Greco-Sinitic ψάμμος / ʃˠa mɑk̚ ("desert")

« previous post | next post »

[This is a guest post by Chau Wu]

The psammo- component of the winning word in this year's Scripps National Spelling Bee, psammophile, is of interest to me because it is a good example of European-Sinitic lexical correspondence. The Ancient Greek word psámmos (ψάμμος) means ‘sand’.  When used together with a definite article (ἡ ψάμμος), it also means ‘the sandy desert’. Examples can be found in Herodotus: ‘the sandy desert’ of Libya (4.173), Ethiopia (3.25), and Egypt (3.26). In Sinitic, ‘sandy desert’ is 沙漠 (MSM shāmò / Tw soa-bô·). From psammos to shāmò, it is easy to see three processes of simplification that may have taken place to transform the Greek loan: simplification of the initial cluster ps- > s-, that of the medial -mm- > -m-, and the loss of the final -s. The simplification of ps- > s- is also seen in Greek derived English words such as psyche, pseudo-, and psalm.

The correspondence between Greek psammos and Sinitic shamo has been mentioned in my SPP-262 paper, p. 99, in the section describing the pattern of sound correspondence: the loss of final -s.


Selected readings

And dozens of other Language Log posts in this vein.


  1. Michael Carasik said,

    June 8, 2023 @ 4:28 pm

    If there's really a link between this Greek word and a Sinitic one, I'm encouraged to wonder whether Hebrew צמא 'thirsty' is also related.

  2. Victor Mair said,

    June 8, 2023 @ 5:03 pm


    צָמֵא • (tsamé)

    to thirst, be thirsty
    to long, yearn

    Compare Arabic ظَمِئَ‎ (ẓamiʔa, “to thirst, be thirsty; to desire, long”)


  3. Martin Schwartz said,

    June 8, 2023 @ 6:26 pm

    Re the Greco-Sinitic speculation at hand, I wonder how the sand blew from the Greeks to China, etymologically. I know of the Silk Road, but was this a sand road, so to speak? For Gr. psammos, there's an interesting etymological issue which can be seen from
    psamathos wiktionary. By the way, the latter word yielded the name
    of a district in Constantinople, whence the name of the singer
    Psamathianos–if you like old exotic music, check out the
    of songs–mutually quite different–under that name on YouTube.
    I reissued a "bordello" song by him, with obscene lyrics,
    pirated under Psamatianos (sic) YouTube–nice for fns of old tangos.
    @Michael Carasik: Have a look at Psalm 63:2-3, where your Heb. word is connected textually with a dry and weary land without water. But I don't think the Heb. and Gr. words are any more related than "psalm" and "psammos".

  4. Victor Mair said,

    June 8, 2023 @ 9:13 pm

    As someone who has entered the Taklamakan numerous times to study the Bronze Age and Early Iron Age Europoid mummies that were preserved there, I can vouch for its being a vast, arid desert full of endless sand dunes.

  5. Victor Mair said,

    June 8, 2023 @ 9:16 pm

    ψάμᾰθος • (psámathos)


    Usually taken as a cross of ψάμμος (psámmos, “sand”) and ἄμαθος (ámathos, “sand”). Furnée adduces Proto-Germanic *samdaz (“sand”), so it could also directly continue a word from Proto-Indo-European *sámh₂dʰos (“sand”).

    IPA(key): /psá.ma.tʰos/ → /ˈθos/ → /ˈθos/


    ψάμᾰθος • (psámathos) f (genitive ψᾰμάθου); second declension

    sand of the seashore
    (in the plural) grains of sand
    countless multitude

  6. Martin Schwartz said,

    June 9, 2023 @ 1:02 am

    To put my chief question more clearly: The Gr. word at issue
    as candidate for source of the Sin. does not really qualify as a Kulturwort and is not found as a borrowing among the very few Gr. LWW in Central Asiatic languages; why/how did Chin. pick up the Gr word, when it's likely that there were Sin. alternatives closer at hand?

  7. Victor Mair said,

    June 9, 2023 @ 2:36 pm

    "it's likely that there were Sin. alternatives closer at hand"

    None that I know of.

  8. Philip Anderson said,

    June 10, 2023 @ 5:32 am

    I first met the word “psammos” in ‘Five Children and It’ (E Nesbit), where “It” was the Psammead, a sand fairy, where the name was modelled on “dryad”, “naiad” and “oread”.

  9. Jonathan Smith said,

    June 10, 2023 @ 1:57 pm

    Q: Given onset correspondences Middle Chinese nasal : Hokkien voiced obstruent, which value is relevant for the "Long Shot Comparisons" promoted on this corner of LL?
    A: Whichever makes the favored comparison look better, duh: for 'cow' choose Hok. g- not MC ng-, for 'desert' choose MC m- not Hok. b-, and so on.
    Q: Doesn't Sin. 'desert' historically end in -k, lost in both Mandarin and colloquial Hokkien (cf. Cantonese saa1mok6)?
    A: … Are you saying Chinese is impervious to borrowing? OR Are you saying current reconstructions of early Chinese are that reliable? OR [your favored straw man]

  10. Wolfgang Behr said,

    June 12, 2023 @ 4:11 am

    Hard to see any plausible path from OC *sˤraj+mˤak (in the BS system) to ψάμμος, whether from older *ψάφ-μος and therefore related to ψῆφος ‘pebble’, cf. Lat. sabulum ‘sand’ (as suggested by Chantraine, Dict. étym., p. 1286) or from Pre-Greek *sam- ‘sand, mud’ (as per Louis Deroy, "La valeur du suffixe préhellénique -nth- d'après quelques noms grees en -νϑος", Glotta 35, 1956, 171-195 @183, cited in Beekes).
    Shamo 沙漠~幕 is not a pre-Qin word, it seems, and first safely attested in texts like the Yantie lun 鹽鐵輪 and Hanshu 漢書. It may well be a loanword in Han Chinese. Given the predominantly Xiongnu, Wuhuan 烏桓 (Avar), and Xianbei 鮮卑 contexts of the early textual attestations, I would look into the direction of Old or Middle Iranian and Yeniseian for possible points of contact (if any).

  11. Victor Mair said,

    June 12, 2023 @ 6:55 am

    From a scholar of Central Asian Hellenism:

    Reading the comments under the sand-ψάμμος post I have different thoughts. They seem correct and the period of the Qin-Han also corroborates with the events of a contact.

    Most of the Orientalists absolutely refuse to see a contact between the Chinese and the Greco-Bactrians. But the Greeks were there for about 200-300 years based in walled cities and having mixed armies in Central Asia expanding during the time of the Qin State. Why is it so difficult for them to accept that?

    My only possible explanation is that they impose their actual views on the Greeks as a modus operandi for their analysis. Or maybe that they don't have any Central Asian Hellenistic period studies background? To me it has more to do with a mental pathology or a mental block… because anybody could understand that the distance was NOT far. And with the analysis of all the texts and the archaeological discoveries "ça saute aux yeux"

  12. Victor Mair said,

    June 12, 2023 @ 7:14 am

    The Iranians had intense contacts with the Greeks from at least the 5th c. BC

  13. Chris Button said,

    June 12, 2023 @ 9:36 am

    I'm skeptical, but the final -k in Chinese does not rule out an association. Debuccalization of final -s is well known cross-linguistically (and in Chinese in the origin of qu-sheng). Given that final -k would have had an incomplete release via a glottalic co-articulation, a case could be made for the coda somehow reflecting a debuccalized original -s as part of the loan process.

    One additional note is that Min languages (whichever one is chosen) should never be used as a proxy for a "good" Old Chinese reconstruction. Random aside: I still have not figured out why Pulleyblank's proposals for the "extra" distinctions in Min onsets was never taken seriously.

  14. Victor Mair said,

    June 12, 2023 @ 7:12 pm

    From a Latinist colleague:

    I think most of us classicists know about psammos because we read Herodotus as part of our intro to the canonical authors; but have we ever met a psammophile?

