the singular through back-formation (see Salvioni 1900). 29 Cf. words like pregio, servigio, ragione, and the suffix -agione with a «weak» outcome that were common in Old Tuscan, probably as loanwords from Lombard varieties (Rohlfs 1966: 409). 114 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 3) Word-final ‹ s › always represents the / s/ outcome of - SS -, - Sc -, x -, pS in a quite regular fashion: The imperfect subjunctive (from the Latin pluperfect) as A631 haves ‘(they) had’ < * haBeSSet < haBuiSSet (cf. word-internal -ssas in A863 havèsson ‘(they) had’), A298 gros ‘big’ < late Lat. groSSuM ; I97 pes ‘fish’, < piSceM , Pr.40 nes ‘(he) goes out’ < * ne+es < inde + exire , ades ‘now’ ( passim ) < ad ipSuM etc. In other cases, world final ‹ s › can also stand for the / z/ outcome of weak - S -, c -/ g - + e , i , and -* dj that came to be at the end of the word; but in these cases, there is a great deal of oscillation between ‹ s › and ‹ z › and sometimes ‹ x › 30 . This is clearly influenced by French usage (Nyrop 1899: 353), in which they were both used for final / s/ after the loss of the phoneme / t͡s/ . B346 gris vs A63 griz ‘grey’ < Germ. *gris ( FEW 16: 80); E438 jus vs E319 juz ‘judge’ < JudiceM ; A873 fis vs. A784 fiz ‘(he/ she) did’ < Feci / Fecit ; A592 vis vs. A241 viz ‘seemed’ < ViSuM 31 . The sequence / us/ or / uz/ is always spelled ‹ oux › to oppose it to ‹ oz › that represents / ɔs/ or / ɔz/ . Some rhymes confirm the interchangeability of ‹ s ›, ‹ z › and ‹ x ›, such as nariz ‘nostril’ : Beatrix ‘Beatrix, proper name’ (B560-61; see also D351-53, F449-50, F607-08). Moreover, / s/ from Latin - SS -, - Sc -, x -, pS is almost never found in rhyme with / z/ (stemming from - S -, c -/ g - + e , i , dj -) with only two exceptions 32 . So, it is not entirely clear if the / s/ and / z/ phonemes were neutralized word-finally in OAD, and maybe they were only partially so. However, it is quite clear that - Sc -, x and weak c -/ g - + e , i , which, potentially, could have developed in / ʃ/ and / ʒ/ (see above), are not represented by graphemes different from ‹ s › ‹ z › that are employed for / s/ / z/ and then they are indistinguishable from the outcomes of - S -. Therefore, it can be concluded that no / ʃ/ and / ʒ/ were present, and that they were merged with / s/ and / z/ word-internally. 30 Sometimes an etymological or pseudo-etymological ‹ x › can be found: A42 sex ‘six’ < Sex , C160 dex ‘ten’ < deceM , A640 pax ‘peace’ < paceM , cf. nom. pax , along with C77 paz, I254 pas . 31 Some functional specialization can be found in dis vs diz . Both represent the present form ‘(I) say, (he/ she) says’, but seemingly only the first is used also with the meaning of the perfect ‘(I) said, (he/ she) said’ (unambiguously at least in B398, C124, G352, H50, H 597, I252, I254, M62, M105, M118). 32 Cas ‘case’ < caSuM , rhymes regularly with / s/ -ending words as bas ‘low’ < BaSSuM (A189-90, A718- 19, E19-20), two imperfect subjunctive forms anganas ‘(he) deceived’ (F357-58), anterficas ‘(it) meant’ (G309-10) and the French loanword Ypocras ‘hypocras, drink’ ( Gdf. dict ., s.v. hypocras ); meys ‘month’ < Me ( n ) SeM rhymes with deis ‘(he) gave’ (imperfect subjunctive) < * deSSet < dediSSet in B405-06. A counterexample is cas that rhymes with pas ‘peace’ in A582-83. 115 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 2.2.2 Alveolar affricates Up to this point, the status of the alveolar fricatives poses no significant issue, as their phonemic status is almost the same as the contemporary mainstream Piedmontese dialects. More questions arise with the interpretation of the graphemes ‹ c › + ‹ e ›, ‹ i › and ‹ z › both word-initially and word-internally (and in some cases word-finally), and ‹ cz › word-finally. Bottasso (1953: XI-XII) and Clivio (2003) thought that these spellings were merely graphical in nature and that they had the phonemic value of the contemporary Turin dialect, i.e., the alveolar fricatives / s/ and / z/ while Giacomino (1901: 405, 422) believed that ‹ c › + ‹ e ›, ‹ i › represented / s/ 33 , while cz was / t͡ s/ and ‹ z › both / t͡ s/ and / d͡ z/ . Salvioni (1905, I 157), reviewing Giacomino’s article, argued that ‹ c › in ‹ ce › and ‹ ci › actually represented / t͡ s/ too. Indeed, Salvioni’s hypothesis is the most well-informed, because it better fits the phonological history of Gallo-Italian dialects, while the others are too much influenced by the phonology of contemporary Piedmontese dialects (viz. Turin) and do not consider the vast internal evidence that / t͡ s/ and / d͡ z/ were not merged with / s/ and / z/ in OAD (see below). In fact, the use of ‹ c › + ‹ e ›, ‹ i › for / t͡ s/ , typical of many medieval Romance languages (Spanish, French) was very well present in Lombard spellings in Northern Italy (often overlapping with ‹ z ›) until the Early Renaissance 34 as Sanga (1984: 154-64; 1985: 19-26; 2006: 379-80) has abundantly shown. This is clear from the evidence presented by OAD too. For the outcomes of the voiceless Latin sounds c + e , i (> * cj ) * tJ -, *tJ - ( ttj ) in a strong position the following spellings are found: 1) word-initially ‹ ci › + ‹ a ›, ‹ o ›, ‹ u › , ‹ c › + ‹ e ›, ‹ i › but sometimes ‹ z ›; 2) word-internally ‹ ci › + ‹ a ›, ‹ o ›, ‹ u › , ‹ c › + ‹ e ›, ‹ i › but also ‹ z ›; 3) word-finally ‹ cz ›. 1) In the word-initial position, ‹ ci › + ‹ a ›, ‹ o ›, ‹ u › ; ‹ c › + ‹ e ›, ‹ i › are the most common spellings. Ciò ‘that’ ( passim ) < * cjo < ecce hoc , E31 cena ‘supper’ < cenaM , A19 cincq ‘five’ < * cinque (< quinque ), G460 ciogna ‘stork’ < ciconiaM , C110 ciop ‘lame’ < * tsopp - (REW 13/ 2, 347, cf. [t͡sop] in contemporary southern Piedmontese and northern western Lombard dialects, AIS 191 and Giamello 2007: 45). Sometimes ‹ z › and ‹ ci › alternate on the same lexical item: G363 ciambel / A677 zambegl ‘troubles sg. / pl.’ < cyMBaLuM ( FEW 2: 1611), cià ( passim ) / za (B99) ‘here’ < * cja < ecce hac 35 , E58 ciascun / A94 zascun ‘everyone’ < * ciSque unuM < quiSque unuM 36 . 33 The same value is attributed by Rossebastiano and Papa (2011) in their survey of Medieval Piedmontese text, which, however, does not include the OJ . 34 This norm is alien to the concurring central Italian spellings that would replace the Lombard ones from the Renaissance period, where ‹ ce › and ‹ ci › represent a postalveolar affricate. 35 In several instances the cià/ za overlap is problematic since za also represents the JaM outcomes, as the very common discourse marker ‘already’; the minimal pair / t͡sa/ ~ / d͡za/ is still well attested in contemporary Alta Langa dialects (Giamello 2007: 29) 36 Cf. also in the OJ the French (or Italian? ) influenced form A266 chiascun (Ghinassi 1976: 97). 116 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 In some other cases, the voiceless sound is always represented by ‹ z ›. I263 zuch ‘log’ < * tsukk (cf. It. zucca ‘squash’ rep 1428; [t͡syk] in some southern Piedmontese dialects) , A398 zorgn ‘deaf ’ (etymology unclear, [t͡sɔrɲ] in conservative contemporary dialects, see AIS 190) , P45 zocre ‘clogs’ < SoccuM (but / t͡s/ in It. zoccolo and contemporary southern Piedmontese dialects, AIS 1569) etc. 2) In the word-internal position, the graphemic representation of the phoneme / t͡s/ oscillates between ‹ ci › + ‹ a ›, ‹ o ›, ‹ u › , ‹ c › + ‹ e ›, ‹ i › and ‹ z ›. Occasionally this is problematic, as will be seen, because the ‹ z › grapheme is used to represent the voiced counterpart / d͡z/ . In this environment only consonants that were originally in an internal strong position are found (see the reconstruction above). It is important to note that an intermediate *ccj - (< ce -, ci -) must have developed from a lengthening of the consonant in the suffixes aceaM (augmentative/ pejorative), uceaM , iceaM (diminutive; cf. Italian -accia, -uccia, -iccia ) as well as in words containing such segments. B286 ‘big pants’ < BracaS + *accja ; E65 tortuce ‘small pies’ < * tortaM + < *uccja ; see also C333 pellicer ‘furrier’ < * pelliccja + ariuM , facia ‘face’ ( passim ) < * faccja < FaciaM / ‘(he/ she) do’ (subj.)’ and ‘face’, G74 limaza / E58 limace ‘snail/ snails’ < * limaccja < * LiMaceaM , cf. [liˈmat͡sa] in contemporary southern Piedmontese dialects, AIS 459, Giamello 2007: 151; E66 nizoere ‘hazelnuts’ < * noccjole < nuceoLaS , cf. It. nocciola , and [niˈt͡søɹe] in some contemporary southern Piedmontese dialects ( AIS 1302, Giamello 2007: 50). Other instances of strong *cj appear after consonant. A173 cace ‘tights, trousers’ < * calcjas < caLceaS , A172 caucer ‘boot’ < * calcjariu < caLceoLariuM , A226 dolza ‘sweet’ < duLceM , A503 pancia / P49 panza ‘stomach’ < * pancja < panticeM etc. The outcomes of *ttj or strong tj are well represented 37 . O74 cancion ‘song’ < cantioneM , B196 commenza ‘(he) begins’ < * cuMin ( i ) tiare , A278 spucia ‘smell’ < ex - + * putJare < putiuM (cf. Italian puzza ), F403 stancia ‘room’ < StantiaM , G275 forcia ‘force’ < * FortiaM , I788 moce ‘cut off’ < * muttjas (cf. it. mozzo ), G441 haucer ‘to rise’ < * aLtiare , G232 sacier ‘to satiate’ < * Satiare , F167 lincoeu ‘bed sheet’ < * lentjolo < LinteoLuM . As previously mentioned, in several cases there is a very significant oscillation between ‹ c › and ‹ z › on the same lexical base, a proof of the identity of the phoneme represented by these graphemes. A500 cacer ‘to hunt’ , O3 cacià / E6 cazà ‘hunted’ or ‘stuck’ < * captiare , or I654 maza / I263 macia ‘club’ < * mattja < MateaM , I839 amazà killed’ / E256 amacer ‘to kill’ < * adMattJare ‘to kill’, D250 anciuma / D263 anzuma etc. ‘on top’ < in + cyMa (cf. [ənˈt͡syma] in contemporary Alta Langa dialects, Giamello 2007: 120) and more. 37 Cf. also the outcome of * pti in H275 noce ‘marriage’ < nuptiaS and the suffix *itia , that was strengthened to *ittja (cf. it. -ezza ): E181 drueza ‘pleasure’ < * druto + itia ( FeW 3: 164) , A252 freza ‘hurry’ < * FrictiaM ( FEW 3: 784). 117 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 3) Word-finally, the most common spelling is ‹ cz ›, a digraph typical of medieval French and Spanish writing systems (that in the 16 th century evolved in ‹ ç › and was used to represent / t͡s/ , see Nyrop 1899: 104). Outcomes of internal *ccj and *ttj are also found here due to the fall of word-final Latin unstressed vowels typical of Gallo-Italian dialects. In many cases they are masculine-inflected forms of many words and suffixes seen in 2) above: G256 vegliacz ‘bad old man’ < VecLuM < VetuLuM + *accjo < aceuM (cf. -acia / -aza ), G507 meystrucz ‘little master’ < Ma ( g ) iStruM + * -uccjo < uceuM (cf. -uza , -ucia ), siacz ‘sieve’ < * setaccjo < SaetaciuM ; cf. also the apparently nominatival form G293 fecz ‘dregs’ < Faex (cf. fetz in old French and fez in old Lombard texts, FEW 3: 366) and the positionally strong * tj in A317 soencz ‘often’ < * sovente <* SuBentiuS , A517 Laurencz ‘Lawrence’ < L aurentiuM . A morphologically relevant case is F69 porcz ‘pig’ in which the singular form is modeled on the plural porci through a back-formation (as amisi above, see Salvioni 1900). Likewise, derivation processes and different types of inflections show well that ‹ cz › is a positional variant of internal ‹ ci › + ‹ a ›, ‹ o ›, ‹ u › , ‹ c › + ‹ e ›, ‹ i › and ‹ z ›: B513 dolcz / B73 doucz ‘sweet’ (m. sg.) vs. A226 dolza / F57 doza ‘sweet’ (f. sg.) 38 , B259 facz ‘I do’ vs. facia ‘I do’ (subj.), A534 nicz ‘overripe’, by extension ‘bruise’ vs. I534 annicia ‘to make blue’, ‘to offend’ (cf. [nit͡s] in contemporary dialects, AIS 1259, REP 1002-1003, FEW 6/ 2: 182-83 < MitiuM ? cf. it. mézzo ), H161 pelicz ‘fur’ < * pelliccjo < peLLiceuM vs. C333 pellicer ‘furrier’ + ariuM , A173 cace ‘britches’ < caLceaS vs. B34 cacz ‘I put on (shoes)’ < * calcjo etc. However, some cases are apparently problematic. Words as Pr.4 capacz ‘able’ and E353 sagacz ‘shrewd’ show a strong *ccj - > / t͡s/ outcome that is incompatible with the etymological bases capaceM and SagaceM , whose expected outcomes should have been * cavas and *saas, with intervocalic / z/ and the voicing and weakening of intervocalic Latin voiceless plosives typical of Piedmontese dialects (Ferrarotti 2022: 25- 28), if OAD words like C176 pas ‘peace’ < paceM , C39 louf ravas ‘werewolf ’ < LupuM rapaceM are taken into account. Indeed, these are not indigenous words, but instead learned borrowings coming from a prestige or high variety: this case is comparable to that studied by Sanga (1984: 158, 1985: 21), who showed that capazz in Fabio Varese’s texts in the 16 th -century dialect of Milan do not show the expected / ʃ/ outcome, because it is actually borrowed from Italian through an adaptation strategy that traces back to the older Northern Italian vernacular tradition. The same goes with words ending with the iciuM / itiuM suffix: we find -‹ ici › as ending as A110 benefici ‘benefit’ < BeneFiciuM , A161 offici ‘duty’ etc. < oFFiciuM , A752 propici ‘favorable’ < propitiuM ; cf. A903 ociose ‘idle (f. pl.)’ < otioSaS , especially if compared with -isi ending words (as B269 servisi ) that show no such strengthening, and the Western 38 In this case ‹ l › was present as just an etymological «silent» grapheme, as these words were probably realized as / dut͡s/ , / ˈdut͡sa/ (cf. the dolza : coza ‘gourd’ rhyme in I270-71). 118 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 Romance intervocalic voicing instead (see again offizij in Fabio Varese’s texts in Sanga 1984: 158). Another problematic case is C304 chisi ‘chickpea’: in fact, / ˈt͡ʃizi/ (attested also in contemporary dialects, REP 428) does not respond to the Latin bases cĭcer or * cĭcer eM , as one would expect * ceis(e) / ˈt͡sei ̯ z(e)/ . This asymmetry was attributed to a dissimilation process by Giacomino (1901: 422). In Old French there was an alternation too between the etymologically linear ceire (< * cĭcerem ) and the non-linear chiche (Nyrop 1899: 319-20). Chisi , however, seems to be a singular form back-formed on the plural, as amisi (see above, §2.2.1 N 28), due to the prototypical occurrence of the referent as a plurality. It is possible, following in part Fouché (1966: 55), that a base *cici or more likely * cisi evolved to / ˈt͡sizi/ and then dissimilated to / ˈt͡ʃizi/ (in which the stressed vowel and / z/ are probably the result of a folk-etymologization with * pīsi ‘green peas’) 39 . Additionally, several minimal or subminimal pairs can be identified that involve / t͡s/ vs. other alveolar and postalveolar phonemes. - with / t͡ʃ/ : C137 chiri ‘cleric’ < cLericuM ~ I171 ciri ‘church candle’ < cereuM , A622 peg ‘breast’ < pectuM ~ A438 pecz ‘piece’ < * pettjo < * pettiaM , F175 lacet ‘small lace’ < * laccjo < * LaceuM + ittuM ~ I53 laget ‘sweetbread’ < LacteM + ittuM ( REP 849); - with / s/ : B298 cerf : ‘deer’< cerVuM ~ A297 serf ‘servant’ < SerVuM , Pr. 7 pacz ‘crazy’ (etymon unclear, but cf. It. pazzo ) ~ C177 paz ‘peace’ < paceM , A293 civil ‘civil’ < ciViLeM ~ A294 sì vil ‘so coward’ < Sic ViLeM , B199 cent ‘one hundred’ < centuM ~ A410 sent ‘I hear’ < * sento < Sentio , E106 masser ‘farm owner’ < * MaSSariuM ~ I629 macer(gle) ‘to kill’ < * adMattJare . It is also noteworthy that words bearing final ‹ cz › never rhyme with anything else than ‹ cz ›-ending words, a further clue that rules out any merger. Finally, the riddle in N64 k5q , should be interpreted as ka cincq qu = * cacz in cu / kat͡s in ky/ i.e. ‘kick in the ass’ (< * calcju < caLceuM , Bottasso 1953: 295), that seems to work well only if / t͡s/ is a separate phoneme from / s/ (and, again, shows the identity of initial / t͡s/ in / t͡sink/ and the final / t͡s/ in / kat͡s/ ). The representation of the voiced counterpart / d͡z/ is clearer. It is rendered through ‹ z › word-initially and internally, and with ‹ cz › or ‹ z › word-finally. The following diachronic sources can be isolated. 1) Word-initially, the outcomes of g + e , i and J are found: 39 Some minor issues can be found in the given name Cacian < c aSSianuM , possibly folk-etymologized with cacé ‘to hunt’ and giandousse ‘small glands’, ‘bubonic plague’ (* gLandaM + uceaM ), where the suffix -ousse does not appear elsewhere and does not respond as the outcome of uceaM (cf. tortuce ‘small pies’ in E65): for this, it could be a loanword from Venetian giandussa ( GDLI 765, Boerio 1856: 305). 119 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect za ‘already’ ( passim ) < JaM , B623 zóvon ‘young’ < JuVeneM , B43 Zohan ‘John, given name’ < JohanneM , F201 zuré ‘to swear’ < Jurare , G119 zenougl ‘knee’ < genuc ( u ) LuM , G512 zanzive ‘gums’ < gingiVaS , E529 zué ‘to play’ < Jocare , C181 zobia ‘Thursday’ < * JoVia etc.; see §2.1 for a different dialectal outcome in the OJ . 2) Word-internally, outcomes of -* gg -, and strong position -* dj are found: A97 reze ‘to hold’ < * reggere < regere , F25 astrenza ‘(he) tighten’ (subjunctive) < ad + Stringere , I186 foza / C136 feuza ‘fashion’ < * foddja < FoVeaM (cf. It. foggia ), B16 manezé ‘to handle’ < ManuM + *eddjare < idiare ; it appears also in numerals like H445 treze < * tredci < tredeciM , H329 quinze < quindeciM ; / dz/ is found in the learned word I419 scandalizà ‘to be scandalized’ < ScandaLizare as well. 3) Word-finally, outcomes of strong -* gjand -* dj are found 40 : B235 loncz ‘far away’< Longe , B179 broncz ‘bronze’ < possible loanword from It. bronzo , I60 Giorcz ‘George’ < g eorgiuM , and B306 foncz ‘mushroom’ < Fungi (which is a backformation on the plural form, Salvioni 1900, see porcz above) 41 . In conclusion, the distribution of the graphemes used to represent the alveolar and postalveolar affricates is position-dependent, not only as far as the spelling is concerned, but also as regards the representation of contrasts based on voicing. In the word-initial position there are few overlaps between the various spellings for / t͡s/ and / d͡z/ because in this position the voicing opposition is quite relevant in terms of functional yield. Word-internally, the graphemes for / t͡s/ and / d͡z/ are more interchangeable, as contrasts are limited in this environment: ‹ z › and ‹ g › can often represent a voiceless sound, but not the other way round. Finally, in the word-final position, almost no contrast is marked graphemically: ‹ cz › and ‹ g › are the only spellings employed, and this is due to the general devoicing of obstruents that was probably a feature of OAD (see §4.2). 2.3 Palatal lateral and palatal approximants In contemporary Piedmontese dialects the palatal approximant / j/ stems from different Latin sounds: 1) word-internal - LJ and cL - (as a delateralization of / ʎ/ , and only in Western and Southern Piedmontese, see Ferrarotti 2022); 40 Because this graphemic representation is the same as the voiceless counterpart, it is possible that the two sounds were merged in this position due to the general devoicing of word-final obstruents (see §4.2). 41 In other cases ‹ z › can be employed: A335 mez ‘half ’ < * meddjo < MediuM , that contrasts clearly with mes , both A626 ‘messenger’ < MiSSuM and E107 ‘harvest’ < MeSSeM and G21 crez/ F156 crecz ‘(I) believe, (he/ she) believes’ < * creddjo . Also A155 lez ‘law’ must come from the strengthened * legge < LegeM , as it rhymes once with G447 recz ‘net’ (from nominative retiS ; cf. Old French retz FEW 10: 331; see fecz above). 