Chair: Cormac Anderson
Part-of-speech (POS) tagging forms an important part of computational linguistics. Comprehensively tagged corpora are vital for refining the diachronic linguistic profiles of languages. Chronologicon Hibernicum aims at a chronological framework of the linguistic changes (phonology, morphology, syntax) that can then be used to date literary texts within the Early Irish period. So far, however, a fully functional POS-tagger is still lacking for Old Irish. This paper will discuss the theoretical and technical framework for building an automatic tagger. It will then focus on the steps involved, where issues arise and how to possibly solve them and give an outlook on future applications.
This paper discusses oblique subjects marked with doin Old Irish copular expressions involving ‘adverbial’ predicates. This kind of construction is exemplified by (1) and (2), below. Such oblique subjects are widespread in the Old Irish with various kinds of predicates. For instance, Le Mair et al. (2017) discuss possible oblique subjects in nominal/adjectival predicates, as in (3). This analysis, however, depends on the view that nominative marked items such as cobirin (3) are not subjects. In contrast, examples like (1) and (2) are intrinsically interesting because there is no other item in the sentence beside the predicate and the oblique phrase that could be analyzed as a subject. This paper develops a formal syntactic approach to understanding the oblique subjects with such adverbial predicates. The paper makes use of data tagged within the ChronHib project such as the Poems of Blathmac, the minor glosses, Wb., Ml., Sg., and various Old Irish prose and poetic texts attested in contemporary manuscripts (see Thes. i and ii).
(1) huare ní in óen diaithir doib
'since they are not in one orbit' (Thes. ii 13, fol.18d5)
(2) Is arafia dom
'I have it in my power' (Thes. i 3.9, fol. 4a)
(3) isgnáthdocobir cach lobir hifochidib
'He is wont to help every feeble one in tribulations' (Wb. 16a31)
This paper will look at a few linguistic variations that began to occur in the Old Irish period, especially those related to the passive and deponent verbal forms, for instance, non-deponent inflection for originally deponent verbs, deponent forms for active inflectional forms, and pres.ind.pass.sg. -ar/-(a)ir > -thar, -th(a)ir in strong verb inflection. These variations are tagged in the corpus built by the Chronologicon Hibernicum project that hosts more than 100,000 tokens from Old Irish texts, including the Milan and St. Gall Glosses, Stowe Missal, Poems of Blathmac, the Book of Armagh, etc. Quantitative data are retrieved from the occurrence of the variations across different Old Irish texts, and statistical analysis will show when these variations happened and how they spread during the Old Irish period. The result will contribute to the overall endeavour of the ChronHib project to achieve a refined diachronic linguistic profile of the Old Irish language.