Linguistic research in the early decades of the 21st century is increasingly characterised by the use of ‘big data’ such as digital corpora. However, the analysis of digital corpora can also be informed by the large-scale, largely manual, research that characterised the 20th and even the 19th century, namely the scholarship that produced the Oxford English Dictionary and its offshoot the Historical Thesaurus of English (KAY; ROBERTS; SAMUELS; WOTHERSPOON 2009). These vast lexicographical resources are particularly useful in the semantic analysis of English language corpora. The present paper illustrates the application of these lexicographical resources to corpus analysis by tracing one particular example of linguistic prejudice - the signification of ‘Irishness’ in English’ in the Oxford English Dictionary, the Historical Thesaurus of English, and the Hansard Corpus of British Parliamentary Speeches (ALEXANDER; DAVIES 2015).

At a meeting of the Philological Society on 15th January 1965, Professor Michael Samuels of the Department of English Language at the University of Glasgow announced his intention to produce a historical thesaurus of English. He later revisited his reason for undertaking this formidable project in Linguistic Evolution (SAMUELS 1972: 180):

In 2009, forty-four years, nine months and one week after the meeting of the Philological Society, a team, then led by Professor Christian Kay, brought this vision to completion, and the online Historical Thesaurus of English (HT) was launched, alongside a print version, the Historical Thesaurus of the Oxford English Dictionary (HTOED). The HT classifies almost 250,000 concepts, based on a common-sense or ‘folk taxonomy’ that was developed over the four decades of the project’s duration. Every sense of every word in a series of multi-volume historical dictionaries was collated and categorised by a team of lexicographers under the directorship of Samuels and then Kay. The conceptual domains expressed by these senses are first divided into The External World, The Mental World and The Social World, and then the words are sorted by increasingly fine distinctions according to their senses. The first two levels are shown in Table 1 below.

Table 1:

First level
01 The world	02 The mind	03 Society
Second level
01.01 The earth	02.01 Mental capacity	03.01 Society & the community
01.02 Life	02.02 Attention & judgement	03.02 Inhabiting & dwelling
01.03 Health and disease	02.03 Goodness & badness	03.03 Armed hostility
01.04 People	02.04 Emotion	03.04 Authority
01.05 Animals	02.05 Will	03.05 Law
01.06 Plants	02.06 Possession	03.06 Morality
01.07 Food and drink	02.07 Language	03.07 Education
01.08 Textiles and clothing		03.08 Faith
01.09 Physical sensation		03.09 Communication
01.10 Matter		03.10 Travel & travelling
01.11 Existence & causation		03.11 Occupation and work
01.12 Space		03.12 Trade and finance
01.13 Time		03.13 Leisure
01.14 Movement
01.15 Action
01.16 Relative properties
01.17 The supernatural

Source: (http://ht.ac.uk/downloads/HT_Semantic_Hierarchy.pdf)

Each semantic domain is coded, and the three primary categories (01 The World, 02 The Mind and 03 Society) afford 377 further categories, each of which allows successively fine distinctions. The concept ‘Irish’, for example, is coded as:

The coding distinguishes between the collective sense of ‘Irish’ as referring to (a) the nation collectively, and (b) any inhabitant of that nation. The coding assigned to the word (01.04.07.06.02|03) thus indicates that the word has been categorised successively as:

As Samuels foresaw, once each sense of every dictionary headword had been categorised, the nature of conceptual domains that had been dispersed throughout the dictionaries by their conceptual domains can be brought together and arranged chronologically, so their historical development can be examined. The collective sense of ‘Irish’ (attested from c1205), then, occupies a conceptual domain alongside other terms, including ‘Scottas’ (OE), ‘Irishry’

(1375-1827), ‘wild Irish’ (1399-1547), ‘mere Irish’ (1612), and ‘black Irish’ (1888-). Those interested in how the words relate to each other within this conceptual and chronological ecology need to refer back to the historical dictionaries from which the entries are taken. The OED, in this case, notes that ‘Irishry’ refers to ‘Irish people collectively; the Irish nation, specifically the native, Catholic Irish as opposed to Protestant settlers from England and Scotland’; ‘wild Irish’ referred to ‘Gaelic-speaking Irish people inhabiting the areas of Ireland not under English control, and subsequently also Irish people regarded (especially by the English) as uncivilized, primitive, or unruly’; ‘mere Irish’ is a late survival of a use of ‘mere’ meaning ‘Of a people or their language: pure, unmixed’; and ‘black Irish’ is an ‘adjective and noun, sometimes derogatory, describing an Irish person, or one of Irish ancestry, having dark hair and a dark complexion or eyes’.

By examining the members of the semantic domain together, as a lexical set, the Anglophone construct of Irishness is clearly evident: the Irish are a primitive people, dark-haired and dark-complexioned, adhering to their Gaelic language and Catholic religion. The HT also shows the chronological reach of these aspects of ‘Irishness’ in Anglophone written culture (Table 2).

Table 2:

The timeline visualisation is suggestive of those characteristics of Irishness that hold sway on the Anglophone imagination over time. The Old English term ‘Scottas’ is replaced by the term ‘Irish’; the expression ‘Irishry’ that is associated particularly with Catholicism is most attested from the 14th to the 19th century; the notion of the Irish as wild and primitive has its most productive period from the 14th to the 16th century; the notion of the Irish people and language as being ‘mere’, as in undiluted and pure, is relatively uncommon and historically archaic (altough as we shall see, it returns in the 19th century); and the conception of the Irish as being physically distinct, dark-eyed and dark-complexioned, dates from the late 19th century and continues to the present day.

There are, of course, other senses of ‘Irish’, not only as referring to the collective but to the individual, the native or inhabitant of Ireland. In this sense there is a much wider variety of expressions falling within the conceptual domain (see Table 3). A subcategorisation of this domain shows several patterns. The patron saint of Ireland is invoked in nicknames like ‘Paddy/Pat/Paddywhack/Patlander’ with ‘Mick’ and ‘mack’ as alternatives, the latter being the Celtic patronymic ‘Mac’. As for the former, according to Oxford Dictionary of Christian Names (s.v. Michael), ‘The name was not used in Ireland much before the 17th cent., but is now one of the commonest Irish names, often abbreviated to Mick, Micky, or Mike.’ The OED entry adds, ‘Use of the word was perhaps reinforced by the numerous Irish surnames in Mc-, and Mac-.’ The saints’ names ‘Patrick’ and ‘Michael’ also invoke a Catholic heritage.

Table 3:

Irishman c1205-	Teaguelander 1689-1732	West Briton 1816-
Ireis 1297	colloq.	Patlander 1820-
Yreis 1297	boglander 1730/6-1755	Greek 1823- slang
Irish 1613-1828	bog-stalker a1758	potato-eater 1823-
Scots mack 1617-c1688	brogueneer 1758-1840	Grecian 1853- slang
hooded man 1621 + 1641	Milesian 1771- joc.	Mick 1856- joc.
Teague 1661-1900	Paddy 1780- freq. derog.	Harp 1904- US slang
colloq.	Pat 1806-
bog-trotter 1682-1855	Irisher 1807-
dear joy 1688-1710	Paddywhack 1811-1846
colloq.

Other nicknames include ‘Teague’ (which the OED identifies as the ‘Anglicized spelling of the Irish name Tadhg, variously pronounced /teɪɡ/, /tiːɡ/, /taɪɡ/, fancifully identified with Thaddeus and its familiar form Thady), and ‘Teaguelander’. This name, often in the form ‘Taig’, can also be a ‘Protestant term of contempt for a Roman Catholic.’ Rather more classical nicknames are ‘Milesian’, from ‘Milesius, Latinized form of Irish Míl Espáine, the name of a legendary Spanish king whose sons (as reported in Lebor Gabála) are reputed to have conquered and reorganized the ancient kingdom of Ireland about 1300 b.c.’.

The remaining expressions are more or less explicable. ‘Dear Joy’ is described in the OED simply as ‘a familiar appellation for an Irishman’; the term is so defined in a dictionary of cant from the late 17th century, but, as with so many slang expressions, no clear etymology is given. In the 19th century, we learn, Irish immigrants to England referred to themselves, or were referred to, as ‘Greeks’ - the OED gives a citation from a dictionary of slang published in 1823 that informs us that ‘Irishmen call themselves Greeks-none else follow the same track to the east; throughout this land, many unruly districts are termed Grecian.’ A less classical set of epithets is ‘bog-trotter’, ‘boglander’, ‘bog-stalker’, picturing the Irish as those who inhabit or set foot across an unfertile and inhospitable wasteland. Again, a less than positive view of Irishmen is evoked by ‘hooded man’, a term that is explained by the poet, Spenser, around 1599: ‘For a Theife it [sc. the Irish mantle] is so handsome..he can in his mantle passe thorough any town or company, being close hooded over his head..from knowledge of any to whom he is indangered.’ Less common expressions, like ‘brogueneer’, refer to perceived characteristics of the Irish, for example that they speak with a strong accent or ‘brogue’, and ‘potato-eater’, a 19th century term that alludes to the staple food of many Irish people, and (in the century in which the term was coined) simultaneously evokes the blight that resulted in a catastrophic famine. The two remaining terms are worth remarking on: ‘West Briton’(1816-) draws on a unionising strategy also evident a full century earlier, in Scotland, to obscure traditional national nomenclature with ‘regional’ alternatives. The OED cites a letter from one J. Giffard to Sir Robert Peel on 19th March, 1816, (Brit. Library Add. MSS. 40,253, f. 258) in which he argues, ‘Had the whole [of the United Kingdom] been called by one common name Britain-we should have had the Inhabitants proud of the glorious Title Britons and we West Britons would have been as much conciliated and attached as the North Britons are.’ ‘Harp’ is a more recent Americanism, attested from the early 20th century, whereby the inhabitants of Ireland are referred to by one of its national symbols. The OED citations inform readers that, when tossing coins, ‘harp’ could mean ‘tails’ in Ireland (‘from Hibernia being represented with a harp, on the reverse of the copper coins of that country’) and the word was also used to refer to Irish women, probably as a reduced form of ‘harpy’.

