\u201cModern Irish\u201d is a cover-term for several Celtic dialects belonging to the Goidelic branch\u2013 including the standardised variety Caighde\u00e1n Oifigi\u00fail (\u201cOfficial Standard\u201d)\u2013, spoken nowadays as inherited languages in the non-continuous Gaeltacht areas of Ireland, and as learned varieties throughout the Island.
\nModern Irish varieties have nominative-accusative alignment and preserve two cases. Nominal case-marking, however, opposes an all-purpose unmarked case, used for A/S/P and and after prepositions, to an adnominal genitive case also used for arguments of non-finite forms. Pronominal arguments, on the other hand, partially preserve a distinction between nominative (A/S with active forms) and accusative (P, and S with passive and copular forms) and residual verbal indexing of A/S (see relevant sections below). Therefore, the label \u201cnominative\u201d has been used in the coding frames for A/S subjects of active forms, although this sort of marginal case-marking always combines with word-order coding.
\nThe coding frames given here include all arguments or adjuncts of the pre-defined role frame, plus a few additional ones that were found regularly in the sources. The database includes as coded alternations passive and various non-finite constructions, plus a plethora of uncoded alternations (such as, e.g., reflexive, oblipatient, reciprocal). Each alternation is included for a verb meaning if it is attested for that counterpart verb in one of the two major corpora (see below Source of the data).
\nThe major resources for argument/adjunct identification are word order and prepositions. Although case-marking applies regularly to all noun phrases, the two case-forms are an adverbal/adpositional form (that follows verbs and prepositions) and an adnominal, genitive form that follows nominal heads. Therefore, case-marking is not the main device to flag A/S/P arguments with finite forms.
\nOnly a few case-marked pronominal forms distinguish A/S vs. P, namely 2SG t\u00fa vs. th\u00fa, 3SG masculine s\u00e9 vs. \u00e9, feminine s\u00ed vs. \u00ed, 3PL siad vs. iad. On the other hand, depending on dialect, first persons better preserve A/S indexes with active verbal forms, which usually alternate with independent subject pronouns after non-personal forms, e.g. t\u00e1im and t\u00e1 m\u00e9 \u2018I am\u2019, chuireas and chuir m\u00e9 \u2018I put (PST)\u2019 (but not *t\u00e1im m\u00e9 or chuireas m\u00e9 with both person index and independent subject pronoun). Unmarked word order is VSOX and A/S arguments regularly follow the verb and precede P and prepositional arguments.
\nThe genitive case is used to flag either P or, more rarely, A/S with non-finite forms. Given widespread homophony of forms in some inflectional classes, case-distinctions are more easily overt with definite noun phrases, due to the different initial consonant mutations triggered by the definite article, e.g. tithe \u2018houses\u2019 (unmarked or genitive case), but na tithe \u2018the houses\u2019 (unmarked case) vs. na dtithe with initial /d/ \u2018of the houses\u2019 (genitive case). Note that while nominal genitives always follow their syntactic head, pronominal genitives (possessives) are proclitic to their nominal heads, on which they trigger initial consonant mutations. Since this is a general paradigmatic rule and exceptionless, the two strategies have been considered the same for the sake of coding frames.
\nPrepositions trigger initial consonant mutations on a following indefinite noun, while a following definite article may fuse with the preposition but is not affected by the same mutation as nouns, and triggers its own mutation on the following noun depending on case, e.g. do dhaoine with initial /\u0263/ \u2018to people\u2019, do na daoine with initial /d/ \u2018to the people\u2019. Most prepositions are followed by the unmarked case, but a few prepositions which have nominal origin require the genitive case, e.g. chun \u2018towards\u2019. The prepositions most frequently used to flag arguments are: the superessive and malefactive preposition ar \u2018on\u2019, which is also used for experiencers with some predicates; the instrumental, comitative and adessive preposition le \u2018with, by\u2019; the locative and allative preposition i \u2018in, into\u2019; the datival preposition do \u2018to\u2019 (regularly used for recipients but sometimes also to flag A/S in non-finite constructions). Less frequent ones are: the ablative prepositions as and \u00f3; the locative and agentive preposition ag \u2018by\u2019, which also flags some experiencers; the locative (and temporal) preposition roimh \u2018before\u2019 which may flag stimuli; and the subessive preposition faoi \u2018under, about\u2019. Note that pronominal complements of prepositions are obligatorily expressed as indexes as in Old Irish; only a few prepositions have lost pronominal inflection (see below Characterization of indexing resources).
