4th Workshop on Sound Change (Salamanca, October 18, 2019) PROGRAM
9:00‐9:30 Opening session 9:30‐10:15 Henriksen, Nicholas Exploring the production‐perception link and sound change in Andalusian Spanish: Focus on /ptk/ lenition 10:15‐11:00 Renwick, Margaret E. L. Structural, lexical, and social factors in Romance vowel changes 11:00‐11:45 Cangemi, Francesco Redundancy, Specificity and Sound change Break 12:15‐13:00 Recasens, Daniel On the historical evolution of voiced palatoalveolar fricatives and affricates in Catalan and other Romance languages 13:00‐13:45 Abete, Giovanni Internal and external factors contributing to the emergence and spread of /r/ metathesis in Neapolitan Lunch time 16:00‐17:45 Poster session García‐Amaya, Lorenzo/Henriksen, Nicholas, Speech rate and pause variation in two varieties of Peninsular Spanish Niculescu, Oana/Vasilescu, Ioana, Examining the acoustic and articulatory contrast of hiatus – diphthong pairs in contemporary standard Romanian in terms of formant trajectories Nkollo, Mikołaj, Towards the phrasal attachment of EP proclitics. An empirical inquiry into a diachronic sound change Rost Bagudanch Assumpció, Algunos apuntes sobre el rotacismo de /n/ en la evolución al español Ryan, John M., The differentiated outcomes of Classical Latin open and checked syllable structure into Spanish, Italian and Neapolitan Santos, João Paulo Moraes Lima dos, External sandhi processes in vowel sequences in Brazilian Portuguese and peninsular Spanish: change and variation 17:45‐18:30 Hansen, Anita Berit French “E caduc” in word‐initial syllables – engaged in a lexically diffused process of stabilization? Evidence from repeated real time studies of read and spoken Parisian French 18:30‐19:15 Calamai, Silvia & Nodari, Rhotic degemination in marginal Tuscan speech: temporal analysis in legacy speech data Rosalba 19:15‐20:00 Sánchez‐Miret, Fernando Elision of the definite article /‐l/ in Romanian: setting out the issues
(Oral presentations of 35 minutes each with 10 minutes for discussion.) Abstracts (in alphabetical order)
Abete, Giovanni Internal and external factors contributing to the emergence and spread of /r/ metathesis in Neapolitan Calamai, Silvia & Nodari, Rosalba Rhotic degemination in marginal Tuscan speech: temporal analysis in legacy speech data Cangemi, Francesco Redundancy, Specificity and Sound change García-Amaya, Lorenzo/Henriksen, Speech rate and pause variation in two varieties of Peninsular Spanish Nicholas Hansen, Anita Berit French “E caduc” in word-initial syllables – engaged in a lexically diffused process of stabilization? Evidence from repeated real time studies of read and spoken Parisian French Henriksen, Nicholas Exploring the production-perception link and sound change in Andalusian Spanish: Focus on /ptk/ lenition Niculescu, Oana/Vasilescu, Ioana Examining the acoustic and articulatory contrast of hiatus – diphthong pairs in contemporary standard Romanian in terms of formant trajectories Nkollo, Mikołaj Towards the phrasal attachment of EP proclitics. An empirical inquiry into a diachronic sound change Recasens, Daniel On the historical evolution of voiced palatoalveolar fricatives and affricates in Catalan and other Romance languages Renwick, Margaret E. L. Structural, lexical, and social factors in Romance vowel changes Rost Bagudanch Assumpció Algunos apuntes sobre el rotacismo de /n/ en la evolución al español Ryan, John M. The differentiated outcomes of Classical Latin open and checked syllable structure into Spanish, Italian and Neapolitan Sánchez-Miret, Fernando Elision of the definite article /-l/ in Romanian: setting out the issues Santos, João Paulo Moraes Lima dos External sandhi processes in vowel sequences in Brazilian Portuguese and peninsular Spanish: change and variation
Internal and external factors contributing to the emergence and spread of /r/ metathesis in Neapolitan
Giovanni Abete, University of Naples Federico II
Abstract
This paper investigates the emergence and spread of /r/ metathesis in Neapolitan by discussing both historical data and laboratory experiments. The aim is to develop an evolutionary model of metathesis which takes into account phonetic and phonological factors, as well as the influence of social structure and language use.
Two main types of /r/ metathesis can be identified in Neapolitan: metathesis of post‐consonantal /r/ (e.g. crapa ‘goat’ < Lat. CĂPRA(M), and metathesis of pre‐consonantal /r/ (e.g. fròffece ‘scissors’ < Lat. FORFĬCE(M), with the second type being systematically accompanied by compensatory lengthening (Abete 2015). Such metathetic forms are widely attested in the history of Neapolitan (cf. Ledgeway 2009: 110), however the phenomenon is far from being regular and the words that look like unaffected are legion. Furthermore, metathetic forms are often in synchronic alternation with non‐metathetic variants, over which they can sometimes prevail, although in many cases they seem to have very low prestige and tend to exit from use (cf. Tuttle 1997: 273). The goal of this paper is precisely to understand why some words look like more prone to undergo metathesis and how such phenomenon can spread through the lexicon and through the community.
After arguing in favor of a perceptual basis for /r/ metathesis in Neapolitan (cf. Blevins and Garret 1998; 2004), I will focus on the hypothesis that the phonotactic patterns of this language are a relevant predictor of the process, following the model developed by Hume (2004). In order to do this, I will get statistics on Neapolitan consonant clusters and their relative frequency in different prosodic positions (cf. Coleman and Pierrehumbert 1997; Frisch et al. 2000). By relying on these data, it will be possible to verify if the actuation of metathesis is probabilistically related to the phonotactics of the metathetic forms, or, in other words, if the phonotactics of Neapolitan metathetic forms is in some way more similar to the most common phonotactic patterns attested in this language. This hypothesis will be tested on a corpus of Neapolitan metathetic forms (Abete 2015), as well as by a production experiment aimed at eliciting metathesis in nonwords (cf. Lunden and Renoll 2017; Müller 2014). Results will be discussed in relation to what we know about the evolution of /r/ metathesis in Neapolitan, with special reference to its lexical and social distribution, by exploiting a variety of sources including ancient texts, dictionaries and linguistic atlases.
References
Abete, G. (2015). The role of the syllable in the metathesis of /r/ in Neapolitan. In Russo, D. (ed.), The notion of syllable across history, theories and analysis. Cambridge: Cambridge Scholars. 241‐263.
Blevins, J. and Garret, A. (1998). The origins of consonant‐vowel metathesis. Language, 74: 508‐556.
Blevins, J. and Garret, A. (2004). The evolution of metathesis. In Hayes, B., Kirchner, R. and Steriade, D. (eds.), Phonetically‐based phonology. Cambridge: Cambridge University Press. 117‐156. Coleman, J. and Pierrehumbert, J.B. (1997). Stochastic phonological grammars and acceptability. In Coleman, J. (ed.), Computational phonology: Third meeting of the ACL special interest group in computational phonology. Somerset, NJ: Association for Computational Linguistics. 49‐56.
Frisch, S.A., Large, N.R. and Pisoni, D.B. (2000). Perception of wordlikeness: Effects of segment probability and length on the processing of nonwords. Journal of Memory and Language, 42: 481‐496.
Hume, E. (2004). The indeterminacy/attestation model of metathesis. Language, 80(2): 203‐237.
Ledgeway, A. (2009). Grammatica diacronica del dialetto napoletano.Tübingen: Niemeyer.
Lunden, A. and Renoll, K. (2017). Position and stress as factors in long‐distance consonant metathesis. The Linguistic Review, 34(4): 615‐634.
Müller, D. (2014). Liquid metathesis. Poster presented at Sound Change in Interacting Human Systems, 3rd Biennial Workshop on Sound Change, UC Berkeley, May 28‐31 2014.
Tuttle, E.F. (1997). Preferential and pseudo‐metathesis (in Italo‐Romance dialects). In Bertinetto, P.M., Gaeta, L., Jetchev, G. and Michaels, D. (eds.), Certamen phonologicum III: papers from the Third Cortona Phonology Meeting, April 1996. Turin: Rosenberg & Sellier. 267‐91.
Rhotic degemination in marginal Tuscan speech: temporal analysis in legacy speech data
Silvia Calamai (Università degli Studi di Siena, Italy)
Elba Island speech appears to be the least vigorous among Tuscan speech varieties: according to Giannelli (2000), there are only 20,000 actual speakers, after excluding those living in Portoferraio, where a Leghornese variety is spoken. Elba Island speech presents several conservative features, showing at the same time influences from Southern and Corsican dialects. While it is well investigated on the lexical lexical point of view (Cortelazzo 1965, Nesi 1997), it appears barely investigated from a phonetic-phonological point of view. Its phonematic system is the same as that of the Leghornese system, but with a different distribution of several phonemes. According to Giannelli (2000), the Elba Island stressed vowel system is devoid of the Leghornese allophones. As for the consonantal system, on the eastern side of the island intervocalic plosives show a lenition process, while younger generation speakers pronounce Florentine variants [x] and [h] in lieu of intervocalic /k/. A relevant feature of Elba Island speech is the degemination of /rr/ (e.g. te[ra] It. terra ‘land’), which appears to be rather vital, at least in middle-aged speakers, and whose geolinguistic distribution appears to be still rather unknown in Tuscany (Rohlfs 1966; Giannelli 1997: 300). The degemination of /rr/ is attested to in the western side of the region but it is occasionally documented in the Florentine area as well, especially in elderly and rural speakers. In both areas the degemination of /ff/ is occasionally attested to (e.g, di[f]erente It. differente ‘different’). This research aims at investigating the degemination of /rr/ in Elba Island speech, starting from the analysis of a corpus collected (for other purposes) by the historian Tiziana Noce (Noce 2003). The speech archive is made up of about 30 hours of interviews, dealing with life histories of more than 30 male and female speakers, mostly miners, seafarers, laborers, and housewives. Their stories portray a colorful picture of Elba society from the end of the nineteenth century to the beginning of the twenty-first, with particular attention to the coal-mining communities located on the eastern side of the island (Rio Marina, Capoliveri, Porto Azzurro, and Portoferraio). The speech archive – originally recorded on analogic compact cassettes – has been digitized and catalogued by the Scuola Normale Superiore and Università di Siena joint project Grammo-foni. Le soffitte della voce (Calamai & Bertinetto 2014). The records describing contents, speakers, and settings of every single interview are retrievable at https://grafo.sns.it, after the user’s authentication. Although the speech archive appears to be highly multi-faceted, given the type, number and variability of factors involved, it turns out to be particularly suited for an initial socio-phonetic research on the degemination of /rr/, considering the fact that this particular sound change is likely to occur mostly in spontaneous speech. In the experimental analysis, durational measurements (both raw and normalized) of the preceding vowel and of the target consonant occurring in words with phonological geminated rhotic were collected. Wherever possible, they were also compared to durational measurements of the preceding vowel and consonantal duration of words with intervocalic singleton /r/, in order to verify whether, for the very same vowel, a significant difference among the two preceding vowels is documented, being synchronically both in an open syllable. In addition, the role of stress was considered as well, in order to verify whether degemination was favoured in unstressed syllables. Given the particular status of the R variable in conveying socio-indexical information (Scobbie 2006), the research also analyses whether speech about older events elicited older variants (Hay & Foulkes 2016). Since the speech archive is collected for historical purposes, and contains several narratives comparing the past and the present days, it allows us to verify whether rhotic degemination was apparent in words that had connotations of ‘the olden days’.
