More on Fillers.

We’ve discussed filler words a number of times (2009, 2017, 2021), but Anne Delaney’s JSTOR Daily piece has some things that weren’t in the previous posts, and I thought it was worth bringing to your attention. Some excerpts:

In spoken language, we see that many elements are universal, one being the way speakers listen and take turns in a conversation. These markers or thinking sounds (uh, uh huh, huh, hmm, er, like, right?) may be collections of sounds with meaningless lexical value, yet they pack a pragmatic punch.

They can be perceived with a range of filters, neutral or positive ones such as creating connection, agreement, and unity; or with a more negative view, such as a crutch, tic, parasitic word, or distracting habit. These exist in every language. The French utter eh bien; Portuguese have então, ta, pois; Japanese えーと (“eeto”), and なんか (“nanka”); Spanish – mira, vale, among others.

Understood by several labels, verbal fillers and hesitation markers are some of these universal elements, a type of discourse marker. Interjections and rejoinders also come to mind—words a listener uses to keep the conversation going and show the speaker she understands and even sympathizes with them; for language learners, using them adeptly (with the right syntactic placement, intonation, and timing) may demonstrate further fluency. […]

When the speaker is not visible, verbal “signposts,” can be helpful, as they take the role, in part, that nonverbals do. Politicians, public speakers, and those making a formal presentation, on the other hand, might gather their thoughts and hold the floor with a “Soooo…,” a “Look,” or an “Ummm” in between points. Learners of more than one language might hear these signals early on, and start incorporating them, purposefully or not, to create added fluency and confidence in the new target language.

Judgments of and reactions to these elements of speech range from perceiving a mark of personality—it’s simply what the speaker does—to a more visceral judgment (distracting, inept, unprofessional, unpracticed). Examples of these utterances include the well-known “you know” and “like,” and the increasingly more frequent “right?”

You know creates an exchange structure focusing the listener’s attention on a specific piece of information provided by the speaker. That seems like a lot to pay for a coffee, you know?

Like is a big one for protecting oneself from potential disagreement: “Do you, like, want to have Indian food for lunch?” perhaps really means: “Will you have Indian food for lunch with me?,” or functions to reduce uncertainty and perhaps avoid rebuttal or being wrong in an utterance:

It’s, like, supposed to rain all afternoon.

Right: Has this one crept in quickly or slowly? The peppering of a “right(?)” in the telling of a story or sharing new information can take the place of “Do you know what I mean?” Additionally, depending on placement and tone, it can connote a spoken yet implied request that the listener agree, while the listener might be thinking, “Must I agree? (I don’t, actually: I just learned this; this is his story; or I don’t agree at all.”) […]

Imagine how challenging it is to use verbal fillers accurately in a second or third language. Or, um, rather, since there is already a more spacious canvas, maybe one should not even bother sprinkling them in. Indeed, interpreters are trained to omit fillers, so as not to be perceived as uncertain or create doubt with the accuracy of an interpretation. Ironically, using filled pauses naturally, “accurately,” where they are expected in conversation, can result in the speaker being perceived as fluent, or at least communicatively competent in a second or third language. […]

Mark Liberman discovered, though his parsing of 14,000 phone conversations, that “uh” increases with age, but that at every age, male speakers use it more than female speakers do; whereas, “um” decreases with age, but female speakers use it more than male ones do at each stage in life.

The state of affairs is actually simple and one of a common ground: we each have our own idiolect, a term coined by Bernard Bloch from the Greek idio- (personal, distinct) + -lect (social variety of a language), which is the unique speech of an individual. This term refers to the theory that, while being influenced by or sharing a common dialect, sociolect, culture, and environment, no two individuals have the exact same linguistic tastes and idiosyncratic features in their personal inventory of variants.

Obligatory gripe: I’m tired of the ubiquitous “right?” and wish it would go away. Thanks, Martin!


  1. cuchuflete says

    Two reactions: (1) mira in Spanish is not really a filler word. It’s more a preamble to a statement to follow. The English “Look” serves the same rôle, demanding attention to what the listener is about to hear. (2) The quoted material brought back fond memories of my elderly landlady in Santander, Spain in 1968. When Doña Carmen wanted to join a conversation on a topic that involved social disquiet, such as politics—this was Spain under Franco, and another of her tenants was a female Guardia Civil agent—she would utter a string of filler terms: Pues (pause), Bueno pues (another long pause), Pues nada.