120 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 2) as a result of the weakening of c -. In OAD - LJ and cL seem to keep the older phase / ʎ/ , represented by ‹ gl ›, that is lost nowadays in almost all Piedmontese dialects 42 . Therefore, it is not yet merged with / j/ stemming from the lenition of c that is represented in the OJ by intervocalic ‹ i › or ‹ y ›. Several minimal and subminimal pairs can identify this contrast. - A76 vey ‘true’ < VeruM ~ A82 vegl ‘old’ < * VecLuM < VetuLuM - A74 braye ‘britches’ < BracaS ~ C39 braglia ‘(he) shouts’ < * Brac ( u ) Lare ( FEW 1: 490) - A852 mey ‘physician’ < MedicuM or A215 ‘mine’ (pl.) < Mei ~ A78 megl ‘better’ < MeLiuS - A71 voyé ‘(she) emptied’ (pf.) < * Vo ( ci ) tuM + are ~ B12 avogler ‘needle case’ < acuc ( u ) LaM + ariuM - ay ‘to the’ (articulate preposition a + y , passim ) ~ A910 agl ‘garlic’ < aLLiuM - A852 pey ‘hair’ sg. or B409 ‘pear’ < piLuM or piruM ~ G42 pegl ‘hair’ (pl.) < * peil < piLi - voy ‘(you) want’ ( passim ) < VoLeS ~ B482 vogl ‘(I) want’ < *voljo < * VoLeo . Through several rhymes with French, it can be ascertained the identity of the sound represented by OAD ‹ gl › and French ‹ ll › (that, at that time was still / ʎ/ , see Picoche/ Marchello Nizia 1994: 199, 206), as in faldigle ‘crinoline’ (OAD) : habille (French) ‘(she) wears’ (D195-96) and in the French loanword B79 chiaglia < chaille ‘it bothers’ (pres. subj. of chaloir ) . / ʎ/ appears too as a palatalization of final -l in the plural forms B572 gagl ‘roosters’ , C307 cavagl ‘horses’, probably from *gail, *cavail (cont. Piedmontese [gai̯], [kaˈvai̯]) and in the plural form of the article or the locative clitic i before vowels: A100 gl’eyg ‘the others’ , F541 gl’èron ‘there were’ . There are some cases in which a partial reduction of / ʎ/ in favor of / j/ could be seen, mainly in the plural forms of col ‘that one’ and tal ‘such’: coy/ choy (12+24 tokens) vs. cogl (6); tay is the only plural ( tagl has 3 tokens, but only with the meaning of ‘cut’ < * taLiuM ). Salvioni (1905) sees as exceptions D163 fya ‘faith’, that he deems the outcome of < FiLiaM , but it comes probably from * FidaM , because H118 figla is well attested in OAD. A true exception is M15 coy ‘cabbage’ < * cauLi ? < cauLeS , in which we would expect * cogl if it actually descended from * cauLi . in this case, either a reduction of the coy-tay type occurred or some other origin must be posited. 42 Today it is found mainly in some Occitan dialects of Piedmont (High Susa, Chisone, Germanasca and Pellice valleys). It was attested in conservative Valsesia dialects at the beginning of the 20 th century (Spoerri 1918: 684). 121 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 2.4 Alveolo-dental trill and alveolo-dental lateral approximant In many southern Piedmontese dialects and in some conservative Ligurian ones (but also in Genoese before the 18 th century), a merger of intervocalic / l/ (from Lat. - L -) and / r/ (from Lat. r -) phonemes is attested (it was common also in old Lombard, see Debanne, Delucchi 2013), whereas long / rː/ and / lː/ remained separated. Still, at some point, due to a possible tap [ɾ] articulation of intervocalic / r/ (cf. Spanish caro ‘dear’ vs carro ‘carriage’) and the shortening of long / rː/ and / lː/ , a new system of distinctions arose, that was based mainly on the manner of articulation. Intervocalic / r/ became then articulated as / ɹ/ (< lat. r -, - L -) with a notable lack of vibration and an approximant (and in some descriptions retroflex) articulation. This new phoneme got to contrast both with / l/ (< lat. - LL -) and / r/ (< lat. rr -). Berruto (1974: 32) cites as example of this phenomenon the minimal pair / saˈra/ ‘closed’ < SerratuM ~ / saˈɹa/ ‘salted’ < SaLatuM , and Miola (2013: 65) / ˈfola/ ‘fool’ (f.) < * folla < FoLLeM (with inflectional class change) ~ / ˈfoɹa/ ‘outside’ < ForaS . In OAD this situation is reflected, in several instances, by a significant oscillation in writing between ‹ r › and ‹ l › on the same lexical item 43 as the outcome of an etymological - L -: E170 teyra - A762 teyla ‘cloth’ < teLaM , B35 oly - E256 ory ‘oil’ < oLeuM , H354 bayla - C103 bayra ‘nanny’ < * BaJuLaM etc. This is noticeable also in some subject clitic (SC) pronouns, e.g., r’è / l’è SC.3 Sg =be.3 Sg . preS , ra/ la Sc.3SG.F ( passim ). Moreover, some lexemes bearing the outcome of etymological - L are found exclusively with written ‹ r ›: A132 erbe orente ‘parsley’ < herBaS oLenteS , ([arbuˈrenti] in contemporary dialects, see AIS 1385, Ferrarotti 2022: 215), A203 zeer ‘frost’ < geLuM , E520 carier ‘shoemaker’ < caLigariuM , A584 candeyra ‘candle’ < candeLaM etc. The contrasts identified by Berruto and Miola (see above) are well attested in OAD too, and they rendered as H438 sarrà ‘closed’ ~ I97 (pes) sarà ‘salted (fish)’, A617 fola ‘fool’ (f.) ~ B231 fora ‘outside’. In any case, it is possible that the graphical oscillation reflected some uncertainty in the representation of the / ɹ/ phoneme and not a variable realization between / r/ and / l/ . This is shown by several rhymes, in which the identity of the sound between etymological r and - L can be observed: hora ‘hour’ : gora ‘throat’ F381-82 (< horaM : guLaM ) , relory ‘clock’ : ory ‘oil’ g367-68 (* relorjo ? FEW 4: 484 horoLogiuM : oLeuM ) , seyra ‘evening’ : candeyra ‘candle’ i611-12 ( SeraM : candeLaM ); see also relicore ‘relics’: baricole ‘glasses’ G457-58 ( reLiquiaS + oLaS and a learned loanword from old French béricle < BeryLLiuM , FEW 1: 339), in which a word bearing ‹ r › rhymes with another bearing ‹ l ›. Furthermore, the presence of a non-trill, approximant articulation of this phoneme is explicitly attested for the dialect of Asti in a metalinguistic poem of 1786 43 Giacomino (1901: 416-17) thought that this kind of alternation in the text was an attempt by the author to reproduce some class variation, whereby ‹ r › = low and ‹ l › = high. Although this hypothesis is plausible and historically informed, Giacomino’s claims were based mainly on cherry-picking and not on quantitative correlational data. 122 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 by the priest Stefano Incisa 44 . The same is attested for the dialect of Genoa in 1745 45 (and its then ongoing deletion underlies a much older preexistence). The only two exceptions in the diachronic development described above are found in the outcomes of Latin gaLLinaM ‘hen’, (H320 garina ), and poLLiceM ‘thumb’ (I89 pori ) which unexpectedly exhibit the rhotacism, because usually - LL evolves regularly in / l/ in OAD (see above). This is still attested in the contemporary dialects in the surroundings of Asti ( AIS 1122 «hen», 154 «thumb»). Also, O3 steyra ‘star’ exhibits rhotacism, but, in this case, it is expected because the etymological base is * stēlam (as in French étoile , FEW 12: 254 and in most contemporary Piedmontese dialects, see Ferrarotti 2022: 64-68) and not SteLLaM . 2.5 Other sound classes 2.5.1 Velar nasal Giacomino (1901: 405, 419-40) believed that the presence of double ‹ nn › (often marked with a superscript tilde on the preceding vowel and single ‹ n › in the 1521 print, viz. bo͂ne ) could represent the peculiar Southern Piedmontese evolution of Latin intervocalic and posttonic n consisting in a [ŋn] sequence (Ferrarotti 2022: 84-87) or, at least, in a simple intervocalic velar nasal [ŋ] still typical today of many Piedmontese and Ligurian dialects (Rohlfs 1966: 311-12). The alternation that is present in some rhymes shows that this feature was not systematically notated: bonna ‘good’ (f.) : gona ‘gown’ E182-83 (< gunnaM , cf. old French gonne , FEW 4: 325- 27). However, ‹ nn › appears in other contexts: A176 bonnet ‘hat’, F567 ordonnanza ‘decree’, D300 lyonneyse ‘from Lyon’, in which it is probably influenced by French usage ( GGHF 567-72). While Giacomino’s claim is well-informed, the diversity of environments in which ‹ nn › is present cannot unambiguously tell if a [ŋn] or [ŋ] articulation actually occurred, as no minimal pairs can be found. 44 A «sweet» and «barely audible» sound, see Gasca Queirazza (2010: 198, « cossì doss ch’appenna appenna a ’ss sent» ). A further but much more debatable hint could be a passage in the OJ (H188- 96), in which the comic effect can be fully understood only with an approximant articulation due to the presence of repeated postverbal and postadverbial object clitics -ro and -ra ( veytro laro, veytra quira ‘see him here’, ‘see her there’, diro ‘say it’); some sorts of these shibbolets are still widespread in areas of southern Piedmont whose local dialects have this kind of realizations (see Ghia 2015). 45 Parodi (1902-1905: 305), reports that in the preface to the 1745 edition Çittara Zeneize by Giangiacomo Cavalli (1630 ca.) there is a clear description of an approximant realization of this phoneme that is strikingly similar to the sound described in the case of Asti («non si pronunzia, o, per meglio dire, o si pronuncia così dolce, che appena se ne oda un legger mormorìo» ‘it is not articulated, or, better to say, it is articulated so sweet that just a slight murmur is heard from it’). 123 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 2.5.2 Labiovelar approximant In the OJ corpus, the ‹ gu › digraph is commonly used to represent the outcomes of Latin qu - 46 , gu or Germanic *w, which had been adopted as / gw/ in late Latin and early Romance (Rohlfs 1966: 230-31, Ferrarotti 2022: 125-27): A617 lengua ‘tongue’ < LinguaM , A798 guarder ‘to look’ < Germ. *wardon ( FEW 17: 510), G519 agueyreu ‘sink’ < aquariuM + oLuM . However, it is possible that ‹ gu › does not represent two segments (/ g/ + / w/ ), but that is a mere graphical device to represent the labiovelar approximant / w/ , as in contemporary rural Montferrat dialects (Ferrarotti 2022: 125- 27). In a few cases, an alternation with the digraph ‹ vu › is discernible: A409 vuary - C63 guary ‘much’ < *waigaro ( FEW 17: 469-71), D334 vuagn - I331 guadagn < *waidanian ( FEW 17: 461-69), B392 vuardte ‘look yourself ’ - A798 guarder ‘look’ < *wardon ( FEW 510-24). Moreover, ‹ gu › is found in typically dialectal words that are still found in contemporary rural dialects with no learned equivalent, such as O32 guachion ‘peek’ < *wahtan ( FEW 17: 451-57), and C155 guecha ‘limp’ < *thwaírh ( FEW 17: 410-12) and Pr. 9 angual ‘peer, equal’ < * in - + aequaLeM ( rep 57). At least one minimal pair can be identified: A231 guant ‘glove’ < Germ. *want ~ C172 vant ‘pride’ < deverbal noun from Vanitare (it. vanto ). the contrast between initial / v/ and / w/ seems to be relevant in the case of the western-Piedmontese speaking character, who employs I690 vardia and I593 vayre instead of I587 guardia and vuary/ guary (cf. Ferrarotti 2021). 3. Vowels 3.1 Mid front unrounded The phonological evolution of Piedmontese unrounded front mid vowels / e/ and / ɛ/ is quite complex and varies noticeably depending on the variety examined (Ferrarotti 2022: 64-68, 71-77). No particular remark can be made about OAD, because in the OJ corpus only ‹ e › is employed, and this gives no clue about the disputed contrast between / e/ and / ɛ/ (Berruto 1974: 15-17), leaving us clueless also about the presence and phonological status of / ə/ (< closed syllable ē , ĭ , which often is not present anyway in the surroundings of Asti, see Ferrarotti 2022: 71-77). From a diachronic perspective, in this class sound only the diphthong ey (< open syllable ē , ĭ ) is observable 47 . 