Selected references
Calamai, S., Bertinetto, P.M. 2014. Le soffitte della voce. Il progetto Grammo-foni, Manziana: Vecchiarelli Cortelazzo, M. 1965. Vocabolario marinaresco elbano, L’Italia Dialettale 28: 1-124. Giannelli, L. 1997. Tuscany. In M. Maiden & M. Parry (eds.), The dialects of Italy, London and New York: Routledge: 297-302. Giannelli, L. 2000. Toscana, 2nd edn. Pisa: Pacini. Hay, J. & Foulkes, P. 2016 The evolution of medial (-t-) over real and remembered time. Language 92(2): 298-330. Nesi, A. 1997. I dialetti e il mare: area alto tirrenica. In G. Marcato (ed), I dialetti e il mare. Atti del Congresso Internazionale di Studi in onore di Manlio Cortelazzo, Chioggia 21-25.IX.1996, Padova: Unipress: 33-48. Noce, T. 2003. Voci di vita elbana. Un secolo di memorie tra vigna, miniera e mare, Coop Toscana Lazio. Rohlfs, G. 1966. Grammatica storica della lingua italiana e dei suoi dialetti. Fonetica Torino: Einaudi. Scobbie, J. 2006. (R) as a variable. In K. Brown (ed.), Encyclopedia of Language & Linguistics (2nd ed.), Oxford: Elsevier: 337–344.
Redundancy, Specificity and Sound change Francesco Cangemi University of Cologne
In natural languages, contrasts and meaning are encoded redundantly: units are often different from each other in more than one way, with obvious benefits for memory performance (Guérard 1921). For example, at the lexical level, natural languages behave very differently from the first extensively researched analytical language (Wilkins 1668), where individual units of sound map directly onto individual units of meaning, and thus “a means animal, ab mammal, abo carnivore, aboj feline, aboje cat” (cf. Borges 1952). At the sentence level, languages might conjure different means to encode a given contrast, and “divide the labour” of signification accordingly. For example, if a language uses sentence particles to encode pragmatic meaning, we might predict its intonation to be less used for that purpose. This seems to be the case for the expression of epistemic meaning in the two Romance languages studied by Prieto & Roseano (2016): whereas Catalan makes extensive use of different intonation patterns, Friulan mainly uses sentence particles, with only limited use of different intonational categories. Redundancy is also attested at the phonological level, where speakers encode contrasts using multiple cues (Lisker 1986), which are often distributed across different time domains (Coleman 2003). More importantly, when redundancy makes multiple cues available to speakers, different speakers are known to capitalise on different cues (Repp 1983, Niebuhr et al. 2011). This suggests the hypothesis that individual-specific strategies in the encoding of contrasts might be more frequent when contrasts are particularly redundant. Both individual specificity and redundant encoding are relevant to sound change. This applies both in an ontogenetic sense (over the timescale of decades in individual acquisition, Shattuck- Hufnagel et al. 2015) and in a phylogenetic sense (over the timescale of centuries in language change, Winter 2014). If a contrast is encoded redundantly, we can expect its phonetic realisation not only to be more speaker-specific (see above), but also to be less conventionalised – at least until speakers’ behaviour converges onto one of the possible phonetic realisations of the contrast.
We illustrate the interplay between redundancy, specificity and sound change by focussing on intonation in Northern Vietnamese (Cangemi et al. 2016a). As most tone languages, NV encodes non-lexical or sentence-level through the use of non-phonetic devices, such as sentence final particles (e.g. đi for imperatives, Thompson 1965). However, recent studies also show the existence of intonational patterns in Vietnamese, through the interplay of pitch, duration, intensity and voice quality (Đỗ et al. 1998 for sentence modality, Michaud and Vũ 2004 for emphasis, Jannedy 2007 for information structure). Research on spontaneous interaction shows that some discourse functions (e.g. backchannel, repair initiation) not only are encoded through intonation, but that the intonational use of f0 in these contexts can even override the expected lexical tonal patterns (Hạ & Grice 2010). However, for Southern Vietnamese, Brunelle (2015) suggests that while sentence modalities have marginally different intonational properties, it is not possible to argue for a fully grammaticalised intonational system, due to the large amount of speaker-specific behaviour. In this talk, we reanalyse a corps from Brunelle et al. (2012), featuring Northern Vietnamese read speech with various sentence modalities and affect values. We show that a classical analysis based on pooled data from all speakers yields no discernible trends. However, by applying methods for targeted speaker-specific analyses (Cangemi et al. 2016b), we show that all speakers use their own strategy to encode intonational meaning via intonation. More importantly, we show that some of these strategies might be on a path to grammaticalisation: in colloquial texts published on the internet, young speakers consistently apply (lexically inappropriate) tonal diacritics to express discourse-level meaning. We take these findings to suggest that redundancy in a language co-occurs with specificity in an individual. These can be explored to unveil strategies which are ultimately relevant for the study of ongoing sound change. Speech rate and pause variation in two varieties of Peninsular Spanish Lorenzo García-Amaya & Nicholas Henriksen University of Michigan
This study explores oral-fluency differences between two varieties of Peninsular Spanish: North- Central Peninsular Spanish (NCPS) and Western-Andalusian Spanish (WAS). Our broad research goal is to understand the extent to which cross-dialect speech-rate differences motivate phonological change in historically-related language varieties.
Speech rate is one of multiple phonetic sources that are posited to trigger incipient sound change, and its related sociophonetic variation (e.g., Nadeu & Hualde, 2015). Function words are also known to undergo deletion in casual speech depending on the velocity of the surrounding speech rate (e.g., Dilley et al., 2010). Regarding cross-dialect variation, Schwab and Avanzi (2015) examined data from seven varieties of European French, with Parisian speakers showing faster speaking rates than Belgian and Swiss-French speakers. Factors such as age, gender, and speaking style additionally influenced talker variation (for English, see Jacewicz et al., 2009).
In this study, we examined speech data from 20 speakers, 10 from Salamanca Spain (NCPS), and 10 from Jerez de la Frontera, Spain (WAS). All speakers were university students, between ages 18 and 25, with five males and five females per dialect group. For each speaker, we performed a fluency analysis of their first ten minutes of conversational speech, coding speech (with syllable counts), filled pauses (e.g., eh, um), and silent pauses above 250ms. We then calculated a series of fluency metrics: speech rate, articulation rate, filled-pause rate, mean filled-pause duration, silent-pause rate, and mean silent-pause duration. For all metrics, we fitted analysis of variance models (ANOVA) using the “aov()” function within the statistical software package R (version 3.5.1) (R Core Team, 2018) to compare the two speaker groups (i.e., NCPS and WAS).
The ANOVA results showed a significant difference for filled-pause rate only, with WAS speakers showing higher filled-pause rates than NCPS speakers (p=.049, see Figure 1). For measures of speech fluency, such as speech rate, we did not uncover a significant difference (p=.677, see Figure 2); nor did we uncover a significant difference for silent-pause rates (p=.502, see Figure 3). These results thus do not offer credence to the common stereotype that Andalusian speakers talk faster than speakers of northern Peninsular varieties. On the contrary, they suggest that Andalusians use nonlexical pauses (e.g., uh, um) more than NCPS speakers. Altogether, these results offer preliminary evidence that phonological differences between northern and southern varieties of Peninsular Spanish are not attributable to oral-fluency differences. Finally, we also observe substantial inter-speaker variation in the data sample.
Figure 1. Filled-pause rates between NCPS speakers (Salamanca) and WAS speakers (Jerez). First letter of speaker code is regional origin (Salamanca or Andalusia); second letter is gender (male or female); final number is dialect/gender-specific per speaker.
Figure 2. Speech-rate differences (syllables/sec) between NCPS speakers (Salamanca) and WAS speakers (Jerez). First letter of speaker code is regional origin (Salamanca or Andalusia); second letter is gender (male or female); final number is dialect/gender-specific per speaker.
Figure 3. Silent-pause rates between NCPS speakers (Salamanca) and WAS speakers (Jerez). First letter of speaker code is regional origin (Salamanca or Andalusia); second letter is gender (male or female); final number is dialect/gender-specific per speaker. Selected references Schwab, S., & Avanzi, M. (2015). Regional variation and articulation rate in French. Journal of Phonetics, 48, 96–105. Nadeu, M., & Hualde, J. (2015). Biomechanically conditioned variation at the origin of diachronic intervocalic voicing. Language and Speech, 58(3), 351–370. French “E caduc” in word‐initial syllables – engaged in a lexically diffused process of stabilization? Evidence from repeated real time studies of read and spoken Parisian French
Anita Berit HANSEN
French “E caduc” or schwa – here pragmatically defined as the letter “e” (without accent) in graphically open syllables – has traditionally been described, in non‐meridional varieties, as an element that drops or is unstable in VC_C contexts (samedi [sam ˈdi], école publique [e kɔl py ˈblik] ; mais l(e) garçon [mƐ lə gar ˈsõ]] / [mƐl gar ˈsõ]], la p(e)tite fille [la pə tit ˈfij] / [lap tit ˈfij]), but retains a lot better in CC_C contexts (for example obligatorily pronounced in sept petites filles [sƐt pə tit ˈfij], vendredi [vã drə ˈdi]). As for the unstable positions – in VC_C monosyllabic words and word‐initial syllables – two contradicting hypotheses for change in Parisian French have seen the light in the second half of the 20th century, that of a progressively spreading elision (Malécot 1976; Péretz‐Juillard 1977), prolonging the vowel reduction in unstressed open syllables that took place in the evolution from Latin, and that of a process of stabilization of the vowel (Walter 1977; 1988; 1990; Fónagy 1989). The latter hypothesis is said to be noticeable in Paris since 1950 (Walter 1988) or 1970 (Fónagy 1989), and it is accompanied by observations of words with stable E caduc (melon, cheval for instance) that seem to add to an already existing list: Historical linguists like Morin (1978) have shown that the vowel has been involved in a slow and incomplete stabilization process in French since the 16th/17th century in initial syllables of polysyllabic words, and have become fully stable in some of these words already (belette, femelle, peler, rebelle, serein, etc.), cf. also Walker (1996) who claims, on the basis of recent dictionary data for 400 relevant words, that “roughly half of them” have stable or stabilizing E caduc.
