Dog is in :

*Indo European languages

  • Latin/Roman Languages

  • Latin: canis

  • Chien in French

  • Cane in Italian

  • cão in Portuguese

  • cane in corsican

  • câine in Romanian

  • Armenian -շուն (shun) in Armenian

*Afro Asiatic Languages

  • Hausa : care

  • Aramaic ܟ݁ܰܠܒ݁̈ܶܐ kalba

  • Arabic : كلب Kalb

  • Hebrew :כלב Kelev

  • Akkadian : kalbu

we know that the sound "L" changes frequently to "N" There is a clear ressemblance between the words in Latin and in Semitic languages.

how to explain this ressemblance ?

Is it a common root ?

or a borrowing ?


2 Answers 2


The Latin is from the PIE word *ḱwṓ "dog" which is also the source of Sanskrit श्वन् śván, Irish cú, Greek κύων kýōn, Armenian շուն šun, English hound (from an extended form), and possibly Russian сука súka (amongst other cognates).

The PIE word has a plausible (but not certain) etymology within PIE, from a ∅-grade n-stem extension of *péḱu "livestock" (presumably the original sense would have been "thing of the livestock" and have referred to herding and flock guarding dogs) making it plausibly native.

The Hausa word cited does not seem obviously related to the Semitic words to me. Semitic k & l generally correspond directly to Hausa k & l, rather than to c & r, so I will not take it into account.

The Semitic word appears to be old as it appears in all branches, including Ethiosemitic which is far from direct Indo-European influence. It also shows a final -b which has been suggested to be a relic of a class suffix for harmful animals. As such it probably dates to no later than early Proto-Semitic which predates Proto-Indo-European.

So a borrowing from PIE into Proto-Semitic is impossible,but a borrowing from an even earlier stage of Pre-PIE cannot be ruled out on chronological grounds.

In other early loanwords between IE & Semitic n is generally rendered faithfully (see English wine vs Arabic وَيْن‎ wayn- "black grape" & Ge'ez ወይን wäyn "wine") making a borrowing unlikely.

Cross-linguistically, words for dogs often feature a dorsal stop (or postalveolar affricate) often ending in a continuant. Proto-Hmong-Mien has *qluwˣ, Proto-Sino-Tibetan has *d-kʷəj-n (which has sometimes been connected to the PIE word which would then be a loan), Classical Nahuatl has chichi, Quechua has allqu, and Malay has kuyuk. It seems plausible there is some onomatopoeia going on here, with the words being related to a dog's bark.

So neither word appears to be a borrowing of the other, neither is there evidence of both being inherited from a common ancestor of both language families (as no such common ancestor is well-proven).

  1. The word for "dog" is actually one of the most common didactic tales in historical linguistics these days:

    "... it turned out that the Mbabaram word for "dog" was in fact dúg,[2] pronounced almost identically to the Australian English word (compare true cognates such as Yidiny gudaga, Dyirbal guda, Djabugay gurraa and Guugu Yimidhirr gudaa, for example[3]). The similarity is a complete coincidence: there is no discernible relationship between English and Mbabaram

    Wikipedia: Mbaram language § Word for dog,

    [2]: Dixon, Robert M. W. (1966). "Mbabaram: A Dying Australian Language".

    [3]: Black, Paul (2004). "The Failure of the Evidence of Shared Innovations in Cape York peninsula".

    The necessary assumptions in this argument are:

    • those cognates follow regular patterns that extend to other words of comparable shape in these languages,

    • the fact that dog "remains one of the great mysteries of English etymology" (etymonline.com: dog) is irrelevant as long as it cannot be possibly related to those Australian Pama–Nyungan languages.

    This is often presented in introduction and thus oft repeated uncritically by novices, and experts in various fields alike. NB: Mark Rosenfelder (passim, zompist.org: chance).

    Dixon, however, later rejected the tree model of language descent, "unable to find anything that reliably set Pama–Nyungan apart as a valid genetic group", whereas Bowern et al. are using phylogenetic methods to bolster the case, "that the non-binary-branching characteristics of Pama Nyungan languages [...] are precisely what we would expect to see from a language continuum in which dialects are diverging linguistically but remaining in close geographic and social contact. (Wikipedia: Pama–Nyungan_languages).

    The merrit of this technique is questionable. Bowern argues:

    "As is well known in traditional, qualitative subgrouping, only reconstructed shared innovations are diagnostic of subgrouping (see, for example, Lass 1997). Shared retentions are diagnostic of common membership of a family, but not of the closeness of relationship. (Review: The Indo-European controversy and Bayesian phylogenetic methods. in: Diachronica 2017)

    And in particular,

    "For example, if cognates for both dog and hound were present in several languages and were viable coding choices, but Languages A and B were coded only with dog and C only with hound, A and B would appear as closer to each other than to C

    This is remarkable because Australia never had pet or hunting dogs. This meaning elicited by Dixon was definitely transfered from English.

    Albeit, the Chirala database, on which their 2018 work is based, which "found clear support for a Pama-Nyungan origin around 5,700 years ago in an area south of the Gulf of Carpentaria," (MPG: press release) contains no such word for "dog" because it uses a collection of small Swadesh lists counting no more than 200 words per language, which is the usual fair (so eg. Donald Ringe, Georgiy Starostin, Damian Blasi, etc.) The idea of Swadesh is based around core lexicon which is assumed to be stable, ie. resistant to change and therefore going contrary to the very idea of shared innovations. It seems thus perhaps not unable but difficult to falsify the naive notion that Yidiny gudaga "dog" e.g. had to be from shared innovation. More over, it is generally believed that these methods are unable to detect more distant relationships, so-called macro families.

    For e.g., I can offer two and a half independent examples from Mandinka wuloo-ndingo (dog-young "puppy; liar, little shit") vs. Min Bei 狗仔 /ě-ciě/ (dog-young "dog"), Japanese inu ("dog"), koinu (young-dog, "From 子 (ko, “child”) +‎ 犬 (inu, “dog”)"). Comparing Japanese and a languages from Taiwan has been done before to limited success, but West-Afrikan Mandinka is just too much of a stretch. Anyway, the problem is, what are you going to do with something as short as e? Or, How do you know that first person I is short for *ik when it should technically be *eka in Proto-Germanic (or there abouts, cf. Sihler, New Comparative Grammar of Greek and Latin; Lühr et al., Etymologisches Wörterbuch des Althochdeutschen).

  2. Pretty much the same type of argumentation will hold for PIE and PSem doges. There is no plausible scenario in the history and archaeology of these cultures that would allow connecting them. The Indo-Iranian branches' ideology stands out for defending national Aryan theories of various kinds--the politics that the heritage of this issue has to bear is unmistakable.

    However, there are definitely Semitic substrates in Indo-European languages. Those are technically adstrata, insofar as there is no commonly agreed on identification of pre-Indo-European substrates (and the Early European Farmer aDNA does not seem to prove markers of Semitic descent).

  1. For example, one might note slang words:

    • German Klafte 'woman with inappropriate behaviour' appears to be from Jiddish klawta 'she-dog', and similarly Kelew (with further references c.f. Althaus, Kleines Lexikon deutscher Wörter jiddischer Herkunft).

    • Formally, this does not include German Kläffer "dog", because kläffen "to bark, yap, rant" is onomatopoeic with to clap and cetera (DWDS/Pfeifer: Kläffer; EWAhd: klaffôn, klapfôn).

    • Moreover, Lühr et al. see no evidence of a Jewish determinant in the Old High German corpus and include Voralberg'ish kläpfe "Karfreitagsratsche, Kuh-Schelle, Schwätzerin", Elsasian klepf[e] "Schwätzerin" (ie. 'gossip [woman]', see in that sense Ratsche, Tratschtante, Schnatterente, Klatschbase, Schandmaul, etc. p.p.) simply under klapfa "Peitsche; scutica" ('whip'), "die Knallende" (ie. the whip as cracker-upper, which cracks, see also cracker "white trash" sometimes explained by similar means), which they want to derive one way or another from klaffôn (EWAhd: klapfa), from Proto-Germanic *klap-, yet without Indo-European comparison. Old Frankish clap- in the by-name Claptant is remarkable (of whom?).