\nDitransitives usually adhere to the indirective alignment.
\nReflexive and reciprocal constructions resort to the addition of independent lexical means, namely f\u00e9in added to a pronominal direct or prepositional object for reflexives and (a) ch\u00e9ile, on its own or preceded by a preposition, for reciprocals.
\nWhile Old Irish was radically pro-drop (no pronominal argument could be overt except as an index), Modern Irish requires A/S independent subject pronouns, and in some dialects, including the standard variety, even expletives (for example see RAIN, but also examples 46, 398). This holds with the exception of a few relic 1st, 2nd and 3rd plural person-inflected forms, as mentioned above (see e.g. example 143). No trace of object indexing is on the other hand preserved in Modern Irish varieties. Pronominal objects and the pronominal S arguments of passive and copular forms tend to be displaced in clause-final position (VSXO, VXS).
\nRelativized A/S/P arguments, whose syntactic function in the relative clause is coded through a distinct relative particle and a specific initial mutation on the verb, are considered the same as non-relativized ones.
\nPassive forms, termed autonomous forms in grammars, qualify as desubjectives, since pronominal S arguments surface as object forms, i.e. they are coded the same as P. With transitive and intransitive verbs, passive forms may behave as impersonals, i.e. the A or S argument is suppressed, interpreted as non-referential indefinite, and only prepositional arguments are overt.
\nPrepositions inflect for person, i.e., pronominal dependents of prepositions are obligatorily fused with them (head-marking), e.g. to Conall = do Chonall, to him = d\u00f3 or dh\u00f3, to me = dom or dhom, to you (sg.) = duit or dhuit. Cross-reference of nominal objects of prepositions is generally prohibited, although a few prepositions adopt the 3SG masculine inflected form before definite noun phrases, e.g. le duine \u2018with a person\u2019, leis an duine \u2018with the person\u2019, cf. leis \u2018with him\u2019.
\nWord order is the main device that keeps A/S and P apart, through VSO order, in particular with nominals, since there is no nominative/accusative case distinction. As mentioned above, pronominal objects and the pronominal S arguments of passive and copular forms, whether they have different case-forms for A/S and P or not, tend to be displaced in clause-final position (VSXO, VXS).
\nArguments may be fronted for pragmatic reasons, mostly in left dislocations (with resumption through a pronominal form) or clefting, which, broadly speaking, are topicalising and focusing constructions respectively.
\nDifferent word orders occur with non-finite verbal forms; they may be grouped into two main constructions (VS/O vs. S/O do V):
\nStrictly speaking, only in the second construction is word order the crucial element for argument identification. The two main non finite-constructions have been labelled as follows: the first one as verbal noun/infinitive and the second one as verbal noun with preposed argument/infinitive 2.
\nAs note above, while nominal genitives always follow their syntactic head, pronominal genitives (possessives) are proclitic to their nominal heads, on which they trigger initial consonant mutations.
