/s/-Lenition and the Preservation of Plurality inModern Occitan1

Laurel MacKenzie*

University of Manchester

This paper examines the variableweakening and deletion of /s/ in the Languedocian variety ofModernOccitan,with particularattention to how it has affected the system of plural marking in noun phrases. Using data from linguistic atlases,I demonstrate that /s/-lenition in this variety involves a stage of vocalization to [j]. I find that, where /s/ on the definite articlehas vocalized to [j], the immediately-preceding vowel of the definite article has undergone concomitant raising to [e]. This raisingappears to preserve the difference between singular and plural despite the plural’s weakening /s/. I argue that these resultssupport Labov’s (1994) hypothesis that the meaning of a weakening element may be transferred to a stable, co-occurrent one.

1. Introduction

This paper examines the variable weakening and dele-tion of /s/ in the Languedocian variety of ModernOccitan, with particular attention to how this hasaffected the system of plural marking in noun phrases.As is common in other Romance languages, the dialectof Occitan under study marks plurality on a nounphrase by affixing an -s plural morpheme to the article,the noun, and any adjectives, as in (1).

(1) [la-s ˈkãmbo-s peˈlydo-s]DEF.FEM-PL leg-PL hairy-PL‘hairy legs’

However, as in Spanish and Portuguese (see, e.g.,Guy, 1981; Hernández-Campoy and Trudgill, 2002;Poplack, 1980; Ranson, 1991; Terrell, 1977), the /s/of the plural can variably undergo weakening anddeletion. This paper uses data from dialect atlases(Ravier, 1978, 1982, 1986, 1993) to examine what hap-pens to the singular-plural distinction in Occitan when/s/ is lost: is there a tendency to preserve meaning? Bywhat route?

I demonstrate that /s/-lenition in Languedocianinvolves a stage of vocalization to [j]. When the /s/ pluralmarking on the definite article vocalizes, the article vowelis raised to [e], preserving the difference between singularand plural despite the plural’s weakening /s/. I arguethat these results support Labov’s (1994) hypothesis thatthe meaning of a weakening element may be transferredto a stable, co-occurrent one.

This paper is structured as follows. Section 2 pro-vides background information on the language under

study, including previous work on /s/-loss in Occitan,and the source of the data. Section 3 examines theweakening of the plural -smarker on nouns and definitearticles, followed by an examination of concomitanteffects on the vowel of the definite article. I concludewith a discussion of the role of meaning preservation inlanguage change (Section 4).

2. Background

2.1. Language and Data

Occitan is the generic name used to refer to a collectionof varieties, distinct from French, spoken throughoutSouthern France. Occitan is also known as Langue d’Ocand, historically, Provençal, though today this termrefers properly only to a particular Occitan variety, thatspoken in Provence. The variety of Occitan examined inthis paper is Languedocian.

Ethnologue places both Occitan and French in theGallo-Iberian branch of Romance languages (Lewis,Simons & Fennig, 2014); the two followed differentpaths of development beginning in the 5th centurywith the invasion of Gaul by foreign tribes. NorthernFrance sustained invasions by Britons, Saxons, andparticularly Franks, who lent their name to the nation,while Southwest France, with the exception of Gascony,which was invaded by the Basques, received minimaloutside influence. The Latin language and Romanculture thus remained more intact in Southernthan Northern France, given the different levels ofoccupation in each half of the country. By the 12thcentury, the vernaculars of Northern and SouthernFrance were considered different languages, known asthe Langue d’Oïl in the North and the Langue d’Oc inthe South, after each language’s affirmative particle(Pope, 1952).

1An earlier version of this paper appeared in Penn Working Papers inLinguistics 16.2.*Address for correspondence: Laurel MacKenzie, School of Arts,

Languages and Cultures, University of Manchester, Oxford Road,Manchester M13 9PL, United Kingdom. + 44 (0)161 275 3152.(Email: laurel.mackenzie@manchester.ac.uk)

Journal of Linguistic Geography (2014) 2, 59–73. © Cambridge University Press 2014 ORIGINAL RESEARCHdoi:10.1017/jlg.2014.7

mailto:laurel.mackenzie@manchester.ac.uk

Anglade ([1921] 1977) placed the number of speakersof Occitan at 12 or 14 million; the 1999 French censusfound fewer than twomillion people who said that theirparents had spoken to them at least occasionally inOccitan in their childhood, and only 250,000 who hadspoken it to their own children (Héran, Filhon, andDeprez, 2002). Ethnologue places the number ofspeakers in France at only 110,000 (Lewis et al., 2014).As a reviewer observes, the difficulties inherent inpolling speakers on their use of minority languages—particularly in France, where minority languageshave been aggressively marginalized for centuries(Schiffman, 1996)—mean that we should interpret thesenumbers with some caution, in case some speakers ofOccitan were reluctant to reveal themselves to census-takers. Still, sources agree in recognizing the languageas endangered (Lewis et al., 2014; Moseley, 2010).

The data used in this paper come from the Atlaslinguistique et ethnographique du Languedoc occidental

(abbreviated henceforth as ALLOc; Ravier, 1978, 1982,1986, 1993). The area sampled in ALLOc is bordered bythe Dordogne river to the north and the Garonne riverto the west; the eastern boundary of the sample wasdetermined based on the sample sites used in otherlinguistic atlases (the Atlas linguistique et ethnographiquedu Massif Central, the Atlas linguistique du Languedocoriental) as well as topographical, cultural, and ethno-graphic considerations (for instance, to avoid dividingwhat was felt by natives to be a coherent region).The sampled area encompasses, in whole or inpart, the present-day departments of Ariège, Aude,Haute-Garonne, Tarn, Tarn-et-Garonne, Lot-et-Garonne,Aveyron, Lot, Gironde, Dordogne, and (with one pointrepresented only) Corrèze. It includes the major city ofToulouse but not that of Bordeaux. Data were gatheredbetween 1967 and 1972, collected from 131 localities15–17 km (approx. 9–11 mi) apart, as plotted in Map 1.(Subsequent plots will zoom in on the region under study.)

Map 1. The 131 localities sampled in ALLOc.

60 L. MacKenzie

Approximately 1200 lexical items were elicited forthe atlas, with each lexical item elicited once per locality.73% of the 265 primary informants were aged 60 orolder, and 72%were male. Manyworked in agriculturalprofessions. This is, of course, reminiscent of the ‘non-mobile, older, rural males’ (NORMs) sought out asinformants in many dialectological studies (Chambers& Trudgill, 1998:29), though the atlas gives no indica-tion of whether this was a deliberate choice on the partof the researchers or an inevitability due to a difficultyin finding Occitan speakers in more urban areas.Indeed, a respondent from the city of Toulouse wasincluded in the sample, as were women and youngpeople (though these non-NORM informants areoften identified as having been family members of theolder males).

To elicit forms, researchers provided a French wordand asked the informant for the Occitan translation,so all informants had some level of bilingualism.Data were collected by semantic category and werephonetically transcribed.

2.2. Linguistic Background

We can see a historical parallel to the /s/-lenition ofLanguedocian Occitan in a similar process that occurredin the evolution of French. Old French had two cases, anominative and an oblique; the oblique case, whichmarked plural by affixing /s/ to articles and nouns,was the source of the Modern French single-casesystem. Preconsonantal /s/ was lost by the 12th or13th century, and /s/ before pause by the 17th century,which led to the system French has today, in which anorthographic < s> is attached to plural nouns andarticles but is not pronounced in the vast majority ofcontexts. Instead, plurality in Modern French is con-veyed by another mechanism: change in the vowel ofthe definite article from singular /la/ (feminine) or /lə/(masculine) to plural /le/ (both genders). The onlyplace where /s/ still manifests itself is in the process ofliaison before a vowel: for instance, the vowel-initialnoun l’heure [lœʁ] ‘the hour’ has as its plural les heures[lez œʁ]. In the absence of liaison, though, pluralmeaning for nouns that form their orthographicplural in < s> is expressed only by the vowel of thedefinite article.

The chain of events that gave rise to this plural-conveying vowel alternation is not known. Seklaoui(1989:8) refers to the /e/ in the plural definite article asthe result of “an obscure phonetic evolution.” Spence(1976) proposes that its source is as follows: The vowelsof the Vulgar Latin plural definite articles los and laswere reduced to [ə], which was then variably length-ened compensatorily after loss of the following /s/, andthis [ə] then raised to [e]. He admits, though, that there

is no clear phonetic mechanism by which this couldhave occurred, and that the need to disambiguate sin-gular from plural must have been an influencing factor.In this paper, I propose that the patterns found in therelated language of Occitan may elucidate the question.

This is not the first study of /s/-deletion and itseffects on plurality in Modern Occitan. Eckert (1969,1985) performed a similar study using data gatheredfrom the Atlas Linguistique de la France (henceforth ALF),for which fieldwork was carried out between 1897 and1901. She found evidence for two sound changes inSouthern France: One deleting the plural /s/ marker,which was restricted only to the northern half of theregion she examined, and a concomitant change of*a> [o], which was more widespread, but whichshowed grammatical conditioning. Specifically, in thenorthern half of the region under study, where plural/s/ had been deleted, the *a> [o] change had gone tocompletion in singulars only; in the area where /s/ wasretained, the *a> [o] change had gone to completion inboth singulars and plurals (as sketched in (2)). Directlybetween the /s/-deleting North and the /s/-retainingSouth, Eckert identified an area of “number dis-turbance,” in which /s/ was variably retained; thisregion was bisected by the isogloss delimiting thechange of *a to [o] in plurals. The upshot of all this isthat where the singular-plural distinction was in dangerof being lost due to /s/-deletion, the contrast wasmaintained by the simultaneous *a> [o] change.

Region of /s/-deletion: Region of /s/-retention:

(2) *a > [o] in singulars onlysg. lo bello pullo pl. labella pulla ‘the beautifulhen(s)’

*a > [o] in both pluralsand singulars sg. lobello pullo pl. los bellospullos ‘the beautifulhen(s)’

The data provided in ALLOc were collected 70 yearsafter Eckert’s ALF data, and reveal that /s/-deletionhas continued to spread, even into the area where the*a > [o] change went to completion in both singularsand plurals. Floricic (2010), Barra-Jover (2012), andSauzet (2012) all use the data in ALLOc (or as-yetunpublished data collected as part of theALLOc survey)to investigate the status of plural marking in Langue-docian in the absence of this “recourse” of final vowelquality. All three authors identify a range of phonolo-gical realizations of plurality, which I corroboratebelow. The present paper additionally expands on theseworks by providing a comprehensive geographicalaccount of the situation in Languedocian, examining therealization of plural marking on multiple lexical itemsacross the entire region sampled in the atlas. Doing so

/s/-Lenition and Plurality in Modern Occitan 61

allows us not only to examine the synchronic state ofaffairs, but, in comparing neighboring varieties, tohypothesize as to its diachronic development.

3. Data

The findings presented in this section are drawn fromthe realization of plural on the definite article and thenoun of eight lexical items.1 These eight items wereselected based on two criteria: For the majority oflocalities surveyed, they (a) are of feminine gender and(b) contain a noun beginning with a voiceless stop.These restrictions were made to control for processesindependent of /s/-lenition which are known to affectthe phonological realization of the definite article. Forinstance, as in other Romance languages, Languedocian

Occitan marks noun gender in the vowel of the definitearticle, so nouns were restricted to a single gender (here,the feminine) in order to ensure that all would have thesame definite article. Additionally, because the initialsegment of a Languedocian noun has been shown toinduce allophonic variation in the final segment ofthe definite article that precedes it (Barra-Jover, 2012;Ronjat, 1932; Sauzet, 2012), the manner and voicing ofthe initial consonant were restricted as a control (here,to voiceless stops only).

Any data point in which the noun was not feminine,not voiceless stop-initial, or not plural (due to, forinstance, a location having a collective singular term foran item) was omitted from this study, and is repre-sented in the maps below as an ‘N/A’ point. When anALLOc informant provided two possible realizations for

Ø (deletion) [j] (vocalization) [s] (retention) N/A

Map 2. Realization of /s/ on the noun of les traces ‘animal tracks’.

62 L. MacKenzie

a locality, the first of these was selected for analysis,based on the remark inALLOc’s front material that “in avery general fashion, the order of forms on the map isthe order in which they were provided by the informant”(Ravier, 1978:xiv, translation mine).

Findings below will frequently be exemplified with asingle lexical item, but reference will be made to thatitem’s typicality vis-à-vis the other seven items examined.

3.1. Realization of /s/ on the Noun

Map 2 plots the realization of /s/ on the noun only,with one point for each location surveyed in ALLOc,for the lexical item les traces ‘animal tracks’ (ALLOc’smap 352). Points represented with a blue dot show no

marker of plurality on the noun, e.g. [ˈpjado] (point24.30), which I term ‘deletion.’ Points represented witha yellow dot are those in which /s/ on the noun sur-faces as [j], e.g. [peˈnadɔj] (point 46.17), which I term‘vocalization.’2 Points represented with a red dot arethose in which /s/ on the noun surfaces as [s] (or, lessfrequently, some other fricative with a buccal articulation,such as [ʃ ç x]), e.g. [peˈzados] (point 81.05), which I term‘retention.’3 Points represented with a black ‘x’ were dis-carded as not applicable for the reasons elucidated above(i.e., the informant’s response was not feminine, notplural, or did not contain a voiceless stop-initial noun), orbecause the informant was unable to provide a response.4

Map 2 reveals that vocalization of /s/ to [j], whilenot particularly prevalent in the area under study,

Ø (deletion)

100% 40%

[j] (vocalization)

[s] (retention)

Map 3. Percent realization of most common variant of /s/ on the noun across eight lexical items. More saturated colors indicatea higher representation of that variant across the eight items.

/s/-Lenition and Plurality in Modern Occitan 63

is well-represented in a tight cluster of localities in thenortheast sector. Vocalization is not attested as a stageof /s/-loss in Ferguson’s (1990) typology of [s] > [h]changes, and likewise is not mentioned as a variant inthose synchronic studies of /s/-weakening in Spanishand Portuguese cited earlier. In Occitan, it is frequentlyidentified as an allophonic variant of /s/which appearsbefore voiced consonants (Barra-Jover, 2012; Ronjat,1932; Wheeler, 1988), but the fact that the data points inMap 2 have been restricted to only those in which thearticle precedes a voiceless stop-initial nounmeans that[j] must be a stage on the /s/-weakening trajectory.Indeed, Seklaoui (1989) does list vocalization as a pos-sible, though rare, outcome of /s/-weakening, citingexamples from Italian as well as Occitan.

To show that the patterns of /s/-realization depictedin Map 2 are not specific to the lexical item les traces,I calculated the percentage of each variant of /s/ foreach locality across all eight lexical items examined. Themost common variant for each locality is plotted inMap 3, with a color saturation indicating how commonthat variant was across the eight lexical items (thedarker the color, themorewell-represented the variant).For instance, a locality for which [s] was retained onthe noun in all eight lexical items examined will beplotted as a dark red dot; a locality which had retentionof [s] for one item, vocalization to [j] for four items, anddeletion of /s/ for three items will be plotted as a lightyellow dot (indicating that [j] was the most commonvariant across the eight items, but that it was only

Ø (deletion) [p] (gemination) [h] (aspiration)

N/A

[j] (vocalization)

[s] (retention)

Map 4. Realization of /s/ on the article of les traces.

64 L. MacKenzie

represented in 50% of the observed forms).5 As in Map 2,buccal fricatives other than [s] (e.g. [ʃ ç x]) have beenplotted as instances of retention. Map 3 indicates that, forall but a few border towns, the pattern demonstrated byles traces holds for the other lexical items examined.

Map 3 allows us to identify the northwest sector ofthe region under study as being the most advanced in/s/-weakening (having no overt plural marking on itsnouns), the central/southern portion as being the mostconservative (having not yet begun to weaken /s/ inany way), and the northeast sector as being somewherein between (having lenited its realization of /s/ to [j]but not gone so far as to delete it). These three realizationscould be viewed as points on a single trajectory (with /s/first weakening to [j] and then deleting entirely);

alternatively, there could be two different trajectoriesof /s/-lenition, one in which /s/ deletes with no inter-mediate stage of [j], and another in which /s/ vocalizes(potentially to delete later on). The present data don’tobviously argue for one or the other, and the existenceof a unified trajectory of /s/-lenition is not crucial to thepresent study, so I remain agnostic on this point.

3.2. Realization of /s/ on the Article

Map 4 shows the realization of /s/ on the article of lestraces. In addition to retained (e.g. [las peˈzados], point81.05), vocalized (e.g. [lej ˈpjado], point 47.02), anddeleted6 (e.g. [la: pjɔˈda], point 24.10) variants—plottedas in the previous maps—aspiration of /s/ to [h]

[j] (vocalization)

100% 40%

Ø (deletion)

gemination

[h] (aspiration)

[s] (retention)

Map 5. Percent realization of most common variant of /s/ on the article across eight lexical items. More saturated colors indicatea higher representation of that variant across the eight items.

/s/-Lenition and Plurality in Modern Occitan 65

(e.g. [lah ˈpjados], point 82.12), plotted in green, isobserved; /s/ may also fully assimilate to the followingconsonant (e.g. [lap ˈpjado], point 47.10), resulting in ageminate (plotted in purple).

Map 4 shows that the southern half of the region understudy again behaves conservatively, with [s] retained onthe article as it was on the noun. Also, more generally, /s/on the article has not advanced as far in the weakeningprocess as had /s/ on the noun, with only one pointshowing total deletion of /s/. More advanced weakeningon nouns than articles has been similarly found in studiesof synchronic /s/-lenition in Spanish and Portuguese(Cedergren, 1973; Guy, 1981; Poplack, 1980).

Map 5 shows the same type of plot asMap 3, with themost attested variant across the eight lexical itemsplotted in a color saturation indicative of how commonit is. There is somewhatmore inter-item inconsistency inthe realization of article /s/ than there was for noun

/s/, as indicated by the greater number of lighter-colored points inMap 5 than inMap 3; indeed, themeanproportion of use of the most common variant is 0.977for noun /s/ and 0.891 for article /s/. Still, the patternevident in les traces is largely shared by the other lexicalitems under examination.

Again, the northeast, northwest, and southernregions show differing realizations, but again, thereis no obvious indication of whether they representsuccessive stages on a continuum of [s] > [h] > [j], ordistinct paths of lenition ([s] > [h]; [s] > [j]).

3.3. Article Vowel Quality

The final factor to examine is the quality of the vowel inthe definite article. As outlined in Section 2.2, definitearticle vowel quality plays a crucial role in disambiguatingsingular from plural nouns in the related languageof French.

Map 6. Vowel quality of the singular definite article.

66 L. MacKenzie

Map 6 shows the vowel quality of the singular definitearticle in the region under study (based on ALLOc’smap 1), revealing that the region can be bisected into a [lɔ]region in the northeast and a [la] region in the southwest.The default assumption is that this vowel quality will bemaintained in the plural, as it is in traditional dialects ofOccitan (Wheeler, 1988) and in other Romance languages.

Map 7 superimposes isoglosses indicating theobserved vowel quality of the plural definite article onthe plot from Map 4 depicting the realization of /s/ onthe article of les traces. The black isogloss reproduces the[lɔ]/[la] split in the singular definite articlefrom Map 6. Map 7 reveals that this same split ismaintained throughout much of the region in the plural

article as well, with the crucial exception of the north-west sector, where the vowel of the article is, excep-tionally, [e]. This region of article [e] coincides almostperfectly with the region in which /s/ on the article hasvocalized to [j]. The observed vowel raising is thusalmost certainly due to coarticulation of the vowel [a]and the following palatal glide.

