Lenition

As mentioned before, lenition occurs directly after a vowel, at the end of a word (eg. aδkwī́eθ "she sees"), at the beginning of a word (eg. indā δelwā "that shape") or inside (kloχā "stone"). It also affects consonants at the beginning of consonant clusters (sēβrah "phantom"), with a few exceptions noted below.

Note 1: In later Primitive Irish /h/ would have only occured in stressed position and in pausa (at the end of an utterance). For example mo hulih "my eye" and nīhákwieθ "does not see", but also dīaμalaθuh "disguise'' < dī- "dis-" + saμalaθuh "resemblance". However, in an earlier period intervocalic -h- must have existed in some form if only for a brief time. The presence of this lenited h (even if it is inaudible) has consequences for some forms of poetry as well as for the process of elision, which is discussed below. For this reason a diaeresuis < ̈ >, is sometimes placed between two vowels where an earlier -h-, and even earlier -s- had existed. Some examples of hiatus are from other lost consonants, e.g. krāäh "stable, enclosure" < P.C *krāφos, thought it is sometimes unclear if some cases are hiatus or simply of diphthong, e.g. oyβā, oïβā ("beauty") < P.C. oφibā.

Note 2: /sk/, /skʷ/ and /sm/ resist lenition as do clusters containing them, eg. tō smeχah "your chin", roteskrī́bāθ "it can scratch you".

Note 3: Lenition of n, l, and r seems to have arisen slightly differently than other sounds. According to McCone "Lenition would appear to have applied to r, l, n except in word-in[i]tial position or in contact with a homorganic consonant (i.e. before t, d and before or after s, r, l, n…"

Nasalisation

Nasalisation occurs directly after /n/. For the sake of simplicity the examples below use ean "their", but any word ending in -n can cause nasalisation. Note how the -n changes value or disappears when combined.

Aspiration

When -h appears at the end of a word, as it often does, it attaches to a following stressed vowel. If the following word begins in a consonant, the -h disappears and the consonant remains unchanged,

Note: If the following syllable is not stressed then aspiration does not occur, as -h- is dropped in unstressed position. Eg. nīëχrédiyū "I do not believe it" < nīh "not" + e "it (acc.)" + kredīyū "I believe", not nīheχrédīyū.

Syncope and Delenition

Though not nearly as common as in Old Irish and later, syncope occurs between broadly homorganic consonants (i.e. consonants with the same or very similar place of articulation) in words of more than two syllables. When this occurs any lenited consonants revert to their normal state. This most often occurs in verbs with a root ending in θ or δ,

aδ-wéδeθ > aδ-wéδθ > aδwḗd "he declares"

to-áre-reθeθ > to-árreθθ > toárred "she overtakes"

But this can occur elsewhere,

Morarīγanī > Morrīγanī "Morrigan"

wohaγeθih > wohγeθih > woχeθih "suffering"

Elision

There are two kinds of elision that are most consequential: 1) consonantal elision, which in this context refers to "disallowed" consonant clusters in the middle or end of a word (aδ-bertā > abertā), and 2) vocalic elision, which refers to the deletion of an unstressed vowel when immediately followed unstressed vowel (sena-aθīr > senaθīr). Stress is an important factor in the process of elision. Most such clusters of vowels or consonants occur when two morphemes come together. If there is a stress boundary (i.e., the second element is stressed) then no elision is necessary. This mostly occurs in verbs (es-réγeθ "it rises"), and sometimes in nouns with productive prefixes (kēdu-éχorih "the first key"). However, as soon as the stress moves away from the second syllable elision takes effect, for example, if the addition of another prefix shifts stress to the left (nīh-és-reγeθ > nīhéreγeθ "it does not rise", wī́ra-χēdu-eχorih > wīraχēdeχorih "the real first key").

Note that in the preceding examples the elision is occuring after the stress, and in these cases it is the first vowel that tends to be dropped. When vocalic elision occurs before the primary stress of a phrase, the second vowel of the pair tends to dropped instead, (wo ea χosī > woa χosī "under his foot"). But not always: embi-en-koμ-áre-ānige > embēgoμárānige ("they encountered one another").

While there is generally no ambiguity about the number of syllables in a word, hiatus between vowels can occur from several historical origins. To clarify that two vowels are part of separate syllables, an umlaut < ̈> is placed over the second syllable. This is also convention in Old Irish resources like eDIL (e.g. O.I. ïaich "salmon (pl.)" < P.I. *eöχeh < P.C. *esokes). The hiatus from intervocalic h is most noticeable in verbs where it occurs often in various endings that would otherwise undergo elision. Hiatus also occurs in cases of historical *φ in P.C. < PIE *p, as in niöh "nephew" < P.C. *neφots < PIE *népōts.

