An extreme case of the etymological fallacy

learn pashtoYesterday on Language Log, Mark Liberman posted about the a curious claim that in the language of the Pashtun people of Afghanistan, “the word for ‘cousin’ is the same as the word for ‘enemy.’”  Professor Liberman cannot find evidence to bear this claim out, and strongly suspects that it is bogus.   What sticks in my mind is this quote Liberman gives from an essay by Louis Dupree collected in Islam and Tribal Societies, edited by Akbar Ahmed and David Hart (Routledge, 1984):

Language sometimes reveals unarticulated (or downplayed) conflicts in a society. The term for cousin in Pashto is turbur [and] the word for the worst kind of hatred is turburghanay which could be literally translated ‘cousin-hatred’. But the non-literate, rural Pushtun deny this interpretation. They say: ‘Turbur is turbur and turburghanay is turburghanay. They are separate words. How can they relate? How could I hate my cousin? I would fight to the death with him. I would never leave his body behind in a fight. I would give him my last crust of bread.’

The overwhelming majority of Afghans and Pakistanis cannot read and write, so showing them that the written turbur is a prefix and -ghanay a suffix, which, when combined create a compound word, fails to impress.

It’s hardly surprising that this fails to impress!  Even assuming that Dupree’s etymology is correct, and that the turbur he hears in turburghanay is the word for cousin, we would hardly be warranted to assume that the currency of the word turburghanay implies that Pashtuns secretly hate their cousins.  As Josh Fruhlinger puts it in a comment on Liberman’s post,    

Particularly instructive and hilarious is the quote from the Ahmed and Hart piece, in which the learned outsiders pity the illiterate Pashtuns for not understanding the underlying etymological-psychological implications of the language that they (the Pashtuns) speak. People are determined to believe that language shapes thought even when the acutal speakers of said language don’t recognize the things embedded in the language that are supposed to be shaping their thoughts.

Here’s a little squib about two kinds of mistakes, either of which can be called “the etymological fallacy.”  Dupree seems to have committed both kinds of mistakes.  A person who insists on using words as if their meanings had to be implicit in the meanings of their etymological roots commits a mistake in semantics that can be called “the etymological fallacy.”  It looks to me as if Dupree approached his Pashtun informants in the spirit of this fallacy.  It’s as if he had gone to English speakers and pointed out that the English word nice comes from the Latin word nescius (which meant “unknowing,”) and proceeded to interrogate them about what it is that nice people aren’t supposed to know.  The word has simply changed its meaning over the centuries, so that it has lost any connection it may once have had with the meaning of its etymological base.   

A person who constructs an argument using one word and then proceeds as if the conclusions of that argument applied to other words derived from it commits a mistake in logic that can be called “the etymological fallacy.”  It looks to me as if  Dupree approached the writing of his essay in the spirit of this fallacy.  Argument 1: The Pashto word turbur means “cousin.”  The Pashtun attach great importance to cousinage, modeling other, more distant relationships in their tribal system on it.  Therefore, turbur is a key term for understanding the Pashto tribal system.  This turns into argument 2: The Pashto word turburghanay is derived from turburTurbur is a key term for understanding the Pashtun tribal system.  Therefore, turburghanay is a key term for understanding the Pashtun tribal system.


  1. lefalcon

     /  July 20, 2009

    More info is needed. The semantic field of “turburghanay” may embrace both “cousinship” and “enmity.” And that is weird. But how did that weirdness come about?

    One explanation is that a word’s basic meaning can branch out in various — even seemingly-contradictory — directions.

    Example: “To sanction” means “to approve, support,” but “to impose sanctions” means “to punish someone’s ass.” And punishing someone’s ass is the last thing you’d do if you truly approved of what they were doing.

    So it can be imagined [and this is pure imagining, as I have no idea what the real story is] that the word “cousinship” could’ve referred literally, in some contexts, to a specific kinship relation; while, at the same time, could’ve been used more metaphorically in other contexts to emphasize distance in a relation: “Those dudes over there are nothing like our close family members; shit they don’t amount to anything more than *cousins*! The only thing between us and them is just *cousinship*!”

    Thus, *literal* cousinship is understood as a potent tie among blood kin; whereas the metaphorical usage of “cousinship” drifts further and further away. It initially suggests mere casual acquaintance, then grades over into vague contempt … and eventually explodes into fevered, full-blown hostility. At those goddam COUSINS!!!!