46 ‹ qu › represents instead / kw/ , both as the primary outcome of Latin qu and a secondary outcome of co -, cu - + vowel: quant ‘when’, ‘how much’ ( passim ) < quando , quantuM , B150 squela ‘bowl’ < ScuteLLaM , B352 quag ‘quiet’ < coactuM etc. 47 It is noteworthy that ‹ ey › has at least one more diachronic source, namely the raising of / ai̯/ (see §4.4). It is possible it had a phonetic realization closer to [ɛi̯] or [æi̯], while ‹ ey › stemming from 124 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 3.2 Mid front and mid back rounded The most intricate grapheme-to-phoneme correspondences in the OJ corpus involve mid back and mid front rounded vowels, because spellings often overlap and are dependent on phonological processes that were not always systematically mirrored in the graphemic system. This does not allow to fully understand how they were distributed in the phonological system, and at which stages of the general phonological change OAD was in the 15-16 th century. From a diachronic point of view, there are three phonemes involved: 1) / ø/ appears only as a stressed phoneme. It is usually thought to have evolved from open syllable ŏ > *ɔ > * wɔ > * wø , e.g., * cŏrem ‘heart’ > * kwɔr > * kwør > / kør/ . This diphthongization was probably originated by metaphony induced by final i and u and in some cases e and o , at least according to Rohlfs 1966: 140-42 and Maiden 1997 (but see Loporcaro 2010: 130-35). However, / ø/ can also appear in an originally closed syllable if it is close to palatal approximants as / ʎ/ or / j/ ( fŏliam > [ˈføja] ‘leaf’) or postalveolar affricates as / t͡ ʃ/ or / d͡ ʒ/ ( nŏctem > [nøt͡ ʃ]). A further source of the development of / ø/ are metaphonic (also metaphonetic or «Umlaut») plurals, e.g,. / grɔs/ sg. vs. / grøs/ pl., where final unstressed -/ i/ that in many cases was dropped, conditioned stressed / ɔ/ to front (nowadays they are attested only in peripheral dialects; see §4.4, Loporcaro 1997: 18-23). Note that the actual internal distribution / ø/ varies greatly among dialects, due to different diachronic developments or dialect contact. 2) / ɔ/ stems from closed syllable stressed ŏ , e.g., / kɔl/ ‘neck’ < cŏllum and open syllable non metaphonic ŏ (i.e. in a ending words, but sometimes e too) e.g. / ˈfɔra/ ‘outside’ < ForaS and au / ɔr/ ‘gold’ < auruM . 3) / o/ or / u/ stems from ō , ŭ > * o : flōrem ‘flower’> / fjor/ > / fjur/ , crŭcem ‘cross’> / kros/ > / krus/ . / o/ in more progressive varieties was raised to / u/ due to the phonological gap left by ū > / y/ . The latter outcome is the most common in contemporary Piedmontese dialects, but several conservative varieties (e.g., around Biella, Mondovì, etc.) maintain / o/ ; this is all but common in several Gallo-Italian dialects, and often this evolution can be conditioned by the phonological context (e.g., for Lombard, v. Sanga 1985). This phoneme in unstressed position stems from unstressed * o < ō , ŭ , ŏ . Giacomino (1901: 407, 409-10) already observed that / ø/ is represented by the French spellings ‹ eu ›, ‹ oeu › and very rarely by ‹ oe ›. ‹ ou › is used to represent / u/ , and it is of French origin too. The problem is that, in every one of these cases, ‹ o › is also found, which, in turn, is used for / ɔ/ too 48 . Late Latin / eː/ (< open syllable ē , ĭ ) had an [ei̯] realization, as one would expect. 48 Giacomino thought that at least the ‹ eu › ‹ oeu ›/ ‹ o › oscillation underlay an attempt to represent sociolinguistic variation by Alione, but his arguments did not convince Salvioni (1905: I 155). 125 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 3.2.1 Mid front rounded To assess the actual distribution of / ø/ , I will first observe the contexts in which / ø/ occurs incontrovertibly, namely when it is represented by the dedicated digraphs or trigraphs ‹ eu ›, ‹ oeu › and ‹ oe ›. 1) Open syllable ŏ > *ɔ regardless of the possible metaphonic conditioning of the final syllable: B145 breu ‘broth’ < Germ. *brod ( FEW 15/ 1: 291-300), A506 coeur ‘heart’ < cŏrem , F177 doeu / B631 doeul ‘mourning’ < dŏlum , A773 oeuf ‘egg’ < ŏvum 49 , A43 leu / B235 loeu ‘place’ < lŏcum ; D416 seur ‘sister’ < sŏror , H416 figleu ‘son’ < filiŏlum , A803 griseu ‘lamp’ < * croceŏlum , B125 faseu ‘bean’ < phaseŏlum etc. Among the nonu metaphonic words: C108 aproeuf / G254 aproef ‘close to’ < ad prŏpe , E66 nizoere ‘hazelnuts’ < nuceŏlas , A732 uncoeu / I491 uncoe ‘today’ < hinc hŏdie (also I476 uncò ) etc. 2) Closed syllable ŏ > *ɔ or *o in contact with palatal approximant: E222 beugl 50 ‘(it) boils’ < BuLLit , A20 eugl / A180 oeugl / I445 oegl 51 ‘eye’ < oc ( u ) LuM , E211 feuglia ‘leaf ’ < * fŏlia ; A132 terfoeugl ‘clover’ < trifŏlium , b9 voeugl / A375 voegl ‘(I) want’ < * voljo < * VoLeo 52 etc. 3) Other cases in which a linear phonetic evolution does not explain the presence of / ø/ , namely: a) heu ‘(I) have’ ( passim ) < * ho . Common in eastern and southern Piedmont (Ferrarotti 2022: 164-68), it was possibly fronted by assimilation to the subject proclitic. It is written simply o (13 tokens) too. The eu ending of the analytic future ( andreu ‘I will go’) is formed with heu (and in general with the inflected forms of the present indicative), as is common in the formation of the future of the Romance languages. Seu ( passim ) ‘I know’ is modeled by analogy on heu . b) B36 euteury ‘help’ (also with apheresis as a call for help: I438 ’eury ) < adiutōrium , possibly a semi-learned word (pron. / eu̯ ˈtøri/ ). c) / ø/ is present also in metaphonic plurals (cf. §4.4) such as B175 groes (sg. gros ) ‘big’ < groSSi , A57 seu ‘his/ her’ < * Soi (sg. so ), I48 beu ‘cow’ < *boi < BoVeS (sg. bo ). 49 Contrasts with of ‘(he/ she) had’ < * ov < * auve < habuit (Giacomino 1901: 423, Rohlfs 1968: 327) 50 Also, bogl in the Western-Piedmontese speaking character in the Farsa del Bracho , as Giacomino (1901) already noticed. 51 Also, ogl twice. It contrasts with the subject clitic 3 Sg . M + dative clitic cluster ogl . 52 Vogl is way more common (92 tokens). The 3 Sg voel (which has no «palatal» conditioning) is attested once in the imitated Western Piedmontese speech and it contrasts with the common OAD vol (58 tokens). This contrast is still visible in contemporary dialects ( AIS 1594). 126 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 As several pairs of words show (see N 50-52), it can be seen that the true distribution of / ø/ is far larger than its graphemic representation through ‹ eu › ‹ oeu › and ‹ oe ›, and that ‹ o › often underlies / ø/ . That different spellings corresponded to the same sound is made quite clear by several rhymes: eugl ‘eyes’ : piogl ‘flea’ < pedic ( u ) LuM (F189-90), vogl ‘(I) want’ : oegl ‘eyes’ (I 529-30). Moreover, rhymes like seu my ‘I know’ : homy ‘men’ (A445-46) shows that probably most metaphonic plurals on / ɔ/ were not reported in writing. Other rhymes like euteuri 53 ‘help’ : purgatory ‘purgatory’ (A497- 98) and ory : euteuri (E257-58) possibly underlie the fact that all ory words were realized as -/ ˈøri/ . In fact, B35 oly / E256 ory ‘oil’ < oLeuM still in contemporary dialects is / ˈøli/ or / ˈøɹi/ (see AIS 1012). See also B180 sory ‘smooth’ < SoLiuM ? (/ ˈsøli/ in REP 1323), G367 relory ‘clock’ < * horolorium ([ˈrløri] in aiS 258), avory ‘ivory’ < * eBoreuM , which rhyme with ory or euteury . This development of / ø/ is probably due to the conditioning of final / i/ (cf. §4.4 and 3c. above). A context in which / ø/ would be expected in OAD is before / t͡ʃ/ , but it is never noticeable in writing. In the OJ corpus, only a few words of this kind are attested: I402 og ‘eight’ < octo , E108 cog ‘cooked’ < coctuM , A800 nog ‘night’ < nocteM 54 , B312 bog ‘hole’ (etymology unclear, but / bøt͡ʃ/ in many contemporary dialects, AIS 857) and few others: these in contemporary southern Piedmontese and Ligurian dialects ( AIS 287, 342, 1018. 857) show a clear presence of / ø/ . Therefore, it is more likely that this phoneme was not notated in writing through dedicated digraphs or trigraphs because there were no graphemic contrasts to highlight. It is less likely that the sound / ø/ was not yet developed in that context at that time, as in OAD it was present in all the phonological contexts in which it appears in contemporary dialects. Lastly, the actual structural reasons for this uncertain graphemic alternation remain to be explained. Although / ø/ was an independent phoneme, Alione did not differentiate it systematically in writing: this could be related to the fact that / ø/ was often in a synchronic morphophonemic relation with / ɔ/ and / o/ -/ u/ at least in two different cases: 1) due to final -/ i/ triggered metaphony in plural nouns (see §4.4); 2) due to the fact that, due to compounding or inflection, / o/ -/ u/ often alternates with / ø/ in paradigms, e.g., stressed proeuf / prøf/ ‘(I) try’ vs. unstressed provà / proˈva/ or / pruˈva/ ‘tried’. To a native speaker (or reader), its presence could be evident either by its natural occurrence on a lexeme or if a phonological process or a morphological rule conditioned its presence. This, paired with the inconsistency that was all but typical of non-standardized writing systems at this chronological stage, probably led to a non-systematic notation of / ø/ . 53 The 1521 print has eutoury , but this is in all likelihood a print error, because ‹ o › for ‹ e › seems to be a common misprint in the OJ given that that word is elsewhere spelled euteury (and ’eury as a call for help). It is also highly likely that to the typesetter the ‹ o › : ‹ ou › rhyme was graphically acceptable, while the ‹ o › : ‹ eu › was not (see §3.2.2 for the interchangeability of these sounds). In general, rhyme coherence seems to be the main concern of the typesetter of the first edition of the OJ which, in some parts, is riddled with misprints. 54 oug one time; possibly a print error for neug , that would confirm the presence of / ø/ , see N 53. 