zompist bboard

I just saw this in Catherine Callaghan's Northern Sierra Miwok Dictionary:

/š/ is a voiceless palatal spirant, found in the exclamation ʔiš·o·! 'Scat' and an occasional loan word from English.

That implies that before the other words were loaned from English, /š/ was found in only that one word. I can think of one other example off-hand of a phoneme in a language that occurs in only one morpheme-- in Crow, the glottal stop phoneme occurs only in the interrogative marker -ʔ (conveniently written <?>)-- but I know I've seen other examples, too.

It is a somewhat interesting phenomenon. Can any of you guys think of any more examples?

In Malayalam, fricatives in general are limited to loanwords. The only exception to that I can think of is in the word [ɕɛˈɾi] 'okay, correct, right'.

AmE /eə/.

Some Austronesian languages, according to Blust:

In Bintulu the general marker of negation is ã [ʔã]. Since other words that begin with a- invariably have an oral vowel, the nasality of the negative marker cannot be explained as conditioned. However, in a field corpus of nearly 800 available lexical items no other examples of unconditioned vowel nasality are known. It might be argued that many languages (English among them) have colloquial negative markers with deviant phonetic properties such as morpheme-specific vowel nasality, but a second example from northern Sarawak cannot be explained in this way. Miri, spoken near the mouth of the Baram River, has the minimal pair haaw ‘rafter’ : hããw ‘2sg’. Here vowel nasality may have arisen through the process of ‘rhinoglottophilia’ (Matisoff 1975), in which glottal and pharyngeal consonants lower the velum, with concomitant vowel nasality. But in this case the conditioning would have to be morpheme-specific, since h reflects *k in both words (PMP *kasaw ‘rafter’, *kahu ‘2sg’). Again, vowel nasality in this form appears to be unique, and the question of how to represent it phonemically presents problems.

Mako:

Mako has a seven-vowel system consisting of the phonemes /i/, /u/, /e/, /o/, /a/, /ə/ and /ɨ/ (nasalisation is also contrastive). However, the /ə/ phoneme appears exclusively in the past tense suffix -tə. It doesn’t appear to be an allophone: every other vowel occurs word-finally after -t, e.g. in bamatɨ ‘stop’, bamati ‘stop!’, bamato ‘it will stop’, bamata ‘it stops’, bamate ‘stop! (reported imperative)’ and õpetu ‘outside’, and stress is always word-final.

Latin:

My Latin dictionary says it (/eu/) appears in ceu ‘as’, heu ‘oh!’, neu = nēve ‘and not, or not’, neuter ‘neither of the two’ and seu = sīve ‘or if’. So they’re all exclamations or shortenings except for ceu, and I would imagine ceu is a shortening of an unattested form.
The Latin diphthongs ei and ui are also very rare: ei appears in ei = hei ‘ah!’ and Pompei, the vocative of the gens Pompēius. (The city is Pompēiī.) ui appears in hui ‘ha!’ huic, dat. sg. of hic ‘this’, and cui, dat. sg. of quis ‘who’ and quī ‘what’.

Dahalo:

Several phonemes can be shown to be recent intrusions into the language through loanwords:[3]
/z/ is only found in recent loans from Bantu and can be nativized as /d̪/.
/tʃʼ/ is only found in loanwords from Swahili.
/ʃ/ is only found in loanwords from Swahili and Somali.
Additionally, several consonants are marginal in their occurrence. Five are only attested in a single root:
/ⁿd̠ʷ/
/ᶮdʒ/, in /kípuᶮdʒu/ 'place where maize is seasoned'
/ᵑɡʷ/, in /háᵑɡʷaraᵑɡʷára/ 'centipede'
/ɬʷ/, in /ɬʷaʜ-/ 'to pinch'.
/j/, in /jáːjo/ 'mother'.
Less than five examples each are known of /ᵑʇˀʷ, tʃ, tsʼ, tʃʼ, kʷʼ, dɮ, ʄ, ⁿd̠, ⁿdz/.

Do almost-but-not-quite single instances count?

Chechen and Ingush have a voiceless trill in the numbers seven /vʷor̥/ and eight /bar̥/, found nowhere else in the language. The Bats cognates are /vorɬ barɬ/. Ingush apparently also has [o:] that pops up in Russian loans (regularly replaced with native /uo/ when asked to repeat), plus in /joax/ "say" when used as a quotative or hearsay evidential [jo:x]. This occasionally extends to a few other words with /oa/. The author of the grammar calls it "not even a marginal phoneme" but that it could be one in the future.