    Another scenario would be that the two meanings are so disparate, because two completely historically-separate words merged into one. For example, if, for some reason, English-speakers of the future started to pronounce “medal” as “metal.” Then somebody might assume that soldiers are awarded objects called “metals” [medals], because they are made of metal.

    I would be interested in knowing what the story is with this “turburghanay,” if / when somebody, someplace, figures it out.

  2. lefalcon

     /  July 20, 2009

    Also: Here are a couple of weird things I discovered about Arabic that supply interesting illustrations of how one word’s meanings can be going off in different directions:

    Al-xaal xaal means “The maternal uncle is empty.” (Xaal means both “maternal uncle” and “empty.”)

    I think that’s funny.

    Raghiba fi al-bayt means “He desired the house.” But Raghiba <an al-bayt means "He detested the house." (The same verb can mean "to desire" or "to detest" depending on whether the following preposition is fi or <an.)

    I think that's odd.

    Finally, just because those Pashto guys thought there was no possible relation between "cousin" and "hatred" doesn't necessarily mean that no historical relation exists. When people say "Gimme your John Hancock," they could have no idea who the fuck John Hancock was or even that there was an historical personage of that name. Native speakers can give accurate info about how a word / expression is used in the present state of their language; but they can be totally full of shit when they try to talk about etymologies.

    Example: "Phat" < "pretty hot and tempting." That's a totally bogus and ridiculous origin that could not possibly be.

    However, this etymology has been verified: The expression "HOT SHIT" — as in "I believe myself to be h.o.t. s.h.i.t." — derives from the phrase "He's On Time and So Handsome In a Tux."

  3. acilius

     /  July 22, 2009

    There’s more info about “turburghanay” in the Language Log posting.

    Good point about “those cousins!” It’s like the way some English speakers use “mother” as an insult. If you didn’t where that usage came from, you might jump to all sorts of conclusions about the attitudes it reflected.

  4. lefalcon

     /  July 22, 2009

    Yeah “mother” is a good example in the context of this conversation.

    To summarize, in a nutshell, I think this guy’s basic mistake is simply: While culture and language go together, inasmuch as you need the one to enhance your understanding of the other, it’s a bit ridiculous to believe you can gain some modest knowledge about a language and then infer a sweeping generality about normative behavior in a culture you’re entirely ignorant of — just upon that brittle fundament of one lexical item possessing two divergent meanings.

    I guess we’re just supposed to accept that Afghanis have this incorrigibly, pathologically, and probably genetically vendetta-prone, violent disposition.

    I also noticed one or two rather naive comments on there, attempting to dispel Sapir-Whorfian theory as absurd.

  5. acilius

     /  July 23, 2009

    I don’t think that the authors and fans of Language Log are unusual in their impatience with the Sapir-Whorf hypothesis. I get the distinct impression that the Sapir-Whorf hypothesis is very much out of fashion among general linguists and historical linguists at the moment. Sentiment runs differently in other fields; I don’t know about other branches of linguistics, but I do know that Sapir-Whorf still seems strong among language pedagogy faculties, and literary scholars often defend their disciplines by making appeals to it. And of course fashions change.

  6. I find the inferences drawn, out of context, from the observation that a word that can mean enemy (turburghanay) begins a word that can mean cousin.

    In English what does “child killer” mean? Someone who kills children OR a child who kill people?

    In the UK a couple of years ago a doctors surgery and residence were vandalized because the attackers did not understand the difference between “paedophile” and “paediatrician”.

    And obviously the English-speaking world

    And of course the old, old anecdote about Americans driving on the parkway and parking on the driveway.

    As someone trying to learn French, I still have difficulty with the same word “personne” being used to mean both “someone” and “no one” depending on context.

    And obviously German-speakers who have different words for “the” (masculine, feminine and neuter) clearly have problems with basic biology when they use the neuter form with “girl” (das Maedchen) and the feminine form with corpse (die Leiche). Does a man’s body become feminine after death?

    Absurd? Yes! Most interpretations based on linguistic fragments taken out of context without deeper/wider understanding are equally unreliable.

    Amusing? Maybe. Thought-provoking? probably.

    Definitive interpretations? Definitely not!

  7. acilius

     /  September 8, 2009

    Thanks for the comment, Robbo! And welcome to Los Thunderlads.

%d bloggers like this: