Gaulish was an ancient Celtic language that was spoken in parts of Europe before and during the period of the Roman Empire. In the narrow sense, Gaulish was the language spoken by the Celtic inhabitants of Gaul (modern-day France, Luxembourg, Belgium, most of Switzerland, Northern Italy, as well as the parts of the Netherlands and Germany on the west bank of the Rhine). In a wider sense, it also comprises varieties of Celtic that were spoken across much of central Europe ("Noric"), parts of the Balkans, and Asia Minor ("Galatian"), which are thought to have been closely related. The more divergent Lepontic of Northern Italy has also sometimes been subsumed under Gaulish.
Together with Lepontic and the Celtiberian language spoken in the Iberian Peninsula, Gaulish helps form the geographic group of Continental Celtic languages. The precise linguistic relationships among them, as well as between them and the modern Insular Celtic languages, are uncertain and a matter of ongoing debate because of their sparse attestation.
Gaulish is found in some 800, often fragmentary, inscriptions including calendars, pottery accounts, funeral monuments, short dedications to gods, coin inscriptions, statements of ownership, and other texts, possibly curse tablets. Gaulish texts were first written in the Greek alphabet in southern France and in a variety of the Old Italic script in northern Italy. After the Roman conquest of those regions, writing shifted to the use of the Latin alphabet. During his conquest of Gaul, Caesar reported that the Helvetii were in possession of documents in the Greek script, and all Gaulish coins used the Greek script until about 50 BC.
Gaulish in Western Europe was supplanted by Vulgar Latin and various Germanic languages from around the 5th century AD onwards. It is thought to have gone extinct some time around the late 6th century.
It is estimated that during the Bronze Age, Proto-Celtic started fragmenting into distinct languages, including Celtiberian and Gaulish. As a result of the expansion of Celtic tribes during the 4th and 3rd centuries BC, closely related varieties of Celtic came to be spoken in a vast arc extending from present-day Britain and France through the Alpine region and Pannonia in central Europe, and into parts of the Balkans and Anatolia. Their precise linguistic relationships are uncertain because of the fragmentary nature of the evidence.
The Gaulish varieties of central and eastern Europe and of Anatolia (known as Noric and Galatian, respectively) are barely attested, but from what little is known of them it appears that they were still quite similar to those of Gaul and can be considered dialects of a single language. Among those regions where substantial inscriptional evidence exists, three varieties are usually distinguished.
- Lepontic, attested from a small area on the southern slopes of the Alps, around the present-day Swiss town of Lugano, is the oldest Celtic language known to have been written, with inscriptions in a variant of the Old Italic script appearing circa 600 BC. It has been described as either an "early dialect of an outlying form of Gaulish" or a separate Continental Celtic language.
- Attestations of Gaulish proper in present-day France are known as "Transalpine Gaulish". Its written record begins in the 3rd century BC with inscriptions in the Greek alphabet, found mainly in the Rhône area of southern France, where Greek cultural influence was present via the colony of Massilia, founded circa 600 BC. After the Roman conquest of Gaul (58–50 BC), the writing of Gaulish shifted to the Latin alphabet.
- Finally, there are a small number of inscriptions from the second and first centuries BC in Cisalpine Gaul (modern northern Italy), which share the same archaic alphabet as the Lepontic inscriptions but are found outside the Lepontic area proper. As they were written after the time of the Gaulish conquest of Cisalpine Gaul, they are usually identified as "Cisalpine Gaulish". They share some linguistic features both with Lepontic and with Transalpine Gaulish; for instance, both Lepontic and Cisalpine Gaulish simplify the consonant clusters -nd- and -χs- to -nn- and -ss- respectively, while both Cisalpine and Transalpine Gaulish replace inherited word-final -m with -n. Scholars have debated to what extent the distinctive features of Lepontic reflect merely its earlier origin or a genuine genealogical split, and to what extent Cisalpine Gaulish should be seen as a continuation of Lepontic or an independent offshoot of mainstream Transalpine Gaulish.
