Well, it’s technically not eclipsis in the Irish sense (as it’s not a grammaticalized mutation – for this speaker every an cat will sound like an gat with voiced g) but you are on to something. ;-)
But first note that Gaelic doesn’t have phonemic distinction between voiced and unvoiced stops. The distinction between p, t, c vs b, d, g is rather in aspiration: p [pʰ], t [tʰ], c [kʰ] vs b [p], d [t], g [k] (or preaspiration after a stressed vowel p [ʰp ~ xp], t [ʰt ~ xt], c [ʰk ~ xk]).
Now, something like eclipsis (but, unlike Irish eclipsis, purely a phonological thing happening to consonants after nasals) is common in many Gaelic dialects. Most commonly it involves voicing unvoiced stops b [p], d [t], g [k], p [pʰ], t [tʰ], c [kʰ] to their voiced counterparts: [b], [d], [g], [b(ʰ)], [d(ʰ)], [g(ʰ)] after homorganic nasals (so t, d, c, g after any n, p, b after m). In some dialects (Lewis, Skye) it’s full nasalization, the result being respectively: [m], [n], [ŋ], [mh], [nh], [ŋh].
That’s why an cat sounds [əŋ gaht] here.
You might want to look at the discussion under a similar sentence: "Chan eil an cù eagalach", at Nasalisation 2 or Why am I married to ə NɯNʲə agam? on the Akerbeltz wiki, and at Scottish Gaelic phonology: Nasalisation on Wikipedia.
But I’d also like to know where is this speaker from, I’m not that familiar with Gaelic dialects to pinpoint which one they’re speaking. :)
Thank you, that's fantastic, I'll read through those!
My default assumption (particularly if the Gaelic-speaking population were a lot larger and widely spread than it actually is) would have been that there would be some dialects in which "non-lenition mutation" survived, in parts, probably closer to Ireland, so that was my first thought on hearing this. I always mentally put the "an t-s" pattern as being a surviving example of eclipsis, because it is sensitive to grammatical context ("an s" m. sg. common case, "an t-s" m. sg. genitive) - was the change from voiced to voiceless required in Old Irish/Common Gaelic? Irish examples like "gall"->"ngall" would suggest not. (I'm not great at phonology in general, though, syntax is more my thing.)
The change s /s/ → t-s /t/ is not eclipsis. That’s just a weird lenition after the definite article because of its historical form (the definite article used to end in d or t which is visible only when the following lenited s disappears; spelling like
ant shùil instead of an t-sùil would be more etymologically accurate and closer to actual Old Irish int ṡúil).
As you noted, the s → t-s change happens in sg. masc. gen. (and in sg. fem. nom., and also in sg. dat. of both genders): an t-sùil the eye, an t-sagairt of the priest, ris an t-sagart with the priest – all those context normally have lenition: a’ bhean the wife, a’ chait of the cat, ris a’ chat with the cat.
Eclipsis in Irish is another mutation, not affecting /s/ in any way¹. It changes voiceless stops (/p/, /t/, /k/) into voiced stops (/b/, /d/, /g/) and voiced stops (/b/, /d/, /g/) to nasals (/m/, /n/, /ŋ/). It originally evolved from something very similar to what happens today in Gaelic but the original nasals were lost and in Old Irish it already worked as a grammatical mutation.
It disappeared in Scottish Gaelic (the words that caused it originally sometimes got their nasal back, eg. an, am their and in (Irish a their, i in, both causing eclipsis), gun, gum that vs Irish go causing eclipsis, or it went out of use (eg. eclipsis after the article in dative in Munster and Connacht Irish: ar an mbord vs lenition in Ulster ar an bhord and Sc. Gaelic air a’ bhòrd on the table – historically both were used; or ecplisis after the numerals 7–10: seacht mbád vs seachd bàtaichean for seven boats).
was the change from voiced to voiceless required in Old Irish/Common Gaelic? Irish examples like "gall"->"ngall" would suggest not. (I'm not great at phonology in general, though, syntax is more my thing.)
I don’t understand the question… there is no devoicing (change from voiced to voiceless) in Irish nor was there any historically that I’m aware of. It’s only that Scottish Gaelic lost the contrast of voicedness and uses only aspiration to distinguish /d/ [t] from /t/ [tʰ].
In Irish, by the way, the voiceless stops are aspirated too, so /p/ is actually [pʰ], /t/ is [tʰ], etc. – so /p/ differs with /b/ in two features: voicedness and aspiration. That’s the reason why the word for story in Irish is written scéal but sometimes used to be also written as sgéal – it has voiceless [k], but it is also unaspirated – so it’s unlike regular /k/ which is [kʰ] and unlike regular /g/ which is [g]… In modern orthography the voicedness is deemed to be the more important feature, so the word is always written scéal as having /k/ there. In Gaelic where there is only the difference of aspiration, it is written sgeul with g since that’s the letter typically representing voiceless unaspirated [k].
¹ except perhaps for Clear Island dialect in Co. Cork where, as far as I know, they voice /s/ to /z/ in contexts requiring eclipsis, so a súile their eyes might sound /ə zuːl’ɪ/.
There is actually a lot of confusion in Irish grammar about whether s /s/ → t-s /t/ is eclipsis or not, caused by the fact that it should be, and by the fact that there are situations in Irish where you get lenition in some dialects and the nasal mutation in others.
The term eclipsis originally referred, just like an eclipse of the sun, to something going and front and hiding the thing behind. Thus in the example above, an gcat, the g is in front of the c and hiding it (i.e. it is not heard) in order to replace it. This spelling convention is the standard way to represent the nasal mutation in Irish (but not any other Celtic language) which is why the nasal mutation is usually referred to as eclipsis in Irish (only).
The problem is that s /s/ → t-s /t/ is (strictly speaking) an example of eclipsis, since the t is indeed going in front of the s and hiding it. It is, of course, nothing to do with the nasal mutation, but when combined with the facts that certain prepositions (that historically took either the accusative (causing the nasal mutation) or the dative (causing lenition) now cause different mutations in different dialects, and that s /s/ → t-s /t/ is not obviously a case of lenition, it is easy to see how the confusion arises.
The only solution is not to use the term eclipsis, at least when discussing anything other than Irish grammar, and to be aware of the mixed messages in Irish grammar. DD
Regardless of spelling, eclipsis in the context of Irish just means the mutation you call nasal mutation (which also is not a perfect name – the mutation indeed nasalizes voiced stops, but it voices the voiceless stops and f – thus voicing-or-nasal mutation would be more descriptive).
But eclipsis is the name for that particular mutation, happening after i ‘in’, after 3rd person possessive pronouns, after the article in dative (historical accusative). And changing /s/ to /t/ (written in Irish as ts, not t-s) is not part of that mutation (but it is an irregular lenition, and it does always happen to s when lenition would be expected after the article an).
Now, because of orthography – it is indeed written similarly to eclipsis – there is a confusion and some people do call it that.