  2. English right isn’t a filler word, either—at least not in the usage Delaney describes.

  3. I would take issue with “nanka” being described as a filler word in Japanese.
    政治なんかに興味はない means “I’m not interested in such things as politics.” Remove the nanka and you get “I’m not interested in politics.”

  4. I recently read Michael Erard’s 2007 book on the subject, Um…: Slips, Stumbles, and Verbal Blunders, and What They Mean. It’s a good treatment of the topic, for anyone interested in an extended exploration.

  5. PatrickC says

    I’ve been wondering why closed captioning on British TV shows uses “erm” where Americans would expect “uh” or possibly “um.” Does this reflect an actual difference of sound, or is it just a convention?

  6. “erm” in a non-rhotic accent is “um”.

  7. @PatrickC: The standard British spellings of uh and um are er and erm, which assume non-rhotic pronunciation. See, for example, here.

  8. David Eddyshaw says

    I actually say er rhotically. It never occurred to me that this was a spelling pronunciation, but I suppose that it must be. Now I will have to practice saying uh.

    I think that I only say it that way in mock-uncertain contexts (“Boris Johnson is … er … not always entirely trustworthy”) but introspection is an unsafe guide in such matters.

  9. Depending on speaker, context, tone of voice, etc., 政治なんかに興味はない could be “I’m not interested in such things as politics,” but it could also be “I’m not interested in, like, politics.”

  10. David Marjanović says

    There’s still a difference between a not too unstressed [ɜ] and a not to unstressed [ʌ], but…

    I actually say er rhotically.

    I’ve also met thoroughly rhotic Americans who write errrrrrrrrrrrrrrr, and I think they say it, too. But I think that can all be blamed on reading.

  11. Kate Bunting says
  12. When I was a kid I thought “tsk” (representing an alveolar click) was pronounced “tisk,” and I think that’s pretty common as well.

  13. David Marjanović says

    That’s amazingly common – I’ve actually heard it in some YouTube video. “Tisk, tisk, ti-isk!”

  14. CuConnacht says

    The Turkish filler yani is a borrowing from Arabic يعني, ya’ni, literally “it means” (the most common filler in Arabic). It surprises me that a filler is a loan word. Are there examples in other languages?

  15. David Eddyshaw says

    Not exactly a filler, but Kusaal “O.K.” is certainly borrowed, probably from Hausa (it’s one of a fair number of such words which are kind of ambient in the region.)

  16. CuConnacht, Welsh borrowed the English filler well as wel.

  17. David Eddyshaw says

    True. Though I kind of associate it with wele di, look you …

    (GPC, though confirming the obvious point that’s it’s a loan from English, actually does suggest that there may have been some influence from that, so it’s not just me. It might be another instance of loan-by-contamination, of the kind that gave Kusaal ti’eb “prepare” the additional sense “heal” by association with the Arabic ṭibb “medicine.”)

  18. David Marjanović says

    The best is the non-loan by contamination: the law was too close to la loi to be replaced by it, unlike the entire rest of English legal language. (Only deem & doom escaped by escaping legal language altogether.)

  19. Kusaal ti’eb “prepare”

    Do Kusaal and Mooré verbs ever represent Arabic masdars? I was wondering if Kusaal ti’eb, Mooré tibi could be from تأهب taʾahhub “preparation, preparedness, readiness, alert”, verbal noun of form 5 taʾahhaba (as here) “equip oneself, provide oneself with the necessary equipment, get ready, prepare”. I couldn’t find anything to correspond in Hausa, though. I doubt Kusaal glottalization reflects the hamza in the Arabic word, but what do I know? Is Kusaal glottalization inherited from way back—and if not always, how does it arise?

  20. David Eddyshaw says

    (a) I don’t think there are any loans directly into Kusaal from Arabic; although there are some Muslim Kusaasi, they are a pretty small proportion (and I think this is a comparatively recent development*); there is no tradition among the Kusaasi of literacy in Arabic, as there is with the Hausa. To this day, most Kusaasi seem perfectly happy with their traditional worldview, and show no great enthusiasm to adopt either Islam or Christianity. This is significant, because you would need to identify a proximate source for Kusaal ti’eb in a language which also had vowel glottalisation, and there are no likely candidates.

    The traditional orthography of words like ti’eb is misleading; the structure is not CVCVC but CV:C, with a long glottalised vowel. Vowel glottalisation can be realised as glottal constriction after the first mora, or it can be just creaky voice, but either way the ‘ never behaves as a consonant: it never begins a syllable.

    Vowel glottalisation in Western Oti-Volta is found only in Kusaal (both dialects), Talni and Nabit (which are Kusaal’s closest relatives) and in Farefare/Gurenne/Nankan/Booni (a dialect continuum.) However, it must have been a feature of Proto-Western Oti Volta: Farefare is not particularly closely related to Kusaal within WOV, and moreover the non-WOV language Nawdm has Vɦ in cognates of WOV words with glottalised vowels, e.g. Nawdm raɦm “bile, gall” = Farefare ya’am, mtaɦ “three” = Kusaal ntan’. Nawdm is not too remote from WOV but the split cannot really have been much less than about 2000 yrs BP; Nawdm is much farther from Kusaal (say) than Romanian is from French, for example.

    No other Oti-Volta language has glottal vowels. In one way, this is hardly surprising: it it was a feature of Proto-OV, it’s not too great a stretch to imagine that it would be repeatedly lost independently given the rarity of vowel glottalisation as a feature. However, I suspect that vowel glottalisation was the product of some VC -> V’ rules at some point (this actually happens synchronically in Kusaal, but the process only accounts for a small proportion of all glottalised vowels, and Farefare has nothing like that anyway.)

    Anyhow, Mooré, which is certainly the proximate source of some Arabic loans in Kusaal (e.g. maliak “angel”) doesn’t have glottalised vowels.

    (b) Loans into WOV languages are (as crosslinguistically normal) overwhelmingly nouns, though there are some identifiable loan verbs (and a surprising number of grammatical particles.) But I can’t see how Mooré tibi could really be derived from taʾahhub, unless there was some pretty ferocious imala going on. In fact, it’s the presence of glottalisation in the Kusaal ti’eb which really convinces me that, at least in the “prepare” sense, the word is not borrowed; glottalisation in loanwards seems to occur only by analogy with native vocabulary (e.g. Kusaal lɔmbɔ’nɔg “garden”, borrowed from Hausa lambu but influenced by bɔn’ɔg “swamp, ricefield.”)

    The vowel length alternation Kusaal ti’eb vs Mooré tibi, is not a problem: this is a recognised thing in Oti-Volta, though the original conditioning factors for it (still discernable in Mbelime) have got thoroughly screwed up by wholesale levelling in WOV.

    (c) the forms in -p- in both Kusaal and Mooré definitely go back to *bb, so those forms really must be traced to ṭibb specifically.

    * The comparative lack of Muslim influence probably correlates with the fact that the Kusaasi (and their Mande neighbours to the north, the Bisa) have repeatedly fought off attempts by the Mamprussi to impose chiefs on them; the chiefly clans of the Mamprussi and the Dagomba seem to have been Muslim at least to some extent for centuries, and there are actually many more identifiably Arabic words in Mampruli than Kusaal, mostly mediated via Hausa. The original capital of Gbewa, the first king of the state which later gave rise to the Mamprussi, Dagomba and Mossi kingdoms, was at Pusiga, which is now deep in Kusaasi territory; the capital had to be relocated southward in Gbewa’s sons’ time because of an untimely Kusaasi/Bisa revolt.
    The king of the Mossi (the “Moro Naba”) has actually sometimes been Muslim in recent times, but the Mossi as an ethnic group are overwhelmingly not Muslim; the Mossi fought off the 18-19th century jihadists.

  21. David Eddyshaw, thank you for this full answer! Especially for the mention of Nawdm with corresponding to glottalization elsewhere. (I am always interested in more data like this because of the spontaneous appearance of ʿayin or pharyngealized vowels in some inherited Iranian words in Kurmanji varieties, like çav ‘eye’, Mardin [tʃɑv] vs. Bitlis, Siirt [tʃæʕv] (cf. Avestan cašman- ‘eye’, Balochi چم‎ čam, Sorani چاو‎ çaw, Zazaki çim).)

  22. Xerîb, do those ‘ayins always correspond to an open back vowel elsewhere?

  23. David Eddyshaw says


    You’re welcome!

    Unfortunately, as far as I can make out from the not very detailed descriptions of Nawdm phonology that I have seen, the ɦ symbol is just used for /ʔ/; I don’t know why they adopted that symbol in particular, but it doesn’t seem to have its IPA value.

