F-word diets

« previous post | next post »

JoAnna Klein, "Did Dietary Changes Bring Us 'F' Words? Study Tackles Complexities of Language's Origins", NYT 3/14/2019:

Thousands of years ago, some of our ancestors left behind the hunter-gatherer lifestyle and started to settle down. They grew vegetables and grains for stews or porridge, kept cows for milk and turned it into cheese, and shaped clay into storage pots.

Had they not done those things, would we speak the languages and make the sounds that we now hear today? Probably not, suggests a study published Thursday in Science.

The author emailed me the day before with a request for comments. I wasn't able to get back to her in time, but she kindly sent me a link to her piece when it appeared, and I responded:

I saw it — seems clear and fair.

I agree with Ray Jackendoff that the idea is "is interesting but not earthshaking" — the contribution is not so much a partial explanation for the distribution of labiodentals, because basically who cares, but rather some support for the general concept that physical population differences in principle might sometimes affect language structure. Which again is obviously true in principle, but it's not clear how often it applies in practice. This result would move the needle from "maybe never" to "apparently once in a while".

The usual line of reasoning is the opposite, that vocal tract anatomy has (co-)evolved over the eons to serve the needs of speech communication. (See e.g. the section on "Vocal tract changes in hominid evolution" in my lecture notes for ling001.) That seems pretty well supported, though as with functional-evolutionary explanations for anything, there are disagreements.

 



24 Comments

  1. Peter B. Golden said,

    March 16, 2019 @ 11:10 am

    Hm. What does this tell us about the (now and relatively recently) extinct Ubykh language with its nearly 80 consonants – and two, not very distinct vowels?

    [(myl) The same thing it tells us about Japanese or Danish or Hawaiian — essentially nothing, except maybe some partial explanation for how likely a given language is to have labiodentals.]

  2. Tim Martin said,

    March 16, 2019 @ 11:42 am

    I had hoped that LL would comment on this! I had read another report on this in the Atlantic, and it seemed like much ado about… well, not "nothing," but just a small "something."

    Tangentially, did it bother anyone else that the article implies that all 'f' sounds are labiodental?

  3. Tom Dawkes said,

    March 16, 2019 @ 11:54 am

    How does this fit in with the case of Australian Indigenous languages, which overwhelmingly have NO fricatives, but have multiple stops [dental, alveolar, retroflex, palatal] and nearly always nasals to match each one, and also have at least two laterals and two rhotics?

  4. James Unger said,

    March 16, 2019 @ 1:10 pm

    I was glad to see that the authors cited Hockett 1985. The linguistic idea has been around a long time; the new contribution is the physiological hypothesis of causation, which, for me at least, is not as interesting.

  5. John Roth said,

    March 16, 2019 @ 3:03 pm

    I didn't see the NYT article (I don't subscribe) but did see the Atlantic one. Language started 50,000 years ago? I suppose there are people still holding on to that idea, but it should have been buried in the back 40 a long time ago.

    It's a "fundamental principle" that language (or at least the grammatical and phonetic structure) didn't change for 50,000 years when the only application of that assumption is to historical linguistics, that needs that assumption for about 6,000 years (back to proto-Indo-European)?

    i think someone didn't read the memo properly.

  6. /df said,

    March 16, 2019 @ 3:09 pm

    This was reported by the BBC and I was sceptical. Reading the abstract from the links here didn't help.

    Do the professionals accept that the "combined paleoanthropology, speech sciences, historical linguistics, and methods from evolutionary biology" in the paper itself "provide evidence for a Neolithic global change in the sound systems of the world's languages" and in particular an increase in the frequency of labiodental fricatives.

    Following @Tim Martin, what was to stop our more prognathous prehistoric forebears from forming bilabial fricatives when they wanted to eff with "30% less" effort? If labiodental fricatives dominate bilabials today, did bilabial fricatives dominate before the "Neolithic global change"? Or for that matter, what about dentolabial fricatives which should also have been easier than the modern variety?

  7. Chris Button said,

    March 16, 2019 @ 4:17 pm

    I'm sceptical too. Why isn't the simple fact that lenition (e.g. /p/ becoming /f/) is a very common process in diachronic sound change relative to fortition a sufficient explanation for the situation?

