Pictish language

The evidence of place names and personal names demonstrates that an insular Celtic language related to the more southerly Brittonic languages was formerly spoken in the Pictish area.[5] The view of Pictish as a P-Celtic language was first proposed in 1582 by George Buchanan, who aligned the language with Gaulish.[6] A compatible view was advanced by antiquarian George Chalmers in the early 19th century. Chalmers considered that Pictish and Brittonic were one and the same, basing his argument on P-Celtic orthography in the Pictish king lists and in place names predominant in historically Pictish areas.[7]

Personal names of Roman-era chieftains from the Pictish area, including Calgacus (above) have a Celtic origin.[8]

Celtic scholar Whitley Stokes, in a philological study of the Irish annals, concluded that Pictish was closely related to Welsh.[9] This conclusion was supported by philologist Alexander MacBain's analysis of the place and tribe names in Ptolemy's second-century Geographia.[10] Toponymist William Watson's exhaustive review of Scottish place names demonstrated convincingly the existence of a dominant P-Celtic language in historically Pictish areas, concluding that the Pictish language was a northern extension of British and that Gaelic was a later introduction from Ireland.[11]

William Forbes Skene argued in 1837 that Pictish was a Goidelic language, the ancestor of modern Scottish Gaelic.[12] He suggested that Columba's use of an interpreter reflected his preaching to the Picts in Latin, rather than any difference between the Irish and Pictish languages.[13] This view, involving independent settlement of Ireland and Scotland by Goidelic people, obviated an Irish influence in the development of Gaelic Scotland and enjoyed wide popular acceptance in 19th-century Scotland.[14]

Skene later revised his view of Pictish, noting that it appeared to share elements of both Goidelic and Brittonic:

It has been too much narrowed by the assumption that, if it is shewn to be a Celtic dialect, it must of necessity be absolutely identic in all its features either with Welsh or with Gaelic. But this necessity does not really exist; and the result I come to is, that it is not Welsh, neither is it Gaelic; but it is a Gaelic dialect partaking largely of Welsh forms.[15]

The Picts were under increasing political, social, and linguistic influence from Dál Riata from around the eighth century. The Picts were steadily Gaelicised through the latter centuries of the Pictish kingdom, and by the time of the merging of the Pictish and Dál Riatan kingdoms, the Picts were essentially a Gaelic-speaking people.[16] Forsyth speculates that a period of bilingualism may have outlasted the Pictish kingdom in peripheral areas by several generations.[17] Scottish Gaelic, unlike Irish, maintains a substantial corpus of Brittonic loan-words and, moreover, uses a verbal system modelled on the same pattern as Welsh.[18]

The traditional Q-Celtic vs P-Celtic model, involving separate migrations of P-Celtic and Q-Celtic speaking settlers into the British Isles, is one of mutual unintelligibility, with the Irish sea serving as the frontier between the two. However, it is likely that the Insular Celtic languages evolved from a more-or-less unified proto-celtic language within the British Isles.[19] Divergence between P-Celtic Pictish and Q-Celtic Dalriadan Goidelic was slight enough to allow Picts and Dalriadans to understand each others language to some degree.[20] Under this scenario, a gradual linguistic convergence is conceivable and even probable given the presence of the Columban Church in Pictland.[21]

Difficulties in translation of ogham inscriptions, like those found on the Brandsbutt Stone, led to a widely held belief that Pictish was a non-Indo-European language

John Rhys, in 1892, proposed that Pictish was a non-Indo-European language. This opinion was based on the apparently unintelligible ogham inscriptions found in historically Pictish areas (compare Ogham inscription § Scholastic inscriptions).[22] A similar position was taken by Heinrich Zimmer, who argued that the Picts' supposedly exotic cultural practices (tattooing and matriliny) were equally non-Indo-European,[23] and a pre-Indo-European model was maintained by some well into the 20th century.[24]

A modified version of this theory was advanced in an influential 1955 review of Pictish by Kenneth Jackson, who proposed a two-language model: while Pictish was undoubtedly P-Celtic, it may have had a non-Celtic substratum and a second language may have been used for inscriptions.[25] Jackson's hypothesis was framed in the then-current model that a Brittonic elite, identified as the Broch-builders, had migrated from the south of Britain into Pictish territory, dominating a pre-Celtic majority.[26] He used this to reconcile the perceived translational difficulties of Ogham with the overwhelming evidence for a P-Celtic Pictish language. Jackson was content to write off Ogham inscriptions as inherently unintelligible.[27]

Jackson's model became the orthodox position for the latter half of the 20th century. However, it became progressively undermined by advances in understanding of late Iron Age archaeology, as well as by improved understanding of the enigmatic Ogham inscriptions, a number of which have since been interpreted as Celtic.[28]

Traditional accounts (now rejected) claimed that the Picts had migrated to Scotland from Scythia, a region that encompassed Eastern Europe and Central Asia.[29] Buchanan, looking for a Scythian P-Celtic candidate for the ancestral Pict, settled on the Gaulish-speaking Cotini (which he rendered as Gothuni), a tribe from the region that is now Slovakia. This was later misunderstood by Robert Sibbald in 1710, who equated Gothuni with the Germanic-speaking Goths.[30] John Pinkerton expanded on this in 1789, claiming that Pictish was the predecessor to modern Scots.[31] Pinkerton's arguments were often rambling, bizarre and clearly motivated by his belief that Celts were an inferior people. The theory of a Germanic Pictish language is no longer considered credible.[32]

Pictish toponyms occur in Scotland north of the River Forth.[35] They are distributed from Fife to the Isle of Skye,[33] although are rare in the extreme north.[37]

Many principal settlements and geographical features of the region bear names of Pictish origin, including:

• Aberdeen, Aberdeenshire. Meaning "mouth of the River Don" (c.f. Welsh aber, "estuary, confluence").[34]
• Cupar, Fife. Meaning "confluence" (c.f. Welsh cymer).[35][33]
• Keith, Banffshire. Meaning "forest" (c.f. Welsh coed).[34]
• Kirkcaldy, Fife. Meaning "place of the hard fort" from caer, "fort" and caled, "hard".[38]
• Perth, Perthshire. Meaning "wood, grove" (c.f. Welsh perth).[35]

Several Pictish elements occur multiple times in the region.[35] This table lists selected instances according to the Welsh equivalent.[35][37][34][38]

Some Pictish names have been succeeded by Gaelic forms, and in certain instances the earlier forms appear on historical record.

• Inverbervie, Kincardineshire. Haberberui in 1290, demonstrates that a Pictish aber, "estuary, confluence" has been supplanted by Gaelic inbhir, with identical meaning.[34]
• Strathtyrum, Fife. Trestirum in 1190, suggestive of assimilation of a Pictish tref, "estate", to (unconnected) Gaelic srath, "a valley".[38]

Although the interpretation of over 40 Ogham inscriptions remains uncertain, several have been acknowledged to contain Brittonic forms.[37]

An Ogham inscription at the Broch of Burrian, Orkney has been transliterated as I[-]IRANNURRACTX EVVCXRROCCS.[36] Broken up as I[-]irann uract cheuc chrocs, this may reveal a Pictish cognate of Old Welsh guract 'he/she made' in *uract (Middle Welsh goruc).[36][39] (The only direct continuation in Middle Welsh is 1sg. gwreith < *u̯rakt-ū in the poem known as Peis Dinogat in the Book of Aneirin; this form was eventually reformed to gwnaeth.[40]) With the fourth word explained as spirantized Pictish *crocs 'cross' (Welsh croes < Latin crux) and the corrupted first word a personal name, the inscription may represent a Pictish sentence explaining who carved the cross.[36][39][37]

The Shetland inscriptions at Cunningsburgh and Lunnasting reading EHTECONMORS and [E]TTECUHETTS have been understood as Brittonic expressions meaning "this is as great" and "this is as far", respectively,[37] messages appropriate for boundary stones.[37]

Transliterated as IRATADDOARENS, it is possible that the Brandsbutt Stone inscription attests a Pictish form cognate with Old Breton irha-, "he lies", in IRA-,[36] occurring at the Lomarec inscription in Brittany.[36]

Etymological investigation of the Scottish Gaelic language, in particular the 1896 efforts of Alexander Macbain,[41] has demonstrated the presence of a corpus of Pictish loanwords in the language.[33][41]

The following are possibilities:

• bad. Meaning "cluster" (c.f. Breton bod)[33]
• bagaid. Meaning "cluster, troop" (c.f. Welsh bagad)[33]
• dail. Meaning "meadow" (c.f. Welsh dôl)[33]
• dìleab. Meaning "legacy"[33]
• monadh. Meaning "mountain, moor" (c.f. Welsh mynydd)[33]
• pailt. Meaning "plentiful" (c.f. Cornish pals, Middle Welsh pallt)[33][41]
• peasg. Meaning "gash, chilblain" (c.f. Welsh pisg)[41]
• peit. Meaning "small area of ground" (c.f. Welsh peth)[33]
• pòr. Meaning "grain, crops" (c.f. Welsh pawr)[33]
• preas. Meaning "bush, thicket (c.f. Welsh prys)[33]

Linguist Guto Rhys has noted the potentially "fiscal" profile of several of the loans,[33] and hypothesized that they could have entered Gaelic as a package in a governmental context.[33]

In addition to these loans, the legal term mormaer may represent a survival of a Pictish compound form, composed of the elements mɔ:r, "large. great" (c.f. Welsh mawr), and majr, "steward" (< Latin major).[33]

Pictish may have influenced the sense and usage of several Gaelic words as a substrate.[35] Srath (> Strath-) is recorded to have meant "grassland" in Old Irish,[35] whereas the modern Gaelic realization means "broad valley", exactly as in its Brittonic cognates (c.f. Welsh ystrad).[35] Dùn, foithir, lios and ràth may, by the same token, attest a substrate influence from Pictish.[35]

Greene noted that the verbal system inherited in Gaelic from Old Irish had been brought "into complete conformity with that of modern spoken Welsh",[42] and consequently Guto Rhys adjudged that Pictish may have modified Gaelic verbal syntax.[33]

Kenneth Jackson propounded that Celtic /xs/ developed /s/ in Pictish, at divergence from the usual Brittonic development to /x/.[33] He attributed the following as evidence:

• Artcois (? < *arto-coxsos), a personal-name in the King Lists[33]
• River Lossie, Moray (identified with Loksa in Ptolemy).[33]

Rhys, however, asserted that to employ either as evidence is "unreliable"; Artcois on the premise either of mediation by Gaelic, scribal corruption, or derivation from Gaelic or Latin elements,[33] and Lossie on a possible incorrect identification,[33] and early forms of Lostyn (1189) suggesting a different etymology.[33]

Toponyms purportedly containing owxselo-, "high", in Pictland showing the regular Brittonic development have been cited as counter-evidence,[33] but are almost certainly invalid because a different derivation is far more likely.[33]

Rhys concludes that evidence is too insubstantial at the present time to ascertain the fate of this cluster in Pictish.[33]

Jackson (1953), Koch (1983), Forsyth (2006) and James (2013) argued for divergence on the basis of /o:/, raised to /u:/ in Brittonic, being retained in Pictish. The item of evidence attributed to this; a hypothesized Pictish toponymic element *ochel, consanguineous to Welsh uchel, "high",[33] is almost certainly invalid, as toponymic parallels are lacking and alleged derivatives (? > Ochil Hills, etc) are less problematically derived from *okelon, "a ridge".[33]

John Koch proposed in 1983 that in Pictish, the case ending -jo/-jos was realised as -ei, at variance with Brittonic in which this case ending was lost. Cited as evidence for this claim was the personal-name Bredei, apparently representing Celtic *brudjos, according to Evans from *brud, "reject, repel". The -ei ending, however, appears unlikely to represent a case ending.[33] It appears in the names of Strathclyde Britons (c.f Dwywei, Uruei, ?Affrei), and may be an agent suffix meaning "seeker", derived from Celtic *sagjo.[33] Or else, Rhys proposed that Pictish -ei may parallel a seemingly-adjectival suffix found in Welsh river-names (c.f. Melai, Menai, Sawddai).[33]

The Welsh bryn ("hill"; < *brunnjo) appears several times in Pictish toponymy (Brinbane > Burnbane, Brenturk > Burnturk, etc) with nothing to suggest Koch's proposed ending.[33]

