Compound agent nouns in English

« previous post | next post »

The thread continues, from which I extract the following:

To explain briefly! An agent noun is a noun derived from another word (usually a verb), meaning "someone who [verb]s." e.g.:

"to travel" => "traveller"
"to rule" => "ruler"
"to direct" => "director"

In Middle and Modern English, agent nouns derived from verbs are almost always constructed using the agentive suffix -er (from German), less commonly from -or (from French).

(Agent nouns derived from nouns usually take -eer or -ist, both from French, but I digress.)

The equivalent suffix in Old English is -a:

ridan (to ride) => ridda (rider)
giefan (to give) => giefa, gifa (giver)

It's superficially similar to -er (i.e. modern speakers tend to pronounce both as a shwa ⟨ə⟩), although they're from different roots.

With me so far?

So *compound* agent nouns are agent nouns that narrow the sense by specifying an object for the verb. In Middle & Modern English they're usually formed by putting the agent noun after the object, e.g.

"to fight fire" => "firefighter"
"to say 'nay'" => "naysayer"

And the same is absolutely true in Old English (OE *loves* compound words), as in these two examples from Beowulf:

"giefan bēag" (to give a ring) => "bēah-gifa" (ring-giver, i.e. a wealthy lord)
"webbian friþ" (to weave peace) => "friþwefer" (peaceweaver, i.e. a hostage bride)

But while these words were formed substantially the same way (aside from the shift from the OE -a to the ME -er) for more than a thousand years, there's this odd period where dozens of compound agent nouns flipped the order and lost the suffix.

These "exocentric" verb-noun compound agent nouns *start* with the verb, without the suffix, and end with the object. Some examples:

"to pick pockets" => "pickpocket"
"to spend thrift (i.e. savings)" => "spendthrift"
"to swash (i.e. strike) a buckler" => "swashbuckler"

And almost all these words were coined between 1550 and 1700 — it's a very rare construction before and after that period.

A lot of these words have *remained* in the language, mostly to baffle and enrage modern speakers, but almost none have been *coined* in three centuries.

But what especially grabs me is how *seedy* these words generally are. Consider the above examples, along with turncoat (traitor), lickspittle (toady), skinflint (miser), turnkey (gaoler), scofflaw (criminal), lackwit (fool), cutthroat (murderer) or sellsword (mercenary).

It seems like, over maybe four or five generations, a whole-ass grammatical word construction appeared, proliferated and died out… and it was used almost wholly for insults and street slang.

And precisely because the sort of people who coin insults and street slang tended not to be the sort of people who write books, we don't really know where that came from or why it was seen in such negative terms.

And I think about far more than I have any business doing. /fin

I sense a Wortgeist.


Selected readings


[h.t. Scott De Brestian]


  1. Mark Liberman said,

    September 15, 2022 @ 6:36 am

    See Stan Carey, "Cutthroat compounds in English morphology", 4/28/2015.

    See also section 2.5, "Compound Nouns whose Heads are not Nouns", in Liberman & Sproat The Stress and Structure of Modified Noun Phrases in English, 1992, which notes that

    Erratically, a verb may combine with a noun to form a noun. One [V NJ pattern produces an exocentric compound whose referent is the (unexpressed) subject. This type is common in Romance languages but not in Germanic ones. It is no longer very productive in English, although it seems to have been popular in earlier times7:
    (35) cut-throat, pick-pocket, pinch-penny, do-nothing, kill-joy, knownothing, dreadnought, sawbones.
    7 Marchand (1969, pp. 37-39) makes the interesting observation that such constructions are almost invariably pejorative, […]

    The reference is to Marchand, Hans. 1969. The Categories and Types of Present-Day English Word Formation. München: C. H. Beck'sche Verlagsbuchhandlung.

  2. Stan Carey said,

    September 15, 2022 @ 6:43 am

    Thanks, Mark. Here's the post, which is based mostly on work by Brianne Hughes:

  3. Victor Mair said,

    September 15, 2022 @ 6:49 am

    Wow! The power of LL!

