ONlwSG

Show / Hide all footnotes

gàrradh 

The following draws upon discussion of the development of ON -rð in Gaelic in Cox 2007b, 55–64.

m. [ˈɡ̊ɑːɍəɣ] etc. (see below), gen. gàrraidh -[i], ‘dyke, wall (usually dry-stone); garden; enclosure; row of freshly cut and stacked peats on top of a peatbank’ (AFB^˄) and the Irish and Manx equivalents (see below) are for the most part derived from ON garðr m. ‘barrier, fence; fenced field; yard, courtyard; farm’ etc. (NO).

Stokes (≈1892, 82): Ir. ‘gardha “garden” ... formed on W gardd, and this from OEng. geard’ (but see fn 5, below); so also Henderson (1910, 215). MacBain (1895, 232) suggests SG gàrradh in the sense ‘garden’ is ‘a borrowing from the English’.

^, 

With a derivation from Old Norse, see for example: (Early Gaelic) Zimmer (1891, 63); Meyer (1891, 461); Craigie (1894, 156); MacBain (1896; 1911); Bugge (1900a, 292–93; 1912, 296); Marstrander (1915a, 61, 78, 94, 102, 116); Sommerfelt (1949, 233; 1962, 74); de Vries (1962, 156); Schulze-Thulin (1996, 105); McDonald (2009, 356); Kelly (2023, 249–50); eDIL^˄; (Irish) MacBain (1896; 1911); Bugge (1912, 296); Lucas (1968, 18); Greene (1976, 79; 1978, 121) ; Mac Mathúna (2001, 77); Kelly (2023: 249–50); (Manx) Greene (1978, 121); PNIM VI, xxvi; McDonald (2009, 356); (Scottish Gaelic) Craigie (1894, 156); MacBain (1896; 1911); Henderson (1910, 215); MacPherson (1945, 36); Sommerfelt (1952a, 231); Greene (1976, 79; 1978, 121); Oftedal (1980, 173; 1983, 155); Cox (1991, 492; 1992, 138); Stewart (2004, 410); McDonald (2009, 356); Kelly (2023: 249–50).

ON garð 

ON r is an apical trill, while ON ð is a voiced dental fricative in this position.

acc. yields EG garrḋa /ˈɡɑʀðə/, 

Also written garda, garrdha, gardha, garrga, garrgha (eDIL^˄, s.v. garrda; modern Ir. garrdha is now generally written garraí); for references, see eDIL and Marstrander 1915a, 116. Marstrander (ibid., 127) rejects Stokes’s derivation from Old English (fn 2, above) on account of the dental spirant in Early Gaelic: Thurneysen (1975, 77) notes that the representation of lenited EG d (ḋ, dh) in other languages, e.g. ð in Old Norse sources, shows that it was a voiced dental spirant; compare, for example, the man’s name EG Donnchaḋ > ON Dungaðr.

with inorganic final -a avoiding what would have been an unfamiliar final cluster.

Marstrander (1915a, 78) notes, garda av on. garðr (rð er uirsk) (‘[EG] garḋa from ON garðr (rð is unIrish)’), and (p. 111) [l]ydforbindelsen rð + kons. er uirsk (‘the cluster rð + consonant is unIrish’). He further writes, (p. 94) [o]ldnorske ord som i singularis viser et vekslende stavelsesantal overtages i irsk i sin korteste form: cnapp, cnarr, margg. I de tilfælde, hvor den korte form ender paa en konsonantgruppe som ingen hjemmel har i irsk utlyd, tilskydes et -/ə/ (-a, -e, -i) fra de længere former. Urigtig er Zimmers opfatning at a i garda skyldes on. -ʀ. (‘Old Norse words which show a varying number of syllables in the singular are taken over into Irish in their shortest form: cnapp, cnarr, margg. In cases where the short form ends in a consonant cluster not found in position final in Irish, -/ə/ (-a, -e, -i) is added from the longer forms. Zimmer’s view that a in garḋa derives from ON -ʀ is incorrect.’). Cf. also ON dorg f. ‘hand-line’ > EG [*]dorġa (O’Rahilly 1976, 241), s.v. dorgh. The addition of -/ə/ may also have been prompted by the analogy of the adjectival ending -ḋ(a)e (Thurneysen 1975, 220–21).