Further data reveal, however, that vowel raising inLanguedocian plural articles is not purely a mechanicalprocess. Maps 9 and 10 present data for the lexical itemles fleurs du vin ‘yeasts that develop on the surface ofwine’ (map 754 in ALLOc). Map 8 shows that the mani-festation of /s/ on the noun is nearly identical to what itwas for les traces in Map 2.

[ ][e]

[j] (vocalization)

[s] (retention)

Ø (deletion) [p] (gemination) [h] (aspiration)

vowel

consonant

fem. sg. la

N/A

[a]

fem. sg. l

Map 7. Vowel quality of the plural definite article (isoglosses) and realization of /s/ on the article of les traces (points). Blackisogloss marks [lɔ]/[la] split in the singular definite article.

/s/-Lenition and Plurality in Modern Occitan 67

By contrast, Map 9 reveals that the realization of /s/on the article of les fleurs du vin differs in a crucial wayfrom that of article /s/ for les traces (Map 4): Much of thesouthern sector of the region, which retained [s] in thearticle of les traces, vocalizes /s/ in the article of les fleursdu vin.

This southern strip of vocalization has a phonologicalsource: It corresponds almost perfectly with the regionin which the noun for this lexical item begins with [f]rather than [k] (as it does elsewhere in the region). Thisis thus an instance of the allophonic /s/-vocalizationmentioned earlier which has heretofore been omittedfrom analysis. The area in which the noun is fricative-initial is surrounded by the green isogloss in Map 10; itcorresponds well to the southern area which vocalizes/s/ on the article to [j].

But unlike the vocalization depicted in Map 7, thisallophonic vocalization does not trigger vowel raising.Map 11 shows a clear difference between /s/-vocalizationin the northwest sector and /s/-vocalization in the south:In the south, vocalization has no accompanying vowelchange from [a] to [e].

4. Analysis

I propose that these findings are interpretable viaLabov’s (1994) observations on the maintenance ofmeaning in systems in which a grammatical marker isbeing lost. Specifically, Labov proposes the following:

When an element of the system is found toco-occur frequently with the signal being deleted,it is increasingly eligible to assume the distinctive

[j] (vocalization) [s] (retention) N/AØ (deletion)

Map 8. Realization of /s/ on the noun of les fleurs du vin ‘yeasts that develop on the surface of wine’.

68 L. MacKenzie

feature representing the semantic feature inquestion (Labov, 1994:596).

Labov cites as an example of this the loss of thenegative marker ne in French, which has been all butreplaced with the particle pas. He proposes that thissituation has its source in the following development.Given that ne was involved in a process of weakeningand deletion and pas was not strictly required for anegative sentence, in the history of French, sentenceswould inevitably surface which were intended by aspeaker as negative but which bore no overt negativemarker. These sentences were ambiguous betweennegative and positive (Labov’s “unsupported zeroes,”1994:589), and some learners may have misinterpreted

them as positive. Over time, as learners attemptedto match their percentage of ne use to that of thenegative sentences they received in their input, theunambiguously-negative input they received becamemore and more skewed toward tokens containing pas,since those that didn’t contain pas were prone to befactored out as positives. This eventually resulted in pastaking over as the negative marker, as the loss of neeventually went to completion.

If an analogous process occurred in Languedocian,we could explain the differing article vowel qualitiesbetween the northwest sector and the south in cases likeles fleurs du vin, where both regions vocalize /s/ on thearticle but only the northwest shows concomitant vowelraising. In the south, /s/ is consistently retained on the

[j] (vocalization)

[s] (retention)

Ø (deletion) [k] (gemination) [h] (aspiration)

N/A

Map 9. Realization of /s/ on the article of les fleurs du vin.

/s/-Lenition and Plurality in Modern Occitan 69

noun (Map 8), meaning that there is no possibility of aplural form being mistaken for a singular, even whenthe /s/ of its article has weakened to [j]. In the north-west, by contrast, the loss of /s/ on the nounmeans thatthe singular comes to bear a close resemblance to theplural: Only the [j] on the article separates them. If this[j] were itself to be variably deleted, singular-pluralambiguity would result and those tokens in which thevowel had not raised to [e] could be misclassified assingulars. At the same time, a variably retained [j]would trigger vowel raising, meaning that the numberof forms unambiguously identified as plurals couldeventually shift to those that bore [e] in their article,despite having no consonantal plural marking. In the

south, where there are no ambiguous forms, learnerswould not have any more evidence of plurality fromforms in which the vowel was raised than from forms inwhich it wasn’t. As a result, any coarticulation-inducedraising would fail to be cemented as part of the lin-guistic system, instead remaining merely a phoneticvariant of the unraised vowel.7

This analysis crucially depends on there beingambiguous forms in the northwest to bemisrepresentedas singular. I have assumed here that such forms arisewhen [j] is variably deleted, but there is no evidencefrom ALLOc of such [j]-deletion actually occurring. Thisis the downside of working with linguistic atlases: wehave no access to the synchronic variation present

[j] (vocalization)

[s] (retention)

Ø (deletion) [k] (gemination) [h] (aspiration)

N/A

[f]initial consonant of noun

final consonant of article

(elsewhere) [k]

Map 10. Realization of /s/ on the article of les fleurs du vin. The area encircled in green has a fricative-initial noun (elsewhere, thenoun is stop-initial).

70 L. MacKenzie

within a speaker or a community at the time the datawere collected. Nevertheless, the atlas has a clearadvantage over other linguistic methods in allowingrapid comparison between multiple regions. In thepresent study, such rapid comparison has revealed datathat appears to support the importance of ambiguous ornear-ambiguous forms for meaning preservation in theface of a process of lenition.

5. Conclusions

This paper has added data from the Languedociandialect of Occitan to the vast typology of cases of /s/-lenition in the world’s languages. The dialect atlas data

considered here have revealed that /s/-lenition inLanguedocian includes a stage of vocalization to [j],which, when it occurs on a definite article, appears totrigger raising of a preceding vowel. This raising, achange which is mechanically actuated, apparentlybecomes used to preserve plurality in regions where/s/-weakening is advanced.

This study raises a number of further questions. Did/s/-loss in French go through a stage of vocalization aswell, and is this the source of the mysterious /e/ inplural definite articles in that language? What hashappened to the realization of plural on verbs in Lan-guedocian, and can that be shown to have had anyeffect on plural marking in the noun phrase? After all,

[j] (vocalization)

[s] (retention)

Ø (deletion) [k] (gemination) [h] (aspiration)

N/A

[e]

vowel

consonant

[a][ ]

fem. sg. la

fem. sg. l

Map 11. Vowel quality of the plural definite article (isoglosses) and realization of /s/ on the article of les fleurs du vin (points).Black isogloss marks [lɔ]/[la] split in the singular definite article.

/s/-Lenition and Plurality in Modern Occitan 71

though the noun phrases in ALLOc were uttered inisolation, natural speech would of course provideadditional sources of plural marking. Finally, pre-liminary analysis of masculine plural articles revealsthat those in the advanced northwest sector take theshape [lu] in the singular but [lej] in the plural. Is thisalso (perhaps less articulatorily likely) a result of vowelmutation triggered by the following [j], or is this theresult of analogy with the feminine article?

As is the case with many dialect atlases, ALLOcwas conceived primarily to study lexical variationand was not strictly designed to examine sound chan-ges: For instance, no attempt was made to targetminimal pairs. Nevertheless, it has revealed a reason-able path by which the otherwise-unexplained /e/in French plural articles may have come about,and has provided data relevant to theories of homo-phony avoidance and the maintenance of semanticdistinctions.

Notes

1 The eight lexical items examined were les traces ‘animaltracks’ (ALLOc’s map 352), les courroies du joug ‘the bands ofa yoke’ (map 516), les ridelles ‘walls of a cart’ (map 552), lesfleurs du vin ‘yeasts that develop on the surface of wine’(map 754), les tranches de pain ‘slices of bread used togarnish soup’ (map 1058), le pantalon ‘trousers’ (map 1131),les jarretières ‘garters’ (map 1136), les incisives ‘incisors’(map 1175).

2 Ravier (1978:xiii, translation mine) transcribes this soundin ALLOc notation as [i]̯ and describes it as the “weakelement” of a “decreasing diphthong”which does not formits own syllable. I follow other researchers (Floricic, 2010;Sauzet, 2012) as interpreting this as IPA [j].

3 Consonants noted by ALLOc as being “weakly articulated”are not differentiated in this or subsequent maps fromconsonants that were articulated normally.

4 One might expect to see included in this list of realizationsof /s/ its aspiration to [h], a common stage of /s/-lenitionacross languages (Ferguson, 1990; Seklaoui, 1989). Indeed,Map 4 will show that aspiration is a common realization of/s/ on the definite article. The lack of aspiration observedon the noun is most likely due to a prohibition on phrase-final [h] (whichwould appear on the noun); the lexical itemselicited for ALLOcwere produced in isolation, with nothingfollowing the noun.

5 Calculations of percentages were done after N/A tokenswere omitted, so for localities where N/As were recordedfor one or more lexical items, the denominator of thefraction was less than eight.

6 Only one locality (24.10) ever shows deletion of /s/ on thearticle; it is accompanied by lengthening of thearticle vowel.

7 Sauzet (2012) similarly notes the correspondence betweenthe region of vowel change in the definite article and theregion of /s/ deletion on the noun, and briefly mentions apossible role of article-final [j].

Acknowledgements

Many thanks to William Labov, who provided invalu-able comments at every stage of this work, audiencesat the University of Pennsylvania and at NWAV38, andtwo anonymous reviewers for their comments.

References

Anglade, Joseph. [1921] 1977. Grammaire de l’ancien Provençalou ancienne Langue d’Oc: Phonetique et morphologie. Paris:Editions Klincksieck.

Barra-Jover, Mario. 2012. L’évolution des marques de plurielnominal roman à la lumière de l’occitan. In MarioBarra-Jover, Guylaine Brun-Trigaud, Jean-Philippe Dalbera,Patrick Sauzet & Tobias Scheer (eds.), Etudes de linguistiquegallo-romane, 201–216. Paris: Presses Universitaires deVincennes.

Cedergren, Henrietta. 1973. The interplay of social andlinguistic factors in Panama. Ithaca, NY: Cornell Universitydissertation.

Chambers, J.K. & Peter Trudgill. 1998. Dialectology, 2nd edn.Cambridge: Cambridge University Press.

Eckert, Dorothy A. 1969. Grammatical constraints in phonologicalchange: The unstressed vowels of Southern France. New York,NY: Columbia University thesis.

Eckert, Penelope. 1985. Grammatical constraints inphonological change: Unstressed *a in Southern France.Orbis 31. 169–189.

Ferguson, Charles. 1990. From esses to aitches: Identifyingpathways of diachronic change. In Thom Huebner (ed.),Sociolinguistic perspectives: Papers on language in society,1959–1994, 200–215. Oxford: Oxford University Press.

Floricic, Franck. 2010. Remarques sur le marquage du nombredans le parler occitan de Veyrines-de-Vergt. In FranckFloricic (ed.), Essais de typologie et de linguistique générale:Mélanges offerts à Denis Creissels, 417–433. Lyon: ENSEditions.

Guy, Gregory R. 1981. Linguistic variation in BrazilianPortuguese: Aspects of the phonology, syntax, and languagehistory. Philadelphia, PA: University of Pennsylvaniadissertation.

Héran, François, Alexandra Filhon & Christine Deprez. 2002.La dynamique des langues en France au fil du XXe siècle.Population et Sociétés 376. 1–4.

Hernández-Campoy, Juan Manuel & Peter Trudgill. 2002.Functional compensation and southern peninsular Spanish/s/ loss. Folia Linguistica Historica 23. 31–57.

Labov, William. 1994. Principles of linguistic change, vol. 1:Internal factors. (Language in Society 20). Malden, MA:Blackwell.

Lewis, M. Paul, Gary F. Simons & Charles D. Fennig. (eds.).2014. Ethnologue: Languages of the world, 17th edn. Dallas,Texas: SIL International. Online version: http://www.ethnologue.com

Moseley, Christopher (ed.). 2010. Atlas of the world’s languagesin danger, 3rd edn. Paris, UNESCO Publishing, Onlineversion: http://www.unesco.org/culture/en/endangeredlanguages/atlas

72 L. MacKenzie

http://www.ethnologue.comhttp://www.ethnologue.comhttp://www.unesco.org/culture/en/endangeredlanguages/atlashttp://www.unesco.org/culture/en/endangeredlanguages/atlas

Pope, M. K. 1952. From Latin to Modern French, with especialconsideration of Anglo-Norman: Phonology and morphology.Manchester: Manchester University Press.

Poplack, Shana. 1980. Deletion and disambiguation in PuertoRican Spanish. Language 56. 371–385.

Ranson, Diana. 1991. Person marking in the wake of /s/deletion in Andalusian Spanish. Language Variation andChange 3. 133–152.

Ravier, Xavier. 1978. Atlas linguistique et ethnographique duLanguedoc occidental, vol. 1. Paris: Editions du CentreNational de la Recherche Scientifique.

Ravier, Xavier. 1982. Atlas linguistique et ethnographique duLanguedoc occidental, vol. 2. Paris: Editions du CentreNational de la Recherche Scientifique.

Ravier, Xavier. 1986. Atlas linguistique et ethnographique duLanguedoc occidental, vol. 3. Paris: Editions du CentreNational de la Recherche Scientifique.

Ravier, Xavier. 1993. Atlas linguistique et ethnographique duLanguedoc occidental, vol. 4. Paris: Editions du CentreNational de la Recherche Scientifique.

Ronjat, Jules. 1932. Grammaire istorique des parlersprovençaux modernes, vol. 2. Montpellier: Société desLangues Romanes.

Sauzet, Patrick. 2012. Occitan plurals: A case for a morpheme-based morphology. In Sascha Gaglia & Marc-OlivierHinzelin (eds.), Inflection and word formation in Romancelanguages, 179–200. Philadelphia: John Benjamins.

Schiffman, Harold. 1996. Linguistic culture and language policy.London: Routledge.

Seklaoui, Diane R. 1989. Change and compensation: Parallelweakening of [s] in Italian, French, and Spanish. New York:Peter Lang.

Spence, N. C. W. 1976. A note on the history of the Frenchdefinite article le/la/les. Romance Philology 29. 311–318.

Terrell, Tracy. 1977. Constraints on the aspiration and deletionof final /s/ in Cuban and Puerto Rican Spanish. BilingualReview/La Revista Bilingüe 4. 35–51.

Wheeler, Max W. 1988. Occitan. In Martin Harris & NigelVincent (eds.), The Romance languages, 247–278. Oxford:Oxford University Press.

/s/-Lenition and Plurality in Modern Occitan 73

Dialect perceptions in real time: A restudy ofMiami-Cuban perceptions

Gabriela G. Alfaraz*

Michigan State University

Perceptual dialectology investigates nonlinguists’ beliefs about their own and other varieties. This paper fills a gap inlongitudinal research in this area with a restudy of the perceptions of Miami Cubans carried out twelve years after the firststudy. Perceptions are examined in relation to social and demographic changes with a sample of 84 participants of Cubanorigin who responded to a questionnaire about the correctness of regional varieties of Spanish. The results showed thatperceptions of non-Cuban varieties remained relatively stable over time, continuing to correlate with race and poverty.Perceptions toward the Cuban Spanish of the Miami community were also stable and, as in the earlier study, were highlypositive, reflecting strong beliefs in its correctness-status. In contrast, perceptions of Cuban Spanish on the island weresignificantly more negative; it was ranked the least correct of the regional varieties evaluated. Factors underlyingperceptions are examined in relation to demographic changes, political ideology, and beliefs about race and poverty. Thispaper highlights the contribution of the longitudinal study of dialect perceptions to the understanding of languageattitudes, intergroup relations, and language change.

1. Introduction

This paper presents a longitudinal study of the percep-tions of Cubans residing inMiami-Dade County, Florida,where 65% of the population is Latino, and over halfof Cuban origin, toward the correctness of varietiesof Spanish, including the Cuban one. The study wasmodeled on an earlier one carried out in 1998 with thesame population. The aim of this second study was toexamine whether perceptions had changed or remainedstable in relation to social and demographic changes inthe community. In what follows, the goals of perceptualdialectology are summarized, the findings of the earlierstudy are outlined, and the Cuban community in Miami-Dade County is described.

1.1. Perceptual Dialectology

The area of study known as perceptual dialectologygrew from Preston’s research on nonlinguists’ attitudestoward regional varieties of U.S. English (Preston, 1986,1988, 1989, 1996, 1999; Long & Preston, 2002). Since itsinception, studies have been carried out in a wide rangeof communities with diverse sociolinguistic situationsand language ideologies, including English in the Uni-ted States (i.e. Benson, 2003; Bucholtz, Bermudez, Fung,Vargas & Edwards, 2010; Fridland & Bartlett, 2006;Hartley, 2005; Preston, 1986, 1988, 1996) and England(Pearce, 2009), French in Canada (Evans, 2002) and

France (Kuiper, 1999, 2005), Spanish inMadrid (MorenoFernández & Moreno Fernández, 2002), at the US-Mexico border (Martínez, 2003) and in Miami (Alfaraz,2002), German after the reunification (Dailey-O’Cain,1999), as well as Hungarian (Kontra, 2002), Turkish(Demirci, 2002; Demirci & Kleiner, 1998, 1999), Japanese(Long, 1999), and Korean (Long & Yim, 2002), to men-tion some varieties and regions.

At the heart of research in this area is an interest inunderstanding the beliefs and attitudes of the folk, oreveryday people, toward language varieties and theirspeakers (Niedzielski & Preston, 2000). Preston (1999:xxv) noted that among the reasons for investigatingwhat the folk believe about regional and social dialectsis that it can reveal “folk dialect areas where there arenone scientifically and vice versa.” Moreover, Prestonpointed out that “instances of language change and so-called language attitudes...might be profoundly influ-enced by folk beliefs about language, particularlybeliefs about the status of language varieties and thespeakers of them” (p. xxiv). Labov (2001: 191) noted that“covert attitudes and beliefs” contribute to the trans-mission of changes “if speakers feel that their adoptionof the linguistic form will lead others to attribute tothem the positive traits of the given group and allowthem to share in the privileges of that group.” Studieson perceptions of dialects can help estimate the like-lihood of convergence or divergence (Giles, 1973), andthus, the direction of linguistic changes in dialect con-tact situations (Auer, Hinskens & Kerswill, 2005; Trud-gill, 1986).

Perceptual dialectology studies have generated abody of work that has contributed to our understanding

*Address for correspondence: Gabriela G. Alfaraz, Department ofRomance and Classical Studies, B-350 Wells Hall, 619 Red Cedar Rd,East Lansing, MI 48824-1027, Office: (517) 884-6343, Cell: (517) 303-4048.(Email: alfarazg@msu.edu)

Journal of Linguistic Geography (2014) 2, 74–86. © Cambridge University Press 2014 ORIGINAL RESEARCHdoi:10.1017/jlg.2014.8

mailto:alfarazg@msu.edu

of dialect perceptions, but longitudinal studies repre-sent a gap in the research. With the exception of Preston(2011) and the research reported on in this paper,studies have not addressed perceptions over time, par-ticularly in relation to social changes. Preston (2011)compared findings from twenty-some years of workon perceptual dialectology in Michigan. At the outset,he asked:

Will Michigander’s language attitudes look thesame twenty years later? Surely more than twodecades of greater media exposure to regionalvarieties, particularly their use by obviously welleducated speakers, and perhaps some growingsense that severely downgrading the speech ofothers might not be the nicest thing one could dowill have ameliorated these harsh views.