In order to grasp the direction of the dynamic tendencies in recent times in Parisian French (spreading elision or spreading stabilization?), and in order to look into the role of the lexicon, a first real time study was undertaken in Paris, comparing data from 1972‐74 and from 1989 (Hansen 1994). This study, however, could not prove that any stabilization of E caduc in VC_C was taking place in spoken language. On the contrary, young people in both corpora dropped the vowel more often than adults, both in monosyllables and in word‐initial syllables. Reading, however displayed very high retention rates, especially in word‐initial syllables, and important lexical differences were found here in the 1989‐material (38% retention in demi, 53% in semaine, but 100% in celui, reconnu, tenait for instance). Factors of frequency and articulation were investigated to account for these differences (ibid.).
Relevant empirical studies, published between 1996 and 2017 (Racine & Grosjean 2002; Geerts 2011; Lacheret et al. 2011; Bürki et al. 2011) are difficult to compare directly to our own (francophone but not French speakers, or French but not exclusively Parisian origin; different delimitation of E caducs to consider; journalistic speech rather than ordinary people’s speech, etc.), but two surprising results in those studies that methodologically come closest to ours are that word‐initial syllables are not seen as specifically favorable to retention compared to monosyllables (Geerts 2011; Lacheret et al. 2011), and that dynamics are still talked of as spreading elision (Racine & Grosjean 2002).
A renewed real time study, comparing the Parisian corpus from 1989 with freshly collected data (2012‐2015) that parallel the first corpus exactly (comparable speakers, same reading exercice and interview scheme, same interviewer) (Hansen in progress), however permits to see the contours of an increased stabilization of E caduc in word‐initial syllables among Parisians, both in speaking and reading style. This movement touches the words irregularly, but can it be termed a “lexical diffusion”? Frequent words seem to change last, but frequency cannot explain all of the variation (there are leaders and laggers even among the most frequently used words). Re‐ words play a particular role, as does the articulatory relationship between first and second consonant. These findings are discussed in connection with Labov’s criteria for speaking of lexical diffusion (Labov 1994), and with statements on frequency effects in different types of sound changes (Philips 1984; Yaeger‐Dror 1996). On a more methodological note, the relatively modest time span covered here (even with repeated real time data collection), and the relatively limited number of different attested lexical items (even with use of an identical reading task), make us question whether incontestable lexical diffusion can only be traced and documented over several centuries and on more abundant data (cf. the 300 years time span, in Ogura & Wang 1996). References
Bürki, Audrey, Mirjam Ernestus, Cédric Gendrot, Cécile Fougeron, Ulrich Hans Frauenfelder (2011) What affects the presence versus absence of schwa and its duration: A corpus analysis of French connected speech, in Journal of the Acoustical Society of America 130, 3980‐3991.
Fónagy, Iván (1989) Le français change de visage ? Revue Romane 24, 2, 225‐253.
Geerts, Twan (2011) Ch(e)va : Schwa français en syllabe initiale, Langue française 169 (Phonologie du français contemporain : usages, variations, structures), 39‐54.
Hansen, Anita Berit (1994) Etude du E caduc – stabilisation en cours et variations lexicales, Journal of French Language Studies, 4, 25‐54.
Hansen, Anita Berit (in progress) Variation et changement dans la prononciation du français de la région parisienne – une approche multiples facettes.
Lacheret, Anne, Chantal Lyche, Atanas Tchobanov (2011) Schwa et position initiale revisités : l’éclairage de la prosodie en phonologie du français contemporain, Langue française 169 (Phonologie du français contemporain : usages, variations, structures), 137‐158.
Labov, William (1994) Principles of Linguistic Change, vol. 1: Internal Factors, Oxford UK/Cambridge USA, Blackwell.
Malécot, André (1976) The Effect of Linguistic and Paralinguistic Variables on the Elision of the French Mute‐e, Phonetica 33, 93‐112.
Morin, Yves‐Charles (1978) The status of mute « e », Studies in French Linguistics, vol. 1, 2, p. 79‐140.
Ogura, Mieko & William S.‐Y. Wang (1996) Snowball Effects in Lexical Diffusion. The Development of ‐s in the Third Person Singular Present Indicative in English, in Britton, Derek (ed.) Papers from the 8th International Conference on English Historical Linguistics, Amsterdam/Philadelphia, Benjamins, 119‐141.
Péretz‐Juillard (1977) Les voyelles orales à Paris dans la dynamique des âges et de la société, unpublished doctoral thesis, Université de Paris V.
Phillips, Betsy S. (1984) Word frequency and the actuation of sound change, Language 60, 320‐342.
Racine, Isabelle & François Grosjean (2002) La production du E caduc facultatif est‐elle prévisible ? Un début de réponse, Journal of French Language Studies, 12, 307‐326.
Walker, Douglas (1996) The new stability of unstable ‐e in French, Journal of French Language Studies, 6, 211‐229.
Walter, Henriette (1977) La phonologie du français, Paris, Presses universitaires de France.
Walter, Henriette (1982) Enquête phonologique et variétés régionales du français, Paris, Presses universitaires de France.
Walter, Henriette (1988) Le français dans tous les sens, Paris, Robert Laffont.
Walter, Henriette (1990) Une voyelle qui ne veut pas mourir, in Green, John N. & Wendy Ayres‐Bennett (eds) Variation and Change in French. Essays presented to Rebecca Posner on the occasion of her sixtieth birthday, London, New York, 27‐36.
Yaeger‐Dror, Malcah (1996) Phonetic evidence of lexical classes: The case of a Montreal French vowel shift, in Guy, Gregory et al. (eds) Towards a social science of language. Papers in honor of William Labov, vol. 1: Variation and Change in Language and Society, Amsterdam, Benjamins, 263‐287. Exploring the production-perception link and sound change in Andalusian Spanish: Focus on /ptk/ lenition
Nicholas Henriksen University of Michigan
As has been well established in works by Ohala (1981, 1993) and Beddor (2009, 2012, 2015), the production-perception link is critical for understanding the role of the individual listener/speaker in sound change. Within this research program, the question of how the listener might initiate change can be approached through two different lenses. On the one hand, Ohala proposes a model of listener-based sound change centered on a listener’s misperception of the articulatory source of contradictory (or overlapping) acoustic information. On the other hand, sound change might occur when a listener may correctly perceive a phonetic effect but attribute it to an articulatory gesture that is different from the sound’s origin, through cue re-weighting (Beddor, 2009, 2012, 2015).
In this project, we test competing theories of sound change, focusing on intervocalic /ptk/ lenition in Andalusian Spanish (spoken in southern Spain). Previous research explores incipient sound change in north-central Peninsular Spanish, focusing on linguistic and social factors that contribute to the lenition of /ptk/ (e.g., Nadeu & Hualde, 2015). In this project, we extend the work on /ptk/ lenition to Andalusian Spanish, where this sound change is considerably more advanced than in varieties spoken in north-central Spain. We conducted sociolinguistic interviews with 30 speakers from Jerez de la Frontera (young, middle-aged, and older), with each age group equally balanced for gender. The acoustic findings (using metrics of %-voicing, C/V- ratio, and duration) reveal that Andalusians show rates of /ptk/ lenition between 50% and 60%, substantially higher than data reported for speakers of north-central Peninsular Spanish (25%, Hualde et al., 2011). We found a gender effect whereby Andalusian men voice /ptk/ at higher rates than Andalusian women, compatible with biomechanical explanations for /ptk/ voicing (Nadeu & Hualde, 2015). However, we did not uncover age-related effects, due to the high rate of inter-speaker variation across the speaker sample.
Given this result, we conducted a follow-up word-identification perception experiment with the original 30 participants to examine more closely the production-perception link through an experimental design, along the lines of Coetzee et al. (2018). We manipulated overall consonant duration and percent voicing, in order to determine which of these two cues is more reliable in word identification of intervocalic /ptk/, and also to determine if those speakers who show higher frequencies of /ptk/ lenition (from the production results) are more likely to perceive /ptk/ based on greater voicing or shorter duration of the acoustic stimulus. Young and middle-aged listeners were generally more likely to rely on voicing (but not duration) than older listeners. Altogether, the perceptual data are consistent with generationally-based cue re-weighting in Andalusian Spanish. The data present evidence for a more advanced sound change (compared to north- central Peninsular Spanish) in which the /ptk/ - /bdg/ contrast is maintained through duration only, but not necessarily through voicing, due to differential cue weighting.
Collaborators: Andries Coetzee & Lorenzo García-Amaya (University of Michigan) References
Beddor, P.S. (2009). A coarticulatory path to sound change. Language, 85, 785–821. Beddor, P.S. (2012). Perception grammars and sound change. In M.-J. Solé & D. Recasens. (Eds.), The Initiation of Sound Change: Production, Perception, and Social Factors (pp. 37–55). Amsterdam: John Benjamins. Beddor, P.S. (2015). The relationship between language users’ perception and production repertoires. In The Scottish Consortium for ICPhS 2015 (Ed.), Proceedings of the 18th International Congress of Phonetic Sciences. Glasgow, UK: The University of Glasgow. Coetzee, A., P. S. Beddor, K. Shedden, W. Styler, & D. Wissing. (2018). Plosive voicing in Afrikaans: Differential cue weighting and tonogenesis. Journal of Phonetics, 66, 185– 216. Hualde, J. I., & Nadeu, M. (2011). Lenition and phonemic overlap in Rome Italian. Phonetica, 68, 215–242. Hualde, J. I., Simonet, M., & Nadeu, M. (2010). Do words or phonemes change? Evidence from lenition pro- cesses in progress. Poster presented at the Workshop on Sound Change. October 2010. Barcelona, Spain. Hualde, J. I., Simonet, M., & Nadeu, M. (2011). Consonant lenition and phonological recategorization. Ohala, J.J. (1981). The listener as a source of sound change. In C.S. Masek, R. Hendrick, & M. Miller. (Eds.), Papers from the Parasession on Language and Behavior (pp. 178–203). Chicago: Chicago Linguistics Society. Ohala, J.J. (1993). The phonetics of sound change. In C. Jones (Ed.), Historical linguistics: problems and perspectives (pp. 237–278). London: Longman.