  8. Chris Button said,

    March 16, 2019 @ 4:20 pm

    I should add… based on the "p, t, k" (almost) universal.

  9. Christian Weisgerber said,

    March 16, 2019 @ 4:35 pm

    I would like to urge people to read the actual paper before discussing this, rather than relying on the bastardizations that are circulating in the general press.
    http://science.sciencemag.org/content/363/6432/eaav3218

    Use Sci-Hub to circumvent the paywall if you don't have an institutional subscription.

    [(myl) Or get it here.]

  10. Y said,

    March 16, 2019 @ 8:35 pm

    Their explanation boils down to that some sounds would require more muscular effort with different facial anatomy. Before making unprovable claims about the Neolithic, I wish the authors would show that such a difference in effort (30% or whatever) would actually exert the claimed evolutionary pressure. How much variation is there in muscular effort among the consonant inventory of any one language? Do children with prognathous faces take longer to learn to produce bilabial and labiodental articulations? Does that matter for the stability of any manner of articulation? Answers to these interesting questions can be had with more certainty, but they aren't going to make it out of the specialist journals into the NYT.

  11. Victor Mair said,

    March 16, 2019 @ 11:03 pm

    From Bob Ramsey:

    Reminds me of that stuff Ono Susumu once said about the physiological change in the human population in Japan from Jomon to Yayoi causing the Japanese to not be able to put their lips together properly, and thus causing changes in the は行 consonant!

    Ono's idea was absolute nonsense in so many ways!

  12. David Marjanović said,

    March 17, 2019 @ 10:58 am

    I'm sceptical too. Why isn't the simple fact that lenition (e.g. /p/ becoming /f/) is a very common process in diachronic sound change relative to fortition a sufficient explanation for the situation?

    One of the questions the paper addresses is why the result is almost always [f] and not the bilabial [ɸ]. Today, [ɸ] seems to exist 1) as a short-lived intermediate stage between [pʰ] and [f] or [h], 2) as an allophone of a generic back fricative (like in Japanese), and 3) in languages that distinguish /ɸ/ from /f/ (that's fairly common in West Africa).

    Another is why [w] > [ʋ] (> [v]) has been such a common sound change in Eurasia in the last 3000 years or so.

  13. Chris Button said,

    March 17, 2019 @ 12:10 pm

    @ David Marjanović

    I haven't had time to read the paper yet so probably shouldn't be commenting. However, at first blush, that is largely a question of maintaining symmetry within the phonological inventory as parallel slots become available (otherwise the surface phonetics shift without the underlying phonology), combined with glides like /w/ sitting atop the sonority hierarchy leaving them as such in coda position where they may be treated as diphthongs on the phonetic surface (but not underlyingly) yet requiring greater distinction in the onset position hence undergoing fortition, or acquiring a preceding glottal thus rendering them vocalic etc.

  14. David Marjanović said,

    March 17, 2019 @ 1:40 pm

    The supplementary information is here (pdf, no paywall).

    glides like /w/ […] requiring greater distinction in the onset position hence undergoing fortition

    Again, why is the result so often a labiodental [ʋ] or [v} and so rarely a bilabial [β]? The paper claims to have the answer.

  15. Chris Button said,

    March 17, 2019 @ 2:03 pm

    Like I said, distinctiveness of the onset. You're verging on a glide otherwise.

  16. David Marjanović said,

    March 18, 2019 @ 4:57 pm

    [ʋ] is a glide, and at least as widespread as [v] in Europe today. Many sounds reported as [v] are really [ʋ], and the transcriptions just date from the time before the symbol [ʋ] was introduced; most of German and Slavic, for instance, has [ʋ] and lacks [v].

  17. Chris Button said,

    March 18, 2019 @ 7:17 pm

    @ David Marjanović

    It's simply a matter of descending the sonority scale regardless of what point in the continuum it ends up.

  18. R. Fenwick said,

    March 19, 2019 @ 3:45 am

    @Tom Dawkes:
    How does this fit in with the case of Australian Indigenous languages, which overwhelmingly have NO fricatives, but have multiple stops [dental, alveolar, retroflex, palatal] and nearly always nasals to match each one, and also have at least two laterals and two rhotics?