The various modern Gaelic reflexes of ON garðr in Scotland, Ireland and the Isle of Man are distinguishable historically chiefly by lengthening of the stressed vowel in the north and by the development of stressed epenthesis (svarabhakti) in the south.

O’Rahilly (1976, 49 fn 1) does not differentiate lengthened and unlengthened forms in Ireland in this way, but sees the one phenomenon (stressed epenthesis) following the other (lengthening); however, see fn 20, below.

The modern Scottish Gaelic reflex of ON garðr is gàrradh m. (sometimes written gàradh), e.g. /ˈɡɑːʀəɣ/ ‘stone wall or fence (between fields)’ (Oftedal 1956, 56, 109, 127).

For other reflexes, see the following table.

The long vowel before medial -rr- appears at first sight to be anomalous, as lengthening is normally expected before original geminates only in pre-consonantal or word-final position.

E.g. geàrr ‘to cut’, but verbal noun gearradh m. See, for example, O’Rahilly 1976, 49–52; Ó Baoill 1990, 131; Cox 2000, 214–15.

Borgstrøm (≈1940, 205) notes the occurrence of the long vowel and states that stressed vowels are lengthened ‘at least partly before -rð in Old Norse loanwords, e.g. gàrradh [ˈɡ̊ɑːʀəɣ] “a stone fence, a garden” < ON garðr, cf. SG geàrraidh (Lewis) [ˈɡ̊´ɑ̈ːʀ̣i] “site of a shieling” < ON gerði ’ (s.v. geàrraidh).

In fact, the modern reflexes of ON garðr in Scotland, the Isle of Man and the northern part of Ireland appear to share a similar development: with lengthening of the stressed vowel before -rð, yielding */ˈɡɑːʀðə/, 

O’Rahilly (1976, 51): ‘It is probable that none of the other important dialectal differences of today has its roots deeper in the past than this [lengthening].’ In contrast (ibid. 201–02): ‘It seems a reasonable inference that the beginnings of the epenthetic vowel in Irish hardly go back beyond the thirteenth century.’ Lengthening of the stressed vowel would prevent the development of epenthesis; cf. ON ármann acc. > SG àrmann ‘warrior, hero; leader’ (s.v.). (Calder’s (1972, 73) analysis of SG àrach ‘battlefield’ as ‘àr-ṁach (àr-ṁaġ), àramhach’ with epenthesis is erroneous: the word is from EG árṁaġ (with lenition of the second element in a closed compound, and a long stressed vowel), yielding árṁach under the influence of árḃach ‘slaughter’ (eDIL^˄, s.vv. árbach and ármag).)

assimilation of the dental fricative to the preceding consonant (cf. SG measarra < EG mesarḋa), and the development in some areas of unlenited /ʀ/ > lenited /r/.

For Scotland, see SGDS, Items 78–80; for the north of Ireland, see LASID I, 180. In Manx, the original four Old Gaelic r-phonemes (i.e. both unlenited and lenited forms of non-palatal and palatal sounds) were effectively reduced to one (Jackson 1955, 116 ff.).

The resultant */ˈɡɑːʀə/, */ˈɡɑːrə/, with the addition of a closing velar fricative, appears to be the base from which the various modern reflexes are derived: SG gàr(r)adh (i.e. -[əɣ] etc.); 

Besides forms in unstressed [ə], note the euphonious forms in Argyllshire [ˈkɑːrɑ(ɣ)] (Holmer 1938, 173), Arran [ɡɑːrɑ̌k] (LASID IV, 206, Item 622) and Rathlin [ɡɑːrɑ] (Holmer 1942, 199); cf. instances of a similar development in SG dathadh (SGDS, Item 298) and gànradh (ibid., Item 458). For discussion of vowel harmony via lowering of unstressed vowels near the sonorants l and r, see Ó Maolalaigh 2006b, 242, and 2008a, 165–66.