EXAMINING THE ACOUSTIC AND ARTICULATORY CONTRAST OF HIATUS – DIPHTHONG PAIRS IN CONTEMPORARY STANDARD ROMANIAN IN TERMS OF FORMANT TRAJECTORIES
Oana NICULESCU1, Ioana VASILESCU2 1“Iorgu Iordan – Al. Rosetti” Institute of Linguistics, Romania 1University of Bucharest, Faculty of Letters, Romania 2LIMSI, CNRS, Université Paris-Saclay, Bât. 508, Orsay, France
Introduction. Previous work has focused on hiatus – diphthong distinction mainly in terms of durational patterns [1]; [2]; [3]; [4] among others. There are few studies which examine these vocalic pairs according formant movements [5]; [6]; [7]. The novelty of our account consists in capturing the dynamics behind hiatus – diphthong contrast in terms of formant trajectories in contemporary standard Romanian. Data and methodology. Among Romance languages, Romanian displays a robust diphthong – hiatus contrast, the heterosyllabic [iV] sequences being the preferred outcome [2]. Romanian has a seven vowel inventory (3 high vowels /i/, /ɨ/, /u/; 3 mid vowels /e/, /ə/, /o/; one low vowel /a/) and two phonologically unary diphthongs (/e a/ and / o a/) [8]. It is a Romance language characterized by a high diversity in both ascending (e.g. /ja/, /je/) and descending diphthongs (e.g. /aj/, /ej/). As for VV-sequences, all 49 can surface as external hiatus (adjacent vowels across word boundary), while internal hiatus (adjacent vowels within the word) is more restricted (based on the lexicographic analysis conducted in [9], only 35 pairs were identified). The aim of this study is to compare formant movements of hiatus – diphthong pairs by means of two controlled experiments. The recordings took place in a quiet room, directly onto a laptop. A Behringer B1 microphone and an external audio interface M-Audio Fast Track were used. This is a pilot study derived from the thesis of the first author [10]. The experiment was twofold. Experiment 1. In the first part of the analysis, the objective was to showcase prototypical representations of hiatus and diphthongs in standard Romanian (Figure 1). The vowel pairs were inserted in logatoms placed in carrier sentences (5 repetitions were elicited). Based on the resulting spectrograms, we designed a transcription system of F1 and F2 trajectories (e.g. F1V1 ~ F1V2 F1V1V2—; F1V1 < F1V2 F1V1V2↗). By comparing the canonical description with the acoustic outputs, the transcription system enabled us to highlight the discrepancies (e.g. F1i F1u, meaning that F1iu —, since /i/ and /u/ are both high vowels, but what we observed was that F1iu ↗, meaning that /i/ is higher than /u/ (observation confirmed also by ultrasound [10]) – aligning Romanian with the results obtained in [11], which concluded that for the 29 out of the 30 languages under investigation, F1 is higher for high back vowels than for high front vowels). Experiment 2. In the second part of the study, we analyzed 8 diphthong – hiatus pairs (/je/ – /ie/, /əj/ – /əi/, /ɨj/ – /ɨi/, /oj/ – /oi/, /uj/ – /ui/, /ew/ – /eu/, /iw/ – /iu/, /ow/ – /ou/) placed in word-final position. The near-minimal pairs (e.g. /al.ˈtoj/ – /al.to.ˈi/; /a.ˈtew / – /a.ˈte.u(l)/, the elision of the definite article -(u)l is common in spontaneous speech [12]) were embedded in carrier sentences (Zic ___ tare ‘I say ___ loud’), and 3 repetitions were elicited. The outputs were manually segmented in Praat starting from the onset of F1 in the first (semi)vocalic unit until the offset of F2 in the second (semi)vocalic segment (Figure 2). Data were collected due to a script provided by M. Renwick – each vocalic sequence is divided into 9 parts, and the formant values (F1, F2, F3) are extracted from 10% through 90% of the pair’s duration. The results are then plotted in R using the ggplot2 package. Results and discussions. In this presentation we focus on data of one Romanian speaker extracted from a larger database of 9 speakers representative of the Southern dialect of Muntenia. Results Experiment 1. From the prototypical representations we observed that hiatus has two steady states, corresponding to each of the vowel in the pair, while diphthongs are characterized by a continuous transition from one segment to another. This line of inquiry is relevant to the study of vocalic sequences by providing numerous spectrograms (70 to be precise, see Chapter 5 [10]) processed by a transcription system offering a coherent interpretation of F1 and F2 trajectories. Results from Experiment 2. In this experiment we looked both at duration and formants. In regard to the temporal domain, hiatus is longer than diphthong (308 ms (st. dev. = 36), compared to 212.7 ms (st. dev. = 32)), both vocalic pairs being longer than previous accounts, effect correlated to word-final lengthening. An Anova test was performed with ‘Duration’ as the dependent variable, and ‘Group’ as an independent factor (with two levels, corresponding to ‘hiatus’ and ‘diphthong’). The results showed that hiatus is statistically longer than diphthong (F (1, 130) = 249,835; p < 0,001). As for the frequency domain, the established view is that hiatus manifests a higher degree of curvature in the F2 trajectory than diphthong [5]; [13]; [14]; [15]. Our analysis identifies and describes, for the first time, all 4 possible outcomes (Figure 3), thus proving that the phenomena is more complex than previously noted. By examining various vocalic pairs, this study opens up discussions in relation to modeling gradient phonetic and phonological distinctions between hiatus and diphthong. Future analyses. These results will be complemented by gathering additional data. On the one hand, other speakers can be recorded in different speech situations. On the other hand, since this procedure is time consuming and only covers a limited number of cases, we suggest looking at large corpora of forced aligned spontaneous speech [16]; [17] which have proven to be of high relevance in acoustic analysis, testing various linguistic hypothesis and exploring sound change and variation [18]; [19]; [20].
Figure 1 Figure 2 Prototypical representation of /i.u/ Prototypical representation of /ju/ The pair /altoj/ – /altoi/ hiatus diphthong
p j u
Figure 3 F1 – F2 – F3 trajectories for hiatus – diphthong pairs (with F2 frequencies highlighted)
REFERENCES [1] Hualde, J. I., M. Prieto. 2002. ‘On the diphthong/hiatus contrast in Spanish: some experimental results’, in Linguistics, 40(2), 217–234. [2] Chitoran, I., J. I. Hualde. 2007. ‘From hiatus to diphthong: the evolution of vowel sequences in Romance’, in Phonology, 24, 37–75. [3] Garrido, M. 2007. ‘Diphthongization of Mid/Low Vowel Sequences in Columbian Spanish’, in Selected Proceedings of the Third Workshop on Spanish Sociolinguistics, 30–37. [4] Hualde, J. I. 2014. Los sonidos del español, Cambridge, Cambridge University Press. [5] Aguilar, L. 1999. ‘Hiatus and diphthong: Acoustic cues and speech situation differences’, in Speech Communication, 28, 57– 74. [6] Alba, M. 2006. ‘Accounting for Variability in the Production of Spanish Vowel Sequences’, in Selected Proceedings of the 9th Hispanic Linguistics Symposium, 273–285. [7] Souza, B. 2010. Hiatus Resolution in Spanish: An experimental study, Ph.D. dissertation, University of Pennsylvania. [8] Chitoran, I. 2002. The Phonology of Romanian: A constraint-based approach, New York, Mouton de Gruyter. [9] Niculescu, O. 2012. ‘Hiatul în limba română. O perspectivă fonostatistică și acustică’[Hiatus in Contemporary Romanian. A Phonostatistic and Acoustic Perspective], in Rodica Zafiu et. al. (ed.), Limba română: direcții actuale în cercetarea lingvistică. Actele celui de al 11-lea Colocviu internațional al Departamentului de Lingvistică (Bucharest, 9 – 10 December 2011), Vol. I. Gramatică, Fonetică și fonologie, Istoria limbii, filologie, Bucharest, Bucharest University Press, 286–294. [10] Niculescu, O. 2018. Hiatul intern şi hiatul extern în limba română contemporană. O analiză acustică [Internal and external hiatus in contemporary standard Romanian. An acoustic analysis], Ph.D. dissertation, University of Bucharest. [11] de BOER, B. 2011. ‘First formant difference for /i/ and /u/: A cross-linguistic study and an explanation’, in Journal of Phonetics, 39 (1), 110–114. [12] Vasilescu, I., I. Chitoran, B. Vieru, M. Adda-Decker, M. Candea, L. Lamel, O. Niculescu. 2018. ‘Studying variation in Romanian: deletion of the definite article –l in continuous speech’, in Linguistics Vanguard, Forthcoming. [13] Chitoran, I. 2002. ‘A perception-production study of Romanian diphthongs and glide-vowel sequences’, in Journal of the International Phonetic Association, 32(2), 203–222. [14] Chitoran, I. 2003. ‘Gestural timing and the glide percept in Romanian’ , in Proceedings of the 15th International Congress of Phonetic Sciences (ICPhS XV), Barcelona, Spain, 3013–3016. [15] Chitoran, I. 2003. ‘Inter-gestural timing between vocalic gestures as a function of syllable position: Acoustic evidence from Romanian’, in Proceedings of the 6th International Seminar on Speech Production, Sydney, Australia, 25–30. [16] Vasilescu, I., B. Vieru, L. Lamel. 2014. ‘Exploring pronunciation variants for Romanian speech-to-text transcription’, in Proceedings of SLTU-2014, 161–168. [17] Niculescu, O., I. Vasilescu, B. Vieru, L. Lamel, M. Adda-Decker. 2017. ‘Semi-automatic analyses of vocalic sequences in Romania’, poster presentation at Phonetics and Phonology in Europe (PaPE), Köln, Germany. [18] Ohala, J.J. 1996. ‘The connection between sound change and connected speech processes’, in Arbeitsberichte (AIPUK 31) Universität Kiel, 201–206. [19] Adda-Decker, M. 2006. ‘De la reconnaissance automatique de la parole a l’analyse linguistique des corpus oraux’, in Journaux d’Etude sur la Parole, France [20] Vasilescu, I., C. Dutrey, L. Lamel. 2015. ‘Large scale data based investigations using speech technologies: the case of Romanian’, in Proceedings of the 8th Conference on Speech Technology and Human-Computer Dialogue “SpeD 2015”, Bucharest, October 14–17, 6 p.
Towards the phrasal attachment of EP proclitics. An empirical inquiry into a diachronic sound change
One of the original features of present-day European Portuguese (EP) is the asymmetry of preverbal and postverbal clitic pronouns. While the former fail to exhibit allomorpy, enclitics interact in a very intricate morphological and phonological way with verb forms (Spencer & Luís, 2012: 205-206). In doing so, enclitics display an affix-like behavior (Luís & Kaiser, 2016). Yet, morpho-phonological structures have not always been mapped onto syntactic level that way. The aim of the poster is to uncover mechanisms responsible for the rise of the contemporary clitic grammar to replace the old one. The central topic is interpolation, i.e. the insertion of an element between a clitic pronoun and the verb [cl X verb]. In Romance linguistics, instances of clitic-verb non-adjacency are assumed to lend support to the post-lexical (syntactic) rather than word-internal nature of clitic attachment (Miller & Monachesi, 2003). Throughout the history of EP, interpolation has been, albeit at varying degrees, a viable linear model (see (1) below). For ca 180 years (mid-17 th up to the early 19 th century), it involved uniquely the não sentential negator (cf. Martins, 2016). The thesis is that the [cl não verb] sequence paved the way to the present-day EP clitic system (as outlined above). In Classical Portuguese (approx. 1550-1700, in line with Castro’s, 2006 periodization) preverbal pronouns were left-aligned to their syntactic governing category (phrasal attachment; Gerlach, 2002). At the same time, if their morpho-phonological properties are considered, they were enclitically linked to the preceding non-verbal sound material (to some of the prepositions, the wh-pronoun quem ‘who’ and the very não negative; see (2)). The interpolation of não helped do away with such an inconsistent architecture. Our corpus, compiled on the basis of 39 texts released between 1614 and 1858, contains as for now 667 manually retrieved examples of both [cl não verb] and [não cl verb] sequences. The average ratio of interpolation to cases of clitic-verb adjacency attains 2.64 : 1 (484 vs. 183 occurrences). Yet, the two models are shown to have not been fully interchangeable. They were different in the type of pronouns given preference in each of them: vowel-initial 3rd person direct objects ( o, a, os , as ) tended to be paired with interpolation. The respective statistics are as follows: 115 occurrences of [cl não verb] vs. 18 occurrences of [não cl verb], i.e. 6.38 to 1. By contrast, the remaining pronouns did not depart significantly from the average ratio. Up to the early 18th century, the não o sequence surfaced occasionally as nãno : as não ends in the [ãw̃] nasal diphthong, it coerced the ensuing vowel-initial pronouns into taking a nasal onset (o > no , as > nas, etc.). The reversal of this order, i.e. interpolation, eliminated such instances of clitic allomorphy from the preverbal field: once não came to separate a clitic pronoun and the verb, the não no , não na , etc. sequences were ruled out. From that period on, phrasal attachment has become exclusive in preverbal position (Luís, 2014) and o, a, os , as eventually lost their morpho-phonological ties with whatever happened to precede them. In present-day EP, clitic allomorphy has been earmarked for postverbal items (que não a viram ‘that they did not see her’ vs. vir am-na ‘they saw her’). As for interpolation, it lost much of its former momentum, although individual dialectal data still evinces a certain amount of variation (Magro, 2010).