    The fact that Australian Aboriginal peoples were predominantly neither sedentary nor engaging in intensive agriculture right up until the European invasion means it's hard to use their languages as a control in this instance. On the lack of fricatives coupled with the richness of stop and sonorant inventories (as well as some other features, like an unusual reticence for assimilation by place of articulation), some years ago Andy Butcher did also muse on a possible environmental cause for this phenomenon, but the cause he hypothesised was unrelated to the one being discussed in this article. Butcher's paper can be read in full on his Researchgate page, but I quote the relevant section here:

    "[W]hat motivates and perpetuates the unusual phonemic inventories [of Australian Aboriginal languages] and hence the place-of-articulation imperative? By this point we are well into the realm of speculation, but I will conclude with the following remarks. Middle ear infection (more precisely: chronic otitis media – COM) develops in almost all Aboriginal infants within a few weeks of birth and up to 70% of Aboriginal children consequently have a significant conductive hearing loss (Coates, Morris, Leach, & Couzos 2002). This commonly affects the low frequency end of the scale (under 500 Hz), but may also affect the upper end of the scale (above 4000 Hz). As we have seen, the vowel systems of Australian languages are in general quite small and the majority of them lack any true close vowels. In other words, these systems have no vowel quality distinctions which depend on formant frequencies below about 400 Hz. The consonant systems of Australian languages are also lacking in contrasts which depend on low frequency acoustic cues (voicing), but in addition lack contrasts which depend on cues at the high frequency end of the spectrum (friction, aspiration). On the other hand Australian sound systems are rich in contrasts which depend on rapid spectral changes in the middle of the frequency range. Sounds with high amplitude lower formants can make use of large areas of the basilar membrane of the inner ear, whereby neurons primarily tuned to higher frequencies may nevertheless phase-lock their firing to these lower frequency sounds because of their high intensity. Sounds of this type therefore have a temporal representation in terms of the firing pattern of neurons over a broad span of the basilar membrane and are therefore highly resistant to noise (Greenberg 1996). Thus it appears that Aboriginal languages are rich in sounds whose differentiation exploits precisely that area of hearing ability which is most likely to remain intact in sufferers of chronic middle ear infection."

  19. R. Fenwick said,

    March 19, 2019 @ 4:07 am

    @Peter B. Golden:
    Hm. What does this tell us about the (now and relatively recently) extinct Ubykh language with its nearly 80 consonants – and two, not very distinct vowels?

    Not a lot. The massive consonant inventories of Ubykh and its sisters are the result of little more than reassignment of frontness and roundness from the vocalic nucleus to the consonantal periphery. (Chris Button and I recently had a pleasant discussion on this where I went into the development of North-West Caucasian, and other vertical vowel systems, in more detail: see here.) It's no different in anything but degree from the development of consonantal palatalisation in Slavic or Irish from earlier vocalic contrasts, and in both Slavic and Irish such systems are relatively recent developments too. If you strip Ubykh's consonantal inventory of palatalised, labialised, pharyngealised, and labialised-pharyngealised contrasts (and treat alveolopalatals and palatoalveolars as palatalised alveolars and retroflexes respectively), you're left with a still fairly respectable but much more pedestrian inventory of 36 units.

    For what it's worth, Ubykh does have labiodentals, f (v?) (one of the most striking facts of Ubykh phonology is that non-pharyngealised v is attested only in loanwords and a single recorded onomatopoeic term, although this last may be inherited). Whether they existed in Proto-North-West Caucasian is not entirely certain. Labiodentals can't be reliably reconstructed for Proto-Circassian (they exist in all modern dialects, but aren't cognate; West Circassian f arose from the labialised velar *, and East Circassian f v f' arose from labialised alveolopalatals *ɕʷ ʑʷ ɕʷ'), though Abkhaz-Abaza and Ubykh seem to share a few cognates with labiodentals that may point back to the protolanguage:

    Ubykh "to eat" : Common Abkhaz *fa "to eat"
    Ubykh fəm "to smell, to sniff" : Common Abkhaz *fəʕʷə́ "smell, odour"
    Ubykh vər "sound of a swift motion" : Common Abkhaz *vərə́ "sound of swift rotation"

  20. Chris Button said,

    March 19, 2019 @ 10:18 pm

    Chris Button and I recently had a pleasant discussion on this where I went into the development of North-West Caucasian, and other vertical vowel systems, in more detail

    The aforementioned Charles Hockett also got a mention there too regarding his two vowel analysis of Mandarin back in 1947.

    Regarding the interesting thoughts by Andy Butcher…

    As we have seen, the vowel systems of Australian languages are in general quite small and the majority of them lack any true close vowels.

    I suppose the ə/a system of Arrernte would be the classic "extreme" case in this regard. Given what we know about "vertical vowel systems" thanks to Northwest Caucasian languages and reconstructed proto-languages like Old Chinese and Proto-Tibeto-Burman, along with the e/o (i.e. ə/a) of Proto-Indo-European, this characteristic of Australian languages is not necessarily something that requires a special explanation in my opinion.

    The consonant systems of Australian languages are also lacking in contrasts which depend on low frequency acoustic cues (voicing), but in addition lack contrasts which depend on cues at the high frequency end of the spectrum (friction, aspiration).

    I personally don't find the broad lack of fricatives that surprising when one considers their generally derived nature once one conducts any sort of in-depth diachronic investigation of language families elsewhere. However, the complete absence of any original single /s/ phoneme at all, regardless of specific surface phonetic realizations, would be highly remarkable in an earlier proto-language (although Edwin Pulleyblank started speculating along those lines for Old Chinese in his 1995 emendation of his 1991 "Ganzhi as Phonograms" paper, although I'm not persuaded by the conclusions proposed in either version). I wonder if there has been any research done in that regard for Australian languages?

  21. R. Fenwick said,

    March 20, 2019 @ 3:26 am

    Given what we know about "vertical vowel systems" thanks to Northwest Caucasian languages and reconstructed proto-languages like Old Chinese and Proto-Tibeto-Burman, along with the e/o (i.e. ə/a) of Proto-Indo-European, this characteristic of Australian languages is not necessarily something that requires a special explanation in my opinion.

    Oh, yes. It's more about how it fits in with broader phonological trends, like the absence of fricatives, the rarity of an oral voicing contrast in stops, the richness of the sonorant inventories, and so forth. Where Australia becomes so complicated is not only that all these things coexist, but that they coexist almost universally across the continent and across the boundaries of all the dozen or so top-level language families identified to date. It's not just a mild tendency; it stretches the idea of Sprachbund to its very extreme. For example, the consonantal inventory of Woiwurrung (of the Pama-Nyungan family, spoken around Melbourne) has the following consonantal inventory:

    stops /t ʈ c t̪ p k/
    nasals /n ɳ ɲ n̪ m ŋ/
    laterals /l ɭ/
    rhotics /ɾ ɻ/
    glides /w j/

    but virtually the same system is seen in the Pama-Nyungan language Jiwarli (plus palatal and dental laterals), the Tangkic language Kaiadilt (minus only the retroflex lateral), and the Southern Daly language Murrinh-patha (minus only the dental nasal), despite the fact that no two of these languages are separated from each other by less than 1,000 km as the crow flies, and only Woiwurrung and Jiwarli are widely accepted to be related on any phyletic level at all.

    I personally don't find the broad lack of fricatives that surprising when one considers their generally derived nature once one conducts any sort of in-depth diachronic investigation of language families elsewhere.

    The Australian case isn't about a "broad lack" of fricatives, though; it's about an absolute lack of any fricatives, even /s/ or /h/. My understanding is that there's literally not a single Australian language known to have developed phonological fricatives independently. The only one that has them, Kala Lagaw Ya, is spoken in the Torres Strait and probably developed its fricatives through influence of the neighbouring Trans-Fly languages of Papua New Guinea.

    I wonder if there has been any research done in that regard for Australian languages?

    Yep, there's been some research done into reconstructing Proto-Pama-Nyungan (which accounts for the majority of Australian languages numerically as well as most of the landmass), some of which was kicked off by the late and great Ken Hale, and continued by Geoff O'Grady and Barry Alpher. Much remains to be done, though. It's a particularly challenging nut to crack because of the wide spread of languages, the poor documentation for many of them, and extensive inter-language borrowing, all of which is complicated by the fact that the unusual similarity of phonologies means it's often impossible to disentangle whether a word of a given form is a loan or not.