His findings led him to conclude that the group’s per-ceptions had not changed: “[P]rejudice against the Southand perceptions of Michigan as the site where the mostcorrect English is spoken have not changed in twentyyears” (p. 7). Moreover, rather than having mitigatedtheir evaluations of other regions, Michiganders gaveharsher ratings to some regions than they had in earlierstudies.

The description of findings from the first study ofMiami Cubans, discussed in the next section, will showthat Cubans share with Preston’s Michiganders a veryhigh regard for the correctness-status of their ownvariety. The evidence Preston found for stability in theperceptions of Michiganders provides a backdrop forthe study of perceptions among Miami Cubans. Itmotivates the question of whether the perceptions ofCubans have remained stable because of their stronginvestment in the correctness-status of their variety,similar to the Michiganders in Preston’s study, orwhether Cuban perceptions have changed, despite thehigh regard for their own variety, in response to localsocial and demographic changes.

1.2. The First Study of Miami-Cuban Perceptions

The original study, discussed in Alfaraz (2002), examinedthe perceptions of Cuban-origin individuals residingin Miami, Florida. It focused on regional varieties ofLatin American Spanish, presented as countries, follow-ing a common classification model that disregards thatisoglosses do not coincide with national boundaries(Lipski, 1994; Penny, 2004), but which, based on evidencefrom a pilot study that included drawing dialect regionson maps, reflects how Spanish speakers think aboutregional dialects. Along with the varieties of LatinAmerican, Peninsular Spanish was included as a varietybecause it is widely accepted as the standard–Spain isthe seat of the institution governing Spanish, La Real

Academia Española, which oversees and approves stan-dard grammar and lexicon published in their officialdictionary and grammar books.

Fieldwork was conducted in Miami in 1998, on theheels of a major wave of immigration from Cuba. Thestudy included a stratified sample of 148 individualswho had immigrated at different time periods, or whohad been born in the United States and actively usedSpanish. The study and all instructions were in Spanish.Using a questionnaire, we asked participants to rateon a seven-point scale the correctness, pleasantness,and degree of difference of the varieties of LatinAmerican and Spain. Although given the opportunityto add regions other than the countries listed, only a fewparticipants added Andalusia, reflecting a generalnorth-south division in Peninsular Spanish. Rather thaneliciting perceptions of the Cuban variety by listing thename of the country, Cuba was listed as two varieties:One representing the variety of the diaspora (Cuba 1),and the other the variety of the island (Cuba 2).This separation of Cuban Spanish into two varietieswas based on observing widely held beliefs aboutintragroup differences based on the dimensions ofspace—here (Miami-United States) versus there (Cuba)—and time—then (pre-revolution) versus now (post-revolution). When taking the survey, participants didnot comment on the division and rated them withouthesitation. Indeed, it was surprising to see automaticresponses given to their own variety. The ratings ofother varieties were straightforward, both of those witha stronger demographic presence in Miami and thosenot commonly encountered, as, for instance, Bolivian orParaguayan Spanish.

The findings for regional dialects in Figure 1 confirmsthat the distinction made for the two varieties of CubanSpanish captured a perceptual dialect boundary thatwas revealed in their ratings. The diaspora variety(Cuba 1) received the highest rating (6.0), second only toSpain (6.34), whereas the island variety (Cuba 2)received among the lowest ratings (4.03). The sig-nificance of the separation between the varieties wasshown with a cluster analysis in which Cuba 1 andSpainwere in one cluster, Cuba 2 in a second, and all theother regional varieties in a third. Although a linguisticboundary between the diaspora and island varietieshad not been demonstrated with empirical data, therewas a perceptual-dialect boundary whose function wasto strengthen intragroup boundaries. Not only did theestablished immigrants give Cuba 1 higher ratingsthan Cuba 2, the newest arrivals also gave the islandvariety lower ratings than the diaspora one. Regardlessof time of arrival to Miami (or birth there), all groupsnoted differences between Cubans here and Cubansthere. For newcomers, the perceptual boundary wasspatial (here versus there), whereas for established

A real time study of Miami-Cubans’ perceptions 75

groups, the boundary was both spatial and temporal(then versus now).

While these results for Cuban Spanish mostobviously stemmed from political ideology, they alsorevealed ideologies of economics and race. For theCuban varieties, wealth and race correlated withcorrectness: The island was perceived as poor, black,and nonstandard, whereas the diaspora was wealthy,white, and standard. These beliefs about wealthand race surfaced in the evaluation of other regionalvarieties as well. A correlation was found betweenpositive ratings and degree of white population and,similarly, between positive ratings and greater wealthor economic development, operationalized as grossdomestic product per capita. Thus, higher ratings weregiven to regions with larger proportions of whitepopulation and higher gross domestic product percapita, for instance, Spain (6.34) and Argentina (5.16),which ranked third after Cuba 1. On the other hand,correctness ratings were lower for varieties from poorerregions that had larger proportions of non-white ormixed-race speakers, particularly (and paradoxically),the Spanish of their two Caribbean island neighbors,Puerto Rico (3.83) and the Dominican Republic (3.73). Insum, this first study confirmed that Cubans in Miamibelieved in the high degree of correctness-status of thediaspora (their) variety. Moreover, the study showed alink between correctness-status, race, and economicdevelopment. Whether it continues to influence per-ceptions will be examined in this restudy of thecommunity.

1.3. Cuban Presence in the Community

The Cuban enclave inMiami-Dade County predates theCuban Revolution, but it was this event in 1959 thattriggered large waves of immigration from the islandto the United States and settlement of heavy

concentrations of Cubans in Miami. According to the2010 Census, over two-thirds of Cubans lived in thestate of Florida and nearly half the Cubans in the UnitedStates lived in Miami. In the Miami enclave, Cubansmade upmore than half (53%) of the Hispanic or Latinopopulation. In the decade from 2000 to 2010, the Cubanpopulation increased by 26%, and immigration fromCuba, as well as from other Latin American countriesfueled growth of the Hispanic population.

There were four major waves of immigration fromCuba from 1959 to 1996, with continuous immigrationof smaller numbers in the interludes. The largest num-ber of Cubans arrived in the 1960s and early 1970s:The first wave (1959–1962) brought 248,000 Cubans, thesecond (1965–1973) brought 297,318 more, and in thethree-year interlude (1962–1965) between these waves,56,000 more arrived. During the Mariel Boatlift in 1980,124,776 Cubans arrived by boat, and during the lastmajor wave, the so-called Rafter Crisis (1994–1996),80,000 Cuban rafters and visa holders settled in theUnited States.

Growth of the Cuban population in Miami continuedat a rapid pace in the ten-year period between 1995and 2005. During this time, over 200,000 Cubans wereestimated to have immigrated to the U.S., among themrafters (balseros), people smuggled in on speedboats(boteros), and winners of the U.S. government visalottery (el bombo) (Henken, 2005). According to the 2010Census, the number of Cubans in the U.S. increasedfrom 1.2 million in 2000 to 1.8 million in 2010. Whileagencies resettled Cuban arrivals to various parts of theUnited States, an overwhelming number settled inMiami-Dade County or migrated there from the placesthey had been sent during the resettlement process.

The Cuban population in Miami-Dade numbered856,007 at the time of the 2010 Census. The attraction ofthe area for newcomers, as García (1996: xi) describedit during the mid-1990s, continues to explain its appeal

Spain

Cuba

1

Arge

ntina

Colom

biaCh

ile

Costa

Rica

Vene

zuela

Urug

uay

Boliv

ia

Ecua

dor

Parag

uay

Guate

mala

Peru

El S

alvad

or

Pana

ma

Cuba

2

Nica

ragua

Mex

ico

Hond

uras

Puert

o Rico

Domi

nican

Rep

.

6.346.00

5.17 5.14 4.97 4.90 4.824.50 4.45 4.44

4.23 4.19 4.17 4.13 4.10 4.03 4.01 4.00 3.98 3.83 3.73

Figure 1. Results for all varieties from original study.

76 G. G. Alfaraz

to Cubans, whether new arrivals or establishedimmigrants:

The new immigrants whowill settle in Miami willfind the city familiar and yet alien. Miami iscertainly not a Cuban city, but it is home to thesecond-largest Cuban population in the world,and new immigrants will find it enough likehome to want to settle there—or return there,if things do not go well for them in other parts ofthe United States. Miami will serve as a prismthrough which to interpret the U.S., and hope-fully a buffer to shield them from the harsheraspects of the adaptation process. They will beamong compatriotas who know what it’s like toleave one’s homeland in search of better options.

AlthoughMiami-Dade has historically been an enclavefor Cuban immigrants, it attracted immigrants from LatinAmerican countries as economic instability and politicalturmoil increased migration from the region. The 2010U.S. Census reported that, between 2000–2010, the num-ber of South Americans grew by 77% and the number ofCentral Americans by 65%. Like Cubans, immigrantsfrom other Spanish-speaking countries often prefer tosettle in Miami rather than venture beyond the bilingualcity to other regions, where theywill, most certainly, needto learn English in order to work and carry out publicinteractions.

The immigration of Spanish-speaking groups hascontributed to a dialect contact situation in whichCubans are increasingly in contact with diverse vari-eties of Latin American Spanish. Moreover, it canbe said that heavy recent immigration from Cuba hascreated a situation in which Cuban Spanish is in contactwith itself, in the sense that the variety of earlier immi-grants is in contact with the variety of newcomers.