    Also, as a Latinist, I'm intrigued by the fact that the Greek word psammos and the Latin word harena manifestly have nothing to do with one another — well, you should never say that, but at first glance… — while psammos and shamo (not to mention sand) at least look as though they might be related.

    For some reason, I am also thinking about the possibility that Sappho's name conceals an underlying *shap-pho. Should we be translating her name as Sandy?

  15. Chau said,

    June 13, 2023 @ 1:52 pm

    @Jonathan Smith
    Thank you for your insightful and penetrating comments. My formal research training is in protein sequence analysis in molecular biology. The technique I used is called “multiple sequence alignment analysis” by which I collected known sequences of the same protein from as many species as possible, lined them up evolutionary-wise, and looked for patterns of similar sequences. I then tried to discern any relationship between sequence structures and protein functions. I applied this technique to lexical comparison between Taiwanese and non-Sinitic languages (the first one I studied being Tocharian which was not a long shot comparison), and so the title of my SPP-262 paper starts with “Patterns of sound correspondences…” Thus, looking for patterns is second nature to me.
    In the following reply to your comments, I will use patterns to support my points. Because space in the Comments section is premium, I will use 3 examples per pattern which is the minimum required number for establishing a correspondence set (see David Branner, The Classification of Miin and Hakka, Mouton de Gruyter, 2000, p. 15).
    Abbreviations: l., literary (reading); v., vernacular (reading); OE, Old English; ON, Old Norse; Gk, Greek; L, Latin; Hok., Hokkien; Tw., Taiwanese; MSM, Modern Standard Mandarin.

    (1) “for ‘cow’ choose Hok. g- not MC ng-…“: Let’s look at the data for 牛, 吳, and 鵝.
    牛niú. As I have mentioned in another thread where 牛 ‘cow’ is discussed, the nasalization of g- > ng- was an innovation in the northern branch of Sinitic from which Mandarin emerged later. Sino-Japanese (Go-on) gu / (Kan-on) gyū do not show the ng- initial, thus, the innovation may have taken place after the loan to Japanese. For 牛, Tw. has, besides (v.) gû, (l.) giû, one nasalized variant ngiu (in 疑母). The proposed source is PIE *gwous.
    吳 wú, the name of a kingdom during the Spring-Autumn period and another during the Three-Kingdom period, the name of a topolect, and also a common surname. The oracle bone inscription shows ‘a man with tilted head + mouth’ with the meaning ‘to shout, speak loudly’. (See 吳 Wiktionary.) Tw has two pronunciations: gô·, and ngô·. The Sino-Japanese (Go-on) gu and (Kan-on) go also show lack of nasalization. My proposed source for 吳 is Gk góos (γόος) ‘howling, groaning, wailing’.
    鵝 é, the goose. The bird is called in OE gōs which I believe is the source for Tw gô (l./v.), and in ON it is gás which may give rise to another Tw (v.) giâ and the Sino-Japanese (Go-on/Kan-on) ga. Tw has a nasalized variant (l.) ngô·.
    Here are the patterns for the initials of pronunciations of 牛 – 吳 – 鵝. The source: g-g-g, Tw non-nasalized: g-g-g, Tw nasalized ng-ng-ng, and Sino-Japanese g-g-g.

    My guiding principle for choosing a match between source and loan is to select the one closest to the source. In this case, I go for g-g-g, of course. This is like buying a used car, the closer the model year to the present and the lower the odometer (closer to manufacture, the source), the better. And you would give a car that has been involved in a wreck (that is “innovation” in linguistics) the least consideration. In this case the ng- nasalization is the wreck.