\nData were collected in 2025 for the 80 core meanings in ValPaL, plus a few additional ones that correspond to the Old Irish added meanings. Counterparts were selected generally among the ones suggested in the New Irish-English dictionary, relying on the Corpas N\u00e1isi\u00fanta na Gaeilge (CNG: Bhreathnach, \u00da., M\u011bchura, M., \u00d3 Cleirc\u00edn, G., \u00d3 Meachair, M., \u00d3 Raghallaigh, B., Scannell, K. & U\u00ed Dhonnchadha, E., Corpas N\u00e1isi\u00fanta na Gaeilge \u2013 National Corpus of Irish. Dublin City University, 2024) and on the dialect repository Taisce Chan\u00faint\u00ed na Gaeilge (Palandri, A., Bhreathnach, \u00da., M\u011bchura, M., \u00d3 Cleirc\u00edn, G. & \u00d3 Raghallaigh, B., Taisce Chan\u00faint\u00ed na Gaeilge - Repository of Irish Dialects. Dublin City University, 2025), which have both been sources for the examples. The Corpas N\u00e1isi\u00fanta na Gaeilge is a broadly lemmatised collection of both written and spoken texts (reported in standard spelling), which span the period 2000-2024. It is a balanced corpus of 100 million words, with written sources however largely prevailing. The repository Taisce Chan\u00faint\u00ed na Gaeilge is on the other hand an audio archive, a collection of spoken texts recorded between the 1940s and 1990s in various Irish-speaking areas, originally broadcast on Raidi\u00f3 \u00c9ireann and Raidi\u00f3 na Gaeltachta, converted into audio files with accompanying adapted spelling.
\nExamples in the database are therefore generally taken from contemporary written sources if from CNG, from contemporary spoken sources if from CNG, TG4 (Irish Television Channel) or if otherwise specified in the comments, and from 20th century spoken sources if from Taisce Chan\u00faint\u00ed na Gaeilge. An attempt has been made to identify basic frames relying on mutual agreement of both corpora.
\nA single example from the Connemara dialect is taken from Caint Chonamara, a collection of everyday conversations recorded in the Connemara Gaeltacht in 1964, as quoted in the Briathra/Verbs section of the dictionary Pota Focal (Wigger, A., Valency dictionary of Irish Verbs).
\nThe main contributor, Elisa Roma, is associate professor of General and Celtic Linguistics at the University of Pavia. Andrea Palandri, Comhalta Taighde/ Research Fellow at Fiontar & Scoil na Gaeilge, Ollscoil Chathair Bhaile \u00c1tha Cliath (Dublin City University), has given fundamental input to employ the repository of Irish dialects, and has suggested counterpart choices, examples and emendations.
", "latitude": 53.22, "longitude": -7.62}, "name": "Irish"}, "geometry": {"type": "Point", "coordinates": [-7.62, 53.22]}, "id": "iris1253"}, {"type": "Feature", "properties": {"values": [{"pk": 4291, "basic_codingframe_pk": 2191, "original_script": null, "simplex_or_complex": "Simplex", "comment": "deponent, denominative from ferc 'anger'; fergaigidir is the eDIL lemma, but the spelling in all occurrences in CorPH is with c (fercaigidir)Old Irish is the Celtic variety of the Gaelic branch documented in various manuscript sources dating from the late 7th to the 9th century AD (see Source of the data below).
\nAny noun phrase (argument or adjunct) is always optional and it was not possible to apply a test using the sentence frame suggested in the ValPaL manual to distinguish between arguments and adjuncts. The coding frames given here include all arguments/adjuncts of the pre-defined role frame with the corresponding Old Irish case-marking, plus a few additional ones that were found regularly in the sources.
\nThe database includes as coded alternations passive and non-finite constructions, but since A/S/O pronominal arguments must be indexed on verbal forms, verb inflection is regularly affected by the nature of arguments (see Characterization of indexing resources below). Each non-finite construction is included for a verb form if it is attested in the corpus for that verb.
\nThe major resource for argument/adjunct identification is case marking, which applies regularly to all noun phrases. Case forms used to flag arguments with finite verb forms are nominative (A/S) and accusative (O). The dative case is used with prepositions, alongside the accusative case, while the genitive is an adnominal case and may be used to flag arguments only in non-finite constructions (i.e., with verbal forms which have nominal rection), or when governed by prepositions which have nominal origin (e.g. dochum + gen \u2018towards\u2019, from a proclitic form of tochim \u2018(act of) stepping to\u2019).
\nThe prepositions most frequently used to flag arguments are: the dative preposition do (regularly used for recipients but also to flag A/S in non-finite constructions), the allative prepositions i+acc and co+acc, the ablative prepositions a+dat and \u00f3+dat, the locative preposition i+dat, the superessive and malefactive preposition for+acc. Less frequent ones are the directive preposition fri+acc, which flags addressees and in general faced participants, and the adessive preposition la+acc, which may flag experiencers, companions and agents with passive and non-finite forms. Note that pronominal objects of prepositions are obligatorily expressed as indexes (see below Characterization of indexing resources).