Manx garey (i.e. -/ə/, 

O’Rahilly 1976, 68.

with loss of final -dh 

Jackson 1955, 95.

); and Ulster and Donegal /ˈɡaːruː/ (and by reduction /ˈɡaːru/) 

O’Rahilly 1976, 66.

– all in regular development of final -adh, according to dialect.

While the lengthening in the northern development of ON garð acc. > SG gàrradh etc. is paralleled in the development of ON gerði > SG geàrraidh, if it is correct that the Lewis village name Garrabost [ˈɡ̊ɑɍə ̩b̥ɔs̪t̪] NB510332 

Taylor’s (1981, 19) Gàrrabost, with a lengthmark, is corrected to Garrabost in Taylor 2011.

derives from ON *Garðabólstað acc. ‘(the) farm of the enclosures’, with gen. pl. of garðr (Oftedal 1954, 396; Cox 2022, 721–26), then post-stress ON -rð has also yielded SG /ʀ/ with a preceding short stressed vowel; cf. SG (Lewis) Garsan [ˈɡ̊ɑʂən] /ˈɡɑʀsən/ (Cox 2002a, 285), perhaps from ON *Garð-sund acc. ‘(the) dyke-sound (i.e. wall-bay)’ or Garðsenda obl. ‘[the] head of the dyke’, and Borraiseadar [ˈb̥ɔɍi  ̩ ʃad̪̥əɾ] NB321194, if one of its suggested derivations holds true, viz ON *Vǫrðusætr ‘[the] shieling of the cairn’ (Cox 2022, 547–48). The distinction between lengthened and unlengthened reflexes of ON garðr may be a question of the degree to which loans are adapted, or the pace at which they are adapted, to the changing phonological system of the target language. In other words, the loan-name Garrabost preserves the quantity of the stressed vowel of its Old Norse antecedent, 

While place-names frequently retain conservative forms, e.g. Bhiondalam [ˈvĩᵰ̪d̪̥ə  ̩ ɫ̪am] NB175417 with a short stressed vowel (< ON *Vind-holm acc. ‘(the) wind-island’), it is not always the case, e.g. Tinndir [ˈtʲʰẽĩɲd̥ʲəɾʲ] NB226461 with diphthongisation (< ON *Tindar ‘(the) teeth’) (Cox 2000).

while the loan-word gàrradh has been treated phonologically in the same way as other elements in the Gaelic lexicon.