The words in the conceptual domain of individual inhabitants of Ireland reinforce and give detail to the patterns evident in the collective domain of ‘The Irish’: from an Anglophone perspective, Irish individuals remain the Other: Catholic, Gaelic-speaking (or marked by a Gaelic accent), criminal, and impoverished (inhabitants of a wasteland, living on a diseased staple). There is a strategy, however, for incorporating the Other into the Self, namely by constructing the imagined nation as Britain and renaming Ireland as its western region. The timeline in the HT (Table 4) again shows the chronological rise (and occasional fall) of the words that articulate these concepts.

Table 4:

Keen collectors of words referring to Irish people will possibly note apparent absences in the HT inventory, one being ‘Fenian’, which the OED defines as, originally, a member of a mercenary army that supported the King of Eire, and, more recently, ‘one of an organization or ‘brotherhood’ formed among the Irish in the United Sttes of America for promoting and assisting revolutionary movements, and for the overthrow of the English government in Ireland.’ Accordingly, this term is found in the HT in the categories ‘pertaining to ancient warriors’ and ‘members of Irish political associations’:

In the former conceptual domain, the word ‘Fenian’ exists in a relatively small group of related terms for ancient Irish warriors, namely ‘kern(e)’ 1422 + 1550, ‘Fein(ne)’ 1782-, ‘Shanavests’ 1811-, and ‘Fenian’ 1816-. In the latter domain, of members of Irish political associations, the term exists in a network of semantic relations with ‘hougher’ 1712-1878, ‘White boy/Whiteboy’ 1762- ‘Steel boy’ 1772-, ‘United Irishman’ 1791-1837, ‘Defender’ 1796, ‘Orangeman’ 1796-, ‘Marksman’ 1800-1813, ‘Orangist’ c1800-1822, ‘Thresher’ 1806-1812, ‘Marchman’ 1813, ‘Rockite’ 1828-, ‘Brunswicker’ 1828-, ‘Terry Alt/Terryalt’ 1831-1898, ‘whitefoot’ 1832-, ‘Repeal Warden’ 1841-, ‘Young Irelander’ 1855, ‘Fenian’ 1864-, ‘Land-leaguer’ 1880, ‘Invincible’ 1883-1887, ‘Leaguer’ 1892, and ‘Prov(v)ie’ 1972-. The point of this example is that the alert user of these lexicographical resources needs to look at different sets of key words and explore different parts of the HT to build up a fuller picture of the cultural world evoked by the lexical items.

In short, then, the examples of ‘Irish’ and ‘Fenian’ illustrate the enormous power of the HT to (to repeat Samuels’ words, quoted above) ‘bring together under single heads all the words, current or obsolete (and all the obsolete meanings of words still current) that have ever been used to express single and related notions’ (SAMUELS 1972: 180). Once this has been done, the user has access to a conceptual domain, a kind of ‘map’ of the way in which related concepts, here ‘Irish’ as a nation and the inhabitants of that nation, have been articulated in Anglophone culture over the past thousand years. As the HT came to fruition, the resemblance of many of the ‘folk taxonomies’ to conceptual domains, as they were increasingly being described in cognitive semantics (eg LAKOFF & JOHNSON 1980; KÖVECSES 2006; KAY & ALLAN 2015) became highly suggestive. The conceptual domains of the HT cannot be understood as representing an individual’s cognitive domain (there are, for example, many English-speaking individuals who would not have any mental category at all for Irish political associations); however, one way of looking at the HT is as a representation of the collective cognitive domains of the Anglophone community over the past millennium. It is not an infallible representation, of course: the multivolume historical dictionaries on which it is based are the process of generations of lexicographers’ labours to record all the senses of the written (and only latterly the spoken) language. The record can never be complete or comprehensive; however, the data that we do have is considerable and, as the example of ‘Irish’ itself shows, it can be highly illuminating of recurrent patterns of thought.