    You two — including the original author — replied early in the morning from who knows wherever you are within about an hour of when I made the post.

  4. ardj said,

    September 15, 2022 @ 7:31 am

    Um…. Shakespeare ?

  5. John Hutton said,

    September 15, 2022 @ 9:27 am

    "In Middle and Modern English, agent nouns derived from verbs are almost always constructed using the agentive suffix -er (from German)"

    No, English -er is from OE -ere which is probably from Latin -arius (reflexes of which are found across other Germanic languages). E.g. leornere "learner", bōcere "scribe" ( ridda (rider)
    giefan (to give) => giefa, gifa (giver)"

    This kind of formation occurred alongside OE -ere and were, I suspect, morphologically unproductive. These are masculine n-stems: sg. N -a, AGD -an; pl. NA -an, G -en-a, D -um. Only the gen pl has a distinct inflection (the rest were lost), and the dative is anomalous.

    Quite a number of masc n-stems were clearly agent nouns in origin (e.g. bana "slayer", lida "sailor"), but their ranks also include nama "name". Mōna "moon" had probably long since lost the sense of "the measurer". The compound agent noun suffix -bora also belongs here (e.g. mundbora "protector").

  6. Robert Coren said,

    September 15, 2022 @ 9:56 am

    My impression is that the "exocentric" type is a (the most?) common way to form such nouns in French (e.g., tire-bouchon = "pull-cork" for English corkscrew; passe-partout = "pass-everywhere" for "latchkey"). Was there a period during which borrowing formations from French was fashionable?

  7. Martin said,

    September 15, 2022 @ 11:06 am

    One non-pejorative (if not quite fully laudatory) exocentric noun though is "do-gooder".

  8. Coby L said,

    September 15, 2022 @ 11:14 am

    I can think of quite a few non-pejoratives of this form: breakfast, breakwater, dreadnought… And "breakfast" goes back to the 15th century.

  9. Bloix said,

    September 15, 2022 @ 11:16 am

    My very favorite example is the wonderful "fuckwind," meaning the common kestrel, for its habit of beating into the wind, thereby maintaining a stationary position in the air. Apparently windfucker was more common – Hopkins' "windhover" perhaps originated as a euphemism.

  10. Jonathan Smith said,

    September 15, 2022 @ 11:18 am

    yes social media / blogosphere citation practices do leave something to be desired cough hack… while we're at it,
    Brianne Hughes Encyclopedia Briannica and her 2012 paper

    Hughes seems to call these "turncoat compounds" which is better than "cutthroat…" as cutthroat the noun is shall we say dated…

    Much writing on similar topics in Mandarin… my memory of Ceccagno & Basciano (2008) "Compound headedness in Chinese: an analysis of neologisms" is mostly that wen2xiong1 文胸 'bra' involves a verb wen2 'conceal' (+ xiong1 'the breast')…. surely not? :D

  11. Victor Mair said,

    September 15, 2022 @ 11:30 am

    The bird is not trying to do THAT to the wind.


    From Middle English *fukken, probably of Germanic origin: either from Old English *fuccian or Old Norse *fukka, both from Proto-Germanic *fukkōną, from Proto-Indo-European *pewǵ- (“to strike, punch, stab”). Compare windfucker and its debated etymology.

    Possibly attested in a 772 AD charter that mentions a place called Fuccerham, which may mean "ham (“home”) of the fucker" or "hamm (“pasture”) of the fucker"; a John le Fucker in a record from 1278 may just be a variant of Fulcher, like Fucher, Foker, etc. The earliest unambiguous use of the word in a clearly sexual context, in any stage of English, appears to be in court documents from Cheshire, England, which mention a man called "Roger Fuckebythenavele" (possibly tongue-in-cheek, or directly suggestive of a depraved sexual act) on December 8, 1310. It was first listed in a dictionary in 1598. Scots fuk/fuck is attested slightly earlier, probably reinforcing the Northern Germanic/Scandinavian origin theory. From 1500 onward, the word has been in continual use, superseding jape and sard and largely displacing swive.

    A range of folk-etymological backronyms, such as "fornication under consent of the king" and "for unlawful carnal knowledge", are all demonstrably false.



  12. ohwilleke said,

    September 15, 2022 @ 11:46 am

    "do-gooder" is usually pejorative, and used in a sense similar to the more modern slang of "a Karen"

  13. Bloix said,

    September 15, 2022 @ 11:58 am

    The bird is beating against the wind, which as your etymology shows is what fuck originally meant. The evolution of a word from meaning "strike" to "have sex with" is not unusual. Compare the contemporary "I'd hit that." (I know that some folks think that etymology is ALWAYS misleading, but in this case I would say probably not.)

    But by Hopkin's day the strike, beat meaning was long gone and the sexual intercourse vulgarism was what was left. The OED's first citation for windhover (spelled windover) is 1674, while its first cite for windfucker is 1599. (Although it tells us to see 'Halliwell' for fuckwind, it has no citations for it.)
    So apparently windhover replaced windfucker/fuckwind a long time ago.
    I'm not sure what is contested about the etymology of windfucker – it pretty clearly originated as meaning windstriker, windbeater – is there anything disputed about that?

  14. Victor Mair said,

    September 15, 2022 @ 12:34 pm



    If the term is a compound of wind +‎ fucker, it may preserve an old sense of fuck (“to beat, to strike”) which is also found in cognates (for example, Bohuslän Swedish fokka (“to fuck; to thrust, to push”)) but was otherwise lost from English, and it can be compared to the regional synonym fuckwind. (Wright's English Dialect Dictionary compares fuck in the latter word to fjúka (“be driven (by the wind); fly”) instead, while Liberman says the Norse word "has no [other?] cognates anywhere in Germanic".) However, the synonym windsucker is almost as old, and was rendered in older texts as windſucker using a long s, so some scholars think windfucker is a misreading of windſucker; others think windſucker is a bowdlerization of windfucker. Compare the later term windhover and the Orkney term windcuffer.

    Modern attestations of the second, vulgar sense may be unrelated to the bird.



  15. klu9 said,

    September 15, 2022 @ 12:45 pm

    Finally I know what to call these kinds of nouns! Moore's thread & Carey's article both fascinating.
    But what about "headscratcher"?
    A headscratcher does *not* scratch your head (so it's not the agent, right?), rather it causes *you* to scratch your head yourself (so you are the agent of the verb "scratch"). Is this still technically a (O.V.-er) compound agent noun? Or something else? And are there other examples that *cause* rather than perform the verb?

  16. CuConnacht said,

    September 15, 2022 @ 1:59 pm

    "Scofflaw", perhaps the most recent on the list, is a rare example of a word coined for new word competition that found an actual place in the language, at least in journalese. 1920, originally for someone who scoffed at Prohibition.

    But it is not regularly formed. A scofflaw does not scoff law, as a killjoy kills joy or a cutpurse cuts purses; they scoff at the law.

  17. CuConnacht said,

    September 15, 2022 @ 2:00 pm

    ^"coined for a new word competition"

  18. James Wimberley said,

    September 15, 2022 @ 2:56 pm

    You would think new words formed by reversing the order in current compounds would be attractive to experimental writers, as they are instantly decodable: smithword, lickass, breakcode, peevegrammar.

  19. John Swindle said,

    September 15, 2022 @ 10:16 pm

    There’s “grab-ass,” also spelled “grab ass” or “grabass,” presumably 20th-century, as in “Stop playing grab-ass and start picking up cigarette butts!” but it’s noncount and refers to rowdy horseplay (“grab-assing”) rather than to one who engages in it.

  20. Biscia said,

    September 16, 2022 @ 2:15 am

    But what about all the famous Gropecunt Lanes? Don't they precede the period in question by a couple of centuries, at least?

  21. Rosemary Kuwahata said,

    September 16, 2022 @ 2:30 am

    How about a hiphopper?