The relationship between Gaulish and the other Celtic languages is also subject to debate. Most scholars today agree that Celtiberian was the first to branch off from the remaining Celtic languages. Gaulish, situated in the centre of the Celtic language area, shares with the neighbouring Brittonic languages of Great Britain, the change of the Indo-European labialized voiceless velar stop /kʷ/ > /p/, whereas both Celtiberian in the south and Goidelic in Ireland retain /kʷ/. Taking this as the primary genealogical isogloss, some scholars see the Celtic languages to be divided into a "q-Celtic" group and a "p-Celtic" group, in which the p-Celtic languages Gaulish and Brittonic form a common "Gallo-Brittonic" branch. Other scholars place more emphasis on shared innovations between Brittonic and Goidelic and group these together as an Insular Celtic branch. Sims-Williams (2007) discusses a composite model, in which the Continental and Insular varieties are seen as part of a dialect continuum, with genealogical splits and areal innovations intersecting.
At least 13 references to Gaulish speech and Gaulish writing can be found in Greek and Latin writers of antiquity. The word "Gaulish" (gallicum) as a language term is first explicitly used in the Appendix Vergiliana in a poem referring to Gaulish letters of the alphabet. Julius Caesar reported in his Commentarii de Bello Gallico of 58 BC that the Celts/Gauls and their language are separated from the neighboring Aquitanians and Belgae by the rivers Garonne and Seine/Marne, respectively. Caesar relates that census accounts written in the Greek alphabet were found among the Helvetii. He also notes that as of 53 BC the Gaulish druids used the Greek alphabet for private and public transactions, with the important exception of druidic doctrines, which could only be memorised and were not allowed to be written down. According to the Recueil des Inscriptions Gauloises, nearly three quarters of Gaulish inscriptions (disregarding coins) are in the Greek alphabet. Later inscriptions dating to Roman Gaul are mostly in the Latin alphabet and have been found principally in central France.
Latin was quickly adopted by the Gaulish aristocracy after the Roman conquest to maintain their elite power and influence, trilingualism in southern Gaul being noted as early as the 1st century BC.
Early references to Gaulish in Gaul tend to be made in the context of problems with Greek or Latin fluency until around 400, whereas after c. 450, Gaulish begins to be mentioned in contexts where Latin has replaced "Gaulish" or "Celtic" (whatever the authors meant by those terms), though at first these only concerned the upper classes. For Galatia (Anatolia), there is no source explicitly indicating a 5th century language replacement:
- During the last quarter of the 2nd century, Irenaeus, bishop of Lugdunum (present-day Lyon), apologises for his inadequate Greek, being "resident among the Keltae and accustomed for the most part to use a barbarous dialect".
- According to the Vita Sancti Symphoriani, Symphorian of Augustodunum (present-day Autun) was executed on 22 August 178 for his Christian faith. While he was being led to his execution, "his venerable mother admonished him from the wall assiduously and notable to all (?), saying in the Gaulish speech: Son, son, Symphorianus, think of your God!" (uenerabilis mater sua de muro sedula et nota illum uoce Gallica monuit dicens: 'nate, nate Synforiane, mentobeto to diuo'). The Gaulish sentence has been transmitted in a corrupt state in the various manuscripts; as it stands, it has been reconstructed by Thurneysen. According to David Stifter (2012), mentobeto looks like a Proto-Romance verb derived from Latin mens, mentis ‘mind’ and habere ‘to have’, and it cannot be excluded that the whole utterance is an early variant of Romance, or a mixture of Romance and Gaulish, instead of being an instance of pure Gaulish. On the other hand, nate is attested in Gaulish (for example in Endlicher's Glossary), and the author of the Vita Sancti Symphoriani, whether or not fluent in Gaulish, evidently expects a non-Latin language to have been spoken at the time.
- The Latin author Aulus Gellius (c. 180) mentions Gaulish alongside the Etruscan language in one anecdote, indicating that his listeners had heard of these languages, but would not understand a word of either.
- The Roman History by Cassius Dio (written AD 207-229) may imply that Cis- and Transalpine Gauls spoke the same language, as can be deduced from the following passages: (1) Book XIII mentions the principle that named tribes have a common government and a common speech, otherwise the population of a region is summarised by a geographic term, as in the case of the Spanish/Iberians. (2) In Books XII and XIV, Gauls between the Pyrenees and the River Po are stated to consider themselves kinsmen. (3) In Book XLVI, Cassius Dio explains that the defining difference between Cis- and Transalpine Gauls is the length of hair and the style of clothes (i.e., he does not mention any language difference), the Cisalpine Gauls having adopted shorter hair and the Roman toga at an early date (Gallia Togata). Potentially in contrast, Caesar described the river Rhone as a frontier between the Celts and provincia nostra.
