Hypercorrection as a Symptom of Language Change: Majorcan Catalan Standard Pronunciation

Abstract

The mutual influence between dialects and standard language in terms of upward or downward convergence, the creation of a diaglossic repertoire, and a typology of relations dialects/standard is of utmost importance for the analysis of language evolution. Majorcan Catalan is an eastern Catalan dialect, traditionally considered as archaic. Data from television programs show a tendency to adapt pronunciation to a standard mainly based on central Catalan dialects, even disregarding concessions granted by the prescriptive grammar. This article is focused on the causes of this centripetal movement and on the function either accelerating or decelerating of (para)prescriptive works in the process. Data from an annotated oral corpus (Corpus Oral de la Llengua Catalana) are used. The results are consistent with a centripetal movement in the construction of the standard Catalan as used in formal registers in the Balearic Islands. A proposal is made for the interpretation of the relevant data as evidence of an evolution from a gliding access to standard to a shifting access.

1. Outset

Let me open this article with a short personal anecdote to illustrate how far from reality popular ideas on standard and prescribed language can lie. During the writing of Gramàtica Essencial de la Llengua Catalana, the online abridged version of Gramàtica de la Llengua Catalana, I was invited to provide voice recordings in my native dialect, Majorcan Catalan. I accepted the invitation and I recorded maybe a few hundreds of words and sentences. Nevertheless, at least in one case I raised an energetic “stop the presses!” after feeling the limits of my language competence openly transgressed. The written sentence to be pronounced and recorded was porta’l (“bring it”), to illustrate a typical phonological process of my dialect by which stress is shifted to the syllable bearing an enclitic pronoun, creating a difference between standard Catalan porta’l [ˈpɔrtəl] and Majorcan Catalan [poɾˈtəl]… the only problem being that the verb portar is to the best of my speaker’s competence a shibboleth of non-Majorcan (and Minorcan) Catalan, a geosynonim of the irregular verb dur, used by the islanders. “I would never produce such a word! The example is unnatural! Moreover, the geolinguistic distribution of portar and dur is aknowledged by at least one issue of our language authority, Institut d’Estudis Catalans”. The reaction of the editors of the grammar was to provide me with several hundreds of examples of “portar” being used by contemporary Majorcan writers. It is of no interest at all to pursue the story to its end because its crucial elements have already been put forward: (i) some language users, at least writers, do not always adhere to their dialects, (ii) in certain cases prescriptive or quasi prescriptive texts concede further than claimed by speakers, and (iii) there exist such things as language corpuses able to put to test advanced speakers’ selfconfidence on the degree of terms appurtenance to regional standards.

These points provide us with the aims of this paper: how should the standard language used by Majorcan/Balearic advanced users be classified? how can centripetal movements be detected in language usage? We shall focus on spoken language for the remain of the paper, with a particular attention on the neglect by advanced speakers of language authority’s concessions for variation.

2. Introduction

Several conservative features of Majorcan Catalan have probably been preserved because of its geographic isolation, which has contributed to the myth of a certain degree of purity of this dialect1. The initial guiding role for the Noucentisme movement of the first decade of the 20th century played by two major Majorcan poets, Miquel Costa i Llobera (1854–1922) and Joan Alcover (1954–1926) is a strong contributor of the myth (Castellanos 1986; Molas 2007), together with the language patriotic task of Antoni M. Alcover (Majorca, 1862–1932) (Julià-Muné 2020, pp. 47–79). Alcover’s campaigning for a modern codification of Catalan reached its zenith in the 1906 Congrés Internacional de la Llengua Catalana, followed by his appointment as the first president of the Philological Section of Institut d’Estudis Catalans, the official academy of the language, founded in 1911. Alongside this perception of Majorcan Catalan as a prestige dialect, two additional facts must be taken into account:

(i) Educated Majorcan speakers have always switched to a more formal code for certain uses (ecclesiastic, public, written) clearly identifiable by features such as definite article, first and second person plural pronouns, alternance between the two first person singular pronoun (jo/mi), and certain other features either phonetic or lexical, as stated in Veny (2001):

“No oblidem, però, el paper decisiu jugat per l’Església en el desenvolupament, al Principat i en especial a Mallorca, d’un estàndard oral ‘avant la lettre’ caracteritzat en aquest cas per l’ús, en la predicació i les tasques d’apostolat, d’una llengua noble, força digna, no divorciada del poble, integradora (que usava nosaltres i no noltros, el, la i no es, sa, etc.) i que ha fet més acceptable per a la societat illenca l’estàndard modern” [It is not to be forgotten the crucial role played by the Church in the evolution, in Catalonia and mainly in Majorca, ‘avant la lettre’, of an oral standard characterised by a noble language, worthy enough, not divorced from the popular usage, integrative (using ‘nosaltres’ [we (standard)] and not ‘noltros’ [we (dialectal)], ‘el’, ‘la’ [the (definite singular article, masculine, and feminine (standard)], and not ‘es’, ‘sa’ [the (definite singular article, masculine and feminine) (dialectal)] and turning the modern standard into something more acceptable for the island’s society] (Veny 2001, p. 32).

(ii) Even if traditionally isolated and archaic to a certain extent, Majorcan Catalan remains an eastern dialect of Catalan and, therefore, a close relative of central Catalan, the ruling dialect in modern standardisation. Not surprisingly, linguistically self-educated writers in times when Catalan was lacking at school fell in disconcertion, and switched to central Catalan giving up any trace of their original Majorcan dialect in their works in either high or low register.

The sociolinguistic situation in contemporary Majorca is quite complex: Spanish appears to be the exoglossic standard variety, although there subsists a traditional endoglossic standard variety. New advanced users of Catalan tend to shift to an alternative endoglossic standard, based on central Catalan dialects2. The theoretical frame here is that of Auer (2011). The basic concepts in Auer’s typology of dialects/standard scenarios in Europe deal with the distance between standard and dialect varieties, ranging from exoglossic standard varieties being used by dialect speakers (by means of language shift in Auer’s Type Zero scenario), to the final disappearance of dialects in favour of an endoglossic standard variety (Type D). In the middle of this scale, Type A and B show endoglossic standards in writing (A) or both in writing and speaking (B), and Type C sees the rise of varieties (‘regiolects’) lying midway between dialects and standard, and replacing, in certain speakers’ usage, the former functions of dialects3. Let us state the question in his terms: whereas a well-established tradition exists of endoglossic standard based on the own dialect—a standard used in Majorcan literature after Catalan Renaissance starting in 1834—adherence to another endoglossic standard mostly based on a neighbouring dialect appears in the 1960s. It can be seen as a fair example of transition from the original gliding access to the new shifting access to standard, all of this in the hands of writers formally educated in the Spanish exoglossic standard. Maybe for this reason, the transition between standards has been explained on grounds of ignorance:

“No voldria fer un judici temerari, però pens que, en molts de casos, en tenia la culpa la ignorància de la normativa. […]. Sia com sia els escriptors d’aquell període no aconseguiren aportar un model de llengua regular, coherent, que satisfés les exigències de la normativa i alhora recollís tot allò que el dialecte mallorquí conserva dels clàssics i del bon parlar tradicional, és a dir, la part digna d’entrar a formar part d’un ‘estàndard composicional’ [I would not like to deliver a reckless opinion, but I think that in many cases ignorance of prescription should be blamed. […]. In any case the writers of that period did not succeed in finding a regular language, coherent, both satisfying grammar and conveying all that inherited by Majorcan from the classics, the traditional way of speaking, i.e., that part of the dialect best deserving to be included in a ‘compositional standard’]” (Grimalt Gomila 1997, pp. 86–87).

These facts are especially important as a recognition of Fabra’s prediction that dialectally based regional standards would finally converge (Fabra 1918). Naturally, Fabra‘s prediction postulated regional standards based on the classical solutions of the language varieties in use in the different areas of Catalan. Failure to comply with this requirement naturally led to the situation described by Grimalt and rephrased here in Auer’s terms.

While the language and the educational authorities (once Catalan took a preeminent place in education) insist in the offer of a standard more closely linked to the island’s dialect, speakers show a tendency to embrace the more general endoglossic but exodialectal standard, the gap between both being small, as explained above, but, even so, still visible enough. As already proposed in Dols (2020a) for morphology, the causes of this adherence are to be found in the speakers’ exposition to Barcelona-based mass media, even if Spanish-speaking media are still far more abundant, the competition framed here not being between Spanish- and Catalan-speaking media, but between two (endoglossic) standardisations.

3. Materials and Methods

The aim of this article is to illustrate how speakers of Majorcan Catalan tend to abandon features of their vernacular pronunciation (even if granted by the official prescriptive grammar and other prescriptive guides) in favour of a more general standard variety of Catalan. To do so, samples of an oral corpus of Catalan will be used as evidence of this tendency. Please note that there is no claim to statistical significance.

The Oral Corpus of Catalan Language (“COLC”, by its initials in Catalan) is a project by the Institut d’Estudis Catalans starting in 2019 and is still under construction. The project aims to compile and annotate samples of oral language from all the Catalan dialects. Proceeding in a top-down direction, the project commenced by gathering non-spontaneous news programs from the public television of the Balearics. For now, that part of Balearic oral standard Catalan subcorpus contains 27 h and 22 min, amounting to 338,191 words. Samples covering 6 h 47 min have already been annotated for grammatical phenomena.

Grammatical annotation is performed making use of section numbers in Gramàtica de la Llengua Catalana (Institut d’Estudis Catalans 2016), 1156 labels in total, with the aim to take advantage of descriptive details of the reference work and to feed a future digital edition of it with oral examples. Every label applied (to a word or to a series of words) receives a qualifier, these being 1 (“predicted” by the grammar), 2 (“contrary to prediction”), and 3 (“unpredicted phenomenon”). Qualifier 3 is of utmost importance for future revisions of the reference grammar. Qualifier 2 informs on non-compliance with grammar regulations or mismatch between observed usage and register ascription according to the grammar.

Let us see two examples:

(a)

In sample 485, fragment 4, a label “4.2.5” has been applied to word “treballadors” (“workers”) after listening to pronunciation [tɾəβəʎəˈðos]. Here, “4.2.5” stands for subsection “4.2.5 Lateral consonants” in the grammar. It can be found there that in Balearic Catalan, a palatal glide ([j]) is found instead of a lateral palatal consonant ([ʎ]) when the Latin etym produces one of the clusters LI, T’L, K’L, G’L (PALEAM > PALIAM > [ˈpajə]) except for in the highest registers, where [ʎ] is produced. Qualifier 1 has been applied to the label because the pronunciation observed matches the register ascription of the phenomena described in the grammar.

(b)

In sample 491, fragment 1579 received label “3.3.4.2a” for the pronunciation of word “seixanta-tres” ([ʃiˌʃantəˈtɾəs] ‘sixty-three’). Section 3.3.4.2a in the Gramàtica de la Llengua Catalana states that the closing of schwa into [i] caused by a postalveolar consonant (as [ʃ] in seixanta [səˈʃantə] > [ʃəˈʃantə] > [ʃiˈʃantə]) can occur in colloquial registers, but it is rejected in formal speech. According to the general style of the recorded speech act, the label applied received a qualifying 2 (“contrary to prediction”).

If attention is paid to fragments receiving labels qualified with a 2, then two types of examples arise: grammar mistakes and instances of inadequacy to a given register. I shall concentrate on two cases of the latter sort, i.e., both deemed to transgress adequacy. Instead of focusing on insufficient compliance with adequacy requirements, I shall illustrate hypercorrect pronunciations4.

4. Three Cases of Hypercorrect Pronunciation

4.1. The Data

4.1.1. Assimilation of Point of Articulation of Coda Stops to a Following Consonant

A prominent feature of Majorcan Catalan, shared with Minorcan and Alguerese, is the weakness of syllable codas when a syllable onset follows. In such a case, the consonant in the coda cannot license a point of articulation different from that of the following (onset-initial) consonant (Dols 1993, 2020b; Wheeler 2005): acte ‘act’ [ˈattə] (Cfr. standard Catalan [ˈaktə]), apte ‘apt’ [ˈattə] (Cfr. standard Catalan [ˈaptə]). The official Gramàtica de la Llengua Catalana (Institut d’Estudis Catalans 2016) describes the phenomenon adding that this assimilation is not linked to a given register or speech tempo. The description is given in Section 4.4.2.4b of the official grammar:

“En mallorquí, menorquí i alguerès, aquesta mena d’assimilacions són sistemàtiques i generals, i no estan condicionades a variables com el registre o el tempo de parla. Així, cap català [kk], cap gorra [gg], poc pacient [pp], poc bromista [bb], cap so [t͡s], capsa [ t͡s ], poc segur [ t͡s ] o accident [ t͡s ] són realitzacions pròpies del mallorquí i el menorquí; i taps, caps, amics, rics, grocs, amb [ t͡s ], són realitzacions característiques del mallorquí i l’alguerès” [In Majorcan, Minorcan and Algherese, these assimilations are systematic and general, and are not conditioned by variables such as register or speech tempo. Thus, cap català ‘no Catalan’ [kk], cap gorra ‘no cap’ [gg], poc pacient ‘little patient’ [pp], poc bromista ‘little joker’ [bb], cap so ‘no sound’ [t͡s], capsa ‘box’ [t͡s], poc segur ‘little safe’ [t͡s] or accident [t͡s] belong to Majorcan and Minorcan; while taps ‘corks’, caps ‘heads’, amics ‘friends’, rics ‘rich (pl.)’, grocs ‘yellow (pl.)’, with [t͡s], are characteristic realizations of Majorcan and Algherese.] (Institut d’Estudis Catalans 2016, p. 80).

It must be added that it is clearly stated in the introduction to the grammar that disapproval is explicit, and simple description implies granting correctness/adequacy: “Quan un determinat cas no es considera acceptable en cap parlar ni registre, es fa constar explícitament” [‘Whenever a given case is not deemed acceptable for any dialect or register, an explicit statement is made]’ (Institut d’Estudis Catalans 2016, p. XXII).

Inspection of a single sample in COLC, 485, produces a consistent transgression of Section 4.4.2.4b of the grammar and, therefore, a label with such a number and qualifier 2 is attached to every segment containing this type of hypercorrect pronunciation (only examples are given here; there are two speakers involved both in expected and unexpected pronunciations, and both of them are speakers of Majorcan Catalan):

• [pɾotəkˈsjo] (protecció ‘protection’) instead of the expected [pɾotətsiˈo]
• [ələkˈsjons] (eleccions ˈelection’) instead of the expected [ələtsiˈons]
• [rəktoˈɾat] (rectorat ‘vice-chancellor’s office’) instead of the expected [rəttoˈɾat]
• [səkˈto] (sector ‘sector’) instead of the expected [sətˈto]
• [məɡˈnɛtikes] (magnètiques ‘magnetic (fem. pl.)’ instead of the expected [mənˈnɛtikes]
• [əbdikəˈsjo] (abdicació ‘abdication’) instead of the expected [əddikəsiˈo]5

In contrast, the same sample produces two instances of the expected assimilations ([ˈfɔts] focs ‘fires’, and [refletˈsjo] reflexions ‘reflections’).

4.1.2. Dissimilation of Sibilants

Another particular feature of Majorcan and Minorcan Catalan is the conversion of alveolar sibilants (/s/, /z/) into stops ([t], [d]) when another sibilant follows, with further formation of a homorganic affricate:

• tres senyors ‘three gentleman’ [ˌtɾət͡sːəˈɲos] (standard Catalan [ˌtɾɛsəˈɲos])
• noves zonificacions ‘new (fem. pl.) zonings’ [ˌnɔvəd͡zːonifikəsiˈons] (st. Cat. [ˌnɔvəzunifikəsiˈons])
• desxifrar ‘to decipher’ [dət͡ʃːiˈfɾa] (st. Cat. [dəʃiˈfɾa])
• dos gelats ‘two ice-creams’ [dod͡ʒːəˈlat͡s] (st. Cat. [doʒəˈlat͡s])

Again, the official grammar grants access of this phenomenon into formal usage of the language in its usual default descriptive manner (see comment on prescriptive technique in Section 4.1.1 above):

“A diferència del que ocorre en la major part de parlars, en mallorquí i menorquí, el contacte d’una fricativa alveolar i una fricativa alveolar o palatoalveolar es resol amb una dissimilació per la qual la primera consonant esdevé oclusiva. Aquesta consonant, en contacte amb la fricativa en segona posició, dona com a resultat una consonant africada geminada (dos sons [tt͡s], dos jocs [dd͡ʒ], dos xalets [tt͡ʃ]). El procés es produeix entre mots (compres sabates [tt͡s]) i entre el prefix i la base (descentrar [tt͡s], desxifrar [tt͡ʃ], desgelar [dd͡ʒ]), i explica la presència d’africades corresponents a les seqüències gràfiques sc en posició interior de mot (ascensor, piscina, escena [tt͡s]). No afecta, però, els mots compostos amb cent, que es resolen amb simplificació: dos-cents, tres-cents [s]” (Institut d’Estudis Catalans 2016, p. 84) [Differently from what happens in most dialects, in Majorcan and Minorcan, the contact of an alveolar fricative and an alveolar or palatoalveolar fricative is solved by a dissimilation by which the first consonant becomes occlusive. This consonant, in contact with the fricative in the second position, results in a long affricate consonant (dos sons ‘two sounds’ [tt͡s], dos jocs ‘two games’ [dd͡ʒ], dos xalets ‘two chalets’ [tt͡ʃ]). The process occurs between two words (compres sabates ‘you buy shoes’ [tt͡s]) and between the prefix and the base (descentrar ‘to decentre’ [tt͡s], desxifrar ‘to decipher’ [tt͡ʃ], desgelar ‘to thaw’ [dd͡ʒ]), and explains the presence of affricates corresponding to the graphic sequences sc inside the word (ascensor ‘elevator’, piscina ‘pool’, escena ‘scene’ [tt͡s]). It does not affect, however, words composed of cent ‘hundred’, where the sibilant sequence is solved by simplification: dos-cents ‘two hundred’, tres-cents ‘three hundred’ [s].]

In sample 485, two segments received qualifier 2 (non-compliance/non-adequacy) for label 4.4.3.4c (dissimilation of sibilants). The cases are més justos ‘fairer (masc. pl.)’ [meʒustos] (expected: [dd͡ʒ]), and escenari (‘scenario’) [əsəˈnaɾi] (expected: [tt͡s]). More interestingly, sample 491 produces three cases of 4.4.3.4c (1) (compliance), and one of 4.4.3.4c (2) in the voice of the same speaker:

• 4.4.3.4c (1): plaques solars ‘solar panels’ ([tt͡s]), disciplines ‘disciplines’ ([tt͡s])
• 4.4.3.4c (2): braços sostindran ‘arms will hold’ ([s])

What is remarkable here is that plaques solars ‘solar panels’ ([tt͡s]) (annotated with 4.4.3.4c (1)) and braços sostindran ‘arms will hold’ ([s]) (annotated with 4.4.3.4c (2)) belong to the same sentence: Els braços sostindran quaranta plaques solars ‘The arms will hold forty solar panels’. The case appears to reflect a phase of instability in a process of language change.

As can be seen above, professional speakers disregard concessions by the official grammar for their dialect and turn to a more general standard Catalan pronunciation.

4.1.3. Stressed Schwa

Stressed schwa6 is a conspicuous feature of Balearic dialects of Catalan: phonemic in nature, it is not reducible by means of regular processes to any other phonological vowel in the same manner as unstressed schwa is in oriental Catalan dialects (Wheeler 2005). Stressed schwa is an archaic feature of Catalan, etymologically linked to Latin /ĭ/ and /ē/surviving only in the islands (PĬRA > Bal. Cat. [ˈpəɾə], other eastern Catalan dialects [ˈpɛɾə]; ECCLĒSIA > Bal. Cat. [əzˈɣləzia], other eastern Catalan dialects [əzˈɣlezia]). This feature leaves Balearic Catalan with eight stressed vowels (/i, e, ɛ, a, ə, ɔ, o, u/), whereas the vowel inventory for the rest of Catalan dialects (except for northern Catalan), both eastern and western, is limited to seven (/i, e, ɛ, a, ɔ, o, u/).

The prescriptive grammar grants the acceptability of all of the traditional pronunciations/é/,/ə́/,/ɛ́/:

“En tots aquests casos són igualment acceptables les diferents pronúncies tradicionals, amb independència que quedin o no reflectides en l’accentuació gràfica” [In all of these cases the different traditional pronunciations are equally acceptable, regardless of their graphic rendering, with or without a stress mark] (Institut d’Estudis Catalans 2016, p. 39).

In sample 485 of COLC, hesitation between stressed/ə́/, /é/, and /ɛ́/ is attested by three labels “3.2.1 (2) Stressed vowel system” attached to words where/ə́/ was expected: passeig ‘promenade’, recerca ‘research’, veim ‘we see’, veurem ‘we shall see’, ple ‘plenary’, terrenys ‘terrains’. In contrast, the same voices produce cases as drets ‘rights’, pateix ‘he suffers’, feina ‘work’, comparteixen ‘they share’, succeeix ‘it occurs’, haver ‘have (aux.)’, irlandès ‘irish (masc. sg.)’, seia ‘he sat’, verd ‘green (masc. sg.), and paradetes ‘booths’, pronounced with the expected/ə́/, and thus labelled with a “3.2.1 (1)”. Except for a stronger adherence to /ə́/ in verbs including the class mark /-ə́ʃ(-)/, there seems to be no regular distribution of the observed tendency to abandon /ə́/. Again, as in Section 4.1.2. above, the most remarkable case of variability is a sentence where both solutions are given:

• ‘Ara el veurem, en color verd’ [Now we shall see it in green], with verd ‘green (masc. pl.)’ producing an /ə́/, and veurem producing an /é/ where /ə́/ was expected.

4.2. Beyond the Data

As the evidence in the sections above demonstrate, the official grammar of the Institut d’Estudis Catalans cannot be held responsible for the behaviours shown as the alternative dialectal pronunciations are not considered incorrect/inadequate by the grammar. Thus, it can be fair to inquire into other language regulations or paranormative texts supporting the observed speech style. Although there is no need for such a search when it comes to explaining incoherences as those pointed out above, there is still some room left for the hypothesis of conflicting rules. Language advisors in the public Balearic TV channel do not interfere in speakers’ pronunciation in details as the ones revised above (Serra 2021). However, a style book of the channel was approved in 2006 (Company and Puigròs 2006). It must be said that the solutions stated in this work are, in general, more restrictive of dialectal features than the Gramàtica de la Llengua Catalana is. The level of restriction can differ for each particular feature. I shall give below a brief summary of recommendations for every case under study with comparisons, when possible, to other paraprescriptive texts published in the Balearics. The examples elicited from the oral corpus do not allow to state diachronic evolutions. Nevertheless, those paraprescriptive texts help understand the contrast between new and traditional pronunciations.

4.2.1. Assimilation in Point of Articulation of Coda Stops to a Following Consonant

Regardless of the traditional unawareness of these phenomena by speakers, as acknowledged by the official grammar, the style book of the Balearic public television and radio broadcasting corporation recommends that it be confined in the area of spontaneity and, thus, this type of consonantal assimilation appears to be discommended for more formal programs as TV news:

“Pronúncies que només es poden emprar en aquells espais que resten oberts a un major grau d’espontaneïtat (debats, entrevistes, retransmissions esportives...):

És admissible assimilar les dues consonants que es troben en contacte en mots com saps, elèctric, parlar. És a dir, aquests mots es poden pronunciar ‘sats’, ‘elèttric’, ‘pal·lar’.” [Utterances to be produced in programs with a higher degree of spontaneity (debates, interviews, sports broadcasts, etc.): It is admissible to assimilate the two consonants in contact in words as saps ‘you know’, elèctric ‘electric’, parlar ‘to speak’, i.e., these words can be spoken as ‘sats’, ‘elèttric’, ‘pal·lar’] (Company and Puigròs 2006, p. 37).

This restrictive account given in the Balearic media style book contrasts with the concession granted by Gramàtica de la Llengua Catalana, and also and more importantly, in the opposite direction, with a previous document issued by the central language authority: a proposal for standard pronunciation published by the Institut d’Estudis Catalans in 1990, where it can be read:

“No són recomanables:

[…]

(5) Certes assimilacions al lloc d’articulació de la consonant següent en casos com contatte per contacte, corrutte per corrupte.” [(Pronunciations) disapproved: (…) (5) Certain assimilations of point of articulation to a following consonant in cases as contatte for contacte ‘contact’, corrutte for corrupte ‘corrupt’] (Institut d’Estudis Catalans 1990, pp. 17–18).

This disapproval must be understood as full. Incidentally, the mechanics of the IEC’s prescription system equals “not recommendable” to “disapproved”. Whichever the status of “not recommendable” may be, comparison to the prescriptive label received by another consonantal process helps understand the meaning of ‘no recomanables’: pruning of consonants in clusters as abs- > as-, ads- > as-, obs- > os- receives the label “restricted to the area of the Balearic dialects” (Institut d’Estudis Catalans 1990, p. 17).

The comparison of Institut d’Estudis Catalans (1990) with Institut d’Estudis Catalans (2016) shows for the process under scope here an evolution towards legitimation (see Section 4.1.1 above for a comment on treatment received in Institut d’Estudis Catalans (2016) by assimilations of the type analysed here). On the other hand, Company and Puigròs (2006) show a self-repressing attitude concerning a dialectal feature. Placing the responsibility for the repression of this feature either on individual speakers or on the style book of a Balearic media system, the paradox remains unchanged: the central authority plays a concessional role while local speakers/media play a more restrictive one.

4.2.2. Dissimilation of Sibilants

As shown in Section 4.1.2. above, the Gramàtica de la Llengua Catalana treats the dissimilation of sibilants as a case of prescription level zero, that is to say, as explicitly acknowledged by the grammar itself that the feature is correct and adequate in all registers: “Totes aquestes variants es consideren adequades amb caràcter general, si no es proporciona cap mena d’especificació sobre el seu ús, o només en l’àmbit d’un parlar o un registre determinats, si així s’indica” [All these variants are considered suitable in general, if no specification is provided on their use, or only in the field of a particular dialect or register, if so indicated] (Institut d’Estudis Catalans 2016, p. XXII). Again, as in Section 4.2.1 above, IEC evolved from restriction to concession, for this feature is dismissed in Institut d’Estudis Catalans (1990):

“No són recomanables:

(…)

L’africació, característica de parlars baleàrics, que es produeix per contacte entre la sibilant final d’un mot i la sibilant inicial del mot següent: tretsenyores per tres senyores, dotxuts per dos xuts” [(Pronunciations) disapproved: (…) affrication, characteristic of the Balearic dialects, occuring by contact between the final sibilant of one word and the initial sibilant of the next word: tretsenyores for tres senyores ‘three ladies’, dotxuts for dos xuts ‘two shots’] (Institut d’Estudis Catalans 1990, p. 24).

No attention is paid to this systematic, consistently non-avoided feature of Majorcan and Minorcan Catalan in any of the paranormative texts generated in the Balearics and focusing either on mass media (Company and Puigròs 2006; Picó and Ramon 2005, 2006), on the school system (Alomar and Melià 1999), or on public use (Alomar et al. 1999). If disapproval in Institut d’Estudis Catalans (1990) was cancelled by Institut d’Estudis Catalans (2016), and there exists no regional disapproval of the feature, then the chances for a guided pronunciation are to be deemed low: the hesitation appears to be genuinely speaker-owned.

4.2.3. Stressed Schwa

The picture here is rather different from that of consonant assimilation or dissimilation of sibilants. Style books explicitly take sides on this issue. Differently from the paradox observed in Section 4.2.1, a consensus has been achieved between the central authority and the regional agents on this specific matter.

Whereas Institut d’Estudis Catalans (2016) again concedes by default, Institut d’Estudis Catalans (1990) meant a challenge against a traditional, prestige feature of the language, a situation worsened by the phonemic character of /ə́/. There exists no tradition of avoiding this vowel in the formal speech of educated persons in the Balearics. The main problem here was the aprioristic basis of Institut d’Estudis Catalans (1990, 1992), requiring the extent of two dialectal groups (out of five: Central, Valencian, Septentrional, Western, and Balearic) to give a feature the status “general” (suitable for addressing a language-wide audience). Features not meeting this condition were confined for usage within a single dialect territory and declared appropriate for standard usage only if judged traditionally prestige (“Són considerats propis d’un àmbit restringit aquells trets característics d’un determinat dialecte però proveïts de prestigi en el seu àmbit i que en conseqüència són recomanables quan la locució va dirigida a un auditori lingüísticament uniforme.” [Features belonging to one dialect but prestigious in their area are, thus, recommendable when speech is addressed to a linguistically uniform audience] (Institut d’Estudis Catalans 1990, p. 12), the emphasis is mine):

“correspon a l’àmbit restringit el sistema de vuit vocals diferenciades propi de parlars baleàrics, les set generals més la vocal neutra: cadena (pronunciat amb vocal neutra en posició tònica)” [the eight-vowel set of the Balearic dialects, comprehending the seven vowels of Catalan and the schwa, belongs to the restricted system].

The IB3 style book declares that “el sistema vocàlic integrat per vuit fonemes […] està totalment acceptat en la pronúncia estàndard, tant en els registres informals com en els formals” [the eight-vowel system (…) is totally accepted in standard pronunciation, both in informal registers and formal registers] (Company and Puigròs 2006, p. 28). There is no information on the agent of such “acceptation”. The word suggests a certain degree of explicitness, not attributable to the language authority at the time.

In their proposal for language usage in the schooling system, Alomar and Melià (1999) also recommend the eight-vowel set, and introduce language change into their considerations:

“La pronunciació de la vocal neutra tònica és tan genuïna i correcta com la pronunciació de e oberta en el seu lloc. Encara que el pas de la neutra a l’oberta no sigui rebutjable, ja que és un estadi d’un fenomen d’evolució normal espontània dins l’idioma, a cada lloc s’ha de recomanar i ensenyar la pronunciació pròpia” [The pronunciation of the stressed schwa is as genuine and correct as the pronunciation of an open e in its stead. Even if the evolution from schwa to open e is not fully rejectable, for it is a stage in normal language change, the local pronunciation should be recommended and taught in every place] (Alomar and Melià 1999, p. 23).

The same call for attention to the local pronunciation of unround middle vowels can be found in Alomar et al. (1999, p. 43).

Similarly to the speakers’ behaviours observed in Section 4.1.2 and Section 4.2.2, even if there is no explicit pressure on speakers for avoiding /ə́/, hesitation is found, and it points at a spontaneous shift towards the central standard variety.

5. Discussion

Some questions should be raised concerning the motivations of professional speakers for shifting models of pronunciation. First of all, evidence put forward in the previous sections point to a certain amount of disorientation, both individual and collective. Up to which point are prescriptive or paraprescriptive regulations to be held accountable for such degree of disconcertion?

Grondelaers et al. (2016) hold an interesting position on language standardisation relating to the different levels of explicit pressure on speakers:

“In view of the different ‘surveillance’ of the Flemish and the Netherlandic norm, one could argue that strict codification is more necessary in communities in which the standard is the result of repression and force, as in Flanders. In communities in which norm acceptance is a matter of consensus rather than insistence, there is less need to explicitly safeguard what has been agreed upon voluntarily. Codification, as a consequence, is probably a more relevant attribute of dominance-based language standards.” (Grondelaers et al. 2016, p. 127).

In this respect, Grondelaers et al. (2016) also point to the existence of language advisors in Flemish media, a position “unthinkable” (their words) in the Netherlands. It appears advisable to set up language surveillance systems wherever the status of a language is disputed, as it is the status of Catalan in its historical territories. In the main part of its territories, Catalan competes with Spanish, with the burden of lacking monolinguals: all of the Catalan speakers in Spain are also fluent in Spanish, but not vice versa. In addition, there is no common broadcasting system covering all the areas of the territory, not even those in Spain. Fragmentation entails the need for supplementary efforts to ensure language unity. Altogether, explicit codification arises as an attempt to maintain the language community united. As hard as it may be to ensure cohesion of the written language via prescription, the dispersion of the spoken language is even harder to restrain. Milroy and Milroy (1985) is already a classic on the matter:

“Attempts to prescribe standard usage in spoken language are never wholly successful, and quite unsuccessful at the level of pronunciation and grammar: however, the prescriptive tradition has succeeded in bringing about broad consensus on the norms of written language” (Milroy and Milroy 1985, p. 52).

The phonological features analysed in Section 4 point at Type B of the relation between dialects and standard according to Auer (2011) and already mentioned in Section 2. The evidence is consistent with that found in the domain of morphology, and already analysed in Dols (2020a). The chances are that the situation will turn into Type C, i.e., a situation in which dialects are displaced by the use of regiolects (as in Cerruti and Regis 20147) for functions previously assigned to them. Although Auer (2011) warns against the oversimplification of attributing the whole responsibility for the generation of regiolects to the dialects-standard advergence, the opening of the relationship between Balearic dialects and their endodialectal standard to a new agent, i.e., the standard of other eastern Catalan dialects, entails consequences affecting not only the shaping of the standard in use in the region, but also the general direction of language change. The evolution of project Corpus Oral de la Llengua Catalana will help us measure the pervasiveness of central Catalan standard solutions into the Balearic dialects. At present, the available data reflect only the formal registers of mass media.

The evolution of Majorcan Catalan as described here is not at all uncommon. On the contrary, alongside the cases from the Low Countries contributed by Grondelaers et al. (2016), and other cases from all over Europe by Auer (2011) (already cited above), the theoretical frame produced by Cerruti and Regis (2014) is based on the observation of Italian and Occitan dialects and regiolects, just to cite three instances of theoretically comparable proposals. The situation of Majorcan Catalan depicted here differs from those in Cerruti and Regis (2014) in a way explainable by means of the concept ‘alternative endoglossic standard’ (see Section 2 above). Our focus here is not on the reshaping of dialects by influence of the standard language, i.e., the downward convergence (Auer and Hinskens 1996; Cerruti and Regis 2014), but on the shift from one endoglossic standard variety to another one, endoglossic as well. In a way, this endoglossic shift is possible only because, in Auer’s words “the structural distance between the dialects and the standard variety is not the same across the dialectal territory. Some dialects (often those of or around the capital) are closer to the standard than others since the standard language was formed on the basis of these dialects” (Auer 2011, p. 492). The increasing exposure of Majorcan Catalan speakers to central Catalan mass media during the last decades may have started a centralisation process of the type cited in Auer (2005, p. 24). Our focus of attention in no case should be interpreted as tantamount to the inexistence of downward convergence in Majorcan Catalan. As mentioned above, the evidence analysed here belongs to a corpus still under construction and, for now, concentrating only on standard varieties.

If we turn now to the difference between exoglossic and endoglossic standards, there is still much to be said on the attested centripetal tendency of standard language in the Balearics, and sight should not be lost of the fact that Catalan is a subordinated language. The question arises on whether the attrition of dialectal features is due or not to new speakers having learnt Catalan as a second language by means of self-teaching material or by exposure to Barcelona-based mass media. As Pradilla Cardona (2020) states,

“[Spanish] plays an important role in the linguistic repertoire of those people whose first language is Catalan and it is the lingua franca of most of the international migrants who arrived around the turn of the millenium. […] The results [of Catalan teaching programmes to immigrants], however, have left a great deal to be desired; Catalan has always been learnt alongside Spanish but a command of Spanish often means that there is no real need to have command of Catalan” (Pradilla Cardona 2020, p. 413).

Not only the shrinking of the speaking community should be borne in mind when analysing social impacts in dialect levelling and standardisation, but also the influence of Spanish, partially acting as an exoglossic standard variety. As Lodge (2013) concludes in his research on the 17th century codification of French,

“It is difficult to see the grammarians operating in an atmosphere of serene rationality and detachment, and avoiding being caught up in the anxieties and insecurities of their time. Their evaluation of particular sociolinguistic variants looks to have reflected pretty accurately their subjective attitudes towards the social groups most associated with them” (Lodge 2013, p. 62).

Whether the present political tensions experienced in Catalonia affect codification anew is a non-trivial matter, though falling out of the scope of the present paper.

Regarding both the horizontal (territorial) and the vertical (explicitness) axes of language prescription, Dols (2021) points out an evolution towards a more open conception of the codification task. Despite the persistence of the sociolinguistic situation outlined here, recent documents of the central language authority in Catalan, Institut d’Estudis Catalans, imply a change of direction in comparison with earlier attempts to lead the standardisation process in an aprioristic manner, as described in Institut d’Estudis Catalans (1990, 1992). The new approach to standardisation coincides with the rise of multi-indexicality in late modern societies, as suggested in Grondelaers et al. (2016). According to the authors this evolution in society impels the speakers towards the search for models based in a new concept of prestige (‘new prestige’) based on dynamism (‘coolness’) and not so much on status (‘traditional prestige’) as in Modernism age (Grondelaers et al. 2016, p. 133).

Whichever reward the speaker receives in exchange for the adoption of a specific standard, the process can be analysed in terms of gain and loss. Coolness, or prestige in a more traditional way, can play that role. A different economy-based approach to the question has been proposed in Costa-Carreras (2020, p. 96) by applying Gazzola’s (2014) concept of equity in language policies. In this approach, the effort deployed by speakers to adhere to the standard variety calls for a counterbalance in terms of the attention paid by the codification process to the dialects. Although the results of codification can be analysed according to this gain-and-loss scheme, attention should be paid to the dynamism of the relation, in the sense that codification can influence dialects, and thus reduce the distance, if the social stimuli are sufficient to attract speakers to the codified version of the language.

6. Conclusions

It seems that the ingredients are ready to attempt an explanation for the observed Balearic Catalan paradox: while the language authority is more open to social multi-indexicality and loosens requirements on unity, trends in formal speech are those of centripetal options disregarding the original dialect. There is enough evidence to hold that Balearic speakers tend towards a more unified variety, due to the perceived dynamism of Barcelona’s dialect just when the language authority responds to evolution in society with closer attention to dialects. If this diagnosis is right, Catalan shows green shoots of internal dynamism in the development of a standard language based on implicit social consensus and not on direct prescription. Further research is needed to determine whether the evidence found in mass media oral language is widely used in other registers and to which point this implies a real case of dialect–standard convergence.