    To the best of my knowledge, Proto-Pama-Nyungan is generally considered to have had no fricatives either. Barry Alpher's PPN reconstruction draws on the following consonantal inventory:

    stops /t ʈ c p k/
    nasals /n ɳ ɲ m ŋ/
    laterals /l ɭ ʎ/
    rhotics /ɾ ɻ/
    glides /w j/

  22. Chris Button said,

    March 20, 2019 @ 9:54 am

    As we have seen, the vowel systems of Australian languages are in general quite small and the majority of them lack any true close vowels.

    Thinking about this again, although the underlying ə/a phonological system of Arrernte is not particularly remarkable, I think I might need to back pedal a little in terms of phonetics since it does seem strange that there would be no true close vowels at all in terms of how things actually surface. As we noted on the Ubykh thread, the underlying phonology does not preclude a multitude of vowels on the surface. A misunderstanding of this is undoubtedly what lies behind the reticence of academics to accept such phonological systems for living or reconstructed languages be that Ubkyh, Inscriptional Burmese, Old Chinese, Proto-Indo-European or whatever (actually if you go back far enough, my hunch is that everything probably ends up as ə/a in the end – sometimes you just hit it at shallower time-depths than others).

    To the best of my knowledge, Proto-Pama-Nyungan is generally considered to have had no fricatives either.

    That is interesting. I do wonder if perhaps it's a question of not being able to go back far enough to find any evidence for an earlier /s/ type phoneme however? It does have a tendency to disappear although usually leaves some kind of evidence in its wake.

  23. Chris Button said,

    March 20, 2019 @ 10:27 am

    Thinking about this again, although the underlying ə/a phonological system of Arrernte is not particularly remarkable, I think I might need to back pedal a little in terms of phonetics since it does seem strange that there would be no true close vowels at all in terms of how things actually surface. As we noted on the Ubykh thread, the underlying phonology does not preclude a multitude of vowels on the surface.

    Given that pervasive middle ear infection must be a fairly recent phenomenon as Butcher notes, I suppose we can interpret his suggestion to mean that it has somehow modified how the underlying phonology of Australian languages is surfacing phonetically today (i.e. in the past there would have been true close vowels on the surface, but that people are no longer producing them as a result of the effects of ear infection). In that regard, he seems to have a good argument.

  24. R. Fenwick said,

    March 22, 2019 @ 1:46 am

    actually if you go back far enough, my hunch is that everything probably ends up as ə/a in the end

    Which, tangentially, is one of the reasons I get so rage-face about many arguments proposed about the phonetic capabilities of language in pre-Homo sapiens hominins. When the minimum number of phonological vowels for a language can be as low as two, and the number of phonological consonants as low as six, basically there's enough evidence from extant human spoken languages to show that any healthy vertebrate with linguistically-capable cerebral architecture should be able to produce enough clear phonetic distinctions to develop a fully-fledged language.

    That is interesting. I do wonder if perhaps it's a question of not being able to go back far enough to find any evidence for an earlier /s/ type phoneme however? It does have a tendency to disappear although usually leaves some kind of evidence in its wake.

    That's entirely possible, of course, and I don't pretend to know enough about Australian Aboriginal linguistics to be able to say much more for certain. As well, from my limited knowledge of the field I believe the reconstruction of Proto-Pama-Nyungan is still in its adolescence, so further work may well show such a phoneme in the protolanguage or some precursor of it. (Nonetheless, it's important also not to fall into Russell's-Teapot-style traps in that regard; as you say yourself about reconstructing Old Chinese, the proper thing to do is just to follow the evidence where it takes you.)

    Given that pervasive middle ear infection must be a fairly recent phenomenon as Butcher notes

    He doesn't say that this is necessarily a recent phenomenon, though, only that the absence of precolonial evidence means we can't say anything about it. He does offer the journal of George Worgan (the ship's surgeon with the First Fleet of European colonists/invaders in 1788), who noted of the Aboriginal people in the region "the constant Appearance of the excrementitious Matters of the Nose which is collected on the upper Lip, in rich Clusters of dry Bubbles, and is kept up by fresh Drippings". No doubt there was a certain amount of ethnocentrism getting involved here, but it may be a sign that upper respiratory tract issues (including middle-ear infections) were already fairly pervasive at the time of initial contact.

RSS feed for comments on this post