    Kusaal itself, as I mentioned, has developed glottalised vowels from VC sequences: this happens with original *ag *ɛg *ɔg, and is recent enough to still play a role in the morphophonemics: e.g. gik “dumb”, plural gigis but zak “courtyard”, plural za’as. In Toende Kusaal, the reflexes were still different from glottalised vowels when Prost described that dialect in the 1970’s, though they are so no longer.

    Most Kusaal glottalised vowels can’t be explained like that though. There are one or two other cases of VC alternating with glottalised vowels, though, one of which happens very systematically; I have to beetle off and do something else just now, but happy to supply further details later …

  24. David Eddyshaw says

    … in fact, Now.

    The systematic case is plurals in the re/a noun class.
    When the stem ends in a consonant, things are straightforward; an epenthetic vowel is usually inserted between the final consonant and the r of the suffix, so you get e.g.

    kugir “stone” (with the usual loss of the final short vowel or the full form kugirɛ)
    kuga “stones”

    The r assimilates to precededing l m n r, so you get e.g. gɛl “egg” (full form gɛllɛ), plural gɛla, but there’s nothing very complicated going on there, obviously.

    The interesting bit is where the stem ends in a vowel; here the usual pattern is e.g.

    nɔɔr “mouth”
    nɔya “mouths” (vowel shortening before word-internal y is an active synchronic rule.)

    However, if the vowel is glottalised, you get this instead:

    yʋ’ʋr “name”
    yʋda “names”

    This is completely regular and works for all cases where the stem ends in a glottal vowel, except those due to the recent changes of *ag *ɛg *ɔg, that I mentioned above, and even they do the same by analogy sometimes.

    Western Oti-Volta languages which have lost vowel glottalisation have no trace of this, but treat all the vowel stems alike: thus Mooré yʋʋre “name”, plural yʋya.

    I don’t have an explanation for this phenomenon at all.

    The plural suffix of the re/a class has historically lost an initial consonant; the evidence of the corresponding 3rd person pronoun in languages with noun-class-based gender agreement suggests that this was *ŋ, a consonant which, obligingly, was not found in Proto-WOV except word-initially, and has generally been lost in word-medial position elsewhere in Oti-Volta too (though some Eastern Oti-Volta languages show -ka or -ha in these plurals.)

    Many of the forms found in re/a plurals across Oti-Volta look like repair strategies to deal with the consequences of stem-final vowels meeting flexion-initial vowels. The -y- of WOV is probably in fact of the same origin as the r in the singular (i.e. Proto-OV *ʎ), and Nawdm in fact does exactly this in this class: noor “mouth”, plural noora, where r is the regular reflex of Proto-OV *ʎ.

    But the -d- is deeply mysterious. I suspect the regularity with which it appears in the plural of glottal-vowel stems is the result of some pretty thoroughgoing levelling, but I don’t have much idea where it came from in the first instance, to provide a model for levelling.

    Less systematic cases are the verb-deriving suffixes -g and -s: when they are added to a root which ends in the stem vowels ɔ or ʋ, the vowel becomes glottalised: “close”, yɔ’ɔg “open”, versus e.g. “dress oneself”, yɛɛg “undress oneself.”
    No idea …

  25. Would the constant use of the adjectival “f-ing” word by some speakers be called a filler? It adds nothing to meaning except possibly a slightly threatening undertone.

  26. do those ‘ayins always correspond to an open back vowel elsewhere?

    No, not always back, but for the insertion of ayin or spontaneous pharyngealization of a consonant, always open: mezin ‘big, large, tall, great’ can be [mɛzˤɪn] and I have heard even something like [mæʕzˤɪn] (cf. Avestan mazant-, maz-; Sanskrit mahant-, mahā-). The letter e represents a vowel that shows great deal of variation along the [æ] ~ [ɛ] spectrum among Kurmanji dialects and speakers and generally reflecting Indo-Iranian short *a.

    There is a full treatment of the problem by Daniel Barry ‘Kurmanji Kurdish Pharyngeals: Emergence and the Perceptual Magnet Effect’ in Current Issues in Kurdish Linguistics (2019).

  27. David Eddyshaw, thank you for all the additional information!

  28. The best is the non-loan by contamination: the law was too close to la loi to be replaced by it

    What are some other examples to illustrate “non-loan by contamination”? I don’t understand how this particular example relates to the actual Middle English situation, since Anglo-Norman lei did in fact enter English.

    The ME word lei “a law; body of law; God’s commandments and teachings”, of Anglo-Norman origin, existed beside Middle English laue (with ME laue itself being of Norse origin and appearing very late in Old English). This ME lei is found in a variety of texts: the Cursor mundi, Chaucer, the Ludus Coventriae cycle, etc. Middle English lei tended to be used in specialized in sense (OED: ‘Law; esp. religious law; hence, a religion, a faith’). Was it lei that was primarily responsible for the retreat of OE ǣ, ME e ‘law, especially divine law’, leaving laue to take over some of the semantic space of OE ġesetnes, dōm, etc.?

    However, the MED has examples of use of lei in non-religious sense (as from the Ludus Coventry cycle: ‘A barn is born… Wolde clymbyn kynge and knytys and lett my lordly lay.’). The last cite in the OED for lei is from 1593 and shows the specialized religious sense:

    G. Peele Famous Chron. King Edward the First sig. B3
    Tis Churchmans laie and veritie, To liue in loue and charitie.

    I wonder if this lay died out because of later competition from its near opposite lay ‘laic; the laity’ (Old French lai).

  29. Only deem & doom escaped

    Just offhand, in courtroom language, there are also oath (beside Middle English serment) and witness (beside rare Middle English tesmoing, testimoigne ‘testimony’) and writ (beside Middle English bref, but with brief surviving in a separate legal meaning) and moot (with much change of meaning). I think it is really cool that bequeath survived only in its legal and quasi-legal metaphorical meaning (with Anglo-Norman lesser ‘leave, bequeath, lease’ with noun les (modern French laisser and legs) being specialized as English lease). It would be fun to find some more.

    Looking at the dates in the OED and the TLFi, I imagine that hearsay is a calque on Old French oïr-dire (modern ouï-dire).

  30. David Marjanović says

    I don’t understand how this particular example relates to the actual Middle English situation

    That’s because I didn’t know the actual Middle English situation!

    I wonder if this lay died out because of later competition from its near opposite lay ‘laic; the laity’ (Old French lai).

    Makes sense to me in any case. The lay of the land is still in use (as a deliberate archaism), though, because it’s contaminated with yet another lay.

  31. David Marjanović says

    Looking at the dates in the OED and the TLFi, I imagine that hearsay is a calque on Old French oïr-dire (modern ouï-dire).

    There’s Hörensagen in German, but Wiktionary and the DWDS don’t trace it farther than a 14th-century hœrsagen and offer no comparisons.

  32. David Eddyshaw says

    thank you for all the additional information!

    Once again, you’re most welcome.

    Another case may be a change of *Vds -> V’Vs in derivation (I say “may be” because there are few examples, and alternative analyses are possible.) An example is the ya-conjugation verb gur “be on guard, be waiting for something.” This would go back to *gudya, but there are no contemporary related Kusaal forms in gud- to confirm this. However, there is the causative gu’us “put someone on guard.”

    There is no such change in flexion in the ga/se noun class: nwadig, “moon/month”, nwadis “months.” However, there is abundant evidence of stem levelling between singular and plural in nouns, so this is not a big obstacle.

    I have played with the idea that this might explain the mysterious yʋ’ʋr “name”, yʋda “names” alternation (that I mentioned above) if you posit *Vdr -> V’Vr in flexion formerly: this might then be the origin of the pattern that got generalised as the plural for re/a class nouns with stems ending in glottalised vowels.

    An immediate problem is that there are actual stems ending in -d in the re/a class, and they don’t do this: kpandir /k͡pãdɪɾ/ “baboon”, plural kpanda etc. However, this can be finessed: most examples are in adjectives, which historically adopted the noun class of the modified head noun (and even now are very often extant with class suffixes from more than one class, now in free variation); it is no great stretch to suppose that the adjective stem before -re would thus be levelled with other noun-class forms of the adjective. And of the nouns, an actual clear majority have meanings where it is plausible that the singular might have been remodelled after the plural form (e.g. “cheek”, “shoulder”, “hip”, “thicket”, “tick.”)

    There is also an interesting adjective meaning “young female that has not yet given birth”, the stem of which actually alternates between sad- and sa’a- in Western Oti-Volta, although the forms in the two Kusaals are disobligingly the wrong way round: Agolle pu’asadir “adolescent girl”, Toende pɔ’ɔ-sa’a; but this can be explained reasonably enough as different stem allomorphs gettting regularised in different languages. Moreover, the fact that the word belongs to the re/a class at all in Agolle Kusaal may be a secondary development: you’d expect one of the more usually “human” classes like ga/se, as in Toende, or a/ba, like the Mooré cognate and synonym pʋg-sada, plural pʋg-sadba. (The first element/head in all these compounds is “woman.”)