\nDitransitives usually adhere to the indirective alignment, with few exceptions (notably TEACH, that allows secundative alignment as well).
\nWhile nominal subjects in the nominative case are always cross-referenced by verbal endings on the verb, all pronominal direct arguments (A/S/O) are indexed on verbal forms and may not be expressed with independent, case-marked pronominal forms. They are indexed as follows:
\nPassive forms with speech-act participants, therefore, qualify as desubjectives. Since the alternation between case-marking and verb indexing for different persons is paradigmatic and exceptionless, the two strategies have been considered the same for the sake of coding frames. In a similar vein, relativized arguments, whose syntactic function in the relative clause is coded on the verb, are considered the same as non-relativized ones.
\nPrepositions inflect for person, i.e., pronominal dependents of prepositions are obligatorily fused with them (head-marking), e.g. to Conall = do Chonall, to him = d\u00f3, to me = dom, to you (sg.) = duit. Cross-reference of nominal objects of prepositions is however prohibited.
\nDespite having regular nominal and accusative case-marking and subject agreement, word order in Old Irish is rather rigid, at least strictly verb-initial, and regularly VSOX. Prepositional pronominal arguments, i.e. prepositions inflected for person, may precede a nominal direct object (VSXO). Arguments may be fronted for pragmatic reasons, mostly in left dislocations (with resumption through a pronominal index) or clefting, which, broadly speaking, are topicalising and focusing constructions respectively.
\nDifferent word orders occur with non-finite verbal forms; they may be grouped into two main constructions (VS/O vs. S/O do V):
\nStrictly speaking, only in the second construction is word order the crucial element for argument identification.
\nNon finite-constructions have been labelled and numbered as follows: 1 = O in the genitive, 2 = A/S in the genitive, 3 = only prepositional argument(s) overt, and 4 = no overt argument (only controlled arguments). The second construction above with preposed argument is labelled do+V(erbal)N(oun).
\nNote that while nominal genitives always follow their syntactic head, pronominal genitives (possessives) are proclitic to their nominal heads. Since this is a general paradigmatic rule and exceptionless, the two strategies have been considered the same for the sake of coding frames.
\nData were collected in the second half of 2023 and in 2024 for the 80 core meanings in ValPaL, plus a few additional ones, through the database Corpus Palaeohibernicum (CorPH). CorPH includes 13123 finite verb occurrences and 5340 non-finite verb occurrences; it comprises ca. 70 texts of different genres and sizes, covering the time-span labelled Early Irish (7th-10th cent.), which lumps together Old and early Middle Irish. Counterparts to verb meanings were selected searching the electronic Dictionary of the Irish Language eDIL. Occasionally, examples were taken from the W\u00fcrzburg Glosses, or from eDIL, when occurrences were missing in CorPH but the patterns were clearly attested elsewhere, including somewhat later sources (Middle Irish). For the meanings COUGH and HUG, however, only tentative corresponding expressions could be provided.
\nMorphological glosses are meant to be as simple as possible. Although Old Irish features many proclitics, including the definite article, prepositions, possessives, copular forms, usually spelt together with the stressed word that hosts them and causing initial mutations, the equal sign is used only for a specific kind of clitics, traditionally termed notae augentes or emphatic particles, namely pronominal enclitics that are attached to any host that includes a pronominal index. Angle brackets are used for pronominal object morphemes that appear between a lexical preverb and a verb root.
\nThe contributor, Elisa Roma, is associate professor of General and Celtic Linguistics at the University of Pavia. Acknowledgments are due to Michele Tron (PhD student in Linguistics at the University of Pavia) for proofreading and format corrections.
", "latitude": 53.0, "longitude": -8.0}, "name": "Old Irish"}, "geometry": {"type": "Point", "coordinates": [-8.0, 53.0]}, "id": "oldi1246"}]}