127 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 3.2.2 Close-mid and open-mid back rounded An overlap very similar to the former is found between ‹ o › and ‹ ou ›. If ‹ ou › represents / u/ (< / o/ ), then assessing the exact occurrences of the digraph will tell in which context it was at least present and if and to which degree the / o/ > / u/ raising (§3.2) was complete. ‹ ou › most often appears in stressed syllables as the outcome of *o < ō , ŭ . A few cases are found in open last syllables and monosyllabic words: B192 autroù ‘elsewhere’ < * aliter ŭbi , A80 prou ‘enough’ < prōdem FEW 9: 417 (also pro ) etc. Still, the plurality of cases is found in stressed closed final syllables and monosyllabic words. The most regular and invariant usage is word-final ‹ oux › for -/ us/ or -/ uz/ (French in origin, Nyrop 1899: 353): B393 privoroux < * pericol < periculum + ōsum , B49 croux ‘cross’ < crŭcem , B394 spoux ‘groom’ < spō ( n ) sum etc. In other cases, the alternation of ‹ ou › with ‹ o › is much more relevant: segnour (13 tokens) segnor (5) ‘lord’ < seniōrem , mour (5) mor (2) ‘face, snout’ < *mŭrr- ( FEW 6/ 3: 236), lour (37) lor (22) ‘they’ < illōrum , sout (5) sot (18) ‘under’ < sŭbtus etc. In a few cases ‹ ou › appears in closed VCC monosyllables: E157 lourd (1 token) - B510 lord (1) ‘dizzy’; the usage of ‹ ou › in stressed penultimate syllables is rare: F445 ascouta ‘listen’ (imperative) < auscŭlta (cf. unstressed B491 ascotè ‘listen’, imperative 2pl) etc. Even if the most typical usage of ‹ ou › in the OJ corpus is in stressed syllables, as previously mentioned, it is also found in unstressed syllables too (from unstressed *o < ŏ , ō , ŭ ). In pretonic syllables its occurrence is limited: A477 favourirme ‘to favor me’ etc. In one case a pun is based on the alternation between pretonic ‹ o › and ‹ ou ›, in the rhyme prou ces ‘enough ass’: proces ‘trial’ ( prōdem + cessum : proceSSuM ) I272-73. Still, it is not clear whether there was an actual phonetic identity between those words. Its employment in unstressed posttonic syllables is more frequent, where it conspicuously alternates with ‹ o ›: dyavou (2 tokens) dya(v)o / diavo (20) ‘devil’ < diaBoLuM . This issue is further complicated by rhymes that seem to point out that ‹ ou › and ‹ o › were employed to represent the same sound: amour ‘love’ : daperlor ‘by themselves’ (H205-06), aloua ‘(I) position’ (subj.) : coa ‘tail’ (M 63-64), chiatroussa ‘Chartreuse’ : rossa ‘red’ (f.) (C100-01) , lour ‘they’ : color ‘color’ (E476-77). Also, the rhyme in A510-11 strassuà ‘sweaty’ < extra - + SudatuM : reprovà ‘tired’ < re + proBatuM shows that before of / a/ the / y/ phoneme could go back to / u/ , as happens in some contemporary dialects ( AIS 177) and as described by Rohlfs (1966: 58) and this points out a pronunciation closer to [u] than to [o] of ‹ o ›. The reasons for this graphemic fluctuation are not entirely clear: one might interpret it as a signal of an ongoing rising of [o] to [u] or perhaps the attempt to render [u] through a dedicated digraph and a more conservative tendency dictated by the usage of conservative spellings (i.e., the use of the original Latin ‹ o ›). Nevertheless, it must be noted that the [o] to [u] change is not well described diachronically but, 128 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 at least in origin, it had to be conditioned by the phonological context: this is evident from the description of some conservative dialects of the beginning of the 20 th century, e.g., the Valsesia dialects in Spoerri (1918: 401-07) and Toppino (1902-1905: 524-26, 547). This could be the case of OAD, as in some contexts, in fact, the digraph ‹ ou › never appears: a) before nasal: laron ‘thief ’ < latrōnem , bon ‘good’ (m.) < bŏnum etc. (before nasal the late Latin phonological opposition *ɔ ~ *o is neutralized); b) before some consonant groups, in part. / l/ or / r/ + consonant: forn ‘oven’ < fŭrnum , torné ‘to return’ < tōrnare etc. c) in several unstressed high occurrent monosyllables: o (3 Sg . M subject clitic, Sg . M definite article < iLLuM ; 2 pL subject clitic < vōs ); lo (3 Sg . M object clitic, 3 Sg . M postverbal subject clitic < iLLuM ) etc. d) before / i/ . This case is particularly relevant because several problematic rhymes are attested that would not be possible if a contemporary [u] realization was present, namely: goy : voy (H533-34), croy : voy (I599-600), croy : coy (C292-93), croy : doy (I631-32), coy : croy (M15-16). At least in principle, in goy ‘joy’ < gaudiuM and coy ‘cabbage’ < cauLi ? , au develops in *ɔ in Early Romance, resulting in / ɔ/ in contemporary dialects ([gɔi̯], [kɔi̯]) so that they cannot rhyme with voy ‘you’ < vōbis , noy ‘we’ < nōbis and doy ‘two’ (m.) < dōi , being commonly realized [vui̯], [nui̯], [dui̯] (< [voi̯], [noi̯], [doi̯]). The same goes with coy ‘those’, the inflected plural form of col < eccuM iLLuM , [kui̯] (< [koi̯]). The case of croy is more complicated, as both Gaulish bases *crŏdios and *crōdios are attested ( FEW 2: 1358) and then possibly [krɔi̯] and [krui̯] (< [kroi̯]) could have existed. In any case, these rhymes could work only if two similar and near-merged vowel sounds were present: then it is possible that / ɔ/ (with a rather high phonetic realization) could rhyme with / o/ (possibly [o̞] ~ [ʊ]), underlying the fact that at that time / o/ was not still generally raised to [u]. In a similar fashion, rhyming schemes as A129-30 hom ‘man’ : com ‘how’ ( hŏmo : quōmŏdo ), G70-71 volp ‘fox’ : corp ‘body’ ( vŭlpem : cŏrpus ), and I667-68 corp ‘strike’ : corp ‘body’ (* cōlpum < coLaphuS : cŏrpus ) can work only if / ɔ/ and / o/ are nearmerged, as in hom and corp ‘body’ / ɔ/ is expected and in com, volp, and corp ‘strike’ / o/ is expected. In most contemporary dialects these lexemes, as already seen, exhibit a starker contrast due to the raising of / o/ to / u/ : / vulp/ , / kulp/ etc. However, again, a different picture is presented by the following bilingual rhyme, French esprouvé ‘tried’ : OAD trové ‘to find’ (D173-74), that seems to point to an identity of sound between French ‹ ou › and OAD ‹ o ›. In summary, it seems that in OAD the raising of / o/ (< * o < ō , ŭ ) to / u/ was not completed, if the evidence provided by the spellings and the rhymes of the 1521 print 129 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect is considered. The great deal of oscillation and the limited usage of the dedicated digraph ‹ ou › could indicate that, at least in some contexts, the change was already underway 55 . Therefore, it could be possible that the phoneme / o/ had different allophonic (both free and conditioned) realizations, mainly [o] and [u] or an intermediate sound like [ʊ], which is attested in older descriptions of Piedmontese (e.g., for the 18 th century Turin dialect in Pipino 1783: 2, 8). 4. Other features 4.1 Other vowel sounds In that period, / y/ < ū was already common in Gallo-Italian dialects 56 and French (Rohlfs 1966: 57-59, Nyrop 1899: 165-66). In the OJ it is specifically confirmed by some bilingual rhymes: OAD creature ‘creatures’ : French injure ‘(he) injures’ (subj.) (D81-82) , French pleü ‘liked’: OAD velù ‘velvet’ (D348-49) and by the pun based on écu ‘shield, type of coin’ el cu ‘ass’ in B81. There is no clue in the text about the possible presence of a southern Piedmontese / ɒ/ phoneme, separate from / a/ and / ɔ/ (Berruto 1974: 31-32, Miola 2013: 55-56). OAD retains the contrast between final unstressed / e/ and / i/ , an important isogloss usually employed to divide Western (contrast generally kept) from Eastern Piedmontese (contrast neutralized in favor of / i/ , see Ferrarotti 2022: 29, 139-42): e.g., B41 antendi ‘(you) understand’ (2sg) ~ A190 antende ‘to understand’, cf. also F280 chinery ‘(I) took’ (perfect tense) vs. A390 auregle ‘ears’, G447 derme ‘to give me’. In the contemporary dialect and already in the 18 th -century dialect of Asti (texts in Gasca Queirazza 2010) this contrast was neutralized in favor of unstressed / i/ , as in Eastern Piedmontese dialects. However, in the OJ , final unstressed / i/ in place of / e/ is employed to represent the speech of some people from Lower Montferrat (Ferrarotti 2021: 158-66), a sign that this kind of positional merger was already common at that time in some dialects. 4.2 Final obstruent devoicing OAD seems to be characterized by an almost general process of devoicing of word-final obstruents that involves / v/ , / d/ , / z/ (§2.2.1), / d͡z/ (§2.2.2) and / d͡ʒ/ (§2.1). In the case of / v/ the devoicing is always represented in writing (1.), while this is done only 55 The change was evidently completed in the urban dialect of Asti in the late 18 th century. The descriptive poem of 1786 in Gasca Queirazza (2010: 198, cf. N 44) clearly states that unstressed ‹ o› was pronounced as «French ou » or «Italian u ». 56 There is no meaningful trace in the OJ of the / y/ > / i/ change that is very common today in a dialect area spanning from Alba to Alessandria (Ferrarotti 2022: 84, Ricca 2021). 130 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 partially with / d/ (2.), as it is observable only when a devoiced phoneme (notated in writing as voiced) rhymes with a voiceless phoneme. 1) / v/ > / f/ : A490 chiaaf ‘key’ < cLaVeM , A358 catif ‘mean’ < captiVuM , I258 of ‘(he) had’ (perfect) < * auve < haBuit , B384 beif ‘I drink’ < BiBo etc. 2) / d/ > / t/ : quant ‘when’( passim ) < quando , B37 monstrant ‘showing’ < MonStrando , the rhymes lord ‘dizzy’ : cort ‘short’ ( LurduM : curtuM , B510-11) , tard ‘retarded’: part ‘part’ ( tarduM : parteM , i575-76) The devoicing of final / d/ , / z/ , / d͡ʒ/ is still quite common in Piedmontese dialects (see AIS 1652, 384, 320), while the devoicing / v/ is rarer ( AIS 889). On the contrary, the contemporary dialects of Turin and Asti show a weakening of / v/ to / w/ ([øu̯] ‘egg’ < oVuM in Ternes 1997: 99), which could be interpreted as an opposite phonological change. No devoicing of final / g/ is attested, which is, in turn, quite common in contemporary dialects (e.g., digh ‘I say’, cf. AIS 1601). 