Archi has a couple words with /ʟ̝/, /náʟ̝dut/ "blue/unripe" and /k͡ʟ̝̊ʼéʟ̝dut/ "crooked." Adjectives very often end in -tːut, -dut, -t:u, etc, so I presume it's an odd reflex when combined with /naˤk͡ʟ̝̊ʼ/ "be.blue" /k͡ʟ̝̊ʼek͡ʟ̝̊ʼ/ "be.crooked." /ʁˤʷ/ is only in three words, /íʁʷˤdut/ "heavy" /ʁʷˤálqˤi/ "fog, light north wind" /ʁʷˤárbos/ "crack of a voice." At a glance, it looks like some of the other labialized sounds are extremely rare as well, though I haven't checked meticulously.

Qiangic languages seem to have a number of vowels that are limited to 1-3 lexemes a piece, though part of that might be incomplete data. Most also have /h/ or /ɦ/ limited to a couple items.

Japhug has /y/ only in /qaɟy/ "fish." The cognate in Tshobdun/Cǎodēng is /qɐɟwɛʔ/.

The only native word in Levei-Drehet with /b/ is /bo/, meaning 'water'.

English [ʇ] in tsk tsk.

Sanskrit /ḷ/ occurs only in one lexeme, kḷp- 'be fit, arrange'.

Xephyr wrote:I just saw this in Catherine Callaghan's Northern Sierra Miwok Dictionary:

[offtopic]are you interested in sierra miwok? i've been looking at sylvia broadbent's grammar of southern sierra miwok and i'd love to correspond about it[/offtopic]

If we're counting interjections and other more purely discourse-y items, English also has /?/ and /@~/ in /?@~?@~/ "uh-uh" and /?@~h@~/ "uh-huh" (apologies if you don't like the SAMPA).

"Huh. / Huh?" and hee-haw (donkey onomatopoeia) are also obligatorily nasalized, at least for me.

Arabic /ɫ/ only appears in Allah's name, no?

As zomp and kodé have pointed out, interjections often contain phonemes that don't show up elswhere in a language, or violate restrictions for those phonemes. I think that can explain at least some of the cases - e.g. the negation marker or past tense marker mentioned may originally have been interjections.

vokzhen wrote:"Huh. / Huh?" and hee-haw (donkey onomatopoeia) are also obligatorily nasalized, at least for me.

That struck a cord with me. It seems that the following broadly transcribed expressions/exclamations contrast for me --

/hə̃Ɂ˧˥/ or perhaps /h̩̃Ɂ˧˥/ (question word, unable to hear, inviting a repeat of the prior statement)

/həɁ˧˩˧/ or perhaps /h̩Ɂ˧˩˧/ (tentative agreement, indeed, etc.)

/əɁ˧˩.həɁ˩˥/ or perhaps /əɁ˧˩.h̩Ɂ˩˥/ (stronger affirmative, agreement; the faster the rise on the second syllable the more conclusive or firm the agreement is)

I am not great at transcribing tone or intonation, but in these expression it is definitely important. Additionally, all three can be made with a closed mouth in which case all seem obligatorily nasalized. For the former two, the tone/intonation/whatever then becomes the contrastive factor.

Onomatopoetic words and exclamations (“shush” to a cat above seems like a prime example) appear belong to a class or classes of words that do not generally fit into the phoneme inventory of languages.

Ubykh /h/ is only found in interjections, loans, and /hənda/ 'now'.

I have [nɛ̃(ː)] as a negative word (I don't know how to spell it, some would spell it <nah>, but <nah> to me is [na(ː)]), and it has a clearly phonemic nasalized vowel on it.

Travis, how do you pronounce <naw>?

kodé wrote:If we're counting interjections and other more purely discourse-y items, English also has /?/ and /@~/ in /?@~?@~/ "uh-uh" and /?@~h@~/ "uh-huh" (apologies if you don't like the SAMPA).

I thought it was pretty common in English to have glottal stops before vowels...

I'm not really sure whether this counts, but I had a professor who worked on Remo, a Munda language spoken in eastern India (Odisha). He said that in Remo, when you want to form a yes/no question, you stick the interrogative marker [ki] (borrowed from Oriya, I think) to the end of the sentence, but apparently, you pronounce it so it sounds pretty much like a bird screeching. I doubt there is any other context in Remo where you'd say something that sounds like a bird screeching. A guy I went to grad school with also pointed out that (some variety/ies of?) Mixtec has a super-low tone that's also only used in a certain context (I think he might have said that it was used in the same context - yes/no questions - but I forget).

Vijay wrote:Travis, how do you pronounce <naw>?

From the orthography I would pronounce it [nɒ(ː)], but it is not really a word in my dialect.

2+3 clusivity wrote:[/həɁ˧˩˧/ or perhaps /h̩Ɂ˧˩˧/ (tentative agreement, indeed, etc.)