    • However, German Kluft ("clothes") is widely believed to be ultimately from Hebrew

    "Rotwelsch Klabot (15. Jh.), Claffot (16. Jh.), Klofft, Klifft, Kluft (17. Jh.) ‘Kleid, Anzug’ [...]

    vielleicht aus hebr. ḥalīfā ‘Kleid’. Oder ist eher von hebr. qālaf ‘schälen’, qelīfā ‘Schale’ (vgl. rotw. Schale ‘Anzug’) auszugehen?" (DWDS/Pfeifer: Kluft 2)

    Actually q-l-f 'peel' is cited as the true etymon more often, because the idea that k must not be derived from h is paradigmatic in Germanist linguistics, definitory of the branch as what it is since Grimm's Law. But see Verner's Law and 'ge-'

    And it does make sense to compare q-l-f because scraps of fur have been traded like money, which is clearly opportunity for a word-of-mouf Wanderword.

    In any other sense, however, Kluft 'rift, gap' patterns with cleave and deserves a solid PIE etymology with undeniable comparisons in claw, cleave, clover, knobloch, clew, clutch, cloud, perhaps clod, klutz, gluteus and other ideas that show no sense of shared innovation and have yet almost no striking, regular comparison beyond the Germanic branch.

Which suggests that the separation is entirely pragmatic post hoc ergo propter hoc. The fact that books have burned and children beaten for speaking the wrong language, is historically precedented and in principle poorly documented. The argument is therefore a fallacy because it is shifting the goal post.

So, although a common cause in ideophones is possible (compare for example German Wau-Wau, English roof-roof, Rufus, my specialty is roofing), this is typically not falsifiable because Rosenfelder's arguments have to hold up to internal reconstruction, too, but they usually don't hold for the wham-bam bing-bong kind of sound word.

That said, a point in the case of n ~ l is Old High German irklāen, cp. knāen 'knowing' (EWAhd); modern German Klötze "nuts", cp. gonads rather than gluteus, albeit not really parsimonious. If you need to know why, see these more general questions:

w ~ l may be salient in the periphery, so it would suffice to compare *ḱwṓ- to *k-l-(?). A better match with kelevelev would be whelp "puppy".

  • 3
    I find it hard to understand what point you're making, but your statement that "...Australia never had pet or hunting dogs." is incorrect. The native Australian dog, the dingo, is known to have been in Australia for at least 4,000 years (there is genomic analysis suggesting an earlier arrival at around 8,000BP). The earliest records we have (early 1600s) show that the dingo was used as both pet and hunting dog, and most likely had been since introduction to the continent. Commented Mar 4, 2023 at 22:15
  • @vectory I rolled back your edit, which was mostly fine but also was an attempt to response; for that, use a comment.
    – Keelan
    Commented Mar 5, 2023 at 10:48
  • @Keelan, it was well within reason, see the help page on editing, "To include additional information only found in comments, so that all of the information relevant to the post is contained in one place", "To correct minor mistakes or add updates as the post ages". So I undid your undoing.
    – vectory
    Commented Mar 5, 2023 at 12:16
  • 1
    @vectory striking out part of an answer, and writing that it’s questionable, especially without giving a source, definitely does not fall under anything you’ll find in the help center as a good faith edit.
    – Keelan
    Commented Mar 5, 2023 at 14:58
  • 1
    I have rolled back to the original answer. When you're editing someone else's answer, please be careful to be supportive of the original author. meta.stackexchange.com/questions/120576/… You can't add comments like "that's questionable" without indications that the original author would agree with that assessment. If you disagree with minor parts, add comments. If you disagree in significant ways, post a new answer.
    – prash
    Commented Mar 17, 2023 at 8:25

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.