4.3 Loss of final / l/ and / r/ In several Southern Piedmontese dialects the loss of word-final / l/ is common (Berruto 1974: 33, Ternes 1997: 89). This feature seems to be present in OAD too, but it was not notated systematically. Some rhymes attest it: acompagnà ‘gone along’: segnà ‘signal’ A87-88 (also E606 segnal ) etc. and some spelling alternations as well: B263 barrý / Pr. 40 barril ‘barrel’, H507 schossà / H61 scossal ‘apron’ etc. In contrast, almost all contemporary Piedmontese dialects show the loss of final / r/ 57 , particularly in well-established derivational and inflectional suffixes such as the outcomes of Latin infinitives are , ēre , ěre , ire , and the agentive suffix ariuM . In the OJ , final ‹ r › is often notated, but it is likely that is an etymological or French-influenced 58 usage, as shown by pairs as A356 ster / Pr.9 sté ‘to stay’ < Stare , A60 haveyr / A135 havey ‘to have’ < habēre etc. A172 caucer / A237 caucé ‘shoes’ < caL ceoLariuM ; cf. also rhymes with French words such as sceurté ‘safeness’ : OAD tanté ‘to provoke’ (d101-02) and esprouvé ‘tried’ : OAD trové ‘to find’ (d173-74). Moreover, the employment of ‹ r›less forms seems contextually motivated, as it is found more frequently before words starting with a vowel, with the specific function to signal a hiatus (synalepha in metrical terms, as final / r/ never matters for metrical purposes in the OJ ): C12 sté accorià ‘to lie down’ vs. Q9 ster lì ‘to stay there’, H206 ster daperlor ‘to stay by themselves’. 57 / r/ was probably lost in some consonant clusters as / str/ but not systematically notated: see the rhyme acost ‘to get close’ : vostr ‘yours’ (* vostro < VeStruM , B374-75). 58 According to contemporary grammarians (Fouché 1966: 666), French preserved final / r/ at that stage, but this is far from clear. 131 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect 4.4 Metaphony In OAD I-metaphonic plurals (Maiden 2020) are well-attested, but they seem to occur in limited contexts. In many cases they are not strictly metaphonic, i.e. there is no change in the quality of the stressed vowel, but just the insertion of a glide after it as an assimilation process. It occurs on low and middle vowels (/ a/ , / o/ , / e/ ) and primarily before nasal: A325 mayn ‘hands’, etc. (but H504 homaycz ‘bad men’, F67 drayp ‘drapes’ etc.), A157 cancioyn ‘songs’ etc., A448 scriveynt ‘writers’ 59 . Frequently, the presence of a glide causes / a/ to raise and front to / e/ , e.g., eyg ‘others’ ( passim ), A280 meyn ‘hands’ (cf. mayn above) , F277 greynd ‘big’ (sg. grand ), etc . (note that the ‹ a › -‹ e › fluctuation in writing could represent an intermediate realization [æ], see N 60). The tendency of / a/ to raise to / e/ before / i/ is a common phonological process of this dialect independent from the metaphonic insertion of / i/ : cf. B552 sgeyr ‘it is necessary’ < ex + cLarire (also A468 schiayr ), C259 meysmament ‘mostly’ < MaxiMa + Mente ( FEW 6/ 1: 563), C237 cheyt ‘fallen’ < * cadituM , B101 Seynt ‘Saint’ (with the unstressed variant Sen ) etc.; the forms C318 citen ‘citizen’ and A269 len ‘clear, easy’ probably come from *citein < * citaìn < ciVitateM + inuM (cf. It. cittadino ) and *lein < * laìn < LatinuM and they show a combination of / a/ raising and the accent retraction that takes place in some diphthongs 60 . The morphophonetic character of these instances of metaphony (in the sense of Maiden 2020) is then very reduced, as it can be found as a simple relic of a fallen final unstressed / i/ , as the case of aveyng shows (< avanti < aB ante passim ), in which no morphological content is present. This kind of metaphony is rather uncommon in contemporary Piedmontese dialects: even the conservative dialects that maintain some kind of metaphonic marking of the plural usually exhibit a quality alteration of the tonic vowel along with glide insertion and do not restrain the presence of it only before / n/ (see some examples in Rossebastiano 1995). In this regard, OAD plurals are similar to those described by Parodi (1901: 13-14, 17-18) for old Genoese texts, and even to those of contemporary Genoese (Forner 1975), that usually has no metaphonic plurals other than those on stressed / a/ and / u/ (< / o/ ): [kaŋ] / [kɛŋ] (< [kai̯n] ‘dogs’ < cani ), [buŋ] / [bwiŋ] (‘good’ < Boni , AIS 1097, 1710). A full-fledged metaphonic plural typical of OAD that is still common among conservative Piedmontese dialect is that on stressed / ɔ/ > / ø/ , which has already been examined in §3.2.1. Similarly, some occasional glide insertions as in D423 poych ‘few’ (sg. poch ) are still common nowadays in contemporary dialects. 59 Subieyt ‘whistles’, soffieyt ‘bellows’ in the imitated Montferrat dialect in D. 60 In OAD a process of accent retraction created several falling diphthongs: 1) / au̯/ from * aór > *àor , e.g., pau ‘fear’ < paVoreM , retagliau ‘retailer’ < retaLiare , tagliau ‘cutting board’ (< taLiare ), ambotau ‘funnel’ (parasyntethic form built with in - + ButteM + atoreM ); 2) / ei̯/ in reyd ‘stiff’ < * reìd < rigi duM , pareys ‘paradise’ < * pareìs < *paraìs < paradiSuM ); 3) / eu̯/ in meura ‘ripe’ (f.) < * màura < * maùra < MaturaM (cf. the semilearned word euteury ‘help’ < * autorio < adiutoriuM ). 132 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 4.5 Syncope The syncope (i.e., deletion) of pretonic vowels and the subsequent formation of prosthetic vowels, so typical of contemporary Piedmontese and several other Gallo-Italian dialects (Ferrarotti 2022: 68-71) is not well attested in OAD, at least from a superficial analysis of the writing system. Giacomino (1901: 410-11) observes that in OAD pretonic reis never reduced to r- (e.g., C318 retorner ‘to return’ vs. contemporary artorné ) and that several words show no deletion of the pretonic syllable: G118 ferrougl ‘deadbold’ (vs. cont. froj ) < * Ferruc ( u ) LuM , A31 derrer ‘behind’ (vs. dré ) < de + retro , G119 zenougl ‘knee’ (vs. znoi ) < genuc ( u ) LuM . However, the deletion is more frequent before / r/ , such as in M125 frua ‘injury’ < * FerutaM , cry ‘sought’ < * quaerituM , A122 pra ‘peeled’ < peLatuM , and it is common in future tense forms such as F626 tornrema ‘(we) will come back’ < * tornerema . Also, if in contemporary Piedmontese elements such as preverbal object and subject clitics are usually reduced to a single consonant segment to which a prosthetic vowel is attached, they never appear in such fashion in OAD: me ‘me’ vs. cont. Piem. am etc. The syncope of posttonic vowels in former Latin proparoxytones may be closely associated with Piedmontese (Clivio 1972), whose sound system usually does not allow CV ́ CVCV syllables, and this is observable in OAD (Giacomino 1901: 412): G150 povre ‘poor’ (f. pl.) < * pauperaS , R30 limosna ‘charity’ < eLeMoSynaM etc. (with oscillation in cases as B612 fomne / I580 fomena ‘woman’ < FeMinaM ). However, Giacomino insightfully shows through the metrical analysis of the octosyllable verse of the OJ that many vowels were written but not actually realized (in brackets): man a r(e)tornerse ay nosg citen (C318) ‘let’s go back to our citizens’ , ne van mia tug p(e)r ofrir candeyre (A59) ‘they all do not go to offer candles’ , ch’e vogl anderm(e)ne ades ades ‘that I want to go away right now’ (B158) , del zov(e)ne chi han necessità ‘of the young ladies that have necessity’ (H44) , te’ bàs(e) lo un poc mal amoreyvo ‘here kiss him, unloving man’ (H457), da governer, tenýv(e)le apè ‘to govern, keep them close’ (H52). It is then possible that OAD had a variable realization of syncope, and its generalization was still undergoing. 4.6 Vowel and consonant lengthening In OAD some traces of phonemic vowel lengthening originated from open syllable Latin vowels (see Loporcaro 2015: 85-115), nowadays all but lost in most Piedmontese dialects, are visible. Still, they are notated in a very irregular fashion: the most significant case is the contrast naas ‘nose’ (14 tokens) ~ nas ‘he is born’ (2 tokens) < naSuM ~ * naScit (but nas ‘nose’ is also present, 14 tokens); cf. also t’èe (6 tokens) vs. t’è (5 tokens) ‘you are’ < * ei < eS vs. l’è ‘he/ she is’ < eSt (in the latter the vowel is never geminated in writing). Some other cases are occasionally notated, as A383 piaas ‘(he) likes’ < pLacet , A490 chiaaf ‘key’< cLaVeM . Some kind of vowel length seems to be used by Alione to portray the speech of the character that speaks the 133 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect dialect of Milan: I372 inamoraad ‘in love’, I319 Mirreen ‘Milan’ 61 etc. Geminate vowels are often used in interjections such as doo, dee, bee, hij, naa ( passim ) . Many Piedmontese and Gallo-Italian dialects exhibit a secondary, unetymological consonant lengthening that involves mainly words that bear the stress on the penultimate syllable, whereby the consonant that immediately follows the stressed vowel is lengthened 62 possibly in a compensatory way, e.g., [siˈgulːa] ‘onion’ < cepuLLaM but [ˈgyd͡ːʒa] ‘needle’ < acuc ( u ) LaM . This has been observed in Valsesia and Montferrat (Rohlfs 1966: 322 N 2, Ferrarotti 2022: 129). In the OJ the graphemic use of geminated consonants varies wildly, sometimes representing etymological geminated consonants and sometimes not. In some cases, one can find them also in a fashion that can hint at the lengthening described above, e.g., B167 Romma ‘Rome’, E135 parpagliolle ‘butterflies, type of coin’ and the verb B500 vèggher ‘to see’. Rhymes as tomma ‘cheese’ : cenoma ‘let us dine’ (D293-94) show that this feature was not notated consistently, bearing no phonemic value. However, data are too scant to give a clear picture, and no definitive assumption can be made. 5. General remarks In general, it is safe to assume that the phonological system of OAD was already very similar to that of the contemporary dialects, although with some important differences. As far as the phoneme inventory is concerned, a richer one is found in OAD if compared to contemporary mainstream Piedmontese dialects, at least as regards consonants. In fact, OAD seemed to maintain not only the phonemes / t͡s/ , / d͡z/ (§2.2.2), but also / ʎ/ (§2.3), which in the contemporary variety of Asti are merged respectively with / s/ , / z/ , and / j/ , as it was the case already in the late 18 th -century dialect (the texts in Gasca Queirazza 2010 show no trace of them). In contemporary dialects, alveolar affricates have all but disappeared in the surroundings of Asti, and they are preserved mainly in an area spanning from the Higher Langhe to Northern Liguria. The disappearance of / ʎ/ was even more thorough, as it is nowhere to be found in contemporary dialects: the only instance of conservation is in some Occitan dialects of Piedmont, while in the dialects of Valsesia it was still retained in some conservative varieties at the beginning of the 20 th century (Spoerri 1918). The delateralization of the palatal lateral approximant (an otherwise uncommon sound in the 61 This can represent a more fronted realization [æ] of / a/ before nasal in comparison to OAD. Such feature is not recorded by the historical phonetic descriptions of the dialect of Milan (Lepschy 1965). A fronted realization of / a/ is attested in the 17 th -century dialect of Milan, but not before / n/ : in this regard, Salvioni (1919: 528 N1) explicitly states that this testimony is irrelevant to the reconstruction of the phonological history of the dialect of Milan, but it could give a hint about a perceived articulatory difference between OAD and the old dialect of Milan by an OAD speaker. 