Logotip de la revista Noves SL





Autumn 2003

"Auliva" or "oliva"? Describing variation in the Segrià, by Josefina Carrera-Sabaté
and Elisabet Freixenet-Esteve

This article looks at processes involved in a sound change in progress in the Segrià (western Catalonia) by exploring the productive and perceptive competence of 30 speakers in the village of Sunyer. (1)


Printing version. Describing variation in the Segrià PDF printing version. 73 k



1. Introduction

2. Objectives and methodology

3. The phenomenon under study

4. Results
4.1 Production
4.1.1 Linguistic variables
4.1.2 Social variables
4.2 Perception of [aw], [o] and [u]
4.2.1 [aw] as an "odd" variant
4.2.2 The contrast between [aw], [o] and [u] and evaluations of [aw]

5. General considerations

6. Bibliography


1. Introduction

Ever since the contribution made by Labov (1966), the analysis of sociophonological material has been approached, often, via two differentiated but complementary dimensions: 1) linguistic production and 2) linguistic evaluation (Moosmüller, 1988). In linguistic production, the incorporation of social and situational aspects to explain the elements of a grammar consitutes one of the most notable developments in variationism. This approach allows a more dynamic, less static view of linguistic systems than found elsewhere (in structuralism and generativism). Furthermore, linguistic perception has been considered essential to round out the explanation of any process of linguistic change since "The people taking part in the change do not just speak: they also listen. Thus one must assume that variation and change occur not only in production but also in perception" (Janson, 1983: 24). Nevertheless, methodological difficulties involved in observing and analysing the behaviour of receptors, compared with the behaviour of producers, have meant there has been comparatively little interest in including a study of subjective perception of speakers in studies of linguistic change.

From a global perspective, elements such as prestige, attitudes or beliefs of speakers associated with individual learning capacity take on great importance. Similarly, aspects like the influence of the written language can become fundamentally important in explaining certain processes involved in change. Indeed, in Catalan, a definite tendency to adjust pronunciation to fit the spelling has been observed. This in turn results from a preconceived ideology which associates knowledge of spelling with knowledge of the Catalan language and love of one's country (Segarra, 1985). Such an ideology, moreover, has been further encouraged by the schools, since one of the prime objectives in primary, secondary and adult education has been acquisition of skill in the written language, to the apparent detriment of oral dialectal forms not reflected in writing (Siguán, 1990).

2. Objectives and methodology

The purpose of writing this article was to think about use and perception of the solutions [aw], [o] and [u] in pre-stress word-initial position in items like oliva (olive), ofegar (drown) or humit (damp) in speakers aged 19 to 85 years in Sunyer (2). The intention is to present an up-to-the-minute assessment of this process of phonological variation in generations born between 1917 and 1983, and observe how the language habits of the speakers have modified while bearing in mind divergences in knowledge of written Catalan and the education received by each generation.

The 30 subjects we selected constitute 9.8 % of Sunyer's population, which has 300 inhabitants all of them Catalan-speaking. Selection of subjects was done at random, but proportional to distribution of the population to ensure representativeness. Variables considered were, education, knowledge of written Catalan, sex, occupation and social class. The sample was divided into the following age groups:

a. 66 to 85 years (born between 1918 and 1937);

b. 40 to 65 years (born between 1938 and 1963); and

c. 19 to 39 years (born between 1964 and 1984). Each group contains 10 individuals who vary in terms of the above-mentioned social variables.

The productive and perceptive competences of these speakers were calculated using the Goldvarb 2001 programme (3) as detailed below:

a) The study of linguistic perception was based on a questionnaire containing closed questions on 14 words that could be realised with either [aw] or [o] or, in two words, [u]; these words feature formal differences determined by the immediate environment, the stressed vowel, the etymology and the placing of the accent. These were: orella (ear), oculista (occulist), olor (odour, smell (n.)), operar (operate), orinal (chamber pot) ofega (drown, choke), oreneta (swallow, (n.)), ocell (bird), ovella (sheep), oliva (olive), oració (prayer) otorinolaringòleg (ear nose & throat specialist) (pronounced [o t o " r i n o]), humitat (humidity, damp (n.)) ufanosa (luxuriant). Owing to the low lexical incidence of the variation we analyse here (4) we decided to obtain realisations of the variable in question by elicitation (5), and to disguise the true purpose of the study, we included in the questionnaire a number of questions on words that were unrelated to the object of our study, but connected in that they belong to the same semantic field.

b) The questionnaire on perception enables us to analyse, firstly the oddness factor, that is, the degree to which subjects saw the [aw] variant as odd, when realised in four words in the following sequence of sentences: L’han operat perquè es va engolir dos olives / Aquell pis estava ple d’humitat / Quantes oracions saps? (He was operated on because he swallowed two olives whole / That apartment was full of damp / How many prayers do you know?) We analysed responses to the question Nota/notes alguna cosa estranya en la pronunciació d'algun mot d'aquesta frase? (Do you (vostè) / do you (tu) notice anything odd about the pronunciation of any word in this sentence). From this we observe the capacity of the inhabitants of Sunyer to detect the diphthong [aw] in pre-stress, word initial position and determine whether maintenance of [aw] is viewed negatively or as a commonly heard sound, pleasant to the ear.

Secondly, the questionnaire on perception enables us to determine: 1) the subjects' capacity to discern phonetic differences and 2) the evaluations made by subjects on pretonic [aw] and [o] in words that behave differently in their selection of [aw] and [o] or [u] in pretonic position. The subjects heard the following words twice: olor, ovella, orella and ufanosa. The first three words were heard once with the [aw] variant and once with the [o]; the last word was heard with the variants [aw] and [u]. The aim here was, firstly, to discern the subjects' ability to discern the vocalic contrast in pretonic position in these words, and secondly, to analyse the connotations of the pronunciations in terms of speaker-characteristics and the maintenance evinced by each of the variants in the linguistic community. The questions we used to obtain data on evaluations of the   different pretonic variants were: 1) which is the vowel that you think is the good one? (6); 2) which is the vowel that you use; 3) which is the vowel that is generally used in Sunyer?

3. The phenomenon under study

In Lleidatà –the dialect of Lleida province– the vowel o of vulgar Latin deriving from classical Latin O, O, U and the diphthong AU, when in a free syllable and in pretonic (pre-stress) word initial position, have gone to [aw] (7) (Badia, [1951]1984: 164). We find a few examples in words like ovella < OVICULA, oració < ORARE or indeed orella < AURICULA. At the same time, classical Latin U maintained the vowel: UFANA> ufanós.

Utilisation of the [aw] diphthong in word initial pretonic position in Catalan has also been explained in terms of vowel agglutination: l[ao]vella (def. art. + sheep) >l[au] vella> l [aw]vella or l[oo]fici (def. art. + trade) > l[ou]fici >l[ow]fici>l[aw]fici. This second analysis finds support in Pueyo (1976), Veny (1982), Recasens ([1991]1996) and Solans (1996). Pueyo (1976) argues that the diphthongisation, which does not arise   in all cases, is an outcome of the etymology, since, in general terms, [aw] derives from O and U of classical Latin, although we find a few exceptions in words like ofrena (offering) or humitat (humidity, damp (n.9). What is more, he observes that the diphthongisation works more efficiently when the word in question is feminine, and less systematically when the word is masculine or higher register .

Recasens ([1991] 1996: 139) explains that the raising or closing of [o] ] "may be motivated by the particularly close realisation of /o/ in north-western Catalan, and by coarticulatory effects upon the vowel owing to the following labial or back consonant." Recasens adds that the lack of prestige accorded the diphthongal variant explains why it is declining in favour of the monothong alternative, and why educated or higher register words least likely to have the [aw] variant.

Lastly, Solans (1996) detects dwindling use of the diphthong among young speakers, who tend to use the standard variants. Note that the Institut d'Estudis Catalans ([1990]1992: 17) explicitly states: "Use of the diphthong au instead of o in cases like aufegar for ofegar, aulor for olor, etc, is not recommended".

4. Results

4.1 Production

Calculating probability values for the linguistic and social variables gives a probability of the maintenance of [aw] ain the village of Sunyer of 0.518. The most significant conditioning factors favouring the maintenance of the traditional realisation [aw] are, on the one hand, the syllabic and accentual configuration of the words in question and, on the other, the interviewees' age, education and knowledge of Catalan (see tables 1, 2 and 3). [Log. similitude: -215.993 / Max. log. similitude: -176.714 / X2 total: 64.0594 / p: 0.0000].

4.1.1 Linguistic variables

a) Syllabic and accentual make up of the words

The words most susceptible to being pronounced with the diphthong are three-syllable words and are acute (stressed on the last syllable, as for example operar. Contrariwise, the words which are most unlikely to be realised with the diphthong are four-syllable acute words, such as oració (prayer).(8)

Table 1. Syllabic and accentual disposition of the word


Probability of maintenance of [aw]

v’v (a) 0,51
vv’v (b) 0,73
v’vv (c) 0,31
vv’vv (d) 0,45
vvv’v (e) 0,09

4.1.2 Social variables

a) Age of informants

Among the oldest speakers we see a high probability the [aw], variant will be maintained, among the middle aged the diphthong alternates with the [o] variant at a roughly equal rate, while among the young, the diphthong becomes more infrequent (see table 2):

Table 2. Age of subject


Probability of maintaining [aw]

1918-1937 (66-85) 0,70
1938-1963 (40-65) 0,47
1964-1984 (19-39) 0,39

b) Education and written Catalan

Interviewees with less education and less knowledge of written Catalan are those that maintain the diphthong [aw] with the highest probability indexes, and the probability of the [aw] variant occurring falls off as speaker's schooling increases. At the same time it should be born in mind that this pattern is strong among interviewees with higher education. The fact is that, as has been observed in other studies (Ferrando and Guirau, 1983; Martí, 1985; Pueyo, 1980; Carrera-Sabaté, 2002) consultants with higher education tend to maintain the local variants of their own speech community more than those with secondary education (see table 3):

Table 3. Interviewee's education and knowledge of Catalan


Probability of maintaining [aw]

No education/ - Catalan (1)


Primary education / - Catalan (3)


Primary education / + Catalan (4)


Secondary education / + Catalan (6)


Higher education/ + Catalan (8)



1 de 3