- In the Digesta XXXII, 11 of Ulpian (AD 222–228) it is decreed that fideicommissa (testamentary provisions) may also be composed in Gaulish.
- Writing at some point between c. AD 378 and AD 395, the Latin poet and scholar Decimus Magnus Ausonius, from Burdigala (present-day Bordeaux), characterizes his deceased father Iulius's ability to speak Latin as inpromptus, "halting, not fluent"; in Attic Greek, Iulius felt sufficiently eloquent. This remark is sometimes taken as an indication that the first language of Iulius Ausonius (c. AD 290-378) was Gaulish, but may alternatively mean that his first language was Greek. As a physician, he would have cultivated Greek as part of his professional proficiency.
- In the Dialogi de Vita Martini I, 26 by Sulpicius Seuerus (AD 363–425), one of the partners in the dialogue utters the rhetorical commonplace that his deficient Latin might insult the ears of his partners. One of them answers: uel Celtice aut si mauis Gallice loquere dummodo Martinum loquaris ‘speak Celtic or, if you prefer, Gaulish, as long as you speak about Martin’.
- Saint Jerome (writing in AD 386/387) remarked in a commentary on St. Paul's Epistle to the Galatians that the Belgic Treveri spoke almost the same language as the Galatians, rather than Latin. This agrees with an earlier report in AD180 by Lucian.
- In a letter of AD 474 to his brother-in-law, Sidonius Apollinaris, bishop of Clermont in the Auvergne, states that in his younger years, "our nobles... resolved to forsake the barbarous Celtic dialect", evidently in favour of eloquent Latin.
- Cassiodorus (ca. 490–585) cites in his book Variae VIII, 12, 7 (dated 526) from a letter to king Athalaric: Romanum denique eloquium non suis regionibus inuenisti et ibi te Tulliana lectio disertum reddidit, ubi quondam Gallica lingua resonauit ‘Finally you found Roman eloquence in regions that were not originally its own; and there the reading of Cicero rendered you eloquent where once the Gaulish language resounded’
- In the 6th century Cyril of Scythopolis (AD 525-559) tells a story about a Galatian monk who was possessed by an evil spirit and was unable to speak, but if forced to, could speak only in Galatian.
- Gregory of Tours wrote in the 6th century (c. 560-575) that a shrine in Auvergne which "is called Vasso Galatae in the Gallic tongue" was destroyed and burnt to the ground. This quote has been held by historical linguistic scholarship to attest that Gaulish was indeed still spoken as late as the mid to late 6th century in France.
Despite considerable Romanization of the local material culture, the Gaulish language is held to have survived and had coexisted with spoken Latin during the centuries of Roman rule of Gaul. The exact time of the final extinction of Gaulish is unknown, but it is estimated to have been around or shortly after the middle of the 1st millennium, in the late 5th or early 6th century, after the collapse of the Western Roman Empire had already happened.
The language shift was uneven in its progress and shaped by sociological factors. Although there was a presence of retired veterans in colonies, these did not significantly alter the linguistic composition of Gaul's population, of which 90% was autochthonous; instead, the key Latinizing class was the coopted local elite, who sent their children to Roman schools and administered lands for Rome. In the 5th century, at the time of the Western Roman collapse, the vast majority (non-elite and predominantly rural) of the population remained Gaulish speakers, and acquired Latin as their native speech only after the demise of the Empire, as both they and the new Frankish ruling elite adopted the prestige language of their urban literate elite.
According to the Recueil des Inscriptions Gauloises, more than 760 Gaulish inscriptions have been found throughout present-day France, with the notable exception of Aquitaine, and in northern Italy. Inscriptions include short dedications, funerary monuments, proprietary statements, and expressions of human sentiments, but the Gauls also left some longer documents of a legal or magical-religious nature, the three longest being the Larzac tablet, the Chamalières tablet and the Lezoux dish. The most famous Gaulish record is the Coligny calendar, a fragmented bronze tablet dating from the 2nd century AD and providing the names of Celtic months over a five-year span; it is a lunisolar calendar attempting to synchronize the solar year and the lunar month by inserting a thirteenth month every two and a half years.