    (2) “choose MC m- not Hok. b-…”:
    Southern Min, the Taiyuan (太原) topolect (traditionally called 晉語) and Sino-Japanese show prominent de-nasalization (Forrest, R.A.D., 1965, The Chinese Language, 2nd Ed., London: Faber and Faber, pp. 177-185). MC m- > b- is one such feature and is an innovation.
    Now let’s look at the data. I will use three glyphs bearing 莫 as the phonophore to make a pattern for comparison. I will excuse 漠 of 沙漠 from consideration for now, and at the end let’s see where 漠 should be placed. The three glyphs plus the prototype 莫 are compared as follows.
    莫 MSM mò // Tw (l.) bō·, bók / (v.) bák
    慕 MSM mù // Tw (l.) bō·, —– / (v.) —-
    寞 MSM mò // Tw (l.) —–, bók / (v,) —-
    模 MSM mó // Tw (l.) bō·, —– / (v) —-
    For 慕 ‘to admire, adore, long for, desire, yearn’, the MSM pronunciation has undergone a vowel shift (*mo > mu), but it does not affect our consideration on the initial consonant. My proposed source for 慕 is L. amor/amō (amāre) ‘love’. It may have undergone aphesis to result in *mor > *mo > MSM mu / Tw bō·.
    寞 ‘lonely, desolate’ is used commonly with 寂 to form 寂寞 ‘lonely, solitary; quiet, still’. My proposed source is Gk mónos ‘alone’ / monachós ‘alone’. From the first syllable mon- one can derive *mok, which is realized in MSM mò and Tw bók.
    For 模 ‘model, standard, example’, a likely source is ON mót ‘model’.

    Now we have the patterns for the initials of pronunciations of 慕 – 寞 – 模. The source: m-m-m, and the reflexes of the supposed loans in MSM m-m-m and in Tw b-b-b. We go for m- as the choice, instead of b-. Ditto for 漠 of 沙漠 from Gk ψάμμος > shāmò.

    “Doesn’t Sin. ‘desert’ historically end in -k, lost in both Mandarin and colloquial Hokkien (cf. Cantonese saa1mok6)?”
    ‘Sandy desert’ is written as 沙漠 and occasionally as 沙幕 in older literature. The pronunciations of 漠 and 幕 in MSM and Tw are given below along with the prototype 莫.
    莫 MSM mò // Tw (l.) bō·, bók / (v.) bák
    漠 MSM mò // Tw (l.) bô·, bók / (v.) bák
    幕 MSM mù // Tw (l.) bō·, bók / (v.) —-
    So, in theory, 漠/幕 can be read as bô·/ bō· or bók. In Taiwan it is idiomatic to say soa-bo· for ‘sandy desert’, differing from Cantonese regarding the final -k. In my post I described what it is. Linguistics is descriptive, not prescriptive.

  16. Philip Anderson said,

    June 13, 2023 @ 6:01 pm

    I don’t understand which languages you are comparing. You have picked PIE *gwous, Gk góos (γόος) and OE gōs, which are three different (albeit related) languages. I can accept Tocharian or Iranian (even at a pinch Greek) words being borrowed into a (northern) Chinese language with cultural loan, but words from Old English or Old Norse are simply not credible.

  17. Victor Mair said,

    June 13, 2023 @ 10:59 pm

    @Philip Anderson

    Whether you agree with the results or not, I think you don't understand Chau's methodology, the foundations of which he carefully spells out in his introductory paragraph. He analyzes available phonological data to arrive at patterns of sound correspondences. It's a matter of interdisciplinary cross fertilization, where scientific principles are held to transcend field boundaries. I don't think Chau's saying that Old English or Old Norse necessarily were directly in contact with Old Sinitic.

  18. Chris Button said,

    June 14, 2023 @ 7:09 am

    @ Chau

    In terms of Sino-Japanese, might I recommend looking at some of the discussion in Pulleyblank’s book on Middle Chinese and Marc Miyake’s book on Old Japanese? That would add some clarity as to why certain forms are the way they are and what the earlier forms were.

    Old Chinese is a little more problematic since competing reconstructions don’t tend to agree. For example, I take huge issue with many of the reconstructions proposed by academics (particularly much of the more recent stuff), but I still hugely respect the work since it is grounded in evidence and based on good logic—even if the conclusions are highly debatable.

    Simply put, no Min language (or any other Sinitic language) should be used as a proxy for some attempt at proper historical reconstruction. Min languages do seem to provide some distinctive insights into Old Chinese, but it’s worth remembering that they are also innovative in their own right.