  22. Stan Carey said,

    September 16, 2022 @ 5:20 am

    @Biscia: Yes. See Brianne Hughes's thesis on the topic, "From Turncoats To Backstabbers: How Headedness and Word Order Determine the Productivity of Agentive and Instrumental Compounding in English" (available here, along with other resources). A bar graph on page 23 shows the historical productivity of these "cutthroat compounds": the range is much broader than the Twitter thread suggests, and will also have changed a bit in the 10+ years since.

    Also this, from Volker Gast (2008), "Verb-noun compounds in English and German", Zeitschrift für Angelistik und Amerikanistik, 58(3): 269-282:

    There was (probably) a certain inventory of relevant items even before the Norman conquest, esp. in proper names and epithets. Under French influence, the pattern was ‘upgraded’, i.e. it became more productive and frequent and was used in more (esp. higher) registers. The productivity of exocentric V-N compounds increased steadily in the 14th and 15th centuries and reached a peak in the 16th century (e.g. kill-courtesy, lack-brain, lack-beard in Shakespeare). From the 17th century onwards, its productivity decreased considerably . . .

  23. GH said,

    September 17, 2022 @ 12:57 am

    I assume daredevil is one of these (originally someone who dares the devil, rather than a daring devil); if so, that's one that is not markedly pejorative.

    Would breakneck count, even though it's used as an adjective?

    Suckup is a little different (it's not someone who "sucks ups," but who sucks up to someone), but the pattern seems similar to me, and I assume the word is somewhat recent.

  24. Madhuri Kherde said,

    September 17, 2022 @ 5:31 am

    In English Grammar, the agent is always a noun of a sentence. We also can call the agent the actor. It is the “doer” of an action, becoming the subject of a sentence.
    Agents generally have the endings “-er” or “-or.” When added to a root word, these suffixes mean someone who does something.

  25. john burke said,

    September 17, 2022 @ 12:10 pm

    A seven-year-old boy, a native speaker of English, is a character in Aldous Hulxey's novel "Point Counter Point". He is described as tending to reverse the conventional order of these compounds: "mow-lawner" is the example given.

  26. john burke said,

    September 17, 2022 @ 12:10 pm


  27. Bloix said,

    September 17, 2022 @ 1:17 pm

    John Swindle mentions grabass but says it's not really part of this pattern because grabass refers to an activity – rowdy horseplay – not a person. But a google search does reveal a few examples that do mean harasser – "He's a grabass," "They're all grabasses." AFAICT it seems to have a specific meaning among the women who use it – a man who is annoying but not truly threatening.
    A similar word is kissass, which has the same meaning as ass-kisser.

  28. Josh R. said,

    September 18, 2022 @ 7:33 pm

    Wait, is "sellsword" attested before Martin's Game of Thrones? That's where I first encountered it, and while the meaning was transparent, I assumed it was a word Martin coined to give his fantasy world some unique flavor, much like "to bend the knee" meaning "to submit."

  29. Andreas Johansson said,

    September 19, 2022 @ 12:45 am

    @Josh R.

    Wiktionary's earliest quotation for "sellsword" is from a 1969 fantasy book.

  30. Lasius said,

    September 19, 2022 @ 4:04 am

    In German there are similar words, that also often have negative connotations, maybe due to influence from French.

    Taugenichts – good-for-nothing

    Habenichts – have-not

    Möchtegern – wannabe

    Haudegen – daredevil

  31. Peter Taylor said,

    September 19, 2022 @ 4:49 am

    I tried to check OED for antedating of sellsword, but it doesn't have an entry.

    On the other hand, bend the knee for submit is in Shakespeare: in Richard II, Aumerle responds to his mother's request to pardon Rutland with the line

    Unto my mother's prayers I bend my knee.

  32. Josh R. said,

    September 20, 2022 @ 7:32 pm

    Peter Taylor,

    Personally, I would make a distinction between Shakespeare's metaphorical use here and Martin's in-world idiom "bend the knee" (with the definite article). I think there are many examples of the former (with various possessive adjectives) before Martin, but I suspect very few of the latter until the success of the Game of Thrones series.

RSS feed for comments on this post