Move back one step  Move forward one step  List of themes 

    Palatalisation and initial mutations

Changes to both ends of words
Irish developments in a broader perspective


In the history of Celtic, inherited inflections were lost due to phonetic reduction of unstressed syllables. Various sandhi phenomena also occurred where the onsets of lexical words were lenited (fricativised) or nasalised if the preceding word ends in a vowel or nasal respectively. These low-level phonetic changes were later functionalised. In pre-Old Irish a system evolved, known as initial mutation in which at least three distinctions were available at the front end of lexical words (gemination may also have been an option, but this was lost later), giving the following options in Modern Irish.


Phonetic situation


Linguistic term


unaltered onset

baile ‘town, home’

zero mutation


fricativised onset

sa bhaile ‘at home’



nasalised onset

a mbaile ‘their town’


Parallel to the system of initial mutation an additional means of grammatical distinction, palatalisation, developed at the end of a word. Palatalisation could affect any lexical base. A base is a root (a single syllable) or a root + a root extension. A simple root would be present-day Irish teach ‘house’, bád ‘boat’ whereas a root plus extension would be marcach (< marc + ach) ‘rider’.


Palatalisation was the result of co-articulation in the environment of high vowels, that it a high vowel – /i/ or /e/ – following a consonant (typically found in inflectional endings) caused this consonant to be pronounced in a position with the tongue raised in the mouth as if to produce the high vowel. Later the high vowels disappeared with the general loss of inflections from earlier stages of Indo-European and palatalisation was left as the sole indication of grammatical categories such as the genitive, consider a modern Irish Irish example like cnoc ‘hill-NOM’ – barr an chnoic ‘top of the hill-GEN’ (palatalisation is indicated in writing by the letters i or e before and/or after a main vowel of a syllable).

After its functionalisation, palatalisation contrasted in the grammar with non-palatalisation. In Modern Irish palatalisation and mutations have two domains.

1) Lexical Palatalisation distinguishes many citations forms, beag ‘small’ bog ‘soft’; siúl ‘to walk’, súil ‘eye’; cás ‘case’ cáis ‘cheese’. Lenition marks the head in compounds seanfhear (< sean + fear) ‘old man’ an-ghnóthach ‘very busy’ (< an + gnóthach).
2) Morphological Palatalisation and mutations distinguish different cases, number, gender, etc. nom. sg. an múinteoir (masc. no mutation) ‘the teacher’, an bhliain ‘the year’ (fem. lenition). In the genitive the reverse is the case (masc. lenition; fem. no mutation). Mutations are also important in signalling coreferentiality and the relationship between dependents and their heads, e.g. adjectives cabhair [fem] mhór (lenited) ‘a big help’ : cúrsa [masc] casta (non-lenited) ‘a complicated course’.

Changes to both ends of words

Taken together, initial mutation and palatalisation form a system of distinctions in Irish which are used to indicate major grammatical categories in the language, e.g. cases, number and possession with nouns and tense and mood with verbs. Because of the way the language developed, initial mutation with verbs goes hand in hand with typical suffixes at the ends of words (inflectional endings). Only lenition occurs with verbs to indicate tense and mood.

It is common to have combinations of initial mutations and palatal/ non-palatal alternations within a single word in Modern Irish. These can be classified according to the changes which can occur on the left and the right margin of lexical bases as evident from the following description and example.

Initial mutation alters the manner of articulation for the consonant affected (place of articulation is rarely affected, but cf. /d/ + L > /ɣ/).

Alterations to the end of a word involve a change in consonant quality, either a palatal consonant becomes non-palatal, e.g. final /n/ in arán ‘bread-NOM’ > praghas an aráin ‘the price of the bread-GEN’ or a non-palatal one becomes palatal, e.g. beoir ‘beer-NOM’ > blas na beorach ‘the taste of the beer-GEN’. All consonants in Irish, except /h/, come in pairs consisting of a non-palatal and a palatal member.