Rhys concludes the view that case endings survived in Pictish has no evidence to sustain it.[33]

The view that Pictish retained the Proto-Celtic /oj/ diphthong until a later time than in Brittonic, in which it developed to /u:/ c. 75 A.D., was favoured by Jackson on the basis of two given-names; *Uroican (? < wrojko, "heather", > Welsh grug) and Onust, (c.f. Welsh Ungust).[33] While survival of /oj/ is possible,[33] there are issues with ascertaining the etymology of *Uroican, and the accuracy of the form Onust is uncertain.[33]

Kenneth Jackson proposed that Celtic /sN-/ and /s-/, which usually became /N-/ and /h-/ in Brittonic, were retained in Pictish.[33] Jackson attributed a single item of evidence to each:[33]

• Cairn Smairt, the name of a moor in Ross-shire, according to Jackson conserved the tribal name Smertae and was evidence for retention of /sN-/.[33]
• Simul (filius Druis) appears in the Annals of Ulster (725),[33] and Jackson took this as a personal name ("Simul son of Drest") preserving initial /s-/.[33]

Rhys dismissed both as evidence;[33] Cairn Smairt is more likely to involve the Scots surname Smairt,[33] and ethnic names are unlikely to preserved in insignificant geographical features.[33] Simul has no satisfactory Celtic root,[33] and has the look not of a personal-name, but instead the Latin adverb simul, "at the same time" (the sentence meaning "...at the same time, the son of Drest...").[33]

If Koch's suggestion that -CUHETT- on the Lunnasting inscritpion is analogous to Welsh cyhyd, "as far as" (-hyd < *siti), is correct, it would imply Jackson's proposal incorrect and development of /s-/ to /h-/ as in the rest of Brittonic.[33]

Rhys (2015) gave the view that Pictish, while partaking in some Latin influence, simultaneously resisted it to a greater extent than the rest of Brittonic.[33] Irrespective, Forsyth (1997),[36] Taylor,[35] and Aitchison (2019)[44] have proposed Latin loans within Pictish lexicon with parallels in Brittonic:

• *Crocs, "cross" (c.f. Welsh croes < Latin *crox), on the inscription on the Burrian Stone, Orkney.[36]
• *Ecles, "church" (c.f. Welsh eglwys < Latin eglesia) according to Taylor in toponyms (Ecclesgreig, etc).[35]
• *Leo, "lion, (figuratively) warrior", (c.f. Welsh llew < Latin leo).[44]
• *Maer, "steward" (c.f. Welsh maer < Latin major).[33]
• *Part, "side, area, region" (c.f. Welsh parth < Latin pars), in the toponym Parbroath according to Taylor.[38]
• *Pont, "bridge, (figuratively) leader", (c.f. Welsh pont), according to Aitchison attested in toponyms (? Pointack, etc), and in a personal-name (Brude Bont in Poppleton MS).[44]

Anderson & Anderson (1961) proposed that the Pictish equivalent of neo-Brittonic aber, "estuary", was an o-grade variant *abor, on the basis of toponymic forms in the Book of Deer, Annals of Ulster, and Vita Columbae (Apor Croosan > Applecross, Abbordoboir > Aberdour, etc).[33] Katherine Forsyth adapted this hypothesis, indicating that aber- and *abor- may both have been current in Pictish lexicon.[33]

Rhys opined, however, that *apor was well paralleled in Gaelic phonology, and represented a post-Pictish Gaelic post-labial rounding of aber rather than a diagnostically Pictish form.[33]

Pictish *pett was loaned into Gaelic as pett, peit, in which it has the meaning "estate, portion of land".[33] Jackson saw this as a divergent feature from Brittonic, in which the cognates (Welsh peth, Breton pezh) generally mean "thing".[33] Rhys instead concluded that the meaning originated from semantic narrowing during the borrowing process into Gaelic,[33] on the basis that, while the loan was abundant in Gaelic toponymy (c.f. Pittentrail, Pitlochry, etc), *pett- was absent in diagnostically Pictish place-names,[33] paralleling the element's almost total non-existence in Brittonic place nomenclature.[33]