Similar lengthening occurs in the development of EG parḋus ‘paradise’ (frequently written pardus, parrdhus, parrthus, < Lat. paradīsus (eDIL^˄, s.v. pardus; Mc Manus 1983, 29, 59) > SG pàrras (Alasdair Mac Mhaighstir Alasdair: Ann a phàrras fèin (post-1738), and voc. a Phàrthais fhaoilidh, in Thomson 1996b: 78.606, 169.2080; M^cAlpine 1832: parras [pârr´-as]; Mackenzie 1847, s.v. paradise; MacBain 1911; Dwelly 1911; MacLennan 1925; Thomson 1996a), but parras in Glengarry (Dieckhoff 1932: [paRas]); cf. northern Irish [Pɑːr̥ɐS] ([Pɑːr̥uS]) (Ó Searcaigh 1925, 7); Teelin, Donegal [pɑːʀhuːs] (Wagner 1979, 65); Donegal [pɑːr̥uːS] (Quiggin 1906, 11); Mayo [ˈpɑːrhəs] [ˈpɑ̜ːʀəs] (Stockman 1974, 215, 263); Erris, Mayo [paːrəs] (Mhac an Fhailigh 1980, 164). Contrast epenthesis in West Muskerry, Cork [pɑrəhəs] (Ó Cuív 1988, 105). (Manx pargeiys (Y Kelly 1866), pargys (Cregeen 1835; Kneen 1978) is probably a Middle English/Anglo-Norman import, parallel to Eng. paradise except for syncope and with intervocalic d palatalised before long ī – a better spelling might have been *parjeeys, although the word in its current form appears to have been assimilated to cargys ‘Lent’ (Y Kelly ibid.) – contrast the 1610 Prayerbook’s older form, parus (Moore 1903, 413c).) However, there are words in which lengthening is unexplained, other than by analogy perhaps, e.g. SG eàrradh m. ‘mark, groove, scar, mark of a wound, groove in the staves of a barrel into which the head fits’ (MacLennan 1925), eàrrach f. in the (Islay) sense ‘chimb of a tub or cask’ (Dwelly 1911) and eàrrach m. in the senses ‘chink of a dish, where the edge of the bottom enters’ and ‘chime (chimb)’ (MacLennan 1925). These senses of eàrradh and eàrrach may simply be specialised extensions of a sense ‘bottom, lower extremity’ from SG eàrr < EG err ‘hinder-part, end, tail’ (eDIL^˄) or may be connected with ON ørr ‘scar, mark from a wound’ (s.v. eàrr); also, SG tòrradh ‘funeral’, Ir. tórramh ‘wake’ (Ó Baoill 1978, 236) < EG torraṁa, e.g. Urris, Donegal, [tɔːrhü] ‘funeral’ (Evans 1969, 128), Rathlin [tɔːrəɡ] (Holmer 1942, 243), also [toːrhəɡ] (LASID I, 196, Point 67), Lewis tòrradh /tɔːʀəɣ/ ‘to heap, to pile’ (Oftedal 1956, 360) and Kintyre [Tɔːʀɔɣ] (Holmer 1962, 8); see also LASID I, 196.

^, 

In weakly-stressed position in Old Norse loan-names in Scottish Gaelic, the stressed vowel of ON garð remains unlengthened, yielding SG -[ə], as in the Lewis village name Bradhagair [ˈb̥ɾa-a ˌ ɡ̊əɾʲ] NB2947, while final ON -rð yields SG -[ɾʲ], presumably via -[ɾ] (with alternation between -/r/ and -/r´/) (Cox 2022, 551–56).

In South Donegal, there appears to have been a certain amount of overlapping of northern and southern forms, so that we have examples both with short -i and with long -ī.

See LASID I, 180. O’Rahilly (1976, 49 fn 1) suggests, ‘[p]ossibly, however [see fn 7, above], these are contaminated forms; gáraí, for example, might arise from a blend of gárgha and garaí.’

The Northern Development of EG garrḋa (< ON garð acc.)

(Bracketed forms appear to be analogous forms.)

Notes: (a) Leurbost, Lewis, ‘stone wall or fence (between fields)’ (Oftedal 1956, 56, 109, 127); Western Isles, [ɡ̊ɑːʀəɣ] ‘stone fence, garden’ (Borgstrøm 1940, 205); Skye, idem ‘garden’ (Borgstrøm 1941. 40); (b) Benbecula, pl. ‘gardens’ (LASID IV, 234, Item 622); (c) Wester Ross, also Aultbea [ɡ̊ɑːʀ̥ʰək] ‘stone fence’ (Borgstrøm 1941; 98), also gàrradh càil ‘vegetable garden’ (Wentworth 2003a, s.v. garden); (d) Glengarry, ‘dyke, enclosure’ (Dieckhoff 1932, 95, s.v. garadh); (e) Argyllshire, len. dat. ‘dyke [wall]’ (LASID IV, 219, Item 182), also [ˈkɑːrə(ɣ)] ‘stone fence; garden’ (Holmer 1938, 173); for Islay (and elsewhere), Grannd (2000, 8) cites the orthographic form gàradh ‘garden’; (f) Kintyre, ‘garden’ (Holmer 1962, 5, 32), also [ɡaːʀəɣ] (ibid, 31); (g) Arran, ‘garden’ (Holmer 1957, 19, 46, 62), also [ˈɡɑːrək] ‘[wall]’ (LASID IV, 203, Item 182) and [ɡɑːrɑ̌k] ‘garden’ (LASID IV, 206, Item 622), also [ɡaːrəɣ] (Holmer 1957, 74); (h) Isle of Man, ‘enclosure, garden’ (PNIM I, 105), also [ɡɛːrə] ɡɛːri] (LASID I, 180, Point 88) (for the development of original long ā > [ɛː], see Jackson 1955, 24, and, for the loss of -/ɣ/, ibid., 95); (i) Rathlin (Holmer 1942, 199), also [ɡɑːrhə] (LASID I, 180, Point 67), also [ɡɑːrɑ] (Holmer ibid.); (j) Ulster (Stockman and Wagner 1965, 80), also [ɡɑːrû] (LASID I, 180, Ppoint 66); (k) Teelin, South Donegal (Wagner 1979, 27); (l) Glenties, South Donegal (Quiggin 1906, 11). (In addition to the above, see LASID I, 180.)