Left margin: 1) Zero mutation 2) Lenition 3) Nasalisation
Right margin: 1) Palatalisation 2) De-palatalisation  

Base onset nucleus coda
C- -V- -C
voiceless stop, voiced stop, fricative, nasal palatal, non-palatal (affects any type of consonant)

Base: bás ‘death’ Onset Coda
a bhás ‘his death’ Lenited Non-palatal
a bás ‘her death’ Neutral Non-palatal
a mbás ‘their death’ Nasalised Non-patalal
am a bháis ‘time of his death’ Lenited Palatal
am a báis ‘time of her death’ Neutral Palatal
am a mbáis ‘time of their death’ Nasalised Palatal

Base: bris ‘break-IMPERATIVE’  
brisim ‘I break’ Neutral
brisfidh mé ‘I will break’ Neutral
brise mé ‘I break-SUBJ’ Neutral
bhrisinn mé ‘I break-PAST_SUBJ’ Lenition
bhris mé ‘I broke’ Lenition
bhrisinn ‘I used to break’ Lenition
bhrisfinn ‘I would break’ Lenition

Irish developments in a broader perspective

When viewed cross-linguistically one sees that the morphology of Irish is unique in its combination of features. While palatalisation/de-palatalisation (Bhat 1978) is a common axis along which to differentiate sounds (this has happened in all the Slavic languages, for instance) the initial changes, typical of Celtic languages, are very seldomly found.

Palatalisation is a natural assimilation phenomenon whereby the feature of highness spreads from a vowel to a consonant, usually preceding. It establishes itself most easily with coronal sonorants and fricatives, probably because the secondary articulation is most easily perceived with these segments, witness the many palatal sonorants in Romance languages (for a comparison with Celtic, see Martinet 1952, Ternes 1977) and the common distinction between /s/ and /ʃ/ in many languages. The functionalisation of palatalisation is not that uncommon: within Indo-European it is found on a wide scale in Celtic, Slavic and Indo-Iranian. With those languages in which it attains a grammatical function it is usual to find a secondary palatalisation of labials (Jackson 1967 and Macaulay 1966) with tense lips and a brief [j] on release of the labial as the phonetic correlates of phonological palatalisation, cf. Irish and Russian (Jones and Ward 1969).

The initial mutations are cross-linguistically much rarer. The reason for this is probably that speakers prefer to maintain the stems of words constant and to add inflections (as prefixes or suffixes) when required. Even if changes do occur at the beginning of word-stems (Andersen 1986) it is quite unusual for these to attain a grammatical function over time. Nonetheless, this has happened in some languages, e.g. in certain Bantu languages of western Africa spoken mostly in Nigeria and Ghana (Sapir 1971; on Fula, see Anderson 1976) and in eastern Siberia with a language isolate from the Paleosiberian group. This is Nivkh, formerly called Gilyak, spoken along the lower reaches of the Amur river and on part of Sakhalin Island (see Panfilov 1962-1965, Jakobson 1971, Gruzdeva 1998). Berber is a language (or group, depending on the interpretation of internal differences) in which there is an alternation at the beginning of nouns depending on syntactic contexts (what is called free and annexed in the relevant literature, see Basset 1952).

The functionalisation of initial mutation implies that it has taken over from other grammatical devices which have been lost or at least defunctionalised in a language. The pre-stage to this state can be seen in several dialects/languages. For instance, the so-called gorgia toscana in Tuscan Italian comprises fricativisation and gemination of initial segments of a noun depending on the original form of a preceding grammatical word. Thus the feminine article la causes fricativisation (la casa /la xasa/) and the preposition a (< Latin ad) triggers gemination (a porta /a pporta/ < Latin ad portam), Lepschy and Lepschy (1990), Maiden (1995), Maiden and Parry (eds, 1997). Here one can see what a mutational system looks like embryonically. In order for the functionalisation of initial mutation to be grammatically adequate at least three distinctions must be possible.

  Irish Tuscan Italian
1) zero mutation    
2) mutation 1 lenition lenition
3) mutation 2 nasalisation gemination

A language may have more than three distinctions, for instance Welsh divides lenition into (i) fricativisation and (ii) stop voicing and has nasalisation anyway, this resulting in three mutations, that is with zero mutation, a four-way system of distinctions.

