Print Friendly

Měchura, Michal Boleslav, Dublin City University, Ireland,


The Placenames Database of Ireland is a bilingual database which records the official names in Irish and English of geographical objects on the island of Ireland. The database contains over 100,000 entries, is accessible to the public for free ( and serves around 150,000 searches per month.

The public-facing website was launched in 2008 in the context of recently enacted legislation which gave more prominence to Irish-language placenames in the Republic of Ireland; for details on the legal and sociolinguistic situation of placenames in Ireland see Mac Giolla Easpaig (2008, 2009). For more insights into the history and role of placenames in Ireland, see Mac Mathúna (1990).

We will not attempt to describe the structure of the Placenames Database of Ireland holisticly here. Instead, we will concentrate on three sub-areas where interesting challenges have arisen. In many cases, the issues presented here are issues that have not been solved satisfactorily in our database yet, or a solution has not been implemented yet. In this sense, the paper is merely a discussion of issues rather than a catalogue of solutions.

Linguistic properties of placenames

Placenames databases differ in how much attention they pay to the linguistic properties of the names themselves, such as their gender, inflection and so on. Multilingual databases such as Geonames usually ignore such aspects completely. Smaller databases, however, often focus on only one or two languages and are conceived mainly as a lexical database rather than a geographic one. Linguistic properties relevant to placenames can be broadly subdivided into the following three areas.

Internal structure. Practically all placenames that consist of more than one word have an implicit internal structure, just like other linguistic expressions. Some aspects of this structure may need to be made explicit in a database. These include definite articles and other textual phenomena that get in the way of alphabetical sorting including initial mutations in Celtic languages; composite names such as Ballaghgowla and Froghan; and names with disambiguators such as Black Lough (South).

Combinatorial properties. These are linguistic properties relevant to how the placename interacts with the text in which it is being used and include language-specific features such as gender, grammatical number and inflection paradigms. In addition, the following combinatorial properties are relevant to placenames in particular: how the name can be combined with categorizers such as ‘town’ or ‘county’ (Donegal Town but County Donegal); which locative prepositions are combined with the placename (i nGaillimh ‘in Galway’ but ar an gCeathrú Rua ‘in Carrarow’, literally ‘on Carrarow’).

Lexical relations to other words in the language and to other placenames. The former include demonyms (terms for ‘people from’) and derived adjectives which are sometimes irregular. The latter include cases where one geographical object has been named after another, such as Ballybeg Road or Lismore Terrace.

Cross-linguistic issues

A separate cluster of issues stems from the fact that the Placenames Database of Ireland is bilingual. This does not seem to pose a challenge at first; the general principle is that every place has two names, one in each language, and this seems to call for a simple data structure: all we need is two text fields. However, that approach would fail to account for the following phenomena.

Borrowing and gaps. Sometimes, a place has the same name in both languages. An example is an area of Dublin called Dún Laoghaire. This is an Irish name which is also used in English with unchanged spelling (but with anglicised pronunciation). An obvious solution would be to simply record Dún Laoghaire twice, once as an Irish name and once as an English name. However, this is unsatisfactory as it fails to account for the fact that this name is not really in English, it is merely used in English.

On the other hand, in some strongly Irish-speaking areas, minor features such as crossroads, fields and wells only have Irish names and there are no known English names. One can only assume that if somebody needed to refer to such a place while speaking English, one would briefly code-switch into Irish to utter the name.

A fairly granular data structure is called for here, one that allows us to capture facts as to whether a name in a given language exists or not, whether the name used in a given language is also a name in that language, and whether the name has been borrowed from another language.

Anglicisation, translation and re-interpretation. Many English placenames in Ireland have been obtained from the original Irish names by a process of writing down an approximated pronunciation (Gaoth Dobhair -> Gweedore). In other cases, when the English name was created first, the method used to coin the Irish name has often been translation, such as Butler’s Bridge -> Droichead an Bhuitléaraigh. In other cases still, the Irish and English names are independent coinages (example: Loch Garman/Wexford).

Translation is sometimes accompanied by re-interpretation. An example of this is an area of Dublin called Barra an Teampaill/Temple Bar.The placename originated in English from the personal name Temple but was later re-interpreted as the common noun temple and hence the Irish name (literally ‘the bar of the temple’). Although based on a misunderstanding, it was decided to keep the Irish name as official because it is commonly used.

An ideal data structure would allow us to record these and other etymological relations between names of the same place in different languages. If such relations are explicated and annotated in the database, then we can not only provide better information to users but also extract interesting statistical observations about the relative proportion of these phenomena in the country’s body of placenames.

Dealing with overlapping hierarchies

Most countries are subdivided into administrative units such as districts, provinces or counties. These units form a hierarchy and such hierarchies are deliberately designed as a nested hierarchy to disallow overlapping and facilitate reasoning.

In Ireland, the situation is far from this computational ideal. The basic units (counties -> baronies -> civil parishes -> townlands) may have originally been designed as a nested hierarchy but are no longer so because of boundary changes that have not been propagated up and down the hierarchy and because of the introduction of competing hierarchies which overlap with the traditional system.

The consequence when building a placenames database is that we must work with an overlapping hierarchy where a child may have more than one parent. This complicates things; for example, reasoning is no longer always possible. If we know that A is in B and that B is in C1 and C2 simultaneously, we can no longer infer whether A is in C1 or C2 or both. In fact, there is no way to know this other than by deriving it from a dataset of geographical boundaries, or by recording it explicitly.

Place as an abstract concept

Consider the placename Dún na nGall/Donegal which can be found in the north-west of Ireland. The question to ask is: how many places called Dún na nGall/Donegal are there in this corner of Ireland? The answer will differ depending on one’s perspective. An outsider will see only one, a county of that name. A local inhabitant will probably see two, the county and its capital town of the same name. A local politician will probably see an electoral division of that name and also a town of that name with its town council. In total, there are five units called Dún na nGall/Donegal in that part of Ireland.

In our database, these are treated as separate objects which, as far as the database knows, only happen to have the same name. That, however, is unsatisfactory as it fails to distinguish a case like this from cases such as the 19 places called An Baile Mór (literally ‘the large town’) which can be found all over Ireland and which genuinely happen to have the same name. Another reason why this is unsatisfactory is that it introduces a potential for inconsistency because data such as the Irish name’s grammatical information need to be recorded five times instead of once.

An ideal data structure would provide a way to connect several places to a single ‘abstract place’. The abstract place would contain all information common to the concrete places, such as names and historical citations, and these would then be inherited by the concrete places.


Many of the issues illustrated here are inherently linguistic and stem from the fact that we conceive of our database as primarily a lexical database and only secondarily as a geographical database. A second cluster of issues is caused by the heritage of conflicting and overlapping administrative hierarchies and from the differences in perspective these hierarchies impose.


A. Placenames databases


Placenames Database of Ireland:

B. Literature

Mac Giolla Easpaig, D. (2008). Placenames Policy and its Implementation. In Caoilfhionn Nic Pháidín, Seán Ó Cearnaigh (eds.), A New View of the Irish Language. Dublin: Cois Life, pp. 164-177.

Mac Giolla Easpaig, D. (2009). Ireland’s heritage of geographical names. Wiener Schriften zur Geographie und Kartographie 18: 79-85.

Mac Mathúna, L. (1990). Ár dTimpeallacht Logainmneacha: Inniu agus Amárach [our placenames environment: today and tomorrow]. Dublin: Coiscéim.