1.4. Goals

The goal of this research was to examine dialect percep-tions in real time. To this end, the first study wasreplicated with a twelve-year time depth to investigateperceptions in relation to changes in the community,including significant growth in the number of immigrantsfrom Latin American countries and an influx of new-comers from Cuba that contributed to the growth of theCuban population in Miami-Dade County by 44%between 2000 and 2010. In the remainder of this paper, thestatus of perceptions toward regional varieties of Spanishand toward varieties of Cuban Spanish are described andlater compared to the findings of the earlier study.

2. Design and Analysis

Classic perceptual dialectology studies have used avariety of techniques to elicit evaluations, including

hand-drawn maps, which have been a rich source ofdata about folk beliefs (Hartley & Preston, 1999;Preston, 1986, 1996), and questionnaires about the degreeof difference, and the correctness and pleasantnessof varieties, terms which correspond to the notions ofstatus and solidarity (Lambert, 1967); in fact, in Preston(2011), we find correctness expanded to correctness-statusand pleasantness to pleasantness-solidarity.

As noted above, the original study of perceptionsamong Cubans in Miami used a questionnaire, andthe same questionnaire was used in the restudy. Thequestionnaire listed the Spanish-speaking countries ofLatin America and Spain alphabetically, with the orderreversed on half the questionnaires, and, as in the firststudy, Cuba was listed as Cuba 1 and Cuba 2. A seven-point scale was used to rate correctness and pleasantness,presented separately, but with correctness beingpresented first. As in the initial study, participantscompleted the taskwithout questioning the options.Onceagain, the separation of Cuba 1 and Cuba 2 was notquestioned. Participants clearly enjoyed filling out thequestionnaire andmost talked aloud as they gave ratings,explaining their choices and justifying them with imita-tions and anecdotes, providing a rich source of qualitativedata similar to that obtained when the think aloudmethod (van Someren, Barnard & Sandberg, 1994) isused. In both this and the first study, the confidence withwhich participants gave ratings was striking. Althoughthe actual numbers givenmay have varied somewhat, thetalk surrounding the ratings was similar and demon-strated the cohesiveness of the group’s perceptions.

The questionnaire was written and delivered inSpanish. Participants were recruited if they activelyspoke Spanish in their daily lives, had acquired Spanishas a first language, or had acquired it simultaneouslywith English. Verbal invitations to participate, ques-tions, and explanations were in Spanish, or in Spanishand English, depending on whether the participant wasmonolingual or bilingual. The study was conducted bya Cuban-American bilingual whose variety during datacollection can be described as a standard variety of USCuban Spanish, with a fair degree of leveling of markedphonetic features, and accommodation to participants’varieties, particularly popular US Spanish (Otheguy,2009, 2010) when situations became more casual.Participants were recruited in public spaces, includingretail stores, as well as through acquaintances, friends,and colleagues. As in the first study, participants wereoften hesitant to give social information, even afterbeing assured of their anonymity.

The sample contained 84 Cuban-origin individuals.As in the previous study, an attempt was made tosurvey Cubans of varying ages and socioeconomicstatuses who had arrived from Cuba at different peri-ods of time, as well as Cuban-Americans born in the US

A real time study of Miami-Cubans’ perceptions 77

of Cuban parents, but these were only included in thestudy if they reported actively participating in theSpanish-speaking community. The social factors stu-died included gender, age, socioeconomic status (SES),social network, and year of arrival from Cuba. Genderwas considered a binomial category made up of maleand female. Age groups were established according toten-year spans, with five groups created: 20–29, 30–39,40–49, 50–59, and 60+ . Socioeconomic status reflected acombination of schooling and type of employment; fourgroups were created representing the upper middleclass, lower-middle class, upper-working class, andlower-working class. Network groupings were basedon participants’ reports about the number of Cubanfriends and coworkers they interacted with, resulting infour groups: All Cuban, mostly Cuban, half Cuban, orcontains few or no Cubans.

We also asked the year in which participants arrivedin the US. Four groups were created according to age ofarrival from Cuba. These groupings generally followedthe major waves of immigration described earlier. Thefirst group contained immigrants who arrived between1959 and 1978, during the first and secondmajor waves,which occurred between 1959–1962 and 1965–1973, andin the slow but continuous immigration that occurredbetween and after the major waves. The second groupconsisted of Cubans who immigrated during the yearsbetween 1979–1992; this group had people from thethird major wave, through theMariel Boatlift in 1980, aswell as others who filtered in during the remainder ofthe 1980s, often from third countries. The third grouprepresented immigrants who arrived during the fourthmajor wave, the Rafter Crisis, and covered the period1994–1999. In this restudy, a fourth group was addedthat included immigrants who arrived between 2000and 2010, after the four major waves.

Statistical analysis of the quantitative data from thequestionnaires was carried out using SPSS, with avariety of statistical tests run on the data, including

t-tests, correlations, multidimensional scaling, andk-means cluster analysis. In the following section, theresults for Cuban Spanish and the other regional vari-eties studied are presented separately. The results forother varieties are presented first, followed by those forCuban Spanish. After these results, a comparison of thefindings for correctness from this study and the earlierstudy is discussed.

3. Results

3.1. Regional Varieties

The means of the correctness-status ratings for regionalvarieties are shown in Figure 2. It was not surprisingthat Peninsular Spanish received the highest rating(6.31), which confirmed that it was regarded as themostprestigious variety, representing a model of correctusage. None of the Latin American varieties reached thesix-point range in their rankings; in fact, the next high-est scores were in the low five-point range: Argentina(5.16), Chile (5.11), and Costa Rica (5.07). These werefollowed with varieties in the four-point range: Vene-zuela (4.83), Colombia (4.80), Panama (4.30), Uruguay(4.23), Peru (4.08), and Ecuador (4.04). The rest of theregional dialects received ratings in the three-pointrange, including Paraguay (3.99), El Salvador (3.92),Mexico (3.90), Honduras (3.81), Bolivia (3.73), Guate-mala (3.64), Nicaragua (3.49), Puerto Rico (3.33), and theDominican Republic (3.23).

These perceptions of correctness followed a generalregional trend that is evident in the multidimensionalscaling in Figure 3. The prestige-correctness of Penin-sular Spanish is reflected in the placement of Spain inthe upper left quadrant, set off from other regions. TheCaribbean varieties of Puerto Rico and the DominicanRepublic, rated the two least correct, are furthest to theright on the x-axis. In comments about these regions,participants indicated that the degree of incorrectnessstemmed from their association with large mulatto and

Spain

Arge

ntina

Colom

biaCh

ile

Costa

Rica

Vene

zuela

Urug

uay

Pana

ma

Boliv

ia

Ecua

dor

Parag

uay

Guate

mala

Peru

El S

alvad

or

Nica

ragua

Mex

ico

Hond

uras

Puert

o Rico

Dom.

Rep

.

6.31

5.16 5.11 5.074.83 4.80

4.30 4.23 4.08 4.04 3.99 3.92 3.90 3.81 3.73 3.64 3.49 3.33 3.23

Figure 2. Ratings of regional varieties without Cuban Spanish.

78 G. G. Alfaraz

African-origin populations. The Central Americanvarieties of Costa Rica, Panama, El Salvador, Honduras,and Guatemala are in the upper section, but Nicaragua,another Central American variety, is in the lowerright section. It is, however, different from the otherCentral American varieties in that it has had a strongpresence in the community. The US’s largest group ofNicaraguans resides in the city of Sweetwater in theMiami-metropolitan area, where signage in Spanishreflects the group’s presence, and a public parkand public middle school are named after a famousNicaraguan poet, Rubén Darío. Sweetwater, or PequeñaManagua (Little Managua), is surrounded by othersuburban metro cities with heavy concentrations ofCubans. Plotted along with the Central Americanregional varieties are the South American varieties ofBolivia and Ecuador; their being grouped with CentralAmerican varieties reflects a perception of incorrectnessthat, as with the insular Caribbean varieties, partici-pants attributed to large concentrations of mestizo andindigenous populations. In contrast, the varieties of theSouthern Cone, Argentina, Uruguay, and Paraguay, arein the lower section, and Chile is toward the bottom ofthe upper one. As noted above, Argentina and Chilereceived the two highest correctness ratings amongLatin American regions. The plotting of Cuban Spanish,discussed below, will provide further insight on thisspatial arrangement of these varieties.

To understand the influence of social factors onperceptions, age, gender, socioeconomic status, socialnetwork, and year of arrival were analyzed usingt-tests. All but gender were found to be significant. Theresults for age (p< .001) showed that the oldest partici-pants, 60 years and older, gave the highest ratings(4.54), and there was a trend for ratings to become loweras age decreased: 50–59 years 4.38, 40–49 years 4.20,30–39 years 4.08, and 20–29 years 3.55. Despite anexpectation that the oldest groups would give lowerratings, demonstrating less tolerance to varieties otherthan their own, it was the youngest group that gave the

harshest ratings. Linguistic evaluations can be subjectto age-effects, according to Garrett, Coupland, andWilliams (2003:85-86), who noted that age may influ-ence the judgments of younger adults, whose attitudesare still developing; furthermore, younger people maybe inclined to oppose the views of the dominant group.The lower ratings of younger people may be related toage, but they can also indicate a change in the commu-nity toward harsher evaluations of non-Cuban dialects.

The results for socioeconomic status (p< .001) indi-cated that participants from the highest social statusgroup (upper middle) gave the highest ratings (4.57),whereas those from the lowest social status group(lower working) gave the lowest (3.92). The groupsin the middle had scores in between these tworanges: Lower middle (4.28) and upper working (4.35).Competition for employment and resources mayexplain why participants in the lowest socioeconomicgroup gave the lowest ratings. In the struggle to secureemployment as unskilled laborers, these Cubans mayfind themselves competing with immigrants fromother parts of Latin America, which may cause themto regard less favorably Latin America varieties ingeneral. Along with this economic component, thelinguistic insecurity of this group may contribute to itdowngrading other varieties in relation to its own.

Along with age and socioeconomic status, the com-position of the social network significantly influencedratings of regional varieties (p< .01). Although thedifferences here were small, the results showed thatCubans with more closed networks had more favorableviews of the correctness of other varieties than Cubanswith more open networks. The most favorable ratingswere given by people who reported networks that wereentirely made up of Cubans (4.52), and the secondhighest were from participants with mostly Cubans intheir network (4.33). Varieties were judged moreharshly if the network was half Cuban (4.00), or if it hadfew or no Cubans (4.09). Thus, rather than mitigatingnegative views of other dialects, these findings suggestthat regular interactions with speakers of other varietiescontributed to negative evaluations. The implicationsfor language change are clear: Open networks are notlikely to result in linguistic accommodation that diver-ges from the Cuban norm. The findings for CubanSpanish, discussed below, will add another layer to thisprediction of divergence from other regional varieties inthis community.

Year of arrival or birth in the U.S. also showedsignificant differences in ratings (p< .001). The groupborn in the U.S. gave the lowest ratings (3.92); thetwo groups of immigrants that arrived after 1994had similar ratings—2000–2010 (3.99) and 1994–1999(4.00)—and earlier immigrants gave higher ratings—arrivals from 1959–1978 (4.45) and 1979–1992 (4.59).

–11.5 1 0.5 0 –0.5 –1 –1.5

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4 SpainCosta Rica

Panama

BoliviaEcuador

El SalvadorHonduras

Guatemala

Dominican Rep.

Chile

VenezuelaColombia

Argentina

Paraguay

Nicaragua

MexicoPeru

Puerto RicoUruguay

0.6

0.8

1

Figure 3. MDS Regional varieties.

A real time study of Miami-Cubans’ perceptions 79

Viewed in terms of competition among groups in thecommunity, the more favorable ratings of establishedimmigrants reflects their greater social and economicstability, whereas the newer arrivals are more likelyto be in direct competition for jobs and resourceswith newcomers from other regions, but this does notnecessarily explain the low ratings of US-born Cubans,who are represented across socioeconomic groups.The evaluations of this group are discussed further inrelation to the results for the two Cuban varieties.

3.2. Cuban Spanish

Including the results for Cuban Spanish in the rankingof regions, shown in Figure 4, reveals that Cuba 1 hadthe second highest score (6.12) after Peninsular Spanish(6.31), which made it the highest ranked for correctnessof the Latin American varieties. The other Cubanvariety, Cuba 2, on the other hand, received lowerevaluations (3.14) than any of the other varieties. Theseresults confirmed the community’s high regard for thecorrectness-status of the variety represented in Miamiand its low regard for the variety of the island.

Multidimensional scaling revealed interesting pat-terns in the spatial representation of the two Cubanvarieties in relation to other regional varieties. InFigure 5 we see that Cuba 1 is on the far left below Spainon the x-axis, whereas Cuba 2 in the far upper right,directly above the Dominican Republic and PuertoRico, and closest to Bolivia. This plot also indicates withcircles the clusters derived from a K-means clusteranalysis with three clusters specified. The first cluster ismade up of Spain and Cuba 1, positioned to the far left,set off from other regions. The second cluster includesall the non-Cuban varieties. The single member of thethird cluster is Cuba 2. These results suggest that per-ceptions center on the opposition betweenthe two varieties of Cuban Spanish. The prestige ofCuba 1, measured against Peninsular Spanish as theprescriptive norm, contrasts with the low status ofCuba 2, whose distinctness sets it off from other LatinAmerican varieties. Thus, the often cited belief thatCuban Spanish on the island has become increasinglyimpoverished over time underlies a perceptual bound-ary that marks the political and ideological separationof Cubans in Miami from Cubans on the island.

In contrast to the influence of social factors on theratings of non-Cuban regional varieties, t-tests showedthat only age (p< .001) and year of arrival (p< .05) weresignificant for Cuba 1, and none were significant forCuba 2. The evaluation of Cuba 1 across age groupsshowed that the youngest group (20–29) gave it thelowest rating (5.0) and the oldest group gave it thehighest (6.49). Other age groups were between thesetwo extremes: 30–39 (6.09), 40–49 (5.67), 50–59 (6.00).The same age distribution was found for the rating ofthe non-Cuban regional varieties: The oldest groupgave the highest, and the youngest, the lowest scores. Assuggested above, this pattern may be age-graded, relatedto changes in individuals over time, as noted in Garrettet al. (2003), or it may reflect a change in perceptions in thecommunity. If this is, in fact, age-grading, then younger

Spain

Cuba

1

Arge

ntina

Colom

biaCh

ile

Costa

Rica

Vene

zuela

Urug

uay

Boliv

ia

Pana

ma

Parag

uay

Guate

mala

Ecua

dor

Peru

El S

alvad

or

Nica

ragua

Mex

ico

Hond

uras

Puert

o Rico

Dom.

Rep

.

Cuba

2

6.31 6.12

5.16 5.11 5.074.83 4.80

4.30 4.23 4.08 4.04 3.99 3.92 3.90 3.81 3.73 3.64 3.49 3.33 3.23 3.14

Figure 4. Ratings of regions with Cuban varieties.

–11.5 1 0.5 0 –0.5 –1 –1.5

–0.8

–0.6

–0.4

–0.2

0

0.2

0.4 SpainCosta Rica

Panama

BoliviaEcuador

El SalvadorHonduras

Guatemala

Dom. Rep.

Chile

VenezuelaColombia

Argentina

Paraguay

Nicaragua

MexicoPeru

Puerto RicoUruguay

0.6

0.8

1

Cuba 1

Cuba 2

Figure 5. MDS with clusters.

80 G. G. Alfaraz

participants are likely to change their attitudes as theyage. On the other hand, if these ratings for Cuba 1 reflectcommunity change, then they are indicative of lessfavorable attitudes encroaching on perceptions of thehighly esteemed variety of the Cuban diaspora commu-nity in Miami.

For younger Cubans who arrived as children or wereborn in Miami, Cuba 1 is a variety of U.S. Spanish thatinteracts with English, and their less favorable evalua-tions of it may be based on a negative attitude towardcontact features and salient bilingual practices, suchas borrowing and codeswitching. Similarly, youngerspeakers who immigrated from Cuba agree that CubanSpanish in Miami is prestigious, but for many of thesespeakers, particularly the ones who are not bilingual,new features derived from contact with English thatthey have encountered are highly salient and markdifferences between their native variety (Cuba 2)and the variety in Miami (Cuba 1). These differencesfigure prominently in conversations about languageand language learning in the U.S. In some of the inter-views, newcomers reported, in a humorous tone, thatthey needed to learn to speak Spanglish in Miami, andthat they were forgetting to speak Spanish but had notyet learned English.

The results for year of arrival revealed that Cuba 1received the highest ratings from the newest arrivalsfrom 2000–2010 (6.38) and the oldest from 1959–1978(6.37). Following these were arrivals from 1979–1992(5.93), 1994–1999 (5.64), and the US-born group (5.63).While it was not surprising that the longest-establishedresidents gave Cuba 1, their own variety, more favor-able ratings, it was unexpected that the newcomers,who cannot claim it as their native variety, gave itsimilarly high ratings. In the ratings for Cuba 2,although not significantly different, Cubans born in theU.S. gave the lowest ratings (2.75), followed by arrivalsfrom 1959–1978 (3.08). The ratings were somewhat

higher for the other year of arrival groups: 1979–1992(3.40), 1994–1999 (3.18), 2000–2010 (3.56). In contrastto Cuba 1, and to other regional varieties, there is con-sensus across groups about the status of Cuba 2.

4. Real-Time Comparisons

4.1. Regional Varieties

A comparison of the ratings of correctness for regionalvarieties from the 2010 and 1998 studies revealed agenerally stable pattern. The results in Figure 6 showthat ratings were somewhat lower in the second studythan in the first—fourteen of the nineteen varieties hadlower scores—and that the regions with the lowestscores were the ones that received harsher ratings in thesecond study, including Bolivia, Guatemala, Nicaragua,Puerto Rico, and Dominican Republic. Increases incorrectness ratings at the second time period wereminimal and limited to regions whose correctness hadbeen rated more highly in the first study. Overall,however, differences in ratings over time were notsignificant.

Interesting differences between varieties were foundwhen ratings were converted into regularized scores.This was done based on a percent calculated from thehighest rating, given to Spain, and a rate change scorederived from the difference between the percentagepoints for the two time periods. The regularized scoresin Figure 7 are of changes in the ratings between1998 and 2010. This figure also shows demographicchanges in Miami during the period studied as apercent reflecting differences in the total population ofindividuals who claimed regional origin in the country(from the U.S. Census Bureau figures for 2000–2010).

The regularized rating change confirms that ratingchanges were negative for all varieties except those ofPanama, Costa Rica, Chile, Venezuela, and Argentina.

Spain

Arge

ntina

Colom

biaCh

ile

Costa

Rica

Vene

zuela

Urug

uay

Pana

ma

Boliv

ia

Ecua

dor

Parag

uay

Guate

mala

Peru

El S

alvad

or

Nica

ragua

Mex

ico

Hond

uras

Puert

o Rico

Domi

nican

Rep

.0

1

2

3

4

5

6

7

2010 1998

Figure 6. Comparison of ratings of regional varieties at two time periods (without Cuban Spanish).

A real time study of Miami-Cubans’ perceptions 81

Negative rating changes were largest for countrieswhose correctness-prestige was lowest in the firststudy: Bolivia (−11.08), Guatemala (−8.40), Nicaragua(−7.94), Puerto Rico (−7.64), and the DominicanRepublic (−7.64). Negative changes were also found forEcuador (−6.01), Colombia (−5.0), Uruguay (−3.94),Paraguay (−3.49), El Salvador (−3.02), Honduras (−2.4),and slightly less negative scores were given to Mexico(−1.28) and Peru (−1.11).