/əɁ˧˩.həɁ˩˥/ or perhaps /əɁ˧˩.h̩Ɂ˩˥/ (stronger affirmative, agreement; the faster the rise on the second syllable the more conclusive or firm the agreement is)

I think it's interesting you don't nasalize these, if I do that it sounds like I'm talking with a cold. Also do you really allow them totally voiceless? Even my super-short ones have a brief period of (creaky?) voicing.

In Hoyahoya, /s/ occurs only in the word /sa/ "bone". All other fricatives are spelled with <h>, so /s/ is spelled <sh>.

Spanish /uj/ only appears in muy 'very', and, in some dialects, cuy 'guinea pig' (plural: cuis or cuyes). Some speakers report having /wi/ for these words, but for me muy and fui 'I was' definitely don't rhyme.

Some male Latin Americans have [ˈsə] as an alternative pronunciation of the interjection sí 'yes', but not the adverb sí 'EMPH' (sí lo hizo 'he did do it'). [ə] is nothing surprising for "highland" dialects, but the thing is that this [ə] for the interjection sí is stressed, and often enlongated in celebrations (i.e. [sə:::]). On the Internet you generally see it spelled "seee".

Mexicans and Central Americans have [ˈbɾuʃo, ˈβ-] as an alternative pronunciation of brujo 'shaman' (also applies to bruja 'witch', but I've never heard it for brujería nor embrujado). [ʃ] occurs elsewhere in these dialects in words from English and Nahuatl / Pipil, as well as the interjection ¡osh! 'urgh! (expressing disgust)', but brujo is its only instance in native vocabulary outside interjections. If you think I'm making this up, here's an attestation from the movie Bruce Almighty: at 2:15 you can hear this is some bru[ʃ]o men, ¡vámonos! 'this one is a shaman, guys, let's get out of here!'

Male Salvadoran Spanish speakers sometimes pronounce the sentence-final particle ¿va? 'don't you think?' with [ɑ]: [bɑ]. (This particle also has the forms veá, vedá (for old speakers), and verdad). Va is sometimes also used as an interjection meaning 'alright!, good!', equivalent to ¡vale! in some other dialects.

In Taiwanese standard Mandarin, the triphthong /jaj/ only occurs in one word: 崖 yái 'cliff' (pronounced yá in PRC standard Mandarin).

Alon wrote:Arabic /ɫ/ only appears in Allah's name, no?

Correct.



Almost every language I've studied seems to has at least one hapax phonoumenon: English yeah (mentioned by Nortaneous above), Spanish muy/cuy, (Taiwanese) Mandarin yái, Arabic ’allāh, Latin cui/huic. It seems like something that should be unremarkable, except linguists don't talk about this.

I can't think of anything good for French, but an unpredictable long /øː/ [øː] is reported by le Trésor de la langue française to occur in less than 10 words (jeûne 'fasting' (contrasting with jeune 'young'), meule 'grindstone', émeute 'riot', veule 'sluggish/lazy [person]', neume 'neume').

If we're including loanwords, than the affricate /dʒ/ occurs in Halkomelem only in the word kinjaj meaning "Englishman" (I think this is may be from Chinook Jargon, ultimately from English "King George").

Very interesting topic, I'll have to see if I can dig up any more.

And while we're in the PNW, Makah contains only one word with g, x̌igi·dit "Clallam bay", a borrowing from Klallam (surprise surprise). The other voiced stops are the denasalised b and d. The nasals do occur in a few words, though, such as na·ni· "grizzly bear".

A number of PNW languages have vowels that only appear in nonce words in ritual music (or music in general, in some cases). I don't have any of my material on hand for examples, though.

Norwegian /ʉi/ only appears in the word hui, which only appears in the expression i hui og hast.

Nortaneous wrote:Norwegian /ʉi/ only appears in the word hui, which only appears in the expression i hui og hast.

I was about to point out this same word, although I would add that there's also the derived verb huie. Huie can also be pronounced [hʊɪə] (in that case typically spelled hoie, but I think it's really just the same lexeme), which is funny because [ʊɪ] also only appears in that single word in most dialects.

Finnish /ey iy/ can be generated by the morphology (e- i- + -U), but /ey/ only appears in one root, leyhyä, and the dialectal form keyri (std. kekri), and /iy/ only appears in the place name Kiysaari and dialectal tiyris. /öi yi/, when not generated by the morphology, only appear in onomatapoeia and before /j/.

Amuzgo has a syllabic, prenasalized, voiced velarized bilabial trill which appears exclusively in the word [ʃa˥m̩ˠʙˠ˥] 'antlion'. (The only other phonemic syllabic consonant is /n̩/, which can be pronounced with any place of articulation, or as a lateral, depending on the consonants that follow it. The non-syllabic nasal phonemes are /m/, /n/ and /nʲ/.)