zompist bboard

Hello everyone,

This is my first thread here. I joined the board because I love talking and learning about language, and you lot seem like-minded.
My question is on the topic of segments. https://en.wikipedia.org/wiki/Segment_%28linguistics%29

I will point out the concept of consonant gradation. This example from Welsh:

Dim ond calon lân all ganu
Canu'r dydd a chanu'r nos.

(This feature seems to have been conditioned by a phonemic environment in Proto-Celtic which has been lost in the present day Celtic languages, so that today, this mutation is grammatically conditioned. For example, all words immediately following a "and" suffer one sort of mutation, as does any first word in a question) (the words come from a famous 19th century hymn called "calon lân")
Another example from (some varieties of) Czech is that a masculine animate noun in the nominative plural (and any attributive adjectives) get their last consonant palatalised before the ending is attached. (drahý pták - an expensive bird. drazí ptáci - expensive birds)

It seems that a segment is to a phoneme what a phoneme is to a phone, in that a segment may have several phonemic (and orthographic) realisations, and a phoneme may have several phonetic realisations. In this way, the Welsh words "canu", "ganu" and "chanu" up there all begin with the same segment, and as for the Czech example, the "k" and "c" represent the same segment, as do "h" and "z".

My question is whether this is the mainstream interpretation of what a segment is.

Segment is more of a cover term for a phone or phoneme as a "thing". For example, a word like <cat> has the phonemic form /kæt/. This form has three "segments" /k/, /æ/, and /t/. The phonetic realization of these can vary according to the phonology of the language, etc.

Segmental theory is somewhat deprecated because it is difficult to state that segments are real things in a human's mind unlike phonemes and syllables (including onsets and rhymes, i.e. nuclei and codas).

But given that in consonant mutation in Welsh the mutation is grammatically triggered (and in some cases marks a semantic distinction!) could we even say that it begins with a segment?

A more common way of describing this I think is to refer to them as "allomorphs" or "morphologically conditioned alternants"

Yng wrote:But given that in consonant mutation in Welsh the mutation is grammatically triggered (and in some cases marks a semantic distinction!) could we even say that it begins with a segment?

A more common way of describing this I think is to refer to them as "allomorphs" or "morphologically conditioned alternants"

Isn't allomorph:morpheme::allophone:phoneme, thus if a single segment in a morpheme changes, the allomorph is not the changed segment - it's the entire string with both changed and unchanged segments in it.

Miekko wrote:

Yng wrote:But given that in consonant mutation in Welsh the mutation is grammatically triggered (and in some cases marks a semantic distinction!) could we even say that it begins with a segment?

A more common way of describing this I think is to refer to them as "allomorphs" or "morphologically conditioned alternants"

Isn't allomorph:morpheme::allophone:phoneme, thus if a single segment in a morpheme changes, the allomorph is not the changed segment - it's the entire string with both changed and unchanged segments in it.

Yeah, I'm not sure what the common analysis of consonant mutation in Celtic languages is, but some sort of allo-something is definitely going on.

Anyway, segments are a useful shortcut to discuss phonemes within a word even though no one can say, "there, that's the segment" (as opposed to the onset or coda or such)

It's a piece of orange innit?

Allomorphone? Morphoneme?

there is 'allomorph' and 'morphophoneme' which... well... what does morphophoneme even mean? have I invented it? anyway allomorphs are the context-dependent realisation of a specific morpheme

but my entire point was that no in fact there is not an allomorphic change going on in celtic consonant mutation because rather than just being context-dependent realisations of morphemes (and talking about allomorphs which are triggered across word boundaries is dubious enough in the first place) - this is like saying that two case forms of a Latin noun (for example) are allomorphs when in reality they are DIFFERENT COMBINATIONS OF MORPHEMES

Isn't allomorph:morpheme::allophone:phoneme, thus if a single segment in a morpheme changes, the allomorph is not the changed segment - it's the entire string with both changed and unchanged segments in it.

yes

I appreciate my original post was badly written to the point of obscuring my meaning so please forgive me

Some allomorphs are clear enough: [s]~[z]~[əz] for the usual plural morpheme in English. But others are less clear. so is the [ən] plural (in oxen, children) an allomorph of the plural morpheme? Really, one might say that the plurals in s/z are just phonological and not really allomorphs (since the alternation is only phonologically conditioned).

Orthography can help sometimes, but it really just depends on the example and the language. So English past tense morpheme /əd/ is spelled <ed> for most words regardless of its pronunciation as [t]~[d]~[əd]. The plural morpheme though often gains an <e> when it appears as [əz] (or loses the <e> when it is pronounced without a schwa...).

To be clear:
Jumped [t]
Jammed [d]
Jutted [əd]

Cats [s]
Dogs [z]
Glasses [əz]

CaesarVincens wrote:Some allomorphs are clear enough: [s]~[z]~[əz] for the usual plural morpheme in English. But others are less clear. so is the [ən] plural (in oxen, children) an allomorph of the plural morpheme? Really, one might say that the plurals in s/z are just phonological and not really allomorphs (since the alternation is only phonologically conditioned).

Orthography can help sometimes, but it really just depends on the example and the language. So English past tense morpheme /əd/ is spelled <ed> for most words regardless of its pronunciation as [t]~[d]~[əd]. The plural morpheme though often gains an <e> when it appears as [əz] (or loses the <e> when it is pronounced without a schwa...).

To be clear:
Jumped [t]
Jammed [d]
Jutted [əd]

Cats [s]
Dogs [z]
Glasses [əz]

except that this phonological allo...thing only occurs in this morpheme and not generally as a phonological rule

Right, so which is why I'm saying it's not really clear often whether its phonological alternation, morphological alternation or some combination of the two.

I forget right now if etymologically different morphemes can be considered as allomorphs so /-s/ plural and /-ən/ plural in English as they both mark only plurality. But yeah, root morphemes can have allomorphs as well, so that may be the best analysis for consonant mutation in the modern Celtic languages since the phonological conditioning has often been lost. It's allomorphy if the roots are the part that is changing, but otherwise don't change meaning, or another morpheme is marking the change in meaning.