Raising, Lowering, and Rounding

During the period of monumental Ogham vowels were in the process of vowel affections which, like palatalisation, were not yet a phonemic distinction, and also like palatalisation, would go on to form an important part of Old Irish inflection.

Lowering is the most common effect. When an open vowel a, ā, e, ē, o, ō stood in the following syllable the close vowels ĭ and ŭ were lowered closer to e and o respectively (perhaps realised intermediately as [i̞~ɪ] and [u̞~ʊ]*). For example wirah "man" might be realised as [ʍɪrah], and in fact in O.I. the lowering was complete, yielding fer. Long vowels were not affected, wīrah "true" > O.I. fír.

Raising is slightly more restricted in its occurrence and the exact rules are not fully known yet. In general it is like the opposite of lowering. The open vowels e and o are raised closer to i and u respectively. For example mori "sea" might be realised as [mʊrʲi], and in fact in O.I the raising completed here as well, yielding muir.

A labiovelar (i.e. kw, gw, χw, γw) can cause the following, short i or a to become rounded to u and o respectively. For example kwariyah [kʷœrʲijah] or [kʷorʲea]. However this would have been a fairly late development un Primitive Irish As it must have occurred after the complete raising of words like gweδiθi "prays" > gwiδiθi > O.I. guidid, and after the complete lowering of words like kwrināθi "buys" > kwrenaθi > O.I. crenaid.

*Though not necessarily! For a comparable situation in Latin, in which intermediate vowel qualities are called into question, see Calabrese, A. (2003)

Palatalisation

Because Primitive Irish did not exhibit syncope or apocope until just before the Old Irish period, there are no minimal pairs involving palatalisation (known in Modern Irish as "slender" consonant as opposed to "unpalatalised", "velar" or "broad" consonants. It is possible, however, that consonants were in fact palatalised before front vowels as is the case in the modern Gaelic languages, but it would not have been a grammatical feature. For Tengwah Wālī, I will not be marking palatals in broad transcription, eg. /iniγenā/, but it may show up in narrower transcription, eg. [ˈi.nʲi.ɣʲe.naː]. The rule for palatal consonants would be slightly different than in modern times and may in fact be too complicated for this introduction (see McCone 1996). Here is a brief rundown of McCones rules for "first palatalisation" which is the stage that had definitely begun by the period of monumental Ogham:

1) "The first palatalisation, then, affected a single consonant (and mb, nd, ng) between two front vowels or between any vowel (except ā; see below) and high front i/ī...

between two front vowels:

• deγwih [dæɣʲih] "flame"
• leγeθi [læγʲeθʲi] "lies"
• bereθi [berʲeθʲi] "carries"
• imbeθan [imbʲeθan] "an abundance"
• rendih [rendʲih] "point, peak"

between back vowel and i, ī:

• tōθiyāh [toːθʲijaːh] "kingdom (gen.)"
• gaβiθi [gaβʲiθi] "takes"

but unpalatalised,

• kaneθi [kaneθʲi] "sings"
• karēh [karɛːh] "friend"
• togeθah [togeθah] "fortune"
• erbīθi [erbiːθʲi] "entrusts"
• sāliyāh [saːlijaːh] "heel (gen.)" (> O.I. sál, gen. sála instead of expected **sáile)

2) "...unlike its voiced counterpart, a voiceless dental stop/fricative was palatalised between â and iy (as opposed to i/î without y)."

• rāθiyā [raːθʲijaː] "season, quarter (of the year)"

but unpalatalised,

• lāθih [laːθih] "warrior"
• rāδīθi [raːðiːθʲi] "says"

3) "When the preceding vowel was stressed and rounded, there was no palatalisation of a labial or guttural con-sonant between this and i/ī but palatalisation of any other single consonant (or nd) did occur…"

• slondīθi [slon͈dʲiːθʲi] "names"
• korīθor [korʲiːθor] "places, puts"
• torīμā [torʲiːṽaː] "recounting, telling"

but unpalatalised,

• tolóɣīθ [toˈloɣiːθ] "forgives"
• kugenā [kugenaː] < Latin coquīna "kitchen"
• oμiyan [oṽijan] "copper, bronze"

4) "Between an unstressed vowel and a back vowel (s)siy-  seems to have been simplified to -sy- and then -s- before palatalisation…" (this mostly applies to relative clauses in verbs and is more of an historical development than a conditioned rule)

• marwasa [marwasa] "[one] who killed" from earlier marwassiya.