62 This happens in the dialect of Turin only after stressed / ə/ (Berruto 1974: 16-17). 134 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 world’s languages, Ricca 2021: 3 N5) is a well-known tendency in Romance languages as it is found in other Gallo-Italian dialects, as well as in other Romance languages (French in the 18-19 th century, yeísmo in several varieties of Southern Spanish etc.). The contact with Standard Italian, which possesses / ʎ/ , was uninfluential to its merger with / j/ in most Northern Italy. In general, the loss and the distribution of the alveolar affricates in OAD are rather explanatory of many internal and external diachronic changes in Northern Italian dialects and Romance languages in general. Firstly, it can be observed that OAD fully takes part in the several different mergers that occurred in Northern Italian dialects in the so-called «system of the sibilants» (Sanga 1984). In particular, OAD is an essential piece of evidence showing that Piedmontese dialects behaved as the old eastern Lombard dialects and old Venetian, partaking in the same kinds of mergers. In a more general Romance perspective, following Bonfadini (1995), who considered the Eastern Lombard and Venetian development of the sibilants as typical of Western Romance 63 (in contrast with Western Lombard and Central and Southern Italian dialects, which did not merge the outcomes of strong c + e , i with strong tj -), we can see a further common tendency to deaffrication: Old French / t͡s/ and / d͡z/ respectively merged into / s/ and / z/ ( GGHF 451) from the 13 th century, while Old Spanish / t͡s/ and / d͡z/ did not merge, but developed in fricatives with a different articulation than / s/ and / z/ (possibly / ʂ/ and / ʐ/ , Dworkin 2018: 24), and then advanced to a dental articulation / θ/ and / ð/ in the late Middle Ages 64 . From a variationist point of view, it is noteworthy that these changes have been very consequential on the entire Ibero and Gallo-Romance varieties, as, in the linguistic atlases of Iberia and France, no trace whatsoever is left of the old / t͡s/ and / d͡z/ (see, e.g., ALPI I 27, 63, ALF 217, 171). In this context, OAD, along with contemporary southern Piedmontese and northern Ligurian dialects are quite an exception, as they preserve consonantal contrasts that have disappeared almost everywhere else 65 . In this context, the dialect of Milan and Western Lombard varieties in general seem to have lost much later in time this kind of affricates, possibly in the late 18 th or early 19 th century (Sanga 1984: 164) and many contemporary dialects of that area (Brianza, Ticino, Ossola) still keep them. Not just the conservation of this archaic phonological feature, but also its distribution in the system is telling. / t͡s/ , apart from its natural occurrence as the native outcome of Latin sounds (see §2.2.2) is employed in learned words such as capacz , 63 OAD behaves as a prototypical «Northern Italian» dialect if we take Barbato’s (2019) classification into account, because it merges the strong Late Latin postalveolar affricates with the alveolar (§2.2.2) and merges the postalveolar fricatives originating from Latin weak postalveolar affricates with the alveolar fricatives (§2.2.1). 64 In some Andalusian Spanish and southern American dialects, a further merger occurs, both / θ/ / ð/ > / s/ / z/ , the so-called seseo , which is similar to what happened in Gallo-Italian dialects. 65 Most Piedmontese dialects (with the exceptions mentioned earlier) and some Emilian dialects share the complete merger to / s/ and / z/ , while Friulian and some Venetan varieties have the Spanish-like reduction to / θ/ / ð/ , thus avoiding a complete merger. 135 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect sagacz , offici etc., possibly borrowings from the so-called «Lombard» vernacular, with the same adaptation dynamics that Sanga (1984) observed for the dialect of Milan of the same period (that in part persisted in the 17 th century). This is relevant from a historical sociolinguistic point of view, as it shows that contact dynamics between high superregional learned vernaculars and local dialects were very similar in Milan and the more isolated Asti. On the contrary, no use of the Italianate / t͡ʃ/ , as described in Clivio (1972) for the dialect of Turin after the 18 th century, is present (see, e.g., the Italianisms [ˈfat͡ʃil] ‘easy’ and [t͡ʃel] ‘sky’ that replaced the indigenous [ˈfasil] and [sel]). This means that the dialect of Milan and OAD shared the same contact dynamics between higher, «learned» vernaculars and local dialects that, at that time, did not involve Italian in the same way as in the following centuries. In general, then, the re-introduction of / t͡ʃ/ as a learned Italianate phoneme was quite an unusual phenomenon still in the 16 th century, a time when it was present only as the outcome of few Latin sounds (§2.1) with a characteristic Gallo-Italian distribution and evolution (compared, at least, to the Italian equivalent / kj/ < cL -, e.g., chiave ‘key’ vs. the Gallo-Romance conservation as in clé ). In the specific case of OAD, the main source of non-indigenous word-initial / t͡ʃ/ (and / d͡ʒ/ ) was extensive borrowing from French, especially through more or less stabilized loanwords bearing them, as seen in §2.1. In any case, if OAD represents an interesting missing piece of the phonological evolution of Western Romance, the present analysis has made clear that some essential continuities and discontinuities with contemporary dialects can be identified. For example, the merger of / l/ and / r/ and the deletion of final / l/ and / r/ (§4.3) typical of contemporary southern Piedmontese dialects are identifiable in OAD. Even if the uncertainty of the graphemic choices of the author does not allow to be clearcut about this, it is possible that OAD had / w/ and / ŋ/ as independent phonemes with a distribution much similar to contemporary southern Piedmontese dialects (§2.5). At the same time, it is not clear whether stressed / ə/ was present (§3.1) and whether / o/ was ultimately raised to / u/ (§3.2.2) but, in this regard, it is quite clear that the process had already started. However, some significant features stand out in comparison to the contemporary dialects. There are traces of the conservation of phonemic vowel length (that could corroborate Loporcaro’s 2015 hypothesis of a more widespread diffusion of this feature in the past in areas where today it is not attested at all, §4.6), and a strong word-final devoicing of obstruents that today is not well attested (§4.2). Also, the phonological structure of the word was probably different, as syncope was less pervasive than today, and a «Ligurian» type of metaphonic plural was present which today is virtually unknown to the local dialects of the Asti area (§4.5, 4.4). Finally, the internal and external distribution of unstressed final / i/ and / e/ shows the (slight) displacement of an important dialect border across the centuries (§4.1). In conclusion, even if the phonological system of OAD is masked by a complex and sometimes incoherent spelling system, some crucial hints about it can be grasped 136 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 if a comprehensive analysis weighing internal and external factors is carried out (see §1). This can be fruitfully done only if this historically isolated dialect is considered in the broader context of neighboring Gallo-Italian dialects and Western Romance languages, with which it shares many common structural changes and socio-historical contact dynamics. Bibliography AIS = J aBerg , K./ J ud , J. 1928-1940: Sprach- und Sachatlas Italiens und der Südschweiz , 8 vol., Zofingen, Verlagsanstalt Ringier & Co. (online version NavigAIS by G. Tisato, http: / / www3.pd.istc. cnr.it/ navigais-web/ [8.10.2023]). ALF = g iLLiéron J./ e dMont , e. 1902-1910: Atlas linguistique de la France , 9 vols, Paris, Champion. ALI = B artoLi , M. (ed.) 1995-2018: Atlante Linguistico Italiano , 9 vol., Roma, Istituto Poligrafico e Zecca dello Stato (vol. 1-8); then Torino, Istituto dell’Atlante Linguistico Italiano (vol. 9). ALPI = n aVarro t oMáS , T. (ed.) 1962-2016: Atlas Lingüístico de la Península Ibérica, Madrid, CSIC. B anFi , e./ B onFadini , g./ c ordin , p./ i LieScu , M. (ed.) 1995: Italia settentrionale: crocevia di idiomi romanzi. Atti del convegno internazionale di studi (Trento, 21-23 ottobre 1993) , Berlin/ New York, De Gruyter: 25-42. B arBato , M. 2019: «La palatalizzazione in romanzo occidentale: Cronologia e tipologia», ZRP 135/ 4: 937-70. B enincà , p./ p arry , M./ p eScarini , d. 2016: «The dialects of northern Italy», in: L edgeWay / M aiden 2016: 185‐205. B erruto , g. 1974: Piemonte e Valle d’Aosta , Pisa, Pacini. B oerio , g. 1856: Dizionario del dialetto veneziano , Venezia, Cecchini. B onFadini , g. 1995: «I sistemi consonantici dei dialetti alto-italiani: il caso dell’Alta Val Camonica», in B anFi et al. (ed.) 1995: 25-42. B oSSong , g. 2016: «Classifications», in: L edgeWay / M aiden 2016: 63-72. B ottaSSo , e. 1953: Giovan Giorgio Alione. L’opera piacevole , Bologna, Libreria Antiquaria Palmaverde. c aMpBeLL , L. 2013: Historical linguistics. An introduction. Third edition , Edinburgh, Edinburgh University Press. c aSaccia , g. 1876: Dizionario genovese-italiano , Schenone, Genova. c LiVio , g. p. 1972: «Language contact in Piedmont: aspects of Italian interference in the sound system of Piedmontese», in: e. S cheraBon F irchoW / K. g riMStad / n. h aSSeLMo / W. a. o’n eiL (ed.), Studies for Einar Haugen , The Hague/ Paris, Mouton: 119-31. c LiVio , g. p. 2003: «Giovan Giorgio Alione», in: g. p. c LiVio / g. g aSca q ueirazza / d. p aSero (ed.), La letteratura in piemontese. Raccolta antologica di testi. Dalle origini al Settecento , Torino, Centro Studi Piemontesi, 133-232. c ryStaL , d. 2016: The Oxford dictionary of original Shakespearean pronunciation , Oxford, Oxford University Press. d eBanne , a./ d eLucchi , r. 2013: «Sulle sorti di -L-, -R-, RR latine. La prospettiva italo-romanza settentrionale», in: E. c aSanoVa h errero / c. c aLVo r iguaL (ed.), Actas del XXVI Congreso Internacional de Lingüística y de Filología Románicas, 6-11 septiembre 2010, Valencia , Berlin, De Gruyter: 579-92. DEI = B attiSti , c./ a LeSSio , G. (ed.) 1950-1975: Dizionario Etimologico Italiano , Firenze, Barbera. d WorKin , S. N. 2018: A guide to Old Spanish , Oxford, Oxford University Press. F errarotti , L. 2021: «La lingua dell’Alione e i dialetti piemontesi contemporanei», ID 82: 141-70. 