Examples: (1) ..., para que a tri ſteza naõ as affligi ſſ e, ... so that DEF .F sorrow NEG ACC -3.PL .F afflict-SUBJ .IPFV.3. SG (1701) ‘(was making all efforts) so that the sorrow does not overcome them’ ( http://purl.pt/346 )
..., que o ſirvaõ, os quaes athegora o naõ ſerviaõ ... DEF .M REL .PL up to now ACC -3.SG .M NEG serve-IPFV.IND .3. PL (1660) ‘… to get hold of Indians so that they serve him, that up to now have not been his servants’ (http://purl.pt/16556 )
(2) ... naõ no fazendo de ſde o dia que lhe puzerem a dita pena ... NEG ACC -3.SG .MASC do-GERUND since DEF .MASC day … (1639) ‘… not doing this starting from the day when he will be inflicted this punishment’ ( o evolves into no under the influence of the nasal nucleus in não ) ( http://purl.pt/30213 ) ... pe llo permitir Deos no ſſ o Senhor a ſsi, ... for-ACC .3. SG .MASC allow-INF God our Lord so … (1614) ‘…, for so was allowed by God, our Lord’ (the por o sequence evolves into pelo under the influence of the alveolar trill coda in por ) http://purl.pt/29503
REFERENCES Castro, Ivo (2006), Introdução à história do português (2nd edn). Lisbon: Edições Colibri. Gerlach, Birgit (2002), Clitics between syntax and lexicon . Amsterdam-Philadelphia. John Benjamins. Luís, Ana (2014), On clitic attachment in Ibero-Romance. Evidence from Portuguese and Spanish, in: P. Amaral & A.M. Carvalho (eds.). Portuguese-Spanish Interfaces: Diachrony, synchrony and contact , Amsterdam-Philadelphia: John Benjamins, 203-235. Luís, Ana & Kaiser, Georg (2016), Clitic Pronouns: Phonology, Morphology, and Syntax, in: L.W. Wetzels, J. Costa & S. Menuzzi (eds.). The Handbook of Portuguese Linguistics ,
West Sussex: John Wiley & Sons, Inc, 471-486. Magro, Catarina (2010), When corpus analysis refutes common beliefs: The case of interpolation in European Portuguese dialects, Corpus 9, 115–135. Martins, Ana Maria (2016), A colocação dos pronomes clíticos em sincronia e diacronia, in: A.M. Martins & E. Carrilho, (eds.). Manual de linguística portuguesa , Berlin: De Gruyter, 410-430. Miller, Philip & Monachesi, Paola (2003), Les pronoms clitiques dans les langues romanes, in: D. Godard & A. Abeillé (eds.). Les langues romanes, problèmes de la phrase simple . Paris. CNRS, 53-106. Spencer, Andrew & Luís, Ana R. (2012), Clitics. An Introduction . Cambridge. Cambridge University Press.
CORPUS: Biblioteca Nacional de Portugal - Biblioteca Digital ( http://purl.pt/index/geral/PT/index.html ) On the historical evolution of voiced palatoalveolar fricatives and affricates in Catalan and other Romance languages
Daniel Recasens
The present study focuses on the word initial and intervocalic endproducts of the voiced (alveolo)palatal stop [ɟ] derived from Latin /gi, ge/, /j/, /dj/, /gj/ and /bj, vj/ in the Romance languages and more especifically in Catalan (e.g., [ʒəˈne] IANUARIU ‘January’, [ʒərˈma] GERMANU ‘brother’, [ʎəˈʒi] LEGERE ‘to read’, [ˈtɾuʒə] TROIA ‘sow’). The convenience of investigating this phonetic development arises from the fact that, as shown in the dialectal scenarios (a) through (d) below, [ɟ] has yielded the voiced palatoalveolar consonants [ʒ] and [dʒ] in complex ways (data for the Pallars and Ribagorça regions are excluded):
(a) [ʒ] word initially and intervocalically in most of the Central dialect, and in northern Western Catalan, Roussillonese and Balearic; (b) [ʒ] and, less so, [dʒ] word initially and [jʒ] intervocalically in central Western Catalan; (c) [dʒ] word initially and [jʒ] intervocalically in southern Western Catalan and southern Central Catalan, (d) [dʒ] word initially and intervocalically in Valencian and Algherese.
In addition, the dialectal zones (a), (b) and (c) have [dʒ] instead of [(j)ʒ] intervocalically next to stress in a considerable number of common words such as, among others, [ˈmidʒə] mitja MEDIA ‘half, fem. sing.’, [kuˈrɛdʒə] corretja CORRIGIA ‘belt’ and [piˈdʒo] pitjor PEJORE ‘worse’. An interpretation of these data is presented which is rooted on the effect of contextual and positional factors on the changes [ʒ] > [dʒ] (affrication) and [dʒ] > [ʒ] (deaffrication). In partial disagreement with previous explanatory accounts, it is hypothesized that scenarios (a) and to a large extent (b) have been achieved through the historical development /ɟ/ > [dʒ] > [ʒ] and thus a [dʒ] deaffrication process, most likely to occur intervocalically. In so far as it cannot be clearly accounted for on pure phonetic terms, the systematic presence of [dʒ] in groups (c) and (d) is attributed to a language contact situation involving the speech of inhabitants from northern Western Catalan, Occitania and Aragon who settled in the center and south of the Lleida province, the Kingdom of Valencian and the town of l’Alguer between the 12th and 14th centuries. A graphemic analysis of the Old Catalan texts carried out by the author shows that [(j)ʒ] shifted to [dʒ] intervocalically in lexical items such as mitja, corretja and pitjor due to the analogical influence of related forms which had an affricate. Through this process, for example, [ˈmiʒə] ‘half, fem. sing.’ MEDIA became [ˈmidʒə] because the masculine form [mitʃ] had an affricate. This affrication process started at about the 13th c. or even earlier and proceeded gradually through the lexicon during the following centuries.
Structural, lexical, and social factors in Romance vowel changes Margaret E. L. Renwick
This paper gathers findings from the phonology and phonetics of three Romance vowel systems to show how phonological contrasts are affected, over time and synchronically, by structural, lexical, and social factors. The contrasts discussed are all marginal, meaning the vowels involved are separated by few minimal pairs, are phonetically similar, and may be subject to phonological conditioning. Such factors could lead to sound change, for instance via phonological merger, over time (Wedel, Jackson & Kaplan 2013). In Romanian, the central vowels /ʌ ɨ/ are historical allophones, whose status as separate phonemes was cemented by subsequent sound changes and the introduction of loanwords (Renwick 2014). This reorganized the structural relationship between the two sounds. Synchronically, the phonetic distributions of these lexically rare vowels are heavily overlapping, particularly in continuous speech; I argue that this overlap is tolerated because of the strong contextual restrictions that still affect /ʌ ɨ/. Speaker-hearers can rely on phonological environment, as much as acoustics, to identify the vowel. In Italian, the contrasts in question are the mid vowels, /e ɛ/ /o ɔ/, whose lexical implementation varies across regions but even within words. Via a laboratory study and a large-scale corpus analysis, I investigate these vowels’ realization across and within speakers and regional varieties. Although the high mid and low mid vowels remain contrastive, variation in the mapping between phonetic categories and specific lexical items weakens that contrast (Renwick & Ladd 2016), and it occurs even within a single regional variety. In Catalan, the mid vowels /e ɛ o ɔ/ are also marginally contrastive (Nadeu & Renwick 2016) and variable across dialects, but these oppositions are posited to be tenuous thanks also to social factors including speakers’ bilingualism in Spanish, which lacks a height contrast in mid vowels. Most Catalan-dominant speakers maintain four mid vowels, even if their intuitions of lexical assignment are poor; however, social factors including language dominance and age affect speakers’ lexical assignment of mid vowel height (Renwick & Nadeu 2018). Although native speakers use these contrasts, each investigation above shows considerable variability in their implementation, across and within speakers, lexical items, or geographical areas. Taken together, these case studies show a range of factors that currently influence Romance vowel contrasts, and which may lead to future phonemic changes in these systems.
References Nadeu, Marianna & Margaret E. L. Renwick. 2016. Variation in the lexical distribution and implementation of phonetically similar phonemes in Catalan. Journal of Phonetics 58. 22–47. doi:10.1016/j.wocn.2016.05.003. Renwick, Margaret E. L. 2014. The Phonetics and Phonology of Contrast: The Case of the Romanian Vowel System. Berlin, Boston: De Gruyter Mouton. Renwick, Margaret E. L. & D. Ladd. 2016. Phonetic Distinctiveness vs. Lexical Contrastiveness in Non- Robust Phonemic Contrasts. Laboratory Phonology: Journal of the Association for Laboratory Phonology 7(1). 1–29. doi:10.5334/labphon.17. Renwick, Margaret E. L. & Marianna Nadeu. 2018. A Survey of Phonological Mid Vowel Intuitions in Central Catalan. Language and Speech 1–41. doi:10.1177/0023830917749275. Wedel, Andrew, Scott Jackson & Abby Kaplan. 2013. Functional load and the lexicon: Evidence that syntactic category and frequency relationships in minimal lemma pairs predict the loss of phoneme contrasts in language change. Language and Speech 56(3). 395–417. Algunos apuntes sobre el rotacismo de /n/ en la evolución al español
Assumpció Rost Bagudanch Universitat de les Illes Balears
En el estudio de la evolución fónica del latín al romance castellano, se ha advertido que /n/, en determinados contextos, habría sufrido una disimilación respecto a otro segmento nasal próximo (habitualmente /m/), proceso que consistiría en su reducción a un elemento de tipo rótico. Esto se ha atestiguado no solo para el grupo [-MIN-], sino también para el que Meyer- Lübke (1890-1906 [1974]: 474) clasifica como /n/+C+/ne/ (en casos como sangre < SANGUĬNEM, liendre < LENDĬNEM o landre < GLANDĬNEM).