By far the greatest rate change was found for Bolivia,whose correctness score dropped 11.08 percent. Thishas clear political underpinnings: The ascent of EvoMorales, who had strong ties to Cuban leader FidelCastro and the deceased Venezuelan leader, HugoChávez, was an event that ran directly counter tothe ideological position of the group. Why then wasVenezuelan Spanish given a slightly more positiverating in the restudy, or at least not given a harsher one,when it also experienced political changes towardthe left with Hugo Chavez, elected to office in 1999, at thehelm? The answer is that Cubans are sympathetic tothe plight of the tens of thousands of Venezuelanswho arrived in Miami after Chavez became president.According to the U.S. Census, the number of VenezuelansinMiami grew 117%, from 21,593 to 46,851, between 2000and 2010. Venezuelan politics are followed as closely inlocal, Cuban-dominated, Spanish-language media outletsas the political events in Cuba have been since the settle-ment of the diaspora community.

Although we can point to the growth of the Vene-zuelan community in Miami in the years between thefirst and second study as a factor that mitigatedthe political one, demographic factors alone did notsignificantly account for ratings of regional varieties.When changes in ratings were examined in relation tochanges in the population of groups in Miami, no

significant link was found between them (R2 = .05,F(1, 16) = .83, p = .38). Demographic presence accoun-ted for only 5 percent of the change in perceptionsobserved between 2000 and 2010. Thus, in the dialectcontact situation, there were factors beyond the inter-play of ingroup and outgroup dynamics, evident indemarcation of group boundaries and competition forresources, that influenced attitudes. Political ideologywas clearly an important factor, and it interacted withideologies of race and wealth to shape perceptionstowards regional varieties.

The earlier study reported that economic develop-ment influenced perceptions. A positive correlationwas found between evaluations and gross domesticproduct per capita (r(19) = .67, p< .01). To examinewhether changes in economic development continued tocorrelate with ratings of the regions, a correlations testwas run on the 2010 ratings and recent GDP per capita,which had increased over the last decade in all regionsexcept Argentina. Once again, a strong positive correla-tion (r(19) = .61, p< .01) was found between perceptionsof regional varieties and wealth of the region.

Racial composition of regions was another factorcorrelating with perceptions in the previous study(r(19) = .58, p< .01). To determine whether race con-tinued to be a relevant factor, a correlations test wascarried out on the 2010 mean scores and racial compo-sition, calculated as the proportion of white versusnon-white population. Again, the results showed apositive correlation (r(19) = .61, p< .01) between per-ceptions of varieties and degree of racial diversity inthe region. As in the earlier study, countries that werepredominately indio or mestizo had lower ratings, andcountries where large segments of the population wereof African origin received the lowest ratings. Thus, thelow ratings given to some Central American varieties,

–12

–10

–8

–6

–4

–2

0

2

4

6

Bol

ivia

Gua

tem

ala

Nic

arag

uaD

om. R

ep.

Puer

to R

ico

Ecu

ador

Col

ombi

aU

rugu

ayPa

ragu

ayE

l Sal

vado

rH

ondu

ras

Mex

ico

Peru

Arg

entin

aV

enez

uela

Chi

leC

osta

Ric

aPa

nam

a

Rating Change Percent Population Change

Figure 7. Rating change and population change.

82 G. G. Alfaraz

such as El Salvador, Honduras, Guatemala, and Nicar-agua, some South American varieties, namely Para-guay, Ecuador, and Bolivia, and the Caribbean varietiesof Puerto Rican and Dominican Spanish, can beexplained as a downgrading based on race. As will bediscussed below, race and poverty are factors that,along with politics, also played a role in shaping beliefsabout the correctness-status of both Cuba 1 and Cuba 2.

4.2. Cuban Spanish

Comparing the perceptions of the Cuban varieties in thetwo studies, in Figure 8, shows that Cuba 1 held itsplace as the second most correct variety. Indeed, therating of Cuba 1 remained generally stable, increasingslightly from 6.00 in 1998 to 6.12 in 2010. Cuba 2, onthe other hand, was ranked higher among thevarieties in the earlier study, where it ranked 6th aheadof Nicaragua, Mexico, Honduras, Puerto Rico, and

Dominican Republic, but in the restudy, it ranked last.Ratings of Cuba 2 dropped from 4.03 in 1998 to 3.14 in2010, when it received the lowest correctness ratings ofall varieties.

The ranking of the change in ratings over time for allthe regional varieties in Figure 9 shows differences inthe stability of perceptions of the Cuban varieties. First,it is clear that Cuba 1, compared to itself over timeand compared to most other varieties, had remainedgenerally stable, with only a small increase of 2.35%.Thus, the belief in the prestige and status of Cuba 1 hadheld strong, its resistance to changefirmly grounded in itsideological function. In contrast, Cuba 2 had the largestchange of all varieties, decreasing 13.8 percentage points.Only Bolivia, which experienced a political move to theleft, had a double-digit decrease in the ratings. It is clearthat Cuba 2 was heavily downgraded because it is onthe opposite side of the ideological divide separatingMiami-Cubans from their homeland.

Spain

Arge

ntina

Colom

biaCh

ile

Cuba

1

Costa

Rica

Vene

zuela

Urug

uay

Pana

ma

Boliv

ia

Ecua

dor

Parag

uay

Guate

mala

Peru

El S

alvad

or

Nica

ragua

Mex

ico

Hond

uras

Puert

o Rico

Domi

nican

Rep

.

Cuba

20

1

2

3

4

5

6

7

2010 1998

Figure 8. Comparison of ratings at two times including Cuban varieties.

–14.00

–12.00

–10.00

–8.00

–6.00

–4.00

–2.00

0.00

2.00

4.00

Boliv

ia

Cuba

2

Gua

tem

ala

Nic

arag

uaD

om. R

ep.

Puer

to R

ico

Ecua

dor

Colo

mbi

aU

rugu

ayPa

ragu

ayEl

Sal

vado

rH

ondu

ras

Mex

ico

Peru

Arg

entin

aV

enez

uela

Chile

Cuba

1

Costa

Ric

aPa

nam

a

Figure 9. Rating change over time with Cuban varieties.

A real time study of Miami-Cubans’ perceptions 83

Moreover, in Miami these perceptions also functionto mark boundaries between established Cubans andnewcomers. Downgrading the variety of newcomers,Cuba 2, is an overt rejection of that group. García(1996:xi), whose book on Cuban immigrants in Miamiwas published during the Rafter Crisis when immigra-tion from camps in Guantanamo was in progress, fore-cast tensions that exist today between establishedimmigrants and new arrivals as a result of their racial,sociocultural, and political differences:

This is not to say that there will not be conflict.Emigrés of the first wave (1959–62), dispro-portionately white and middle class, will find itdifficult to relate to the new immigrants, whomthey consider rough, poor, and uneducated. Thefact that many of the new immigrants are blackor of mixed racial heritage, and were oncethe faithful revolutionary proletariat, widens thecultural chasm. Already, the term balsero hasbecome a pejorative among the older emigrés, away of differentiating themselves from the newarrivals, just as the term marielito (for the Cubansof the third wave) acquired currency in the early1980s for many of the same reasons. The arrival ofmore than twenty thousand new immigrantseach year, most of whom will probably stay insouth Florida, will also exacerbate ethnic tensionin Miami.

Thus, language serves to mark boundaries betweenestablished Cubans and newcomers. Led by the olderimmigrants from the early first wave, for whom Cuba 1is the native variety, established Cubans point tothe variety spoken by new arrivals (Cuba 2), with itsabundance of unfamiliar lexical items, and perhapswith phonetic and grammatical changes that advancedover the course of forty-some years, and note it a cor-rupt form of the language they knew. Certainly, this isnot a surprising discovery, given that older generationstend to believe their language is deteriorating inyounger generations. What makes this an interestingfinding that contributes to our understanding ofgroup relations is the separation of the groups in timeand space and their reencounter in one location, theunfamiliar new group having encroached on the phy-sical and linguistic space of the older group, whichdeploys its power and prestige to maintain its positionand uses its beliefs about the correctness-status of itslanguage to fortify the boundaries separating the two.

What is remarkable in the evaluation of the Cubanvarieties is that newcomers are complicit in upgradingthe diaspora variety and downgrading the island one.In the earlier paper, I noted that this “appears to reflecttheir desire to disassociate from the island in order toavoid the negative sentiments” of established Cubans;

meanwhile, they “signal their loyalty to the Miamigroup by accepting its attitudes toward Cuban Spanishon the island.”Whether as an outcome of hegemony, orperhaps driven by their own linguistic insecurity,newcomers embrace the correctness-status of Cuba 1.Newcomers’ attitudes toward Cuba 1 signal theiracceptance of the community’s predominant ideologyand alignment with the values of established immi-grants, not necessarily political, but in terms of theaspirations of immigrants in general, including eco-nomic stability, home ownership, access to educationfor oneself and one’s children, and other indicators ofsuccessful adaptation.

In a comparison of the means by year of arrival at thetwo time periods, shown in Figure 10, we see that thestability of Cuba 1 contrasts with changes in percep-tions of Cuba 2. Ratings for Cuba 2 of Cubans whoarrived during and after the Rafter Crisis (1994–1999)were considerably higher in 1998 (5.1) than in therestudy (3.18). They appear to have accommodated tothe perceptions of earlier immigrants. When the firststudy was conducted, this group had only recentlyarrived. It was a large group with strong connections,forged in planning and leaving Cuba or in the camps atGuantanamo. They also had strong ties to Cuba—manywere the first in their families to emigrate, leavingparents, spouses, siblings, and children behind. Overtime, however, they brought their families and settledinto life in Miami in the same way earlier immigrantgroups had done years earlier. A decade or so later, themembers of this group had less in common with newarrivals