The Modern Gaulish Language
The revival of the Gaulish language: Galáthach hAthevíu

14. Initial Consonant Mutation
 
     Mutation of initial consonants in Celtic languages
 
Initial consonant mutations have been referred to several times in the preceding sections. It is a phenomenon that is particular to the modern Celtic languages: the first letter of a word systematically changes its sound in certain contexts. These sound changes sometimes, though not always, carry grammatical and semantic meaning.
 
In the context of the modern Celtic languages, it is thought that the systematic mutation of the first consonant of a word (when followed by a vowel) happened when it was preceded by a word ending in a vowel, thus creating an intervocalic environment, when considered across word boundaries. This phenomenon, which is not unusual cross-linguistically, is known as “sandhi” and is perhaps best thought of as “advanced slurring” of one word into another. It is closely akin to the liaisons of the French language, and is widespread in Western Germanic languages. In this context, it is worth bearing in mind that it is precisely those languages that replaced the Gaulish language through most of its erstwhile range, and that therefore may have absorbed a certain amount of substratal influence from the Gaulish language, of which the tendency to engage in “advanced slurring” may have been one aspect.
 
The process is aptly illustrated by an example from the Welsh language:
 
        Brittonic *sindos tatos = the father
        Brittonic *sinda mama = the mother
 
In the first example, the initial t- of *tatos does not find itself in an intervocalic environment, since the article *sindos ends in –s. Therefore, no sandhi effect took place on the word *tatos. In the second example however, the initial m- of *mama finds itself in an intervocalic environment, because the article *sinda ends in a vowel. As a result, the initial m- of *mama was weakened, and became [v]. This gives, in modern Welsh:
 
        tad: father
        y tad: the father
        mam: mother
        y fam: the mother [vam]
 
The occurence of these mutations, once established, was later endowed with grammatical meaning and became systematical.
 
     Mutation in ancient Gaulish
 
It is commonly accepted wisdom that the ancient Gaulish material shows no evidence of the systematic mutation of the first letter of a word depending on context. This is often considered to be an indication that the phenomenon only occurred in the Insular Celtic language group (see e.g. Isaac 2007 for a discussion). However, this is overlooking two important factors:
 
1. It is thought that the process of development of the system of initial consonant mutations did only fully develop in the 5th-6th centuries CE. The latest Gaulish attestation dates to the late 4th or early 5th century; therefore, a development that might still have been in process might not have shown up in the written ancient Gaulish record.
 
2. More importantly, mutations were not graphically recorded or represented in the insular languages until many hundreds of years after they first developed as a phonetic and grammatical phenomenon: Old Irish (500-900 CE) and Old Welsh (500-1100 CE) show no mutations, even though the modern languages manifestly feature them, and even though their incorporation into the languages evidently goes back to a period predating the first written records. Mutations only first started to be included in the written record, and imperfectly at that, in the Middle period of both languages. The Breton language did not start to record its mutations until the 17th century.
 
In view of the above, it is clear that if Gaulish was developing a system of mutations, which are a phenomenon of the spoken language, there is a high likelihood that this would not have been reflected in the written record of the time. Nevertheless, it is worth analysing the attested Gaulish material for possible indications of mutation-like processes. It is also important to bear in mind that some leading scholars in the field believe that the attested Gaulish material does indeed show indications of a process of mutation (e.g. Schrijver on Chateaubleau, from David Stifter 2012, pers. com.).
 
There are a number of instances which appear to indicate a variation in the initial consonant of a word within one and the same document. These are:
 
    1) cutio > gutio (Coligny, Delamarre 2003, p. 133)
        cutio > qutio (id.)
        cantlos > gantlos (id.)
 
    2) flatucia > ulatucia (Larzac, Delamarre 2003, p. 338-39)
        bnanom > mnanom (id.)
 
    3) prinas > brina[ (La Graufesenque, Delamarre 2003, p. 253; Lambert 2003. p. 135)
 
    4) apeni < *ac beni (Chateaubleau, Schrijver in Stifter 2012, pers. com.)
 
The context of the first document, in which the first three examples are found, can not reveal anything about the phonetic environment in which these words were found, as the document is a calendar and does not feature continuous writing.
 
The second document however, the inscription of Larzac, is the longest extant running text in the ancient Gaulish language. An examination of the contexts in which the examples quoted above are found yields the following:
 
        banona flatucias
        potiti ulatucia
 
In the above cases, both words are preceded by words ending in a vowel; however, the case apparently differs.
 
        insinde se bnanom
        biontutu se mnanom
        biontutu indas mnas
        bietutu se mnas
        biontutu se mnanom
 
In the above case, it would be possible to isolate the following:
 
        se bnanom
        > preceding word ending on vowel > intervocalic environment > initial “b-“
 
        indas mnas
        preceding word ending on consonant > non-intervocalic environment > initial “m”
 
However, the word “mnas” also occurs three times preceded by “se”, the same word which precedes “bnanom”, which has not apparently brought on a mutation of the initial consonant. Furthermore, the consonant of the radical “bena”, “woman”, is “b-“, not “m-“, so it would be expected that the situation would be the other way around. It is therefore not possible to state that these examples indicate a systematic mutation of initial consonants analoguous with the model operative in Brittonic.
 
The third document, the potters’ accounts of La Graufesenque, shows an alteration of “prinas” to “brina[“. The alteration of /p/ > /b/ appears on the surface to be analogous to the instances found on Coligny, where /c/ > /g/. However, no information is available on the context wherein this alteration takes place.
 
The fourth document, Chateaubleau, the most important late Gaulish text extant because of its length, the high quality of its preservation and its late date (late 4th-early 5th cent.), features the sequence “apeni”. This has been interpreted as a contraction of “ac beni” (Schrijver 1998). “ac beni” would then mean “and [a] woman”. If that is the case it would show a clear and unequivocal instance where the initial consonant of a word is affected by the last letter of a preceding word, the result being, in this case, a loss of the final consonant of the first word and a devoicing of the first consonant of the second word. Or, in other words, a sandhi effect. This is precisely the sort of context that gave rise to the mutations in the modern Celtic languages. However, the parsing and interpretation of the phrase is contested, with an alternative reading being “ape niteme” (Mees 2010), where “ape” is held to be an eroded form of the preposition “ambi”.
 
 
In view of the examples analysed above it is possible to suggest that there were situations in ancient Gaulish where the exact phonetic value of initial consonants was considered either ambiguous or unsure, and was subject to change under some circumstances. This situation appears comparable with the situation prevailing in the Brittonic and Goidelic languages in the 4th-5th centuries CE, which brought forth the systematic mutation of initial consonants. Therefore, it is posited here that Gaulish would have evolved a system of initial consonant mutation similar to those found in Brittonic and Goidelic.
 
     Mutation changes
 
Based on the apparent voicing of “c-“ to “g-“ in Coligny, of “f-“ to, possibly, ”v-“ at Larzac, and of “p” to “b” at La Graufesenque, it is proposed here that the following changes take place in mutation contexts:

The phonetic values of the above graphemes has been discussed under the section “sound changes”.
 
As regards to “s-“ before a consonant, it is dropped, and marked in writing with an apostrophe. In the spoken language, the appearance of a radical in a mutational context indicates the elision of “s-”, therefore indicating the mutation of “s-“.
 
The various mutations of the modern Celtic languages are categorised according to the phonetic processes which occur, e.g. softening, hardening, nasalisation etc. All of the modern Celtic languages have different kinds and combinations of mutations, although there are deep fundamental resemblances among all of them. The mutations proposed for the modern Gaulish language consist of a combination of phonetic processes, mostly voicing and spirantisation (see above). Therefore, it is easiest to describe this mutation as a “mixed mutation”. However, its description is irrelevant, as there is only one change for each sound.
 
     Contexts for mutation
 
The contexts for mutation have become apparent throughout this document. It is posited that mutation has become established as a system for assisting with the conveying of grammatical meaning in the wake of the loss of case endings and therefore meaning, as discussed previously. As such, they represent a simple switch from marking the end of words for meaning to marking the start of words for meaning. This position is supported by Isaac (2007) who argues that VSO languages have a greater innate tendency to modify words word initially than word finally.
 
There are two kinds of contexts for mutation: contact mutation and grammatical mutation. The contact mutation category is the largest one. In contact mutation, a mutation is triggered by a lexeme immediately preceding the word affected. In grammatical mutation, a mutation is bestowed because of a grammatical condition or requirement, which may or may not be caused by a lexeme immediately preceding the word affected.
 
Contact mutations do not necessarily convey grammatical meaning in themselves; instead they usually assist in the marking of such meaning. However, such meaning would not be apparent in the absence of the lexemes triggering the mutation. As such, it can be said that a large part of the mutations featured in modern Gaulish are phonetic mutations rather than grammatical mutations. This situation is similar to the one prevailing in modern Breton, Welsh and Cornish, where some mutations occur for no apparent reason and carry no apparent grammatical meaning.
 
     Contact mutations
 
1) after preverbal particles
 
The following preverbal particles cause mutation: ré, en, a, né, a
 
        ré: past tense marker > cana mi: I sing
                                                 ré gan mi: I sang
 
        en: ongoing tense marker > delgha mi: I hold
                                                         esi mi en dhelghe: I am holding
 
        a: intentional form marker > depri: to eat
                                                         gwéla mi ái a dhepri: I want to go to eat
 
        né: negation marker > gara mi: I call
                                               né ghara mi: I don’t call
 
        a: interrogative marker > gara ti: you call
                                                    a ghara ti: do you call?
 
2) after adverbial particles
 
The following adverbial particles cause mutation: in, ré, ró, ma
 
        in: adverbial adjective marker > tech: beautiful
                                                               in dech: beautifully
 
        ré: intensive marker > már: big
                                              ré wár: very big
 
        ró: excessive marker > ró wár: too big
 
        ma: conditional marker > gwéla mi: I want
                                                    ma chwéla mi: if I want
 
3) after question words
 
Normally question words are followed by the interrogative particle “a”, which causes mutation on the following word, and is not in itself susceptible to mutation. However, in some cases a question word may be directly followed by a different word, which will then undergo mutation; this only happens if the phrase is used as a question:
 
        pé: what, which
        cun: dog
        pé gun: which dog?
 
        caman: road, way
        pé gaman: which road, which way?
 
        ponch ré hái mi a gáma: when I went to walk > here “ponch” is not used as a question word, so it does not
        cause mutation
 
4) after prepositions
 
All prepositions cause mutation:
 
        gwer: on, at
        mór: sea
        gwer wór: at sea (“on [the] sea”)
 
        gwó: under
        pren: tree
        gwó bren: under a tree
 
        can: with
        caran: friend
        can garan: with a friend
 
     Grammatical mutations
 
1) feminine nouns after the article
 
Feminine nouns both singular and plural mutate when preceded by the article “in”. Thus the gender of nouns is effectively marked by the initial consonant mutation. This is not a contact mutation, because the article does not trigger mutation in masculine nouns.
 
e.g.: ben: woman
        in: the
        in ven: the woman
        mná: women
        in wná: the women
 
        brí: hill
        in vrí: the hill
        in vríé: the hills
 
        grá: sand
        in ghrá: the sand
        in ghráé: the sands
 
2. adjectives qualifying feminine nouns
 
Adjectives qualifying feminine nouns, whether it be one or several, are marked by mutations:
 
        ben: a woman
        tech: beautiful
        ben dech: a beautiful woman
 
        brí: a hill
        ardhu: high
        brí hardhu: a high hill
 
        grá: sand
        gwin: white
        grá chwin: white sand
        grá chwin dech: beautiful white sand
 
3. after possessive pronouns
 
Words following possessive pronouns regularly undergo mutation. As this mutation is triggered by the presence of the possessive pronouns, it could be said that this is straightforward contact mutation. However, these mutations differ from contact mutations in two ways:
 
a) the mutations serve to mark the distinction between the 3rd pers. singular masculine and   
    feminine, and between the 2nd and the 3rd person plural:
 
e.g.: cun: dog
 
        mó gun: my dog                              nó gun: our dog
        tó gun: your dog                              só cun: your (pl.) dog – no ICM
        ó gun: his dog – ICM                      só gun: their dog – ICM
        ó cun: her dog – no ICM
 
b) the mutations are transferred to any words preceding the noun that is
    possessed, but also across them to the possessed noun, even though it is no longer in
    contact with the mutation trigger:
 
        e.g.: cun: dog
                pethr: four
                mó bethr gun: my four dogs
 
 
     Unmutatable words
 
Some words cannot be mutated, and also block mutation from taking place. These are:
 
1. the article “in”
 
The article blocks a contact mutation, and does not mutate itself:
 
e.g.: pen: head
        gwer: on
        gwer ben: on a head
        gwer in pen: on the head
 
2. the possessive pronouns
 
While the possessive pronouns cause mutations (see above), they are not mutatable themselves:
 
e.g.: pen: head
        mó: my
        gwer: on
        gwer mó ben: on my head
 
Additional words that do not mutate are the adverbs “né” (not) and “ma” (if). However, contexts that could potentially lead to their mutation are rare. As a broad general rule, those words that do not mutate are those words that cause mutation. Exceptions are the preverbal particles “ré”and “ró”, which do mutate.
 
     Summary of mutation contexts
 
In summary, mutation is caused by:
 
1. pre-verbal particles
2. adverbial particles
3. question words if used as such
4. prepositions
5. possessive pronouns
6. the article before feminine nouns
7. adjectives qualifying feminine nouns