In German and Spanish (I think), you use the word for 'from'. In Japanese though, I think they use 'ni' (which can either mean 'to' or 'at'). In English we use the preposition 'by', which is rarely used for anything other than passives. Though it can have a locative meaning, which is what I think was its original meaning.

Are there other ways languages do this? To me, improvising with another preposition can potentially cause ambiguity issues. At least in English, if two 'by's would appear in a sentence, we can replace one with 'next to'. I.e:

*"It was put by the fridge by Bob" "It was put next to the fridge by Bob"

In German though, I don't know what you would do. As far as I know, they don't have another preposition they can use to mean 'von/from'. And of course 'von' is far more commonly used than 'by' in English. And Japanese 'ni' is even more commonly used than that! Really the only two things we express exclusively with 'by' is the agent of a passive construction, and the author of a book.

What are some other ways that languages use? I'd prefer answers from languages from different families. Languages that are related to each other are of course more likely to use the same tactic than languages that aren't (its probably why most European languages use the word for 'from').

  • Data point: Classical Greek used ῠ̔πό /hypo/ 'under, through, by' with the genitive to mark the agent. Commented Dec 17, 2017 at 20:17
  • I wouldn't asterisk It was put by the fridge by Bob -- it sounds OK to me.
    – TKR
    Commented Oct 20, 2018 at 3:26

2 Answers 2


In French, par ('by') and de ('from/of') can be used depending on the verb. In Late Archaic Chinese, the more common passive construction that includes the agent was jian + main verb + yu + agent. The morpheme yu normally marks the locative, so it's somewhat similar to the Japanese example you mentioned.

In both of these, and the examples you mention, the agent is marked as oblique. However, there are constructions where this isn't the case.

In Early Middle Chinese, a new kind of passive was innovated that uses a strategy you haven't mentioned:

為  王    所  殺
Wei wang suo sha
COP king SUO kill
'was killed by the king' (Records of the Grand Historian)

This strategy is interesting because it's not 'prepositional' like the ones you mentioned. Wang suo sha alone is means 'he whom the king killed'. You can consider this structure a patient nominalisation or a headless object relative clause; either way, the agent is expressed as part of this nominal phrase. So 'literally', the VP translates to 'was a person whom the king killed'.

In modern Chinese, this is the construction we use. It is attested since Late Archaic Chinese, but without the agent; the agent only appeared around Middle Chinese:

我  被   他  看  見   了
Wo  bei ta  kan jian le
1sg BEI 3sg look see SFP
'I was seen by him'

Although people these days often gloss bei as 'by', this isn't completely accurate because bei didn't grammaticalise from a preposition; its original meaning was undergo. For example, bei sha would mean 'undergo killing' (i.e. get killed). The agent was just put between bei and the main verb sometime in Middle Chinese (why this happened, I'm not sure). I'm not sure what you'd call this strategy; you could say it's unmarked, since people just inserted the agent there.

PS You might want to look up how it works in Austronesian languages - that would be really exciting! It's a pretty complicated phenomenon, so considering that I'm not well-read in it, I won't post anything about it - perhaps others could!

EDIT: I just looked up Keenan and Dryer's chapter in Language typology and syntactic description. I found these:

In Haya, the agent is just unmarked, like the modern Chinese example I mentioned:

Ebitooke bí-ka-cumb-w’       ómukâzi
bananas  they-past-cook-pass woman
'The bananas were cooked by the woman'

In some languages, the agent is incorporated into the verb. Keenan and Dryer cite English examples like 'this project is state-controlled', and mention that this process is more productive in Quechua:

Kuru miku-sqa-mi       manzana-ø  ka-rqa-n
bug  eat-ptcpl-comment apple-subj be-past-3
'The apple was bug eaten'
  • 2
    Most Austronesian languages don't really have passives, or rather, have several different kinds of what we could call "passive" constructions. Passive is a technical term in Indo-European languages and (like the term Subject, which is also localized) should be used with caution, if at all, outside that context.
    – jlawler
    Commented Dec 18, 2017 at 3:34
  • Yeah, only nominative languages really have a 'passive'. I did take a look at Maori (one of the polynesian languages) though, because they do have nominative alignment. But I don't know enough of the language to say how they do it. They seem to use the particle that is normally used to mark one of their tenses as a preposition. (Maori is verb-initial, like all the polynesian languages).
    – user19661
    Commented Dec 18, 2017 at 4:29
  • May I know what definition of passive (as a comparative concept) you're working with though? I think the constructions I answered with were fairly close to prototypical passives, but if you specify nominative languages, then I'm not sure if my modern Mandarin example helps, since Mandarin has lost all traces of case. Commented Dec 18, 2017 at 5:36
  • Mandarin is nominative. So is English. He was just talking about your suggestion concerning the Astronesian languages. It would be hard for me to explain it to you. Look up 'morpho-syntactic alignment' if you want to do research into it. Don't get me wrong. I do certainly find your explanation for how passives word in Mandarin interesting.
    – user19661
    Commented Dec 18, 2017 at 10:39
  • I understand jlawler's comment; my reply was mostly to you, because I don't know what definition of the passive you presuppose. Morphologically, Mandarin and other modern Sinitic langauges are neither nominative nor ergative, as there is no case-marking anywhere in the grammar. You can argue that Mandarin is nominative in the syntax, but a) that presupposes that subjects and objects even exist in Mandarin, which is controversial and b) morphologically ergative languages can also pattern with nominative-accusative languages syntactically. Commented Dec 18, 2017 at 11:49

In German you can either use von (actor) or durch (agent) for the actor, so your premise is wrong. Still, German speakers mostly rely on word order to determine the actor.

Die Kiste wurde von einer Windböe von der Ladefläche geworfen. (preferred)

Die Kiste wurde durch eine Windböe von der Ladefläche geworfen.

The crate has been thrown from the truck bed by a wind gust.

But of course there are still cases with ambiguity:

Die Kiste wurde durch ein Missgeschick durch/von eine/einer Windböe von der Ladefläche geworfen.

The crate has been thrown from the truck bed by a wind gust by accident.

Your Answer

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