    One thing I love about LLog is the openness to accommodate more “out there” proposals, I wonder if by adopting some of what I’ve suggested above, you might then uncover some really convincing connections?

  19. Jonathan Smith said,

    June 14, 2023 @ 10:26 am

    @Chau, Since we pretend to do science here, it must be said that you have neither described nor executed a meaningful methodology. Bizarrely, the DNA-sequence-matching metaphor is perfectly serviceable — the reasons you are (or claim to be) unable to understand how to apply the very simple principles involved to language data are utterly beyond me.

    Given my romantic nature I have always held out hope that Chau, Chris, etc., will eventually have Come to Jesus moments and begin to use their knowledge productively rather than spend lifetimes 攑蠓捽仔行雲頂 'waving fly swatters roaming among the clouds.' And yet they 硬不聽勸 'refuse to listen to reason.' But of course, Smith, loss of objectivity is itself a classic symptom of the "wandering ghost" pathology that affects so many promising would-be historical linguists! …damn it all, damn it all…

  20. Chris Button said,

    June 14, 2023 @ 11:13 am

    @ Jonathan Smith

    Your stuff (at least the couple of articles I’ve read) are really out there too, you know.

    Personally I love “out there” stuff in this field. But, as a minimum, it needs a solid grounding in linguistics and palaeography. When someone supplies that, I’m all ears.

  21. Philip Anderson said,

    June 14, 2023 @ 5:10 pm

    @Victor Mair
    You are right, I don’t understand Chau’s method; it seems to mean looking for the same sound, e.g. ‘g’ or ’m’ in different languages. I don’t expect sound correspondences between languages to always be identical, and the further apart they are, the less likely that case is. Grimm’s law was derived from recognising systematic correspondences between d GoGermanic languages and other Indo-European languages, but not identical sounds.
    I do interpret “The bird is called in OE gōs which I believe is the source for Tw gô (l./v.)” as claiming (indirect) transmission from one to the other.

  22. Chau said,

    June 14, 2023 @ 9:42 pm

    @Chris Button
    Thank you for your recommendation of Pulleyblank’s and Miyake’s books. I will check them out. I agree with you that Old Chinese reconstructions have been beset by problems. I do not mean to use Southern Min / Taiwanese as proxies for the reconstruction of Old Chinese. I cite them extensively because, as Taiwanese is my mother tongue, they are the Sinitic topolects I know best. Although they have retained more archaic features than Mandarin, they have their own innovations that one should be cautious about, as you have mentioned. Thank you for your suggestions and advice.

  23. Chris Button said,

    June 15, 2023 @ 10:34 am

    @ Chau

    I hope you enjoy them!

    Pulleyblank's is typically very heavy going and does not even have an index! But it is worth the immense slog to figure it all out. Good luck!

    I love Miyake's book for two reasons. Firstly it seems to be the 'tour de force" on Old Japanese reconstruction ( i am not an expert innsuch matters). Secondly, Miyake seems to have really taken the time to understand Pulleyblank and the connections between Middle Chinese and Sino-japanese.

    Needless to say, Sino-Japanese represents only a portion of each book. Pulleyblank's focus is Middle Chinese and Miyake's is Old Japanese. Sino-Japanese is just an incredibly informative component of each (if you can avoid the inherent circularity when coming at the evidence from both sides and then reconstructing things so they fit both at the same time–as both authors point out).

  24. Chau said,

    June 15, 2023 @ 12:59 pm

    @ Chris
    Re: Pulleyblank's book. Which book: the one on Early Middle Chinese, or the one on Late Middle Chinese, or another?

  25. Chris Button said,

    June 15, 2023 @ 4:38 pm

    "Middle Chinese: A Study in Historical Phonology" (1984). That contains the theory.

    There is also an accompanying lexicon that came out a few years later: "Lexicon of reconstructed pronunciation in early Middle Chinese, late Middle Chinese, and early Mandarin" (1991). That is essentially a list of characters with reconstructed forms.

  26. Chau said,

    June 15, 2023 @ 9:46 pm

    @ Chris
    Thank you so much.
    I have the Lexicon (1991).

RSS feed for comments on this post