El rotacismo histórico de /n/ no es desconocido en la historia de las lenguas románicas: se produce en rumano, en los dialectos de Savoya y del Vaud, así como en otras variantes alpinas e, incluso, del Piamonte (cf. Meyer-Lübke 1890-1906 [1974]: 403-406); sin embargo, se suele dar esencialmente en posición intervocálica. Este hecho encaja con la naturaleza del fenómeno, que se relaciona con un proceso de lenición por reducción temporal (Bauer 2008, Kingston 2008, Lahoz 2015). Precisamente esto es lo que choca con la evolución que tradicionalmente se ha propuesto para los grupos antes mencionados, en que se contempla la síncopa de la vocal postónica como paso previo a la disimilación de nasales (Menéndez Pidal 1926 [1972], Pensado 1984, Ariza 2012). Esta posición ha sido discutida por trabajos recientes (Gutiérrez 2015, Ryan 2018), aunque también venía puesta en duda por algunos autores clásicos (Hanssen 1913: 64).
Esta investigación viene a dar argumentos desde la fonética experimental a favor del rotacismo como primer paso en el cambio desde los contextos tardo-latinos tomados en consideración. En efecto, siguiendo la línea de investigadores como Blake (1987), Wright (1993), Labov (1994) o Blevins (2004), se asume que los fenómenos de variación sincrónicos presentan paralelismos con los procesos de cambio diacrónicos, por lo que el estudio del habla actual puede arrojar luz sobre la evolución desde el latín. Así, partiendo de Rost (2016, 2018), se presentan evidencias acústicas que muestran que la disimilación de /n/ en una consonante rótica tiene mucho más sentido en posición intervocálica y átona que tras consonante. Para ello, por una parte, se han analizado muestras de habla espontánea en seis castellanohablantes madrileños y, por otra, se han estudiado también muestras de habla de laboratorio en tres castellanohablantes norteños. Los resultados indican que entre las variantes alofónicas de la nasal alveolar se encuentran sonidos róticos. Para determinar sus condiciones de aparición se han tenido en cuenta el acento y el contexto fonético (intervocálico, postconsonántico o postpausal).
Asimismo, interesa determinar si los oyentes pueden llegar a confundir casos de /n/ reducida con /ɾ/, que es lo que realmente habría desencadenado el cambio (Ohala 1981, 2012, Pierrehumbert 2001, 2002). Por este motivo, se ha llevado a cabo una prueba perceptiva de discriminación. Los resultados preliminares apuntan a que el reanálisis es posible.
Referencias bibliográficas
Ariza, M. 2012. Fonología y fonética históricas del español. Madrid: Arco Libros. Bauer, L. 2008. Lenition revisited, Journal of Linguistics 44, 3, 605–624. Blake, R. J. 1987. Scribal worries and sound changes in Medieval Spain. En C. Neidle y R. A. Núñez Cedeño (eds.), Studies in Romance Languages, 15-26. Dordrecht: Foris. Blevins, J. 2004. Evolutionary Phonology. The Emergence of Sound Patterns. Cambridge: Cambridge University Press. Gutiérrez, C. 2015. La evolución de las secuencias latinas [min] en español, Zeitschrift fur Romanische Philologie 13, 1, 57-93. Hanssen, F. 1913. Gramática histórica de la lengua castellana. Halle: Max Niemeyer. Kingston, J. 2008. Lenition. En L. Colantoni y J. Steele (eds.), Selected proceedings of the 3rd Conference on Laboratory Approaches to Spanish Phonology, 1-31. Sommerville, MA.: Cascadilla Press. Labov, W. 1994. Principles of Linguistic Change. Volume I: Internal Factors. Oxford: Blackwell. Lahoz, J. M. 2015. Fonética y fonología de los fenómenos de refuerzo consonántico en el seno de unidades léxicas en español. Madrid: Universidad Complutense de Madrid. Tesis doctoral. Menéndez Pidal, R. 1926 [1972]. Orígenes del español. Estado lingüístico de la Península Ibérica hasta el siglo XI. Madrid: Espasa-Calpe. Meyer-Lübke, W. 1890-1906 [1974]. Grammaire des langes romanes, vol. I. Phonétique. Marsella/Ginebra: Laffite Reprints/Slatkine Reprints. Ohala, J. 1981. The listener as a source of sound change. En C. S. Masek, R. A. Hendrick y M. F. Miller (eds.), Papers from the Parasession on Language and Behavior, 178-203. Chicago: Chicago Linguistic Society. Ohala, J. 2012. The listener as a source of sound change: An update. En D. Recasens y M. J. Solé (eds.), The Initiation of Sound Change: Perception, Production and Social Factors, 21- 35. Amsterdam: John Benjamins. Pensado, C. 1984. Cronología relativa del castellano. Salamanca: Ediciones de la Universidad de Salamanca. Pierrehumbert, J. 2001. Exemplar dynamics: Word frequency, lenition and contrast. En J. Bybee y P. Hopper (eds.), Frequency and the Emergency of Linguistic Structure, 137- 158. Amsterdam/Philadelphia: John Benjamins. Pierrehumbert, J. 2002. Word-specific phonetics. En C. Gussenhoven y N. Warner (eds.), Laboratory Phonology VII, 101-139. Berlín/Nueva York: Mouton de Gruyter. Rost, A. 2016. El extraño caso de la rótica infiltrada: reflexiones preliminares. En A. M. Fernández Planas (ed.), 53 reflexiones sobre aspectos de la fonética y otros temas de lingüística, 109-117. Barcelona: Servei de Publicacions de la Universitat de Barcelona. Rost, A. 2018. Flapping of coronal /n, l, d, s/ in Spanish: A transversal lenition process. 51st Annual Meeting of the Societas Linguistica Europaea. Tallinn (Estonia), August 29th to September 1st 2018. Ryan, J. M. 2018. The dual anomalous trajectories of the Latin sequence -MIN- into Spanish: An analysis involving both full syncope and weakening. En: Daniel Recasens y Fernando Sánchez Miret (eds.) Production and Perception Mechanisms of Sound Change, 157- 171. Múnich: Lincom. Solé, M. J. 1992. Experimental phonology: The case of rhotacism. En W. U. Dressler, H. C. Luschützky, O. E. Pfeiffer y J. R. Rennison (eds.), Phonologica 1988, 259-271. Cambridge: Cambridge University Press. Wright, R. 1993. Status quaestionis: El estudio diacrónico del español, Lingüística (ALFAL) 5, 77- 126.
“The differentiated outcomes of Classical Latin open and checked syllable structure into Spanish, Italian and Neapolitan”
John M. Ryan, University of Northern Colorado
The purpose of this study is: 1) to conduct a comparison of the distribution between open and checked syllables in the same short text among three modern Romance languages, i.e., Spanish, Italian and Neapolitan, with that of their progenitor Classical Latin (CL); and 2) to further explore the comparative effects that four historical phonological processes (or their absence) would ultimately determine the extent to which both syllable types are preserved in the modern languages. These processes are: 1) elision vs. retention of word final consonants (e.g., A-MA-TIS > It./Nap. a-ma-te ‘you love’ versus Old Sp. a-ma-des > a-máis); 2) apocope vs. retention of CL morphologically light, word-final unstressed –E (e.g., SOLE ‘sun’ > Sp. sol versus It./Nap. sole); 3) reduction vs. retention of geminate consonants (e.g., PAS-SUM ‘step’ > Sp. pa-so versus It./Nap. pas-so); and 4) movement of -S- from syllable-initial position of a consonant cluster to final position of the preceding syllable (e.g., NO-STRUM ‘our’ > Sp. nues-tro versus It. no-stro or Nap. nuo-ste.)
The first part of the presentation is a brief syllabic analysis of The Lord’s Prayer for each language of the study (each approximately fifty words). Figure 1 (on the following page) provides the percentage frequencies for the distribution between open and checked syllables both within each text overall and for word final syllables only. The most noticeable difference is the dramatic increase in open syllables among the daughter languages, particularly in word final position. The figure also suggests that in terms of overall syllable count, CL exhibits a near equal ratio of 56%/44% in open to checked syllables, while Spanish places second with 62%/38%. This contrasts with Italian and Neapolitan, both displaying a much more significant reduction in checked syllables than Spanish, Italian with a ratio of 81%/19%, and Neapolitan with 83%/17%. These relationships between CL and the daughter languages become even more dramatic when one disaggregates the data for word final syllables alone, revealing sharper contrasts for both CL and Neapolitan than were found for these languages in their texts overall, with CL at 22% open versus 78% checked, and Neapolitan at 98% open versus 2% checked. Word final syllable data for both Spanish at 57% versus 43% and Italian at 86% versus 14% displayed similar distributions with those for their texts overall.
It is no wonder that checked syllables were found to overwhelmingly dominate word final position over open syllables in the CL text, given that this was the locus of inflection as a synthetic language with many forms ending in consonants. Nor is it surprising that among the modern languages studied here Spanish would have a much higher representation of checked syllables than do Italian or Neapolitan given that Spanish, unlike these other two languages, has exhibited a much greater historical tolerance for word final consonants (e.g., preserving CL final -S throughout the verbal paradigm as well as adopting word final -s as the all-purpose plural morpheme). Although checked syllables were found to be rare in word final position in both Italian and Neapolitan, this was not true for initial or medial position, due in part to the abundance of geminates in this position.
The second part of the study, which focused on the effects of historical phonological processes on overall syllabic structure, revealed a recurrent theme that was particular to the development of Spanish from Latin, which was the complex restructuring of original CL syllabic architecture, otherwise extensively retained in Italian and Neapolitan. The analysis shows that among the three languages, Spanish appears to have been the most extensive in the historic application of all three processes of apocope, syncope, and degemination, which in turn set in motion a drastic restructuration of original CL syllable architecture. In the case of apocope, original CL syllable initial consonants became word final consonants. Syncope of post tonic vowels produced illicit combinations of internal consonants which in turn were reanalyzed as new, more licit combinations. Degemination prompted the reanalysis of previously cross syllabic contiguous consonants as single syllable initial palatals or trills. Contrastively, although apocope and syncope have also played a role to some degree for both Italian and Neapolitan as well, they were not impactful on overall CL syllabic structure as has been demonstrated for Spanish. Moreover, degemination was found to not occur at all in these languages, and in fact proactive gemination is the norm for Neapolitan.