A further offshoot of the HT is the online Hansard Corpus of British Parliamentary Speeches. Available as a distinct contribution to the Brigham Young University suite of online corpora, the Hansard Corpus (HC) was designed by Professor Marc Alexander of Glasgow University, the current director of the ongoing HT. The corpus contains 1.6 billion words of British Parliamentary Speeches from 1803-2005, tagged grammatically and also - uniquely - according to the categories of the HT. It is therefore possible to explore a massive corpus of political discourse about a range of topics, including Ireland, using the categories of the HT. This article concludes with a few illustrations of how this kind of exploration might begin.

A simple search of the frequency of the word ‘Irish’ in the HC shows that it appears frequently, particularly in the 1880s and 1890s, during a period of civil unrest known as the ‘Land War’ (Table 8).

Table 8:

During the 1890s there are a number of occurrences (precisely 10) of the expression ‘wild Irish’, mainly in a particular debate in which the term is bandied about between speakers. Most of the occurrences are in fact attributed to a single speaker, Lord Clonbrock, an Irish peer, who also uses it in conjunction with ‘mere Irish’ in a response to Viscount Clifden, during a debate on land legislation in Ireland, recorded in the House of Lords on 31st May 1897:

As with the citation from Swift, we can see Irish speakers drawing upon the stereotyped conceptual attributes of Irishness, more or less ironically, as the exchange develops and the speaker plays with categories and preconceptions. The several occurrences of ‘mere Irish’ in Lord Clonbrock’s speech contribute to a total of 8 occurrences of the collocation in HC in the 1890s. There are earlier occurrences in the HC; however, the alert user should be aware that not all collocations have the sense defined in the OED: compare these two occurrences from the 1840s:

(a) It contained a clause making void the grant, if the Duke should alienate or demise the premises to the mere Irish, or to any who should not have taken the oath of supremacy within a year previous.

(b) This was no mere Irish question; it was an English question…

Only in the first instance does the collocation refer to a particular type of Irishness, that is, ‘pure and undiluted; In the second instance it is the nature of the question, not the race, that is at stake.

The HC tags the corpus using the HT semantic categories, which can be useful and stimulating, though the results must be treated with caution as the tagging of a corpus of this scale is done at least partly automatically, and so the results can be counter-intuitive. However, even given the necessary constraints on the accuracy of coding, searches can be revealing. For example, the association between Irishness and foolishness can be explored by choosing the search item Irish and using a semantic code {AR:24:b} Foolishness/Stupidity to identify collocates (up to four items on each side of the search item) that fall into that conceptual category. The one word that appears in the results is fool (in its nominal and verbal forms), which appears in 22 occurrences, largely from the land legislation debates, a selection of which are exemplified below:

The interesting point about these actual corpus examples in which ‘Irish’ collocates with terms associated with foolishness is that the stereotypical foolishness is regularly denied. Even so, the underlying stereotype informs the discourse even while the discourse tokens repudiate it. The examples from the

corpus demonstrate the value of using the resources - the OED, the HT, the MME and the HC - in combination, and with a certain degree of critical flexibility.

This article has sought to introduce readers to the kinds of linguistic and cultural research that can be undertaken using relatively recent and undeniably powerful lexicographical resources. The Historical Thesaurus of English project has turned the multi-volume historical dictionaries, principally the OED, inside-out, painstakingly recategorizing the entire lexicon of English, past and present, according to semantic categories. The outcome of this massive project allows users to share Samuels’ vision of tracking concepts over time, and observing how they fit into the evolving lexical ecology that articulates a network of concepts characterising cultural assumptions at any period of time. The example of ‘Irishness’ has been used as an illustration. As well as referring to a nation and its inhabitants, the term ‘Irish’ engages in a series of conceptual shifts that illustrate how Anglophone culture has imagined the Irish over time: as wild, unconstrained, unintelligible Gaelic speakers, superstitious Catholics, a pure and undiluted race, dark-eyed and swarthy-complexioned, foolish and hot-tempered. The Mapping Metaphor project shows how terms that refer originally to races and nations gradually transfer their meanings into conceptual domains associated with personal attributes, such as sexual incontinence, untrustworthiness, foolishness and anger. Diachronic, semantically tagged corpora such as the Hansard Corpus demonstrate discourse in action. We see by looking at the results of corpus searches that the meanings and conceptual shifts indicated by the HT and the MME projects exist in contention: that participants in discourse draw on cognitive associations such as ‘Irish people are foolish persons’ not only to portray members of a category in this way, but also to challenge and repudiate such conceptions. In short, the dictionary, thesaurus, metaphor map and corpus must be considered as complementary resources in any research that combines lexicographical and cultural studies.