Oxford University Press's
Academic Insights for the Thinking World

It is hard to stop thief

By Anatoly Liberman


The title of this post is meant to warn our readers that the origin of the word thief has never been discovered. Perhaps an apology is in order. I embarked on today’s seemingly thankless topic after I received a question from Denmark about the possible ties between Danish to “two” and tyv “thief.” Although our corresponded knows that they cannot be related, the implications of the to / tyv case and the attempts to discover the etymology of thief are worthy of a short essay.

To and tyv begin with the same consonant (t), but from a historical point of view the identity of t1and t2 is misleading. In the past, the relevant forms were tvau and þjóf, similar to Modern Engl. two and thief (the letter þ designates the same sound as Engl. th). For some reason, th has been lost in most of Modern Germanic (but not in Icelandic or English). In the continental Scandinavian languages it turned into t, while old t remained unchanged. In German old th became d. That is why the German for thief is Dieb. Hence the rule: English (or Icelandic) th corresponds to Danish, Swedish, and Norwegian t and German d. I have dwelt on this uninspiring subject because it is exactly such correspondences that bear the grandiloquent name “phonetic laws.” Since theory is always more straightforward than practice, the “laws” often give researchers trouble. When words are obviously related but something goes wrong with phonetics, the deviation has to be explained.

For instance, Engl. thousand = Danish tusind (everything is fine!) = German tausend. But the German word was expected to begin with d! Why doesn’t it? The reason, which I won’t discuss here, came to light long ago, and the integrity of phonetic laws was saved. In other cases we may be out of luck. Thus, the vowels of heath and heather are incompatible (again I’ll skip the explanation why). Can we venture the conclusion that heather and heath look almost like homonyms by chance? It seems we should! The game has to be played according to the rules, for, if we disregard them, the game will stop. Historical linguists hope to win a fair wrestling match with the material, rather than participating in a skirmish. Sounds change more rapidly than non-specialists think. That is why etymologists always try to deal with the oldest forms recorded in texts. Danish to and tyv look close enough, but as long as we realize that their t’s have different sources, we won’t even try to compare them.

In Broad Daylight

The main Germanic word for “thief” is old. Gothic had þiufs (spelled þiubs), and with Gothic we are in the fourth century CE. The other related languages had similar forms, none of which resembles any non-Germanic word designating a person who steals. Given such evidence, the etymologist faces at least three possibilities.

  1. Perhaps the root of þiufs (to be more precise, of its protoform) existed in Sanskrit (Greek, Latin, Celtic, Slavic—one or all of them) but had a different sense. If so, we should remember that Germanic f corresponds to non-Germanic p (as in Engl. father versus Latin pater) and look for words with the root teup- (in þiufs, i goes back to e, and -s is an ending) or even teu-, because -p may turn out to be a suffix.
  2. Or þiufs, from the unattested þeofs, is a Germanic coinage and never had cognates in other languages. Considering the meaning of the word thief, it could come into existence as slang. Perhaps that is how thieves once called themselves, but with time the word gained respectability and became part of the Standard. To cite a parallel: In the middle of the eighteenth century, Samuel Johnson, the author of a famous English dictionary, called the noun slum low. Slums are still slums, but the word is no longer “low”: it is neutral.
  3. The word may have been borrowed from another language (not necessarily as part of international thieves’ cant).


Despite the absence of unquestionable cognates (nouns or verbs that refer to stealing) students of Germanic tried to find some phonetically acceptable words that could have been related to thief. The most successful find was Greek typhlós “blind,” with the idea that þeof- meant either “hidden” or “imaginary.” Perhaps a more appropriate gloss would have been “a thing unseen; secret.” The Gothic adverb with the root of þiufs meant “clandestinely” (compare Engl. steal and stealthily). But several circumstances make this etymology suspect. Most important, the oldest Germanic sense of thief was not “someone who steals things under cover of darkness” (in Dickens’s days they said under cover of the darkness), but rather “criminal, violator” and “robber.” (The distinction between “thief” and “robber,” attested in Greek and Latin, doesn’t seem to have existed in the oldest Germanic society, while burglars in our sense of the word were unknown.) As far as we can judge, the ancient Germanic þeof- was not the proverbial thief in the night. The overtones of secrecy inherent in our thief and steal do not predate the introduction of Christianity.

Putative cognates, such as mean “cower,” “strike,” and “violence” (all of them have been offered), match the earliest sense of “thief” tolerably well, but one wonders why they occur in Lithuanian, Greek, and Avestan (an Iranian language). Not that distant and isolated connections among words are impossible. It just so happens that we cannot reconstruct the path from Lithuanian “press together” and “attack,” Greek “strike,” or Iranian “violence” to Germanic “thief.” If those words meant “robber” or if Germanic had words obviously akin to them, there would have been no problem. But even with the written history of a word for “thief” at our disposal, we often wonder at the zigzags in its development. Russian vor “thief” (with congeners elsewhere in Slavic) is probably related to the verb vrat’ “to lie, tell falsehoods,” and the noun’s oldest recorded senses were “cheat, swindler; adulterer.” French voleur “thief” is a metaphor borrowed from falconry. In other cases, the origin of the word for “thief” is as obscure as it is in Germanic. For example, the Romans connected latro “thief” with Greek látron “payment, compensation” (other words aligned to it meant “service; servant, slave”) and Latin latus “side” (compare Engl. lateral). The second derivation was definitely, and the first, quite possibly, a tribute to folk etymology. In Latin, latro “thief” was opposed to fur “robber,” a borrowing from Greek, as though the Romans could not coin their own noun for someone who sat in an ambush and waylaid them. It seems that honest people (and etymologists’ honesty has never been called into question) find it hard to follow thieves’ ways.