Many inscriptions consist of only a few words (often names) in rote phrases, and many are fragmentary. They provide some evidence for morphology and better evidence for personal and mythological names. Occasionally, marked surface clausal configurations provide some evidence of a more formal, or poetic, register. It is clear from the subject matter of the records that the language was in use at all levels of society.
Other sources also contribute to knowledge of Gaulish: Greek and Latin authors mention Gaulish words, personal and tribal names, and toponyms. A short Gaulish-Latin vocabulary (about 20 entries headed De nominib[us] Gallicis) called "Endlicher's Glossary", is preserved in a 9th century manuscript (Öst. Nationalbibliothek, MS 89 fol. 189v).
Some Gaulish loanwords are found in the French language. Today, French contains approximately 150 to 180 words known to be of Gaulish origin, most of which concern pastoral or daily activity. If dialectal and derived words are included, the total is approximately 400 words, the largest stock of Celtic words in any Romance language.
Gaulish inscriptions are edited in the Recueil des Inscriptions Gauloises (R.I.G.), in four volumes:
- Volume 1: Inscriptions in the Greek alphabet, edited by Michel Lejeune (items G-1 –G-281)
- Volume 2.1: Inscriptions in the Etruscan alphabet (Lepontic, items E-1 – E-6), and inscriptions in the Latin alphabet in stone (items l. 1 – l. 16), edited by Michel Lejeune
- Volume 2.2: inscriptions in the Latin alphabet on instruments (ceramic, lead, glass etc.), edited by Pierre-Yves Lambert (items l. 18 – l. 139)
- Volume 3: The Coligny calendar (73 fragments) and that of Villards-d'Héria (8 fragments), edited by Paul-Marie Duval and Georges Pinault
- Volume 4: inscriptions on Celtic coinage, edited by Jean-Baptiste Colbert de Beaulieu and Brigitte Fischer (338 items)
The longest known Gaulish text is the Larzac tablet, found in 1983 in l'Hospitalet-du-Larzac, France. It is inscribed in Roman cursive on both sides of two small sheets of lead. Probably a curse tablet (defixio), it clearly mentions relationships between female names, for example aia duxtir adiegias [...] adiega matir aiias (Aia, daughter of Adiega... Adiega, mother of Aia) and seems to contain incantations regarding one Severa Tertionicna and a group of women (often thought to be a rival group of witches), but the exact meaning of the text remains unclear.
The Coligny calendar was found in 1897 in Coligny, France, with a statue identified as Mars. The calendar contains Gaulish words but Roman numerals, permitting translations such as lat evidently meaning days, and mid month. Months of 30 days were marked matus, "lucky", months of 29 days anmatus, "unlucky", based on comparison with Middle Welsh mad and anfad, but the meaning could here also be merely descriptive, "complete" and "incomplete".
The pottery at La Graufesenque is our most important source for Gaulish numerals. Potters shared furnaces and kept tallies inscribed in Latin cursive on ceramic plates, referring to kiln loads numbered 1 to 10:
- 1st cintus, cintuxos (Welsh cynt "before", cyntaf "first", Breton kent "in front" kentañ "first", Cornish kynsa "first", Old Irish céta, Irish céad "first")
- 2nd allos, alos (W ail, Br eil, OIr aile "other", Ir eile)
- 3rd tri[tios] (W trydydd, Br trede, OIr treide)
- 4th petuar[ios] (W pedwerydd, Br pevare)
- 5th pinpetos (W pumed, Br pempet, OIr cóiced)
- 6th suexos (possibly mistaken for suextos, but see Rezé inscription below; W chweched, Br c'hwec'hved, OIr seissed)
- 7th sextametos (W seithfed, Br seizhved, OIr sechtmad)
- 8th oxtumeto[s] (W wythfed, Br eizhved, OIr ochtmad)
- 9th namet[os] (W nawfed, Br naved, OIr nómad)
- 10th decametos, decometos (CIb dekametam, W degfed, Br degvet, OIr dechmad)
The lead inscription from Rezé (dated to the 2nd century, at the mouth of the Loire, 450 kilometres (280 mi) northwest of La Graufesenque) is evidently an account or a calculation and contains quite different ordinals:
- 3rd trilu
- 4th paetrute
- 5th pixte
- 6th suexxe, etc.
Other Gaulish numerals attested in Latin inscriptions include petrudecametos "fourteenth" (rendered as petrudecameto, with Latinized dative-ablative singular ending) and triconts "thirty" (rendered as tricontis, with a Latinized ablative plural ending; compare Irish tríocha). A Latinized phrase for a "ten-night festival of (Apollo) Grannus", decamnoctiacis Granni, is mentioned in a Latin inscription from Limoges. A similar formation is to be found in the Coligny calendar, in which mention is made of a trinox[...] Samoni "three-night (festival?) of (the month of) Samonios". As is to be expected, the ancient Gaulish language was more similar to Latin than modern Celtic languages are to modern Romance languages. The ordinal numerals in Latin are prīmus/prior, secundus/alter (the first form when more than two objects are counted, the second form only when two, note also that alius, like alter means "the other", the former used when more than two and the latter when only two), tertius, quārtus, quīntus, sextus, septimus, octāvus, nōnus, and decimus.
A number of short inscriptions are found on spindle whorls and are among the most recent finds in the Gaulish language. Spindle whorls were apparently given to girls by their suitors and bear such inscriptions as:
- moni gnatha gabi / buððutton imon (RIG l. 119) "my girl, take my penis(?)"
- geneta imi / daga uimpi (RIG l. 120) '"I am a young girl, good (and) pretty".
Inscriptions found in Switzerland are rare. The most notable inscription found in Helvetic parts is the Bern zinc tablet, inscribed ΔΟΒΝΟΡΗΔΟ ΓΟΒΑΝΟ ΒΡΕΝΟΔΩΡ ΝΑΝΤΑΡΩΡ (Dobnorēdo gobano brenodōr nantarōr) and apparently dedicated to Gobannus, the Celtic god of metalwork. Furthermore, there is a statue of a seated goddess with a bear, Artio, found in Muri bei Bern, with a Latin inscription DEAE ARTIONI LIVINIA SABILLINA, suggesting a Gaulish Artiū "Bear (goddess)".
Some coins with Gaulish inscriptions in the Greek alphabet have also been found in Switzerland, e.g. RIG IV Nos. 92 (Lingones) and 267 (Leuci). A sword, dating to the La Tène period, was found in Port, near Biel/Bienne, with its blade inscribed with KORICIOC (Korisos), probably the name of the smith.
- vowels: short: a, e, i, o u long: ā, ē, ī, (ō), ū diphthongs: ai, ei, oi, au, eu, ou
- occlusives: voiceless: p, t, k voiced: b, d, g
- resonants nasals: m, n liquids r, l
- sibilant: s
- affricate: ts
- semi-vowels: w, y
The diphthongs all transformed over the historical period. Ai and oi changed into long ī and eu merged with ou, both becoming long ō. Ei became long ē. In general, long diphthongs became short diphthongs and then long vowels. Long vowels shortened before nasals in coda.
Other transformations include unstressed i became e, ln became ll, a stop + s became ss, and a nasal + velar became /ŋ/ + velar.
The occlusives also seem to have been both lenis, unlike Latin, which distinguished voiced occlusives with a lenis realization from voiceless occlusives with a fortis realization, which caused confusions like Glanum for Clanum, vergobretos for vercobreto, Britannia for Pritannia.
The alphabet of Lugano used in Cisalpine Gaul for Lepontic:
The alphabet of Lugano does not distinguish voicing in stops: P represents /b/ or /p/, T is for /d/ or /t/, K for /g/ or /k/. Z is probably for /ts/. U /u/ and V /w/ are distinguished in only one early inscription. Θ is probably for /t/ and X for /g/ (Lejeune 1971, Solinas 1985).
The Eastern Greek alphabet used in southern Gallia Narbonensis:
Χ is used for [x], θ for /ts/, ου for /u/, /ū/, /w/, η and ω for both long and short /e/, /ē/ and /o/, /ō/ while ι is for short /i/ and ει for /ī/. Note that the sigma, in the Eastern Greek alphabet, looks like a C (lunate sigma). All Greek letters were used except phi and psi.
Latin alphabet (monumental and cursive) in use in Roman Gaul:
G and K are sometimes used interchangeably (especially after R). Ð/ð, ds and s may represent /ts/ and/or /dz/. X, x is for [x] or /ks/. Q is only used rarely (Sequanni, Equos) and may represent an archaism (a retained kw) or, as in Latin, an alternate spelling of -cu- (for original /kuu/, /kou/, or /kom-u/). Ð and ð are used to represent the letter [[INLINE_IMAGE|//upload.wikimedia.org/wikipedia/commons/thumb/6/68/Aa_taugallicum.svg/12px-Aa_taugallicum.svg.png|//upload.wikimedia.org/wikipedia/commons/thumb/6/68/Aa_taugallicum.svg/18px-Aa_taugallicum.svg.png 1.5x, //upload.wikimedia.org/wikipedia/commons/thumb/6/68/Aa_taugallicum.svg/24px-Aa_taugallicum.svg.png 2x|Aa taugallicum.svg|h17|w12]] (tau gallicum, the Gaulish dental affricate).
- Gaulish changed the PIE voiceless labiovelar kʷ to p, a development also observed in the Brittonic languages (as well as Greek and some Italic languages like the Osco-Umbrian languages), while other Celtic languages retained the labiovelar. Thus, the Gaulish word for "son" was mapos, contrasting with Primitive Irish maq(q)os (attested genitive case maq(q)i), which became mac (gen. mic) in modern Irish. In modern Welsh the word map, mab (or its contracted form ap, ab) is found in surnames. Similarly one Gaulish word for "horse" was epos (in Old Breton eb and modern Breton keneb "pregnant mare") while Old Irish has ech, the modern Irish language and Scottish Gaelic each, and Manx egh, all derived from proto-Indo-European h₁eḱwos. The retention or innovation of this sound does not necessarily signify a close genetic relationship between the languages; Goidelic and Brittonic are, for example, both Insular Celtic languages and quite closely related.
- The Proto-Celtic voiced labiovelar gʷ (From PIE gʷʰ) became w: gʷediūmi → uediiumi "I pray" (but Old Irish guidim, Welsh gweddi "to pray").
- PIE ds, dz became /tˢ/, spelled ð: neds-samo → neððamon (cf. Irish nesamh "nearest", Welsh nesaf "next"). Modern Breton nes and nesañ "next".
- PIE ew became eu or ou, and later ō: PIE tewtéh₂ → teutā/toutā → tōtā "tribe" (cf. Irish túath, Welsh tud "people").
- PIE ey became ei, ē and ī PIE treyes → treis → trī (cf. Irish trí "three").
- Additionally, intervocalic /st/ became the affricate [tˢ] (alveolar stop + voiceless alveolar stop) and intervocalic /sr/ became [ðr] and /str/ became [θr]. Finally, when a labial or velar stop came before /t/ or /s/, the two sounds merged into the fricative [χ].
There was some areal (or genetic, see Italo-Celtic) similarity to Latin grammar, and the French historian Ferdinand Lot argued that it helped the rapid adoption of Vulgar Latin in Roman Gaul.
Gaulish had seven cases: the nominative, vocative, accusative, genitive, dative, instrumental and the locative case. Greater epigraphical evidence attests common cases (nominative and accusative) and common stems (-o- and -a- stems) than for cases less frequently used in inscriptions or rarer -i-, -n- and -r- stems. The following table summarises the reconstructed endings for the words toṷtā "tribe, people", mapos "boy, son", ṷātis "seer", gutus "voice", brātīr "brother".
In some cases, a historical evolution is attested; for example, the dative singular of a-stems is -āi in the oldest inscriptions, becoming first -ăi and finally -ī as in Irish a-stem nouns with attenuated (slender) consonants: nom. lámh "hand, arm" (cf. Gaul. lāmā) and dat. láimh (< lāmi; cf. Gaul. lāmāi > lāmăi > lāmī). Further, the plural instrumental had begun to encroach on the dative plural (dative atrebo and matrebo vs. instrumental gobedbi and suiorebe), and in the modern Insular languages, the instrumental form is known to have completely replaced the dative.