    The “name” word itself, Kusaal yʋ’ʋr, plural yʋda, may be cognate with the synonymous Ditammari (di)yetidi and/or Mbelime hetide and/or Nawdm hidr, but there are several problems with the correspondences: only the Ditammari and Mbelime match without any real difficulty at all.

  33. I don’t have an explanation for this phenomenon at all.

    Comparing with the non-glottalized case would seem to predict e.g. yʊʔʊr : *yʊʔ(ʊ)ya … Getting from *ʔy to d doesn’t seem impossible, e.g. via *ʔɟ > *ʄ > *ɗ.

    (Eschewing “ʋ” which is the labiodental approximant, as far as I’m concerned; its use for /ʊ/ in African linguistics seems to me like some kind of a broken telephone development, maybe from an analogy eεuυ and then further treating ‹ʋ› as if it were a “Latin small upsilon” rather than a rounded ‹v›.)

  34. David Eddyshaw says

    Unfortunately the sequence of glottalised vowel + y does actually occur, in the ya-conjugation, e.g. sɔn’eya “be better than” (reduced by the usual Kusaal apocope of final short vowels to sɔn’e in most contexts.) Here the underlying stem vowel is ɔn’ɔ (i.e. /ɔ̰̃:/), as seen in the corresponding “agent” noun sɔn’ɔd “surpasser.” Unlike modal vowels, glottal vowels don’t shorten before y, but they do undergo fronting of the second vowel mora.

    However, one possible escape from this conundrum is the fact that Kusaal/WOV y is of (at least) three distinct Oti-Volta origins; the one in the ya-conjugation suffix is certainly from Proto-OV *ʎ, but it’s not altogether clear what’s going on with re/a plurals. The Nawdm evidence suggests that that is also *ʎ, but as I said a lot of the forms in the different OV branches seem to show independently developed strategies to get round the collision of stem-final root vowels with vowel-initial flexions, and it is very likely that the Nawdm strategy does not reflect the protoloanguage.

    Mbelime doesn’t introduce /d/ (which is its reflex of *ʎ); thus e.g.

    yóóde “tail”, plural

    This word (as far as the singular goes) corresponds precisely (including tonally) with the Kusaal cognate zʋʋr, plural zʋya. The fronting seen in the plural turns up in many words in this class and certainly suggests a lost *y.

    In Nateni the flexion-initial consonant is preserved, as /k/ (probably from *ŋa via *ga; the Atakora Sprachbund does that kind of thing a lot. It does not like voiced stops, and is only prepared to accept nasals on Thursdays and Fridays.)

    Ditammari seems just to tolerate the vowel hiatus.
    Waama, in which Proto-OV *ʎ -> r, inserts y, like WOV:

    nɔre “mouth”, plural nɔya (= Kusaal nɔɔr, plural nɔya.)

    What may be going on here is an alternation in the initial consonant of the suffix between ŋ and ɲ; there are some cases of this happening root-initially elsewhere in Oti-Volta (in Mampruli, initial *ŋ has actually regularly become ɲ.) The reflexes as y would be quite regular; you would expect nasalisation of the following vowel, but most OV branches that preserve contrastive vowel nasalisation at all have lost it in non-root syllables.

    There is no actual /ʔ/ in yʋ’ʋr [jʊ̰:ɾ], BTW: the orthography is misleading. However, the vowel may be derived historically from Vʔ, so that doesn’t vitiate your idea.

    Eschewing “ʋ” which is the labiodental approximant

    It’s so used in Ewe, you’ll be glad to hear.

    I think the use of “ʋ” for [ʊ] is not really an “Africanist” thing, in the sense that academic linguists specialising in African languages prefer it: they mostly don’t, in fact. It’s more a local orthographic choice, often along with “ɩ” for [ɪ]. It seems to be particularly a Francophone thing, at least in origin.

    (I use it even in my own in-house orthography for Kusaal because I deliberately chose to keep that as close to the standard orthography as possible consistent with accurately mapping to the real sounds of the language; here I generally cite Kusaal forms in the standard orthography, though with some scruples. It really doesn’t mark nasalisation nicely …)

Speak Your Mind