I am inclined to think that þeof- was a native coinage, possibly a slang word, perhaps even a taboo alternation of some other well-known noun (unless it was a borrowing from another language whose speakers were famous for their dishonest ways). That this supposition is not entirely groundless can be seen from the history of Old Icelandic þjófr. It should have been þjúfr, and no one knows who and when violated the “right” form.

Anatoly Liberman is the author of Word Origins…And How We Know Them as well as An Analytic Dictionary of English Etymology: An Introduction. His column on word origins, The Oxford Etymologist, appears here, each Wednesday. Send your etymology question to him care of blog@oup.com; he’ll do his best to avoid responding with “origin unknown.”

Subscribe to Anatoly Liberman’s weekly etymology posts via email or RSS.
Subscribe to the OUPblog via email or RSS.

Image credit: “Stop thief! stop thief! a highwayman!” from Randolph Caldecott’s picture books, series. Source: NYPL Digital Gallery.

Recent Comments

  1. Henno Brandsma

    Westerlauwer Frisian used to have “tsjeaf”, which is no longer in use, and has been replaced (during the 20th century) by the Dutch loan “dief”. Frisian, like Scandinavian, changed th to t, as in trije (three), tsjok (thick), tonger (thunder) and many more. Tsjeaf comes from Old Frisian thiâf, quite regularly.

  2. Anniina Jokinen

    “Th” turning into “t” in Scandinavian languages was not across the board, or was it? Did it not sometimes into “d”? For example, the Swedish “du” (English “thou”) and “din” (English “thine”). I’m not a linguist or an etymologist; perhaps it is a case of something different. I just wondered if the th->t in Scandinavian languages was a hard and fast rule.

  3. Roland Schuhmann

    The meaning of the lith. verb čiùpti ‘take, seize, touch’ in fact doesn’t seem that distant …

  4. John Larsson

    Gothic “thief” in the fourth century.
    2. (“Thief” was slang.) Like many other social misbehaviors, many of them dealt with in the Ten Commandments, stealing must have been observed in the earliest cultures and within all ethnic groups, so how plausible is it actually, that a slang word used among thiefs themselves, could be accepted as the standard word in the language? Isn´t it more normal that the development goes the other way round, that “standard words” turn in to more or less unmentionable slang words?
    3. (“Thief” imported in Gothic from unknown language.) Yes, but from where? At this time the Gothic culture was not exactly declining, so where should one look for an exporting language with such a dominating power?

  5. Roland Schuhmann

    Of course Old Icelandic þjúfr does also exist – as second compound member in Personal Names.
    The Old Icelandic output of Proto-Nors *ew as iú before f, g, k, p and ió in the other positions is in fact not that strict as is often stated in the grammars; there is also Middle Icelandic stiópmódir, -son, Middle Norwegian riófa and also Old Norwegian liús. Also in Old Swedish there are results that are not in accordance with the ‘normal’ sound laws – the situation can most likely be explained through different developments in the Old Norse dialects.

  6. [...] Thief again. One comment on thief referred to an apparently admissible Lithuanian cognate. It seems that if we were dealing with an Indo-European word of respectable antiquity, more than a single Baltic verb for “cower” or “seize” would have survived in this group. I also mentioned the possibility of borrowing, and another correspondent wondered from whom the Goths could learn such a word. Since thief has been attested in all the Old Germanic languages, it belongs to the Common Germanic stock and must have been coined or borrowed before the fourth century, when Wulfila translated the Bible into Gothic. Here it is important to take into account the historical situation. Germanic speakers living several millennia before Wulfila were nomads. Presumably, they would not have been above stealing cattle and horses (with the latter process requiring good, trusted friends) or robbing people. Yet myths reflect this situation sparingly. In Greece (stepping for a moment outside Germania), Hermes became famous because, while a child prodigy of one day old, he stole fifty head of cattle from his half-brother Apollo (their father was Zeus). In Scandinavia, Odin stole the mead of poetry from a giant, and, according to an obscure allusion, Loki stole a precious necklace. Stealing usually presupposed wresting a treasure of cosmic importance from a mighty adversary. We don’t know the age of those tales; the northern myths are hardly very old. Nor have the laws of the nomadic Teutons come down to us. Tacitus’s admiration for the unspoiled barbarians should be taken with a grain of salt, the more so as we have no idea who his informants were. The Old English, Old Frisian, and other similar laws that deal with thieves were recorded centuries after Tacitus. House breaking could not be a common crime among nomads, and keys (very primitive keys) were mainly used for locking doors against stray oxen and such. [...]

Leave a Comment

Your email address will not be published. Required fields are marked *