Chair: Herve Le Bihan
This paper examines mutation in Breton following the numerals daou/div ‘two’ and tri/teir ‘three’. Descriptions of Breton (e.g., Press 2009; Ternes 1992) state that lenition is used following daou/div and spirantisation following tri/teir. However, it has also been noted that lenition is expanding at the expense of spirantisation, including following tri/teir. Hennessey (1990) argues that this is not language attrition, but rather normal language change, attested in Le Roux’s (1924-1963) linguistic atlas, and beginning in the eighteenth century. Today, Standard Breton maintains spirantisation after tri/teir, but regional varieties may not. Fieldwork was undertaken in southwest Brittany to investigate this variation, and to examine the usage of younger Breton speakers who have acquired Breton through immersion schooling. Three groups of speakers were interviewed: older, traditional speakers, younger adults, and teenagers currently in Breton-medium education. Lenition following daou/div is largely well-maintained by all adult speakers, but there is more variation among the teenagers. Mutation following tri/teir shows a much greater degree of variation. Older speakers use lenition, as expected from the literature. Younger speakers may use the Standard Breton spirantisation, or use lenition (especially if they have older family members who speak Breton). Alternatively, they may use a combination of both mutation types. The teenaged speakers omit the mutation entirely in a larger proportion of cases than the adult groups. This high degree of inter- and intra-speaker variation is not surprising given the variation that younger speakers acquiring Breton receive in their linguistic input, and the tendency for lenition to replace spirantisation.
Because of the collapse of family transmission of Breton over the last few decades, bilingual schooling through the medium of Breton now appears to be the main opportunity for the language to be preserved and to gain speakers in a substantial number (Broudic 2009, Moal 2009). Fortunately, a societal demand for such a Breton education does exist: as of September 2018, some 18,000 pupils attend an immersive or bilingual institution, from kindergarten to high school. Furthermore, this trend is supported by public administration: the regional council of Brittany has implemented a language policy since 2003, which was strengthened in 2015 by a new agreement between the Region and the French central State.
Nonetheless, even though more and more researchers are working upon the revitalization of Breton, most studies focus on quantitative and external aspects (Adam 2015, Chantreau 2017, Chauffin 2015, Planche 2002). There are just a handful of studies analysing Breton as it is actually spoken by children in the educational system (Davalan 2000, Kennard 2014, Madeg 2010). This is a particularly important field of investigation however: the Canadian experience has revealed that due to the pressure of the children’s first language, there may be significant differences between the language spoken in a classroom and the language heard among native speakers (Baker 2011: 268). Moreover, even when the correctness of the target-language is not in question, Mari C. Jones (1998) has proved that schooling, because it reinforces standardization, can result in a dialect-loss situation, where local peculiarities are lost in the youngest’s’ speech. To the extreme, young educated and traditional local speakers may not be able to speak one to another, which in turn challenges the legitimacy of such bilingual schools in public opinion.
With this theoretical framework in mind, I have chosen to study the Breton spoken by a group of children attending a bilingual class in central Brittany. In a region where Vannetais Breton is still spoken on a daily basis, the local speakers are indeed facing pressure on two fronts: that of language death, to the benefit of French, and that of dialect replacement by a new School-standard variety. Following a few preliminary attempts, during the spring of 2017 a limited group of eleven pupils, aged 7 to 11, passed a broad series of tests aimed at describing internal aspects of their Breton, on the grounds of phonology and grammar. For comparison, the same tests were later given to an aged traditional speaker who had no knowledge of written Breton. The children were also interviewed about their perception of Breton spoken by dialect or standard speakers, to evaluate their ability to socialize through the medium of Breton. Not surprisingly, the first results show the clear influence of French, through the dominance of S+V syntactic patterns or in the loss of opposition between long and short vowels, a thing that schoolteachers may not be fully aware of.
Depending on the degree of progress in data processing, a sample only of the general study may be presented.
Determining the limitations of phonological and morphophonological rules has long been a driving force in linguistic research. One such proposed curtailment is the restriction of (morpho)phonological rules to apply only to locally adjacent elements. Breton masculine animate plural nouns appear to violate this locality.
Masculine animate plural nouns are lenited following the definite article:
paotr boy paotred boys ar baotred the boys
However, masculine animate plurals which take the plural ending -(i)où are not lenited:
pried husband priedoù husbands ar priedoù the husbands
The definite article and the plural desinence are not in a local relationship and therefore one should not be able to influence the other.
Adopting a Distributed Morphology framework (Halle & Marantz 1993, Embick 2015), a solution is possible. Data from Latin (Embick 2000; 2015: 100-9) show that during Spell Out, the syntacticosemantic features are still present on a terminal node even after it has been given phonological substance via Vocabulary Insertion. Specifically, by the time the Latin pluperfect laud-ā-ve-rā-m I had praised selects the 1sg. past exponent -m (as opposed to the 1sg. non-past exponent -ō), the other exponents have already been inserted: laud-ā-ve-rā-. However, -m is conditioned by the syntacticosemantic feature [+past] and not by the phonological form -rā- since -m occurs in the imperfect too: laud-ā-ba-m I was praising. Therefore, the feature [+past] must still be present in an adjacent node at the time that the person/number terminal node is spelled out.
Since syntacticosemantic features are still present after Vocabulary Insertion, it is reasonable that they be subject to readjustment rules. It is proposed that the feature [+animate] is deleted when immediately local to the plural exponent -où.
Readjustment Rule: [+animate] > ø / ______-où
With the loss of the feature [+animate] from the noun, it no longer satisfies the environment for lenition triggered by the definite article.
Heretofore, Readjustment Rules have been triggered by and have targeted only the phonological content of morphemes and not their features (Embick 2013, Shwayder 2015). Impoverishment Rules, on the other hand, affect features and not phonological content. It is argued that the Breton rule cannot be an Impoverishment Rule since such a rule would miss the generalization that it is not a random assortment of nouns which fail to undergo lenition, but precisely only those with the plural ending -où.
Few animate nouns take the plural ending -(i)où. It is proposed that these nouns were so anomalous on the animacy-inanimacy cline that in time they were viewed as sharing the morphological qualities of the large class of nouns with plurals in -(i)où. As a result, they lost the syntacticosemantic feature [+animate] in the plural and failed to undergo lenition after the definite article.