For o-stems, Gaulish also innovated the pronominal ending for the nominative plural -oi and genitive singular -ī in place of expected -ōs and -os still present in Celtiberian (-oś, -o). In a-stems, the inherited genitive singular -as is attested but was subsequently replaced by -ias as in Insular Celtic. The expected genitive plural -a-om appears innovated as -anom (vs. Celtiberian -aum).
There also appears to be a dialectal equivalence between -n and -m endings in accusative singular endings particularly, with Transalpine Gaulish favouring -n, and Cisalpine favouring -m. In genitive plurals the difference between -n and -m relies on the length of the preceding vowel, with longer vowels taking -m over -n (in the case of -anom this is a result of its innovation from -a-om).
Gaulish verbs have present, future, perfect, and imperfect tenses; indicative, subjunctive, optative and imperative moods; and active and passive voices. Verbs show a number of innovations as well. The Indo-European s-aorist evolved into the Gaulish t-preterit, formed by merging an old 3rd personal singular imperfect ending -t- to a 3rd personal singular perfect ending -u or -e and subsequent affixation to all forms of the t-preterit tense. Similarly, the s-preterit is formed from the extension of -ss (originally from the third person singular) and the affixation of -it to the third person singular (to distinguish it as such). Third-person plurals are also marked by the addition of -s in the preterit system.
Most Gaulish sentences seem to consist of a subject–verb–object word order:
Some, however, have patterns such as verb–subject–object (as in living Insular Celtic languages) or with the verb last. The latter can be seen as a survival from an earlier stage in the language, very much like the more archaic Celtiberian language.
Sentences with the verb first can be interpreted, however, as indicating a special purpose, such as an imperative, emphasis, contrast, and so on. Also, the verb may contain or be next to an enclitic pronoun or with "and" or "but", etc. According to J. F. Eska, Gaulish was certainly not a verb-second language, as the following shows:
Whenever there is a pronoun object element, it is next to the verb, as per Vendryes' Restriction. The general Celtic grammar shows Wackernagel's Rule, so putting the verb at the beginning of the clause or sentence. As in Old Irish and traditional literary Welsh, the verb can be preceded by a particle with no real meaning by itself but originally used to make the utterance easier.
According to Eska's model, Vendryes' Restriction is believed to have played a large role in the development of Insular Celtic verb-subject-object word order. Other authorities such as John T. Koch, dispute that interpretation.
Considering that Gaulish is not a verb-final language, it is not surprising to find other "head-initial" features:
- Genitives follow their head nouns:
- The unmarked position for adjectives is after their head nouns:
- Prepositional phrases have the preposition, naturally, first:
- Passive clauses:
Subordinate clauses follow the main clause and have an uninflected element (jo) to show the subordinate clause. This is attached to the first verb of the subordinate clause.
Jo is also used in relative clauses and to construct the equivalent of THAT-clauses
This element is found residually in the Insular Celtic languages and appears as an independent inflected relative pronoun in Celtiberian, thus:
- Welsh modern sydd "which is" ← Middle Welsh yssyd ← esti-jo vs. Welsh ys "is" ← esti
- Irish Old Irish relative cartae "they love" ← caront-jo
- Celtiberian masc. nom. sing. ioś, masc. dat. sing. iomui, fem. acc. plural iaś
Gaulish had object pronouns that infixed inside a word:
Disjunctive pronouns also occur as clitics: mi, tu, id. They act like the emphasizing particles known as notae augentes in the Insular Celtic languages.
Clitic doubling is also found (along with left dislocation), when a noun antecedent referring to an inanimate object is nonetheless grammatically animate. (There is a similar construction in Old Irish.)
In an interview, folk metal band Eluveitie said that some of their songs are written in a reconstructed form of Gaulish. The band asks scientists for help in writing songs in the language. The name of the band comes from graffiti on a vessel from Mantua (c. 300 BC). The inscription in Etruscan letters reads eluveitie, which has been interpreted as the Etruscan form of the Celtic (h)elvetios ("the Helvetian"), presumably referring to a man of Helvetian descent living in Mantua.