In contrast, lengthening did not take place before -rð in Galway and the south of Ireland. Here, development of EG garrḋa involves the assimilation of the dental /ð/ to velar fricative /ɣ/, 

‘The earliest examples of this [development] are found (for palatalised δ) about the end of the eleventh century, and the fusion must have been complete by the thirteenth’ (Thurneysen 1975, 77; see also O’Rahilly 1926b, esp. 163–64 and 193–95; 1976, 65).

the development of an epenthetic vowel, 

O’Rahilly 1976, 199–202.

and the development of unlenited /ʀ/ > lenited /r/, 

As the evidence of LASID I, 180, shows, non-palatal unlenited rr /ʀ/ survives sporadically.

yielding */ˈɡɑrəɣə/. Development of final -/əɣə/ yields /ˈɡɑriː/ (and by reduction /ˈɡɑri/) regularly in Connacht 

Stair na Gaeilge 486; de Búrca 2007, 70; de Bhaldraithe 1977, 101.7; Mhac an Fhailigh 1980, 149. See also Ó Maolalaigh 2006a, 56–57, for unstressed -adha yielding [iː] and [uː].

and, with regular stress shift, 

O’Rahilly 1976, 86ff.

/ɡɑˈriː/ in the south of Ireland.

An alternative possible development of EG garrḋa, involving metathesis of final -ḋa, seems less likely on account of the fact that an EG *garraḋ */ˈɡɑrəð/ would be expected to yield */ˈɡɑrə/ in the south of Ireland (O’Rahilly 1976, 65–66). Marstrander (1915a, 116, fn 1) writes ‘[m]ørkt rð > ri, sml. Munster dorgha (dorī) av on. dorg, Munster corīəs av corgas (‘the development of rð > ri is unclear, cf. Munster dorgha (dorī) from ON dorg [s.v. SG dorgh], Munster corīəs from corgas)’; this may be because he assumes a unitary treatment of the Old Norse cluster.

In Mayo, there appears to have been a certain amount of overlapping of northern and southern forms, so that we have examples of both short and long stressed vowels.

The Southern Development of EG garrḋa (< ON garð acc.)

(Bracketed forms appear to be analogous forms.)

Notes: (m) Erris, Mayo (Mhac an Fhailigh 1980, 15, 149), also Tourmakeady, Mayo /ə ˈmi sə ŋariː/ amuigh san ngarrdha (de Búrca 2007, 70); (n) Erris, Mayo (Mhac an Fhailigh 1980, 15, 149); (o) Aran Islands, also [ɡɑrɪː] (LASID I, 180, Points 41 and 42, respectively); (p) Waterford (Breatnach 1947, 124), also West Muskerry, Cork (Ó Cuív 1988, 14). (In addition to the above, see LASID I, 180.)

The chronologically later development of ON garð > garaí in the south of Ireland and Galway with its key feature of epenthesis is paralleled in Scotland in the development of ON urð > SG urrdh, s.v.