For initial mutations to become the dominant means of indicating grammatical categories in a language a minimum of three distinctions is necessary which is probably why only two distinctions, say no change and initial fricativisation, do not trigger typological re-orientation of the language in question. Of course a three-way distinction is a necessary but by no means a sufficient condition for functionalisation. Here a look at the phenomenon of consonant gradation in Finnish (Karlsson 1999) is helpful. There are four types, divided into two groups: gradation proper and assimilation. These consist of the following processes: 1) simplification of geminates and consonant clusters, 2) voicing and fricativisation, 3) vocalisation. But the gradation occurs word-internally in Finnish. It is triggered by a closed short syllable which leads to a phonetic reduction of the consonants preceding it. Such a short syllable is typically represented by an inflection, such as the genitive, cf. jalka ‘foot’ : jalan ‘foot’-GEN. The agglutinative suffixes of Finnish are of course still present as opposed to Estonian which, due to the loss of final inflections, has opacified gradation as a sychronic process. The upshot of this is that there is no immediate motivation to functionalise gradation in Finnish and hence its application is not exceptionless. For instance, not all loanwords undergo gradation, cf. auto ‘car’ : auton ‘car’-GEN and not *audon.

It is not possible to predict the functionalisation of changes to the beginning of word-stems once these have arisen in a language. But one can point to certain processes which might trigger this functionalisation. Foremost among such processes would be the decay of inflectional endings which indicate grammatical endings. This may in time lead to a re-interpretation of the initial changes as grammatically significant. This happened in the Celtic languages with first language learners re-analysing (Hickey 2003) the grammatical system in favour of the initial mutations.


Andersen, Henning. 1986. Sandhi phenomena in the languages of Europe. Berlin: de Gruyter.
Anderson, Stephen. 1976. ‘On the description of consonant gradation in Fula’, Studies in African linguistics 7: 93-136.

Basset, André (1952) La langue berbère. Oxford.

Bhat, D. N. S. 1978. ‘A general study of palatalization’, in Greenberg (ed.) 47-92.

Feuth, Els. 1983. ‘Gemination: an Old Irish mutation rule?’, Ériu 34: 143-56.

Fisiak, Jacek (ed.) 1996. Linguistic typology and reconstruction. Berlin: Mouton-de Gruyter.

Greenberg, Joseph. 1978. Universals of human language. 4 vols. Stanford: University Press.

Greene, David. 1956. ‘Gemination’, Celtica 3: 284-9.

Gruzdeva, Ekaterina 1998. Nivkh. Languages of the World: Materials 111. Munich/ Newcastle: Lincom Europa.

Hickey, Raymond 1996. ‘Sound change and typological shift: Initial mutation in Celtic’, In Fisiak (ed.), 133-82.

Hickey, Raymond 2003. ‘Reanalysis and typological change’, in Raymond Hickey (ed.) Motives for language change. Cambridge: University Press, pp. 258-78.

Jackson, Kenneth H. 1967. ‘Palatalization of labials in the Gaelic languages’, in Meid (ed.), 179-92.

Jakobson, Roman 1971. Selected writings. Vol. 2: Word and Language. The Hague: Mouton.

Jakobson, Roman 1971. ‘Notes on Gilyak’, in Jakobson 72-97.

Jones, Daniel and Denis Ward. 1969. The phonetics of Russian. (Cambridge: University Press).

Karlsson, Fred 1999. Finnish. An Essential Grammar. London: Routledge.

Lepschy, Giulio and Anna Lepschy 1990. The Italian language today. 2nd edition. Chicago: New Amsterdam Books.

Macaulay, Donald 1966. ‘Palatalization of labials in Scottish Gaelic and some related problems in phonology’, Scottish Gaelic Studies 11, 72-84.

Maiden, Martin 1995. A linguistic history of Italian. London: Longman.

Maiden, Martin and Mair Parry (eds) 1997. The dialects of Italy. London: Routledge.

Martinet, André 1952. ‘Celtic lenition and Western Romance consonants’, Language 28: 192-217.

Meid, Wolfgang (ed.) 1967. Festschrift für Julius Pokorny. Innsbruck: University Press.

Panfilov, V. Z. 1962-1965. Grammatika nivxskogo jazyka. [A grammar of the Nivkh language] 2 vols. Moscow.

Sapir, J. David. 1971. ‘West Atlantic: an inventory of the languages, their noun class systems, and consonant alternations’, in Sebeok (ed.) 45-112.

Sebeok, Thomas A. (ed.) 1971. Current trends in linguistics. Vol. 7: Linguistics in Sub-Saharan Africa. The Hague: Mouton.

Ternes, Elmar 1977. ‘Konsonantische Anlautveränderungen in den keltischen und romanischen Sprachen’, Romanistisches Jahrbuch 28: 19-53.