Furthermore, per Ryan (2018), it is suggested that Modern Neapolitan and other southern Italian dialects have not gone as far as either Spanish (as alleged) in losing post-tonic Latin vowels or as Italian in retaining them, but instead has reached an intermediate stage and reduced these to schwa. Unlike Tuscan or Northern Italian, in which every vowel of a word, whether tonic or atonic, receives full articulation, Modern Neapolitan employs a system in which only tonic vowels receive their full articulation, and all remaining atonic vowels are reduced to schwa (Clivio et al 2011; Rohlfs 1966). Latin Spanish Italian Neapolitan all syllables: Figure 1. Comparative frequency open 56% 62% 81% 83% distribution of overall and word-final checked 44% 38% 19% 17% open and checked syllables in The Lord’s 100% 100% 100% 100% Prayer for Latin, Spanish, Italian final syllables: and Neapolitan open 22% 57% 86% 98% checked 78% 43% 14% 2% Select References 100% 100% 100% 100%
Azevedo, Milton (2008). Introducción a la lingüística española. 3rd edition. Pearson. Clivio, Gianrenzo, Marcel Danesi, and Sara Maida-Nicol. (2011). An Introduction to Italian Dialectology. Munich: LINCOM, 2011. Lloyd, Paul. (1987). From Latin to Spanish: Vol. 1: Historical Phonology and Morphology of the Spanish Language. Memoirs Series, American Philosophical Society. Loporcaro, Michele. (2011). “Syllable, segment and prosidy” in The Cambridge history of the Romance languages. Vol. I. Structures. Cambridge, Cambridge University Press. Maiden, Martin. (1995). A Linguistic History of Italian. London: Longman Linguistics Library Series, Routledge. Menéndez Pidal, Ramón. (1950). Orígenes del español. Estado lingüístico de la Península Ibérica hasta el siglo XI, (3rd. Ed.) Madrid. Rohlfs, Gehrard. (1966). Grammatica storica della lingua italiana e dei suoi dialetti. Fonetica. Piccola Biblioteca Einaudi. Ryan, John M. (2018). “The Dual Anomalous Trajectories of the Latin Sequence –MIN- into Spanish: An Analysis Involving both full Syncope and Weakening.” In Daniel Recasens & Fernando Sánchez-Miret (eds.), Production and Perception Mechanisms of Sound Change. LINCOM Europa Series in Theoretical Linguistics. Pp. 115-129. Teschner, Richard. (2000). Camino Oral. McGraw-Hill. Wright, Roger. (2002). A Sociophilological Study of Late Latin. Utrecht Studies in Medieval Literacy. Brepols Publishers. Belgium.
2
Elision of the definite article /-l/ in Romanian: setting out the issues Fernando Sánchez-Miret
The definite article /-l/ is usually not pronounced in present-day spoken Dacoromanian1 (lupul [ˈlupu] ‘the wolf’). Most modern descriptions point to stylistic variation, /-l/ supposedly being pronounced only in more formal styles (e.g. DOOM 2005: XCI). Historical grammars of Romanian however have paid little attention to the elision of final /l/. In the brief passages dedicated to this subject, it is always considered a phenomenon restricted to final /l/ of the article (e.g. Tiktin 1888: 447; Tiktin 1905: 88; Densusianu 1938: 168-169; Rothe 1957: 38; Nandris 1963: 141-142). In all other instances, final /l/ is said to have been maintained in Romanian (Nandris 1963: 139; Marin & Rusnac 1991: 102). It is argued that final /l/ of full words is maintained because of its semantic value. The consonant /l/ of the definite article instead has only a morphological function that can be taken over by the vowel /u/ (lup-u ‘the wolf’ vs lup ‘wolf’) or by lateral deletion giving rise to a hiatus (boul [ˈbo.u] ‘the ox’ vs bou [ˈbow] ‘ox’), so that in these circumstances /l/ elision can go its way without any morphological damage (Avram 1959; Nandris 1963). As for the phonetic conditioning of /l/ elision, it is generally attributed to a sandhi phenomenon based on the avoidance of some unusual consonant clusters (Avram 1959; Nandris 1963: 142). The history of this change has not been thoroughly described. The absence of
To begin with, the distinction between ‘normal’ and ‘solemn’ speech (e.g. DOOM 2005: XCI) is too simplistic and fails to catch, among other aspects, the cross-individual variation. More important in what refers to the history of the phenomenon, is the suspicious fact that the supposed phonetic change should have affected from the very start only the definite article but not all the other areas of the lexicon. There is of course no doubt that reduction and elision
1 In other dialects like Istroromanian and Meglenoromanian final /l/ has definitively been lost (Sala 2001: 82, s.v. articol); in Aromanian the definite article is -lu (Puşcariu 1959: 145). In this paper I will concentrate on the Dacoromanian variety of Romanian and for the sake of simplicity I will refer to it as Romanian. /l/ is the masculine (lupul ‘the wolf’) and neuter (ziarul ‘the newspaper’) singular marker of the definite article; words ending in /e/ have the form /le/, e.g. frate ‘brother’, fratele ‘the brother’. 2 Other explanations are based on external, morphological or syntactical factors. For a review of the explanations put forth by Bacinschi, Graur, Weigand, and Drăganu, cf. Avram (1959: 457-458). processes are very likely to operate on frequent content words as the definite article, but one can reasonably assume that the change of interest could also have affected other words. I will discuss some relevant examples next. As for the contextual factors, while all researchers seem to have only looked at the contextual segment following /l/, one should not forget that the elision of the alveolar lateral occurs mainly after the vowel /u/. Cf. examples of final /l/ maintained after other vowels: tatăl, dascăl, al, el. One has to bear in mind that, while the lateral in present-day Romanian is a clear variety of /l/ (Recasens 2012: 380; Popescu & Chitoran 2016). According to Lombard (1935: 109), /l/ in coda position can be slightly velarized in standard Romanian. The presence of dark /l/ has also been documented for some geographical varieties: in Maramureş, dark /l/ has been sporadically attested (Vulpe 1984: 329) and dark /l/ is typical of Ţara Oaşului (Uriţescu 1984: 392, 393-394); in both cases the dark pronunciation is attributed to Ukrainian influence. In these northern regions of Romania, /l/ vocalization is sporadically attested: albă ‘white (fem.)’ [ˈawbə], căldare ‘bucket’ [kəwˈdare], se culcă ‘(s)he is going to sleep’ [sə ˈkuwkə], mult ‘a lot’ [ˈmuwt], mulg ‘I milk’ [ˈmuwg] (Teaha 1966: 156).3 Teaha (1966) assumes that in other times [ɫ] was present in larger areas in the North of Romania and that it has been preserved in Maramureş and Ţara Oaşului owing to the contact with Ukrainian. In spite of being a salient characteristic of speakers from the Republic of Moldova, Lăzărescu (1984) or Spînu (2010) do not mention the velar pronunciation of the lateral in Moldovan. This is in contrast with Verebceanu (2002: 181), who mentions the presence of a slight velarized realization of the consonant in the Romanian part of the historical Moldova and a strongly velarized realization in the Republic of Moldova: [ɫ] in contact with back or central vowels is attested before a consonant and also in final, initial and intervocalic position. One cannot be sure whether [ɫ] has had a broader extension in the past history of Romanian. Would it be the case, /l/ elision after /u/ could be attributed to the darkness of /l/ (cf. Recasens 2014: 141). Turning now to the role played by the context after the /l/, I would expect, as many others do (cf. Lombard 1935; Avram 1959; Vasilescu & Vieru & Lamel 2014; Vasilescu et al. 2019), that a following consonant should favour the deletion process, since it is precisely in this position that one can expect gestural undershoot associated with the fast tongue tip motion and the low precision involved in constriction formation for apical /l/ (cf. Recasens 2014: 55). As a result, the apical gesture can be hidden by the gesture for a C2 in a /lC/ sequence. One can then reasonable assume that the sequence /ul#C/ would favour the /l/ elision process: in this position /l/ can be incompletely articulated and the listener may attribute the acoustic effect of gestural undershoot to the articulatorily similar contextual vowel (Recasens 2014: 140, 155). The relevance of the vocalic context can also be deduced from the history of /-l/ of the genitival article al: this consonant has the same origin as the /-l/ of the definite article (< Lat. ĬLLE) and we find examples of a instead of al since early texts until modern times. However, these cases seem not to be examples of elision but of the bare preposition AD > a, since the etymological origin of the genitival article is AD + ĬLLE (Vasiliu & Ionescu- Ruxăndoiu 1986: 151-152). We have then to ask ourselves why is it the case that the /l/ of the genitival article is not elided. The phonotactic hypothesis cannot explain this difference between /al/ and /ul/. The prediction regarding the impact of the word-initial consonant following /l/ on /l/ deletion, which has been argued to occur in Lombard (1935), Avram (1959), Vasilescu &
3 Cf. the remark in the legend to ALR II SN vol. IV (1965), map 1213 alb ca zăpada ‘white as the snow’, point 346 [aɫb]: “Un informator ocazional, care este intelectual, spune că se zice şi au̯ b” [‘An occasional informant, who has a higher education diploma, affirms that the pronunciation au̯ b also exists’]. Vieru & Lamel (2014) and Vasilescu et al. (2019), was not confirmed by my own data on modern Romanian speakers (Sánchez Miret 2017). This results led me to check more carefully the data presented by Lombard and Avram. The data adduced by Lombard (1935), which are numerically very scant, turned out to be non-significant according to a Chi-Square test. A Chi-Square test run on Avram’s data (see Table 1) after grouping the CC and C contextual conditions under a single C(C) contextual category, yielded a significant difference among the variables subject to analysis (χ2 (2, N = 177) = 7.477, p = .024).
context -/l/ elision CC 5 29%12 71% C 39 45% 47 55% _## 33 59% 23 41% V 13 72% 5 28%
Table 1. Percentages of /l/ maintenance vs elision for each contextual condition in Avram’s (1959) data.
Avram’s data (177 cases) come from three volumes of the ALR II (ALR II 1940; ALR II SN vol. I 1956; ALR II SN vol. II 1956) and Petrovici (1943). To this search I have added data from the ALR II volumes published after 1959 (ALR II SN vol. III 1961; ALR II SN vol. IV 1965; ALR II SN vol. V 1966; ALR II SN vol. VI 1969).4 I also have searched the folkloric texts published by Petrovici (1939).5 Overall, I have come up with a corpus of 444 cases.
context -/l/ elision C 81 31,9% 173 68,1% _## 69 46,3% 80 53,7% V 28 68,3% 13 31,7%
Table 2. Percetages of /l/ maintenance vs elision for each contextual condition in my own corpus.
I have calculated the Chi-Square for my 444 cases (see Table 2). The statistical analysis result also shows that there is a significant relationship between the variables analysed (χ2 (2, N = 444) = 23.088, p < .001). However, my data run against Avram’s conclusions in one respect: contrary to the Avram’s data, in my corpus, /l/ elision (80 cases) occurs more frequently than /l/ conservation (69 cases) before pause. A closer analysis of these 444 examples will be presented at the workshop. Other relevant variables (speaker, type of speech) will also be taken into account.
References
ALR II = Petrovici, Emil (1940). Atlasul lingvistic romîn. Partea II (ALR II). Vol. I: A. Corpul omenesc, boale (şi termeni înrudiţi). B. Familia, naşterea, copilăria, nunta, moartea, viaţa religioasă, sărbători. C. Casa, acareturile, curtea, focul, mobilierul, vase, scule. Sibiu, Muzeul Limbii Române.
