Andalusia was the last region in Spain to be wrested from Moorish control. It is presumably for this reason that a Hispanized version of the old Arabic term Al-Andalus, which originally applied to the whole of Islamic Spain, has survived as the regional name. The actual reconquest of Andalusia was carried through in several stages, with the western and northern cities (Córboba, Jaén, Seville and Cádiz) falling to the Christians in the thirteenth century, but the mountainous kingdom of Granada in the west holding out until 1492.
Despite the extended North African presence in the region, the modern Andalusian population is largely descended from Castilians who settled the area after it was reconquered and from the autochthonous Christian population (referred to as Mozarabs, < Arabic musta‘rab) that had inhabited the region from before the Moorish invasion of the eighth century. The Mozarabs originally spoke Mozarabic, a continuation of spoken Latin, but by the late Middle Ages this variety of Romance appears to have close to extinction, having been displaced by a combination of Arabic on the one hand and other forms of Ibero-Romance, including Castilian, on the other.
Neither Mozarabic nor Arabic appears to have exercised significant structural influence on Andalusian varieties of Spanish, although it seems reasonable to suppose that there was at least some lexical borrowing. It should be noted, however, that virtually all the Spanish words that are of Arabic origin – almohada ‘pillow’, albañil ‘mason/builder’, alcalde ‘mayor’ etc. – are general Spanish words and hence are not specific to the Andalusian dialect.
The distinctive character of Andalusian Spanish should thus be seen as resulting primarily from internal sound change rather than from language contact. The first documented evidence of divergence from Castilian Spanish is from the early sixteenth century, but in all probability the process was underway long before that time.
2.1 Seseo and ceceo
Probably the best-known feature of Andalusian Spanish one that it shares with Canary Island and Latin American Spanish is the merger of the coronal fricative phonemes /s/ and /θ/, which in Castilian Spanish correspond to the letters s and z (or c before i or e) respectively. In the speech of many Andalusian speakers, this merger is manifested as seseo, whereby the sibilant /s/ replaces the non-sibilant /θ/. In such dialects, su aceite his/her oil, for example, is pronounced [swaˈsejte]. Alternatively, ceceo may be observable, whereby /θ/ entirely replaces /s/, as in [θwaˈθejte].
Traditionally ceceo is associated with southern Andalusia, as is indicated in the map below, with seseo predominating in northern Andalusia and in a small island around the metropolis of Seville city. In broad terms, this probably holds true for the present day, but in coastal capitals like Cádiz and Málaga seseo is arguably the commoner of the two phenomena.
The primacy of ceceo
The /θ/ of Andalusian ceceante dialects may not be completely identical with the equivalent sound in Castilian Spanish (corresponding to the letter z, or to c before e or i). However, it is recognizably a non-sibilant articulated at the dental place of articulation, meaning that the best match in terms of the International Phonetic Associations consonantal inventory is [θ] rather than [s], or any subcategory of the latter. Hualde (2005: 153) notates the sound using [θ] modified by the IPA’s dental diacritic, describing it as being ‘dentalized’ and ‘less sibilant’ than the (laminal) /s/ deployed in seseo.
A further reason for associating Andalusian /θ/ with its Castilian counterpart is that both sounds descend from the same phoneme, namely the Old Spanish voiceless dental affricate /ts/, found for example in the word cabeça head, which was pronounced [kaˈbetsa]. The /ts/ corresponding to the letter ç in this word – z in the modern orthography – has evolved to /θ/ in both Castilian Spanish and in those varieties of Andalusian Spanish that have ceceo: [kaˈβeθa] cabeza.
In addition to the voiceless dental affricate, Old Spanish had a voiced dental affricate, /dz/, which was represented in the orthography by the letter z (not to be confused with the modern letter z, which, as just noted, stands for a voiceless phoneme). For example dezir, the forerunner of modern decir to say, was pronounced [deˈdziɾ]. Like /ts/, this voiced affricate was deaffricated at an unspecified time during the late Middle Ages, and its early modern reflex subsequently devoiced, thereby merging with the deaffricated reflex of /ts/. While it is impossible to know for certain how these fricative reflexes of /ts/ and /dz/ were pronounced, it seems safe to assume that they were either fully dental or at least partially dental, rather than purely alveolar.
In Andalusian Spanish, these dental fricatives merged with and displaced the alveolar fricatives /s/ and /z/, which were the sounds represented in the orthography by ss and s respectively. Following this merger, not only were words like cabeça and dezir pronounced with dental fricatives in Andalusian Spanish, as they were in the Castilian dialect, but so too were words like passo step (modern spelling: paso) and casa house, which in Castilian Spanish continued to be pronounced with (apical) alveolar fricatives. Thus modern Andalusian Spanish, unlike modern Castilian Spanish, exhibits no residue of the earlier distinction between dentals (originally /ts/ and /dz/) and alveolars (/s/ and /z/). It is the absence of this distinction which underlies seseo and ceceo, although it is the latter term, albeit used in a more informal sense (and often spelled çeçeo), that was originally used to characterize the underlying phonemic merger.
The term seseo was in fact originally applied to Valencian speech, where it referred to the converse merger to ceceo, i.e. one in which alveolar /s/ displaced the reflex(es) of the medieval dentals. Arguably modern Andalusian seseo is more analogous to the Valencian type of phenomenon than it is to ceceo. For although the /s/ used in Andalusian seseo lacks the retracted or apical character of the Valencian /s/, and ideed the Castilian /s/, it is recognizably a sibilant, whereras ceceo involves the non-sibilant /θ/.
On the other hand, according to one commonly held view (see Lapesa 1981, Penny 2002 and Lispki 1994), both ceceo/çeçeo and Andalusian seseo represent a single phenomenon, synchronically and in historical terms. In that model, /ts/ and /dz/ deaffricated initially to laminal, i.e. denti-alevolar (Spanish: predorso-dental) sibilants, the voiceless member of the pair being essentially the same /s/ that is found in modern French words like façade or poisson fish. This laminal sibilant is then claimed to have fronted in Castilian Spanish to /θ/ during the course of the sixteenth century, but to have remained as a laminal sound in Andalusian Spanish, where, together with its voiced counterpart, it had merged with the alveolar sibilants represented in the orthography by ss and s. From that perspective, modern ceceo is no more than the deployment of a region-specific fronted variant of the laminal sibilant that occurs in seseo and the historical phenomenon of ceceo/çeçeo alluded to by early modern commentators reduces to the use of a laminal sibilant in contexts in which Castilian Spanish had an apical one, both dialects, under the relevant assumptions, deploying the laminal sibilant in contexts in which modern Castilian has /θ/.
There is, however, no concrete evidence to support the notion that the evolution from /ts/ to /θ/ involved an intermediate stage during which the sound in question was articulated as a laminal sibilant. All that can be known with certainty is that the end result is /θ/; indeed, even the existence of the medieval affricates /ts/ and /dz/ falls within the realm of plausible hypothesis rather than established fact. More significantly, and as is discussed in the next section, early Andalusian çeçeo appears to have been seen as quite a striking phenomenon, bordering on an articulation disorder. It is hard to see how the modest linguistic difference between a laminal and an apical sibilant envisaged in the model just described could have attracted quite this level of opprobrium.
An additional factor that may have some bearing on the question is the geographical distribution of seseo in Andalusia, where it is associated prototypically with (i) areas that abut the Castilian dialect area and (ii) the metropolis of Seville city. In both localities, pressure to avoid the marked southern and provincial feature which çeçeo represented can be assumed to have been strong; indeed, there is a long tradition in Spain of systematically disprizing the phenomenon (see Armstrong and Mackenzie 2018: 176, 180). It is, therefore, not completely impossible that Andalusian seseo results from external influence, i.e. dialect contact or the effect of the standard language. Under this scenario, çeçeo/ceceo was originally quite general in the Andalusian region but, over time and in areas that were more exposed to Castilian and/or standardizing influences, speakers marginally adjusted their articulation of /θ/, from an out-and-out dental pronunciation to a denti-alveolar one, in the process converting a non-sibilant into a sibilant. This would reverse the sequencing assumed, for example, by Penny (2000: 119–120), who proposes that while seseo arose in the Late Middle Ages (in Seville) ceceo did not appear until the seventeenth or eighteenth centuries, emerging first in coastal regions of Andalusia.
Were the reflexes of /ts/ and /dz/ sibilants or non-sibilants?
As was mentioned above, nothing can be known for certain about the exact nature of the fricative reflexes, in early modern Spanish, of the earlier dental affricates /ts/ and /dz/. One possibility, at least for dialects that have /θ/ as the end result, is that these sounds were already non-sibilants, i.e. /θ/ and /ð/, at the end of the Middle Ages, as Menéndez Pidal (1958: 113) seems to have assumed. In contrast, Penny (2002: 99) represents them with /s̪/ and /z̪/, which are more clearly suggestive of a sibilant articulation, specifically a laminal (denti-alveolar) one, as exemplified by the /s/ of modern French.
While neither of these two assessments can be ruled out, the way in which commentators in the sixteenth century described Andalusian speech suggests that, in that variety at least, the fricative reflexes of the earlier dental affricates were non-sibilants rather than sibilants. More specifically, Andalusians were known for their tendency to çeçear (also spelled cecear), a verb which then meant ‘to stammer’ or ‘to lisp’, only later acquiring the metalinguistic meaning associated with the modern deverbal nouns ceceo and seseo (see the beginning of 2.1). The original use of the term çeçear/cecear is well illustrated in the extract below from the thirteenth-century General Estoria.
& quemos en somo dela lengua de guisa que dalli tomo por que siempre ceceo despues quando fablaua.
‘And he [Moses] burned the top of his tongue, and to this event he [Josephus] attributed the fact that he lisped/stammered afterwards whenever he spoke.’
(General Estoria I, fol. 137v)
That the tendency referenced by the verb çeçear was assumed to be characteristic of Andalusian speech is indicated by remarks such as the famous çeçeava un poco como sevillano ‘he çeçeo-d a bit like someone from Seville’ made by the conquistador Bernal Díaz del Castillo apropos of an encounter in 1519 with a captain from Sanlúcar de Barrameda (Cádiz province). Given the meaning of çeçear at that time, we can assume that Bernal Díaz del Castillo was alluding to what he viewed as an articulation disorder rather than making a statement about the captain’s phonemic inventory.
In all probability the “disorder” in question indirectly reflected the Andalusian merger of the dental and alveolar fricatives, which is generally assumed to have taken place by that time (or at the very least to have been substantially under way). However, it seems unlikely that simply pronouncing the merged sounds as /s̪/ and /z̪/ would have been perceived as an articulation disorder. More plausibly, the “disorder” in question was the phenomenon known in English as lisping, i.e. the use of the non-sibilants [θ] and [ð] in place of the sibilants [s] and [z]. If so, the implication of Bernal Díaz del Castillo’s comment is that in the relevant captain’s speech, and also in the speech that was characteristic of Seville (presumably the province rather than the city), the fricative reflexes of the earlier affricates /ts/ and /dz/ were indeed non-sibilants, viz. [θ] and [ð], as proposed above.
The latter assumption would also be consistent with the observation below, made by Nebrija in 1492, concerning the pronunciation of the Old Spanish letter ç, which then represented the fricative reflex of voiceless /ts/:
la cual pronunciacion es propria de judios & moros delos cuales cuanto io pienso la(s) recibio nuestra lengua: por que ni los griegos ni latinos que bien pronuncian la sienten ni conocen por suia.
‘which pronunciation is characteristic of Jews and Moors, from whom in my opinion our language received it: because neither the Greeks nor the Latins who speak properly feel or know it as their own’
(Gramática castellana, fol. 9r)
Leaving aside the reference to the Spanish Jews, Nebrija’s conceptualization of ç as a Moorish sound is interesting, as various contemporary or near contemporary data establish a link, either real or perceived, between Arabic speech and the /θ/ sound. First, Lapesa (1981: 376) reports that the ‘moriscos granadinos’ (Moorish converts living in Granada) of the sixteenth century used /θ/ as their best approximation to Castilian /s/, the nearest alternative sound in their phonemic inventory being palato-alveolar /ʃ/. Secondly, the Cordoban scholar Bernardo de Aldrete (1565–1641) appears to have assumed that Arabic speech at that time generally exhibited ceceo. For example, he makes the following observation in respect of the noun Sor:
el Arabe dixo Sar, pero ceceando, çar, que lo usan en su lengua de ordinario
‘in Arabic it appears as Sar, but using ceceo it is çar, which is the usual pronunciation in that language’
(Aldrete 1614: 239; cited Guitarte 1992: 139)
In a marginal comment at the same place in the text, Aldrete further observes (in Latin) of the Italian linguist Angelo Canini (1521–1557) that ‘in θ optime in sua lingua Poenos, intellige Arabe[s] Africanos, balbutire asserit’ (‘he asserts that Carthaginians, i.e. African Arabs, routinely blurt out the θ sound in their language’).
If Nebrija’s observation about Spanish ç reflects a similar type of view, the implication would be that the sound represented by this letter was indeed /θ/ and hence the reflex of /ts/ was already /θ/ by the end of the fifteenth century, Nebrija’s Gramática castellana being published in 1492. In contrast, if ç was pronounced as the sibilant /s̪/, a sound which even at the time would have been familiar from French, it seems surprising that Nebrija should attribute this articulation to Moorish (or Jewish) influence.
It is thus not implausible to suppose that the reflex of /ts/ was already /θ/ by the end of the fifteenth century, implying that the reflex of /dz/ was /ð/, the voiced counterpart to /θ/. Menéndez Pidal (1958: 116) provides an interesting datum which indirectly supports this assumption. Writing in the early twentieth century, he reports that the Extremaduran dialect used in various villages in the north of Cáceres distinguished between /θ/ and /ð/. For example, while cazar to hunt was pronounced [kaˈθal], the word hacer to do was pronounced [haˈðel], a pronunciation captured locally by the spelling jadel. Presumably the voiced member of this pair, viz. /ð/, was a continuation of the voiced dental fricative that existed generally in late medieval or early modern Spanish. That this vestige of that period was the non-sibilant /ð/ rather than a sibilant (e.g. /z̪/) implies that the reflexes of /ts/ and /dz/ were already non-sibilants before the sound change known as the devoicing of the sibilants. Were this not the case, any voiced residue of the earlier affricate /dz/ would have to be a sibilant rather than the non-sibilant /ð/. It follows, given that the devoicing of the sibilants took place over the course of the sixteenth century, that the reflexes of /ts/ and /dz/ became non-sibilants before the sixteenth century or, at the latest, in its early decades.
2.2 Consonantal weakening
Another dominant theme of Andalusian Spanish is consonantal weakening. In syllable-initial position, this is reflected in the realization of /x/ as [h] (as in [ˈpaha] paja straw), in a tendency towards eliminating intervocalic or pre-liquid voiced obstruents (as in [besˈtio] vestido dress and [ˈmaɾe] madre mother) and in so-called CH-lenition, the realization of /tʃ/ as [ʃ] (as in [ˈnoʃe] noche night).
In syllable-final position, the weakening theme is reflected primarily in phenomena relating to /s/ and the liquids.
Syllable-final /s/ may be realized as [h] (as in [ehˈpaɲa] España), it may be elided (as in [laˈola] las olas the waves), or there may be a process of assimilation vis-à-vis the following consonant. When the following consonant is a voiceless obstruent or a sonorant, the output of the process is usually a geminate, as in [eloˈβippo] el obispo the bishop or [ˈmimmo] mismo same. With voiced obstruents, on the other hand, the output is usually a single (voiceless) consonant: [laxaˈʝinah] las gallinas the hens, [laˈɸolah] las bolas the balls.
Among some speakers, elision of final /s/ may be accompanied by a lowering of the tongue position during the articulation of the preceding vowel. As a consequence, the morphological value expressed in standard Spanish by final /s/ comes to be marked by vowel quality. For example, the distinction between mano hand and manos hands comes to be reflected in a vowel quality opposition, viz. [ˈmano] ~ [ˈmanɔ]. In the same way, the distinction between tiene he/she has and tienes you have is indicated by the contrast between [e] and [ɛ]: [ˈtjene] ~ [ˈtjenɛ]. This process, which Spanish commentators usually refer to as a case of desdoblamiento fonológico (phonemic split), is characteristic of Eastern Andalusian varieties of Spanish.
As regards syllable-final liquids, the trueque confusion of [ɾ] and [l] is a notable feature of much unmonitored speech, with the tap [ɾ] often appearing where orthography calls for [l] and, less commonly, the lateral appearing in words which in the standard language have the tap: [baɾˈkon] balcón balcony, [aɾβaˈɲi] albañil builder, [ˈkwelpo] cuerpo body. In word-final position, liquids are frequently elided, as in the albañil example just given.
2.3 Retention of [h] < Latin f
Finally, the retention of [h] as a reflex of initial or post-prefix Latin /f/, as in [ahoˈɣaɾ] ahogar (< Lat. offocare strangle), is sometimes a feature of vernacular Andalusian Spanish. It is certainly one that regularly finds its way into representations of ‘deep’ Andalusian culture. For example the popular term for Flamenco music is cante jondo [ˈkante ˈhondo], which literally means ‘deep singing’. Spanish hondo ‘deep’ comes from Latin fundum, the initial f having no reflex in standard Spanish but surviving as [h], designated orthographically by the letter j, in the relevant speech varieties.
In eastern Andalusia especially in the provinces of Granada and Almería ustedes is usually limited to being a deferential pronoun, as in the Castilian standard.
Armstrong, Nigel and Ian Mackenzie. 2018. ‘Speaker variables in Romance: when demography and ideology collide.’ In Manual of Romance Sociolinguistics, ed. by Wendy Ayres-Bennett and Janice Carruthers, pp.173–196. Berlin: Walter de Gruyter.
Guitarte, Guillermo. 1992. Cecear y palabras afines. In Actas del II Congreso Internacional de Historia de la Lengua Española. Tomo I, Madrid, Pabellón de España, pp. 127-164.
Hualde, José Ignacio. 2005. The sounds of Spanish. Cambridge: Cambridge University Press.
Lapesa, Rafael. 1981. Historia de la lengua española (9th edition). Madrid: Gredos.
Lipski, John M. 1994. Latin American Spanish. London: Longman.
Menéndez Pidal, Ramón. 1958. Manual de gramática histórica española (10th edition). Madrid: Espasa Calpe.
Nebrija, Elio Antonio de. 1492. Gramática castellana (BNE INC/2142).
Penny, Ralph. 2000. Variation and change in Spanish. Cambridge: Cambridge University Press.
Penny, Ralph. 2002. A history of the Spanish language (2nd edition). Cambridge: Cambridge University Press.