137 DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect F errarotti , L. 2022: I dialetti del Piemonte orientale. Contatto e mutamento linguistico . Berlin, De Gruyter. FEW = V on W artBurg , W. (ed.) 1922-1967: Französisches etymologisches Wörterbuch: eine Darstellung des galloromanischen Sprachschatzes , Basel, R. G. Zbinden. F orner , W. 1975: «Metatesi, metafonesi o attrazione nei dialetti liguri? », ID 38: 77-89. F ouché , p. 1966: Phonétique historique du français: Les consonnes , Paris, Klincksieck. g aSca q ueirazza , g. 2010: Saggi minimi di storia del volgare piemontese (1970-2009) , Alessandria, Edizioni dell’Orso. Gdf. dict. = g odeFroy , F. (ed.) 1880-1895: Dictionnaire de l'ancienne langue française et de tous ses dialectes du IX e au XV e siècle , Paris, Vieweg. GDLI = B attagLia , S. (ed.) 1961-2008: Grande dizionario della lingua italiana , Torino, UTET, 23 vol. g enre , a. 1978: «Appunti sulla grafia del piemontese», RID 3: 311-42. GGHF = M archeLLo -n izia , c./ c oMBetteS , B./ p réVoSt , S ophie / S cheer , t. (ed.) 2020: Grande grammaire historique du français , Berlin, Mouton De Gruyter. g hia , A. 2015: «Storia di uno shibboleth in area pedemontana», in: C. M arcato (ed.), Dialetto parlato, scritto, trasmesso , Padova, CLEUP: 337-43. g hinaSSi , g. 1976: «Incontri tra toscano e volgari settentrionali in epoca rinascimentale», AGI 61: 86-100. g iacoMino , c. 1901: «La lingua dell’Alione», AGI 15: 403-48. g iaMeLLo , g. 2007: La lingua dell’Alta Langa , Piobesi d’Alba, Sorì edizioni. h uLL , g. 2017: The linguistic unity of northern Italy and Rhaetia: historical grammar of the Padanian language , Sidney, Beta Crucis. L edgeWay , a./ M aiden , M. (ed.) 2016: The Oxford guide to the Romance languages , Oxford/ New York, Oxford University Press. L epSchy , g. 1965: «Una fonologia milanese del 1606», L’Italia dialettale 28: 143-50. L oporcaro , M. 1997: «L’importanza del piemontese per gli studi di tipologia linguistica», in: Piemonte. Mille anni di lingua, di teatro e di poesia (Vercelli 11-12 ottobre 1997) , Vercelli, Vercelliviva: 11-29. L oporcaro , M. 2010: «Phonological processes», in: M. M aiden / J. c. S Mith / a. L edgeWay (ed.), The Cambridge history of the Romance languages , Cambridge, Cambridge University Press: 109-54. L oporcaro , M. 2015: Vowel length from Latin to Romance , Oxford, Oxford University Press. M aiden , M. 1997: «Vowel Systems», in: M. M aiden / M. p arry (ed.), The dialects of Italy , London/ New York, Routledge. M aiden , M. 2020: Interactive morphonology: Metaphony in Italy, London/ New York, Routledge. M inKoVa , d. 2015: «Establishing phonemic contrast in written sources», in: p. h oneyBone / J. S aL - MonS (ed.), The Oxford handbook of historical phonology , Oxford, Oxford University Press: 72- 85. M ioLa , e. 2013: Innovazione e conservazione in un dialetto di crocevia. Il kje di Prea , Milano, Franco Angeli. M ioLa , e. 2015: «La tirannia della tastiera. Il caso dell’ortografia piemontese», Language Problems and Language Planning 39/ 2: 136-53. n yrop , K. 1899: Grammaire historique de la langue française I , Copenhague, Det Nordiske Vorlag Ernst Bojesen. p arodi , e. g. 1901: «Studj liguri», AGI 15: 1‒82. p arodi , e. g. 1902-1905: «Studj liguri», AGI 16: 105‒51, 333‒65. p arry , M. 2005; Parluma ’d Còiri . Sociolinguistica e grammatica del dialetto di Cairo Montenotte , Savona, Società savonese di storia patria. p arry , M. 2017: «La negazione nelle Farse di Giovan Giorgio Alione», in: d. c acia / e. p apa (ed.), Di nomi e di parole. Studi in onore di Alda Rossebastiano (= Quaderni Italiani di RIOn 8): 575-88. p arry , M. 2023: «Lingua in movimento all’inizio del Cinquecento: la sintassi dei pronomi personali nelle farse dell’astigiano Giovan Giorgio Alione», in p. d eL p uente / t. c arButti (ed.), Dialetti: 138 DOI 10.24053/ VOX-2023-004 Lorenzo Ferrarotti Vox Romanica 82 (2023): 103-139 per parlare e parlarne. Atti del sesto Convegno internazionale di Dialettologia , Potenza, Zaccara editore: 231-47. p etracco S icardi , g. 1992: «Per la definizione dell’anfizona ligure-padana», in: L. M aSSoBrio / g. p etracco S icardi (ed.), Studi linguistici sull’anfizona ligure-padana , Alessandria, Edizioni dell’Orso: 11‒25. p icoche , J./ M archeLLo -n izia , c. 1994: Histoire de la langue française , Paris, Nathan. p ipino , M. 1783: Gramatica Piemontese , Torino, Real Stamparia. r egiS , r. 2013: «Può un dialetto essere standard? », VRom. 72: 151-69. r egiS , r. 2020: «Italoromanzo», RLiR 84: 5-39. REP = c ornagLiotti , A. 2015 (ed.): Repertorio etimologico piemontese , Torino, Centro Studi Piemontesi. r icca , D. 2021: «Typological diversity within the Romance languages», in: Oxford research encyclopedia of linguistics , Oxford, Oxford University Press: 121-45. r ohLFS , g. 1966: Grammatica storica della lingua italiana e dei suoi dialetti. Vol. I: fonetica e fonologia , Torino, Einaudi. r oSSeBaStiano , a./ p apa , e. 2011: «Osservazioni sulla grafia dei testi piemontesi delle origini», in: La grafia della lingua piemontese nei secoli , Vercelli, Vercelliviva: 21-70. r oSSeBaStiano , a. 1995: «Prolessi di -i e metafonesi nel Basso Canavese», in B anFi et al. 1995: 41-44. S aLVioni , c. 1900: «A proposito di amiś», R. 29: 546-58 (repr. in S aLVioni 2008: vol. 4, 121-33). S aLVioni , c. 1905: «Dialetti dell’Alta Italia 1899-1903», KrJber. 5: I-130-61. S aLVioni , c. 1911: «Osservazioni sull'antico vocalismo milanese desunte dal metro e dalla rima del cod. Berlinese di Bonvesin da Riva», in: Studi letterari e linguistici dedicati a Pio Rajna , Firenze, Tip. Ariani: 367-88 (repr. in S aLVioni 2008: vol. 3, 157-78). S aLVioni , c. 1919: «Sul dialetto milanese arcaico», RILomb 52, 2: 517-40 (repr. in S aLVioni 2008: vol. 3, 181-206). S aLVioni , c. 2008: Scritti linguistici , ed. by M. L oporcaro / L. p eScia / r. B roggini / p. V ecchio , 5 vol., Bellinzona, Edizioni dello Stato del Cantone Ticino. S anga , g. 1984: Dialettologia lombarda. Lingue e culture popolari , Pavia, Dipartimento di Scienza della Letteratura, Università di Pavia. S anga , g. 1985 [1986]: «La convergenza linguistica», RID 9: 7-41. S anga , g. 1990: «La lingua lombarda. Dalla koinè alto-italiana delle Origini alla lingua cortigiana», in G. S anga (ed.), Koinè in Italia dalle origini al Cinquecento. Atti del Convegno di Milano e Pavia (25-26 settembre 1987) , Bergamo, Lubrina: 79-163. S anga , g. 2006: «Cosa ci insegnano le grafie italiane antiche», Quaderni di semantica 53/ 54: 371- 90. S poerri , t. 1918: «Il dialetto della Valsesia», RILomb. 51/ 2: 391-409, 683-98, 732-52. t erneS , e. 1997: «Der italienische Dialekt der Stadt Asti. Beobachtungen zu Phonetik/ Phonologie, Morphologie und Lexikon», in: g. h oLtuS / J. K raMer / W. S chWeicKard (ed.), Italiana et Romanica. Festschrift für Max Pfister zum 65. Geburtstag , Berlin, De Gruyter: 87-109. t oppino , g. 1902-1905: «Il dialetto di Castellinaldo», AGI 16: 517-47. V iëtor , W. 1906: A Shakespeare phonology , Marburg, Elwert. V iLLata , B. (ed.) 2007: Giovan Giorgio Alione. Le farse. Testo originale con traduzione italiana e piemontese , Montreal, Lòsna & Tron. V iLLata , B. 2008: Osservazioni sulla lingua dell’Alione . Grammatica dell’astigiano del secolo XVI , Montreal, Lòsna & Tron. DOI 10.24053/ VOX-2023-004 Vox Romanica 82 (2023): 103-139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect Abstract: This article aims to reconstruct some aspects of the phonological system of the 16 th -century Piedmontese dialect of Asti on the basis of the textual corpus provided by Giovan Giorgio Alione’s Opera Jocunda (1521). Its relations with contemporary dialects are examined through internal and comparative reconstruction. If, in many cases, some important continuities have been found, many discontinuities are present as well. At that time, this dialect had a richer phonemic inventory compared to mainstream contemporary Piedmontese dialects, at least as far as consonant phonemes are concerned. In terms of general diachronic evolution, the reconstruction shows that Piedmontese behaved like a Western-Romance language because it shared several historical phonological evolutions with Gallo-Romance and Ibero-Romance, but not with Italian. In this context, it underwent several mergers common to other Gallo-Italian dialects (e.g., the deaffrication of the alveolar affricates). From a language contact point of view, the phonology of the old dialect of Asti shows significant interferences with Middle French (mainly via lexical borrowing), but also with some «high» Northern Italian vernacular, therefore sharing a contact dynamic with Lombard dialects of that time. Keywords: Piedmontese, Asti, Phonology, Reconstruction, Gallo-Italian, Western Romance 139 Reconstruction of the phonological system of a 16 th -century Piedmontese dialect