4 The last volume published (ALR II SN vol. VII 1972) is dedicated to the verb and contains no word of interest for my analysis. 5 The texts from Scarişoara published in Petrovici (1943) are only a selection of those in Petrovici (1939). ALR II SN vol. I = Petrovici, Emil (1956). Atlasul lingvistic romîn. Serie nouă. Vol. I. A. Agricultură. B. Morărit. C. Grădinărit. D. Pomărit. E. Viticultură. F. Cînepa. G. Albinărit. Bucureşti, Editura Academiei. ALR II SN vol. II = Petrovici, Emil (1956). Atlasul lingvistic romîn. Serie nouă. Vol. II. A. Creşterea vitelor. B. Carul, căruţa, sania. C. Păsări de curte. D. Păstorit. E. Lîna, torsul, ţesutul. F. Meserii. G. Pădurărit. Bucureşti, Editura Academiei. ALR II SN vol. III = Petrovici, Emil (1961). Atlasul lingvistic romîn. Serie nouă. Vol. III. A. Plante. B. Cîinele, pisica. C. Animale sălbatice. D. Păsări sălbatice. E. Vînătoare F. Tîrîtoare, amfibii. G. Pescuit. H. Insecte. I. Timpul. J. Configuraţia terenului. K. Ape, umiditate, navigaţie. L. Comunicaţii. M. Naţiuni, categorii sociale, administraţie. Bucureşti, Editura Academiei. ALR II SN vol. IV = Petrovici, Emil (1965). Atlasul lingvistic romîn. Serie nouă. Vol. IV. A. Şcoala. B. Armata. C. Conflicte, justiţie. D. Bani, comerţ. E. Măsuri, greutăţi. F. Bucătărie, mîncăruri, băuturi, alimente, coloniale. G. Tutun. H. Îmbrăcăminte, încălţăminte, farduri, podoabe, lucru de mînă. I. Culori. J. Spălat, călcat, igienă. Bucureşti, Editura Academiei. ALR II SN vol. V = Petrovici, Emil (1966). Atlasul lingvistic român. Serie nouă. Vol. V. A. Fire, caracter, sentimente. B. Petreceri, distracţii. C. Muzică. D. Jocuri de copii. E. Salutul. F. Varia. G. Nume de persoană. H. Superstiţii. Bucureşti, Editura Academiei. ALR II SN vol. VI = Petrovici, Emil (1969). Atlasul lingvistic român. Serie nouă. Vol. VI. A. Substantiv. B. Articol. C. Adjectiv. D. Pronume. E. Numeral. F. Adverb. G. Prepoziţie. H. Conjuncţie. Bucureşti, Editura Academiei. ALR II SN vol. VII = Institutul de Lingvistică şi Istorie Literară al Academiei Republicii Socialiste România Filiala din Cluj (1972). Atlasul lingvistic român. Serie nouă. Vol. VII. Verbul. Bucureşti, Editura Academiei. Avram, Andrei (1959). “Despre cauzele dispariţiei lui l final - articol hotărît”. Studii şi Cercetări Lingvistice 10: 457-463. Densusianu, Ovid (1938). Histoire de la langue roumaine. Tome II. Le seizième siècle. Paris, Leroux. DOOM = Academia Română - Institutul de Lingvistică "Iorgu Iordan - Al. Rosetti" (2005). Dicţionarul ortografic, ortoepic şi morfologic al limbii române. Bucureşti, Univers Enciclopedic. Frâncu, C. (1997). “Morfologia (1640-1780)”. In: Gheţie, Ion, ed., Istoria limbii române literare. Epoca veche (1532-1780). Bucureşti, Editura Academiei Române: 319-346. Lăzărescu, Paul (1984). “Dacoromâna. Subdialectul moldovean”. In: Rusu, Valeriu, ed., Tratat de dialectologie românească. Craiova, Scrisul Românesc: 208-284. Lombard, Alf (1935). La prononciation du roumain. Uppsala, Lundequistska Bokhandeln. Marin, V. & Rusnac, G., eds. (1991). Curs de gramatică istorică a limbii române. Chişinău, Lumina. Nandris, Octave (1963). Phonétique historique du roumain. Paris, Klincksieck. Petrovici, Emil (1939). “Folklor de la Moţii din Scărişoara”. Anuarul arhivei de folklor 5: 111-175. Petrovici, Emil (1943). Texte dialectale: Suplement la Atlasul Linguistic Român II (ALRT II). SibiuLeipzig/Leipzig, Muzeul Limbii Române/Harrassowitz. Popescu, Anisia & Chitoran, Ioana (2016). “Allophonie et position dans la syllabe: Indices acoustiques pour les consonnes latérales”. In: Actes de la conférence conjointe JEP- TALN-RECITAL 2016, volume 1: JEP: 55-63. (https://jep- taln2016.limsi.fr/actes/Actes%20JTR-2016/V01-JEP.pdf) Puşcariu, Sextil (1959). Limba Română II. Rostirea. Bucureşti, Editura Academiei. Recasens, Daniel (2012). “A cross-language acoustic study of initial and final allophones of /l/”. Speech Communication 54: 368-383. Recasens, Daniel (2014). Coarticulation and sound change in Romance. Current Issues in Linguistic Theory, 329. Amsterdam, Benjamins. Rothe, Wolfgang (1957). Einführung in die historische Laut- und Formenlehre des Rumänischen. Halle a. Saale, Niemeyer. Sala, Marius, ed. (2001). Enciclopedia limbii române. Bucureşti, Univers Enciclopedic. [Coordonator: Marius Sala] Sánchez Miret, Fernando (2017). “Elisión de la /l/ del artículo en dos hablantes de rumano”. In: Dragomirescu, Adina et al., eds., Sintaxa ca mod de a fi. Omagiu doamnei profesoare Gabriela Pană Dindelegan, la anivesare. Bucureşti, Editura Universităţii: 361-369. Scriban, Aŭgust (1925). Gramatica limbiĭ romîneştĭ (morfologia) petru folosinţa tuturor. Ĭaşĭ, Viaţa romînească. Spînu, Stela (2010). Graiurile româneşti din nord-estul Republicii Moldova. Chişinău, Acad. de Ştiinţe a Rep. Moldova. Teaha, Teofil (1966). “Despre ɫ (velar) în graiurile româneşti”. Studii şi Cercetări Lingvistice 17: 285-292. Tiktin, Hariton (1888). “Die rumänische Sprache”. In: Gröber, Gustav, ed., Grundriss der romanischen Philologie. Strassburg, Trübner: 438-460. Tiktin, Hariton (1905). Rumänisches Elementarbuch. Sammlung romanischer Elementarbücher: Reihe 1, Grammatiken, 6. Heidelberg, Winter. Uriţescu, Dorin (1984). “Dacoromâna. Graiurile din Ţara Oaşului”. In: Rusu, Valeriu, ed., Tratat de dialectologie românească. Craiova, Scrisul Românesc: 390-399. Vasilescu, I. & Vieru, B. & Lamel, L. (2014). “Exploring pronunciation variants for Romanian speech-to-text transcription”. In: Proceedings of the 4th International Workshop on Spoken Language Technologies for Underresourced Languages SLTU- 2014. – St. Petersburg, Russia, 2014. St. Petersburg, Speech and Multimodal Interfaces Laboratory of SPIIRAS: 161-168. (http://mica.edu.vn/sltu2014/proceedings/23.pdf) Vasilescu, Ioana & Chitoran, Ioana & Vieru, Bianca & Adda-Decker, Martine & Candea, Maria & Lamel, Lori & Niculescu, Oana (2019). “Studying variation in Romanian: deletion of the definite article -l in continuous speech”. Linguistics Vanguard 5/1. (https://doi.org/10.1515/lingvan-2017-0046) Vasiliu, Em. & Ionescu-Ruxăndoiu, Liliana (1986). Limba română în secolele al XII-lea - al XV-lea (fonetică - fonologie - gramatică). Bucureşti, Universitatea din Bucureşti. Verebceanu, Galaction (2002). “VIII. Fenomene fonetice şi fonologice”. In: Bochmann, Klaus & Dumbrava, Vasile, eds., Limba Română vorbită în Moldova istorică. Vol. 1. Leipzig, Leipziger Universität-Verlag: 181-194. Vulpe, Magdalena (1984). “Dacoromâna. Subdialectul maramureşean”. In: Rusu, Valeriu, ed., Tratat de dialectologie românească. Craiova, Scrisul Românesc: 320-354.
External sandhi processes in vowel sequences in Brazilian Portuguese and peninsular Spanish: change and variation
João Paulo Moraes Lima dos Santos
In this study I examine sandhi processes in vowel sequences across word boundaries in Brazilian Portuguese (varieties of Maceió and Fortaleza) and peninsular Spanish (varieties of Salamanca and Ávila). The purpose of this paper is to describe the set of sound change and variation that occurs in these contexts.
Regarding Brazilian Portuguese, many studies have argued that there was a process of /e/ and /o/ raising in an unstressed final position, with /e/ and /o/ becoming /i/ and /u/ (Bisol, 2001, Vieira, 2002). This shift has caused a change from hiatus sequences [ea], [oa] > [ia], [ua] to the diphthongs [ja] and [wa]. Most of my data contain diphthongs (65%), but in a few contexts, the elision of this first vowel is also observed.
Concerning peninsular Spanish, on the other hand, Hualde et al (2008) have found that in nonhigh unstressed vowel sequences, such as /ae/ or /eo/, mid vowels can undergo a reduction and become nonsyllabic vowels. In fact, my results show that the durations of nonhigh unstressed sequences and diphthongs are very similar (both with values between 55ms and 80ms). However, the real tendency is elision: In 74% of the data, I detected the elimination of the mid vowel. In context of two mid vowels, the first vowel is deleted.
Comparing the processes between these two languages, we observe the substantial presence of diphthongs (e.g., ‘mund[wa]zul’) and minimal first vowel elision in unstressed vowel sequences in Brazilian Portuguese. Peninsular Spanish prefers elision, but features significantly fewer diphthongs (e.g., ‘m[ja]buelo’) and nonsyllabic mid vowels (e.g., l[oe]stá). In most of the data, if one vowel is stressed or in the context of stressed vowels, a hiatus is produced in both languages.
References
BISOL, Leda. 2001. Introdução a estudos de fonologia do português brasileiro. 3ª Ed. Porto Alegre: EDIPUCRS.
HUALDE, José Ignacio, SIMONET, Miquel, & TORREIRA, Francisco. Postlexical contraction of nonhigh vowels in Spanish Lingua 118 2008, p. 1906-1925.
VIEIRA, Maria José Blaskovski. 2002. As vogais médias postônicas: uma análise variacionista. In: BISOL, Leda; BRESCANCINI, Cláudia Regina (orgs.) Fonologia e variação: recortes do português brasileiro. Porto Alegre, EDIPUCRS: 127-159.