Elliott
Lash s. xx–xxi
2022
Languages change constantly in all linguistic domains – phonology, morphology, syntax, and lexical use – and their graphic expressions are subject to fashions. Irish, a Celtic language spoken in Ireland, is in no way different. With a written history of more than 1,500 years, Irish is among the oldest attested languages in Europe. Because of its long textual tradition, its development through time is reflected in the huge amount of variation observable in the extant sources, i.e. texts in manuscripts from the 8th up to as late as the 17th and 18th century. The European Research Council-funded project Chronologicon Hibernicum (hereafter ChronHib; 2015–2021) has studied the diachronic evolution of the early medieval Irish language, best known as Old Irish. This article presents the major challenges posed by extant Old Irish texts and introduces two methods developed in the ChronHib project to study synchronic and diachronic variation in the extant material, namely variation tagging and Bayesian language variation analysis.
Languages change constantly in all linguistic domains – phonology, morphology, syntax, and lexical use – and their graphic expressions are subject to fashions. Irish, a Celtic language spoken in Ireland, is in no way different. With a written history of more than 1,500 years, Irish is among the oldest attested languages in Europe. Because of its long textual tradition, its development through time is reflected in the huge amount of variation observable in the extant sources, i.e. texts in manuscripts from the 8th up to as late as the 17th and 18th century. The European Research Council-funded project Chronologicon Hibernicum (hereafter ChronHib; 2015–2021) has studied the diachronic evolution of the early medieval Irish language, best known as Old Irish. This article presents the major challenges posed by extant Old Irish texts and introduces two methods developed in the ChronHib project to study synchronic and diachronic variation in the extant material, namely variation tagging and Bayesian language variation analysis.
2020
2019
2018
2017
The words etar and ceta have a first syllable with a variable vowel: either e (e-variant) or i (i-variant). This paper investigates the diachronic distribution of these two variants. The innovation of the i-variants occurred by the eighth century at the latest in ‘pretonic complexes’: preverbal and prenominal proclitic strings consisting of more than one element (for instance: preverb + relative mutation/pronoun, for example a n-itir·n-ūara ‘when it cools’ Ml. 71b5, or preposition + article, for instance hitar na doinmecha ‘among the adverse things’ Ml. 38a12). A statistical analysis of the Würzburg, Milan, St Gall, and certain minor ninth-century sets of glosses shows that the i-variant of ceta became more common than the e-variant in the late eighth century. Afterwards, in the ninth century, the i-variant of etar became statistically more common than the e-variant. A textual dating criterion is proposed on the basis of these results and comparison with other pretonic raising processes (do > du, ro > ru, tremi > trimi, etc.) is suggested.
The words etar and ceta have a first syllable with a variable vowel: either e (e-variant) or i (i-variant). This paper investigates the diachronic distribution of these two variants. The innovation of the i-variants occurred by the eighth century at the latest in ‘pretonic complexes’: preverbal and prenominal proclitic strings consisting of more than one element (for instance: preverb + relative mutation/pronoun, for example a n-itir·n-ūara ‘when it cools’ Ml. 71b5, or preposition + article, for instance hitar na doinmecha ‘among the adverse things’ Ml. 38a12). A statistical analysis of the Würzburg, Milan, St Gall, and certain minor ninth-century sets of glosses shows that the i-variant of ceta became more common than the e-variant in the late eighth century. Afterwards, in the ninth century, the i-variant of etar became statistically more common than the e-variant. A textual dating criterion is proposed on the basis of these results and comparison with other pretonic raising processes (do > du, ro > ru, tremi > trimi, etc.) is suggested.