LINGUIST List 3.352

Sun 19 Apr 1992

Disc: Eclipsis in Irish

Editor for this issue: <>


Directory

  1. Richard Sproat, 3.341 Queries: Eclipsis

Message 1: 3.341 Queries: Eclipsis

Date: Sat, 18 Apr 92 16:16:38 ED3.341 Queries: Eclipsis
From: Richard Sproat <rwsmbeya.research.att.com>
Subject: 3.341 Queries: Eclipsis

Sean Day writes concerning eclipsis in Irish, the initial consonant
mutation process whereby voiced stops become nasals, voiceless stops
and /f/ become voiced, and vowels are preceded by /n/.

The history is fairly clear. Like other Celtic consonant mutations it
is a result of what was presumably originally a phonetic juncture
phenomenon that became phonologized. In this case the original
juncture effect was nasalization, induced by a nasal ending on the
preceding word. This is pretty certain because, in addition to the
fact that some of the instances of Irish eclipsis still involve
nasalization (i.e., those involving initial /b, d, g, V/), several of
the conditioning environments for eclipsis correspond to conditioning
environments for the nasal mutation in Welsh. For example, we have the
following, where in all cases <mb> is the eclipsed form of <b>):

 Irish			Welsh

 i mBaile Atha Cliath	yng Nghaerdydd (< Caerdydd)
 in Dublin		in Cardiff

 seacht mbad		saith niwrnod (< diwrnod)
 ocht mbad		wyth niwrnod
 naoi mbad		naw niwrnod
 deich mbad		deng niwrnod

 7,8,9,10 boats		7,8,9,10 days

There are certainly grammatical differences in the distribution: the
Welsh numerals given only induce the mutation with the words for day
or year, whereas there is no such restriction in Irish; `fy' (my) in
Welsh induces nasalization, whereas Irish `mo' (my) induces lenition;
Irish `a' (their) induces nasalization, whereas Welsh `eu' (their)
induces no mutation (though it does cause /h/ to be inserted before a
following vowel-initial word); eclipsis also generally occurs in Irish
with the initials of NPs following the sequence P Art, whereas nothing
corresponds to this in Welsh. But on the whole the overlap, plus the
fact that at least some of the conditioning words originally ended in
nasals (cf. Latin `in' "in", `septem' "7", `novem' "9", `decem' "10"),
makes it highly likely that the process was originally a (phonetic)
nasalization junctural phenomenon. Whether it can be traced to common
Celtic is (I take it) still uncertain, but eclipsis was in any event
already extant in Old Irish.

(It can also be noted that some of the eclipsis-inducing elements in
Irish, such as `a' (their) and `na' (genitive plural definite article)
correspond to words ending in nasals in Scottish Gaelic: `an' (their)
`nan' (genitive plural definite article). I do not know whether this
is a later development, or if this is an archaism preserved in
northern Gaelic dialects.)

As for literature on the topic, Thurneysen's "A Grammar of Old Irish"
(Dublin Institute for Advanced Studies, 1946) will contain a
discussion of the history, as does Lewis and Pedersen's "A Concise
Comparative Celtic Grammar" (Gottingen: Vandenhoeck & Ruprecht, 1937)
(there are also earlier German versions of both of these). Lehmann &
Lehmann's "An Introduction to Old Irish" (MLA, 1975), also contains a
little discussion. I am not aware of much in the generative
literature on this topic. Part of the problem is that although several
people have worked on Celtic mutation recently, very little of that
work has been published. Furthermore, most people have concentrated
on the much more pervasive and much more complicated process of
lenition. What little has been said about eclipsis has usually
involved either a straightforward SPE-style phonological rewrite rule,
or (more recently) an autosegmental analysis whereby a nasal
autosegment introduced by the conditioning environment is linked to
the initial of the following word:

 [+NAS] [WHATEVER]

 \ |
 \ |
 \ |
 \ |
 X

If WHATEVER is a voiced stop, then it is nasalized, if it is a vowel
the [+NAS] becomes /n/ (the default nasal?), and if it is a voiceless
stop or /f/, it either changes its voicing feature, or else has the
specification [+VOICED] filled in, and the [+NAS] feature is somehow
prevented from surfacing (e.g., by an inherent [-NAS] specification).

One paper that presents some analysis along these lines is:

Massam, Diane. (1983). ``The Morphology of Irish Mutation.''
MIT Working Papers in Linguistics, volume 5.

There is also a Harvard undergraduate thesis by Rolf Noyer (currently
a graduate student at MIT) that covers a large number of phenomena in
Celtic mutation, including Irish eclipsis.

There must presumably be many studies of Irish dialects that include
discussions of eclipsis and its distribution, but unfortunately, I
don't know anything about that.

Richard Sproat
Linguistics Research Department
AT&T Bell Laboratories			tel (908) 582-5296
600 Mountain Avenue, Room 2d-451	fax (908) 582-7308
Murray Hill, NJ 07974			rwsresearch.att.com
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue