zompist bboard

What are some examples you guys know of phrase-final allophones or allomorphs? That is, when a different form of a word is used right before a pause. This is attested in Standard Arabic as well as in Min Nan.

In Classical Arabic, and by extension in the higher registers of Modern Standard Arabic, many of the inflectional word endings are either dropped or have a reduced form before a pause. For example, the indefinite nominative ending of the "diptote" declension, -u, is dropped in such an environment, so that what would be "2allaahu 2akbaru" (2 = glottal stop) in the middle of a sentence becomes simply "2allaahu 2akbar". An example of a reduced ending would be the nominative dual -aani, which becomes -aan before a pause.

In Min Nan, tones have a different level or contour before a pause, where they effectively become another tone. For a summary of how it works in Taiwanese Min Nan in particular, see this Wikipedia section.

I remember I once read that some Western Romance language (I don't remember which) exhibits a little bit of this, having a few words with a distinct, longer shape before a pause. I remember the article where I read this gave the example pan > pani, but I couldn't find it again.

Any other examples?

Whimemsz has informed me Cheyenne devoices all phrase-final vowels.

In English, I pronounce "to" as /tə/ in most places, but as /tu/ at the end of an utterance. Too boring an example?

Serafín wrote:In Min Nan, tones have a different level or contour before a pause, where they effectively become another tone. For a summary of how it works in Taiwanese Min Nan in particular, see this Wikipedia section.

True of Mandarin third tone, too, which is [21] within a phrase, [213] at the end of one.

zompist wrote:True of Mandarin third tone, too, which is [21] within a phrase, [213] at the end of one.

That's not as nice an example as Min Nan, since people often use [21] at the end of a phrase too... but, yes, it is an example of the phenomenon.

Historically, a number of French words ending in the letter "s" were pronounced with a voiceless final consonant /s/ in pausa and no final consonant before a word in the same phrase starting with a consonant. Before a word in the same phrase starting with a vowel, a form with the voiced fricative /z/ was used. This distribution has broken down to some extent in the modern language though.
Some examples:

• the numeral six "six": /sis/ in pausa, /si/ (in standard French) before consonant-initial word, /si.z/ before vowel-initial word. The word dix "ten" follows the same pattern.*
• the adverb plus "more": /plys/ when it's used as a stand-alone word meaning "more" or in a construction like plus de __ "more of __", /ply/ before an adjective starting with a consonant meaning "more [adjective]," /ply.z/ before an adjective starting with a vowel meaning "more [adjective]"**. [ply] is also used in all contexts for the negative or negative polarity adverb meaning "(not) anymore"
• tous "all (masculine plural)": /tus/ when it's used as a stand-alone pronoun meaning "everyone/all," /tu/ as an adjective before a word in the same phrase starting with a consonant, /tu.z/ as an adjective before a word in the same phrase starting with a vowel

*A similar alternation exists (or at least existed at one point) for some numerals ending in other consonants, such as cinq "five," sept "seven," huit "eight," neuf "nine." The sources I've found indicate the omission of the final consonant is now completely obsolete for sept and neuf, and apparently rare for cinq. The words six, huit, and dix are described as still omitting the final consonant before another word starting with a consonant for speakers of standard French, but apparently they may not for some speakers of Montreal French.

**The "liaison" consonant /z/ between plus and an adjective starting with a vowel is apparently optional in modern spoken French. I don't know much about it since I've only studied French as a student, and textbooks sometimes list certain liaisons as "mandatory" that are apparently actually optional in spontaneous spoken speech. I found a paper that shows it as somewhere intermediate in behavior between articles and possessive determiners (which essentially always have liaison in the relevant environments) and verb forms like "sont" and "ont" (which everyone agrees are only occasionally realized with liaison): http://bernardlaks.info/wp-content/uplo ... r_LCC2.pdf

Another example is Sanskrit-- e.g. the root vac- appears phrase-finally as vak. Unfortunately I haven't absorbed this section of the grammar yet so I can't say much more, but see any grammar for details.

I recall reading somewhere that English actually has ejectives as utterance final allophones of the stops. I actually know a lot of people who do this, so I'm inclined to believe it.

I'm personally not really sure what you're looking for here, tbh. Kinyarwanda also devoices all phrase-final vowels IIRC, and...well, Malagasy just devoices vowels like crazy. Malayalam has an alternation between and [ɯ], and morpheme- (and word-)final vowels are generally deleted before another vowel (I think all of this is true of Tamil, too, but I'm not sure, and I'm not sure how much this is subject to dialect variation in Tamil). I'm not really sure how to explain the distribution, though. Phrase-finally, you would probably be more likely to find [ɯ] than (but of course, you'd find neither before another vowel).

Vijay wrote:I'm personally not really sure what you're looking for here, tbh.

Do you mean my purpose of making this thread? Mostly just to make people aware that phrase-final allophones and allomorphs exist. I don't recall ever seeing a conlang that had this.

Serafín wrote:

Vijay wrote:I'm personally not really sure what you're looking for here, tbh.

Do you mean my purpose of making this thread?

Nah, I think I'm mostly just confused by your Arabic example because I thought that alternation depended on whether the next word began with a vowel or not, not on whether the word was phrase-/utterance-final or not...or am I just confusing Classical Arabic with Modern Arabic varieties?

Vijay wrote:

Serafín wrote:

Vijay wrote:I'm personally not really sure what you're looking for here, tbh.

Do you mean my purpose of making this thread?

Nah, I think I'm mostly just confused by your Arabic example because I thought that alternation depended on whether the next word began with a vowel or not, not on whether the word was phrase-/utterance-final or not...or am I just confusing Classical Arabic with Modern Arabic varieties?

Yeah, the phenomenon I mentioned regarding word-endings is a thing of Classical Arabic / the higher registers of Modern Standard Arabic only. Consider that words in Classical Arabic never begin with a vowel; they always begin with a consonant... The syllable structure is very much CV(C).

In somewhat less high registers of Modern Standard Arabic you do see something about word-endings appearing to avoid a consonant cluster of three consonants (notably when the article l- is involved). Maybe that's what you're recalling.

I was thinking of things in MSA like ibni 'my son' vs. hādhā bni 'this is my son', where ibn does not begin with a hamza. Does Classical Arabic not have the word ibn? Does it have only bin or what?

Sorry, I just don't know much about (Classical) Arabic.

Vijay wrote:I was thinking of things in MSA like ibni 'my son' vs. hādhā bni 'this is my son', where ibn does not begin with a hamza. Does Classical Arabic not have the word ibn? Does it have only bin or what?

Sorry, I just don't know much about (Classical) Arabic. :P

Oh, that, I see. In Classical Arabic that is actually not a phrase-final but phrase-initial alternation... where you have 2ibnu (with a hamza) (or other forms depending on case) at the beginning of a phrase, and bnu/etc. elsewhere. BTW, this also applies to the article (2a)l- and to forms VII-XV ((2i)nfa3ala, (2i)fta3ala, (2i)f3alla, (2i)staf3ala...) (but not to form IV which always begins with 2a- or 2i-, nor most nouns and adjectives beginning with a hamza).

I do wonder what happens when 2ibnun is found at the end of a phrase though, namely how it is pronounced. As in laday-ya bnun 'I have a son'. It can't be just [lædæjːæ bn] because that would seem to violate Classical Arabic's phonological rules. I have no idea myself.

EDIT: I made a thread on WordReference about this. Let's see what they say...

If you have 2ibnu only phrase-initially and bnu, etc. elsewhere, but you can't have CVCC syllables, then how do you say 'the son'? Would you not have something beginning with (2a)l-ibn?

Yeah, al-ibnu.

Actually, when I think about it, you could analyze it as ali-bnu, arguing that the i is epenthetic.

Anyway, R. M. W. Dixon says that the last syllable in a phrase in the Jarawara language is nasalized and bears rising intonation.

Ancient Greek had a different pitch accent for words with the pitch accent on the last syllable when they were found before a pause. E.g. the word for 'god/goddess' would normally be θεὸς within a sentence, but it would change to θεός (notice the acute accent) before a pause.

I'm super late to the game, but Ancient Greek also had "ephelcystic nu" which was an optional -/n/ at the end of some forms which was employed to avoid hiatus or for metrical purposes, but also frequently drops in at the end of clauses or verses.

e.g. Archilochus 6, line 5
... ἀλλὰ θεοὶ γὰρ ἀνηκέστοισι κακοῖσιν, / [verse ends]

Both of the last 2 words contain a dative plural suffix, but the latter includes the ephelcystic nu at the end of the verse.

Wow, no one mentioned Hebrew. Biblical Hebrew has a whole slew of rules for phrase-final allomorphy, mostly involving the lengthening in pausa of short or reduced vowels, and stress movement from the ultimate to the penultimate syllable. Some examples, using the root qṭl as a placeholder:

Nouns:
- qéṭel > qā́ṭel (segolate nouns)
- qǝṭī > qḗṭī, qṓṭī (segolate y-final nouns)

Verbs:
- qāṭǝlā > qāṭā́lā (3Sg.F Perf)
- yiqqāṭǝlū > yiqqāṭḗlū (3Pl.M Impf)

Pronouns/Prepositions/Conjunctions:
- ănī > ā́nī (1Sg.Nom)
- lǝkā > lā́k (to + 2Sg.M)
- wǝ- > wā- (“and”, changes when pretonic)

Examples:
- attā wā-ā́nī “you and I” ~ ănī wā-ā́ttā “I and you”
- qāṭǝlā hī gédī “she killed a lamb” ~ gǝdī hī qāṭā́lā “’twas a lamb she killed”
- nātattī lǝkā ḥăẓī gédī “I gave you half a lamb”
- miggǝdī nātattī lǝkā ḥḗẓī “of a lamb I gave you half”
- nātattī ḥăẓī lā́k “I gave half to you”

For some of speakers of very nonstandard Polish (including me while speaking with them) endings -ę and -ą carrying every possible meaning are [ɜ ʌ] phrase-finally and [ɜ̃m ʌ̃m] in the middle of sentences. Here I give some examples:
Kupię sobie coś nowego.
I'll buy something new.
Coś chyba sobie kupię.
I think I'll buy something.
Widzisz książkę, którą wczoraj czytałeś?
Do you see the book you were reading yesterday?
Chcę tę książkę!
I want this book!
Widzą mnie?
Can they see me?
Nie, nie widzą.
No, they can't.
Zmierz to linijką.
Measure this using a ruler.
Linijką tego nie zrobisz
You can't do this using a ruler.
Remember this is only for some Poles, I personally don't have this allomorphy for -ą (always [ʌ̃m]), some additionally have only [ɜ] for -ę, some have a nasal diphthong for -ą and some even for -ę which are remnants of older pronunciation on which the orthography was based.