I'm not 100% sure of the proper terminology here, so I'm just using the one used in the LCK.

What I mean by a sentential argument is an argument that is a sentence rather than a noun phrase.

Case in point: 'That people still read that book annoys me'. Here, the 'subject' is an entire sentence complete with its own objects: "(that) people still read that book".

But how do languages that mark nouns for case deal with this? You obviously can't apply a case ending to an entire sentence. Yes, I know some languages, like German, don't mark case on the noun itself, but what about languages that do?

  • 1
    Those that I know well (which are IE) do nothing special. (Maybe in some others it is different.) – Adam Bittlingmayer Mar 2 '18 at 22:25
  • In theory there can be ambiguous sentences in languages with free word order including archaic English, but that's language. In practice something like "That ... means that ...." seems to force the SVO interpretation, so says my instinct. – Adam Bittlingmayer Mar 2 '18 at 22:28
  • The translation to mean is a usually good test because it can take sentential arguments for both subject and object. – Adam Bittlingmayer Mar 2 '18 at 22:30
  • 2
    In Japanese a sentential argument generally ends with a nominal (eg 'koto' - "thing, matter"; or 'no' - effectively a pronoun) which can then take case particles such as 'ga' (subject'), 'o' (object), 'wa' (topic). – Colin Fine Mar 2 '18 at 23:37
  • Some agglutinative languages (e.g., Aymara and Quechua) with cases do attach case endings to sentential arguments. But Indo-European languages generally only use complementisers to indicate sentential arguments (e.g., “that” in English, “да” in Macedonian) which is enough to mark the argument syntactically. – Atamiri Mar 3 '18 at 1:24

I don't know if Korean case markers qualify as genuine case markers, but you can absolutely add case markers to a whole clause in Korean. E.g.,

범인이 그 기차를 탔음이 확실하다. It is certain that the criminal took the train.

[beomin-i [geu gicha]-reul ta-ss-eum]-i hwaksil-ha-da.

[criminal-SUBJ [that train]-OBJ take-PAST-(noun making suffix)]-SUBJ certainty-(verb making suffix)-(sentence ending).

Here the verb of the subordinate clause is turned into a noun form, somewhat like "the criminal having taken the train", and then the subject marker attaches to it.

Or you can add a dummy noun 것 (serving the same role as "that" as in "that people still read that book") and attach the case marker there: think of it as analogous to "The fact that people still read that book annoys me."


If one adoopts the Pesetsky's conjecture about the nature of Case, according to which Case is an uninterpretable Tense feature and CP (sentential argument) is the same category as DP - C/DP, complementizers like "that" carry the Case feature in themselves, so it will be enough to only make use of the complementizer to mark the Case. All languages I know stick with this strategy.

Another issue is how to tackle sentential arguments when they come in prepositional objects. Gerund phrases take care of it in English (e.g. "She doesn't believe in getting lost in the wood"). Nevertheless, other languages don't like non-finite complementation to that extend. In Russian, for example, there are so-called pronominal-correlative sentences: the subordinate clause is finite, in its head is a complementizer, the predicate in the higher clause requires a preposition, the preposition assignes a particular case to its object, the complementizer of the lower clause cannot take a case value, so an expletive (demonstrative pronoun in this case) is to be made use of in order to receive a case:

"Ona ne verit v to, čto zabludilaś v lesu"

[She not believe-Pres-3pSg in that (dem.pron.)-AccSg, that (complementizer) get lost-Past-Fem in (the) woods-Preposition.CaseSg].

If there is no expletive, one can simply get rid of the preposition:

"Ona ne verit, čto zabludilaś v lesu".

I am not sure whether this construction is grammatical in English ("She doesn't believe that she has got lost in the woods").

When the complementizer is not "that" (čto), the situation remains the same:

"Ona rasskazala o tom, kak provela leto"

[She tell-Past-Fem about that (dem.pron)-Preposition.CaseSg, how (complementizer) spend-Past-Fem (the) summer-AccSg]

"She told about how she spent the summer".

Once again, the preposition can be ommitted.

Returning to your example 'That people still read that book annoys me' - the Russian translation will be with the expletive too, I mean, not simply "that people read", but "That (dem.pron)-Nom, that people read that book annoys me". Nonetheless, when the sentential argument is not in the subject position, the expletive is not necessary.

The conclusion is that in Russian the case value on sentential arguments is absent, except for when they appear as the arguments of PPs or in the subject position: then the case values are assigned to the expletive. I guess, this expletive pronoun can be substituted by "the fact", so in Russian "fact" dropped and only "the" (which I indicate by "that") is present. With such an analysis the English constructions are essentially parallel to the Russian ones.


I assume you mean overt, traditional case, not abstract case which every language is said to have, in certain theories. It depends on the rules for case marking in the language. In Imbabura Quechua, some subordinate clauses can be case marked, for example can wagra michij-ta ricurcani (you cow herding-acc I-saw) "I saw you herding cattle"; jari atalpata shuwasca-ta-mi yuyani (man chicken-acc stole-acc-witness I-believe) "I believe that the man stole the chicken". The entire embedded S is nominalized and marked accusative or ablative (for "after" clauses). So you can mark case on a sentence. This is because case marking in IQ is an NP-level enclitic. Angas likewise "case marks" embedded clauses, again with an enclitic (the case distinction is rather degenerate: it only marks "is an argument of the verb"). In North Saami and Kalenjin, only nominals are case-marked, so object complement clauses don't have case.


An explanation, by and large coinciding with the one giving by Aharon M. Vermont:

  1. A verb can take another direct object without actually increasing valency (or changing its meaning) thus forcing the listener to identify (or: equate) both direct objects (O1 - the usual direct object, and O2 - the abstract, attached, "metaphorical" (the "id est"-object so to speak)), e. g. (weak O2, since not nominal but an predicative adjective) He is painting the house (O1). > He is painting the house (O1) green (O2).; or (strong O2, since fully nominal) They crowned him. > They crowned him king. (not quite grammatical, I admit)

  2. The sentential argument (I like the phrase) is a predicative adjective in that respect. It can be marked as object by making its subject an object, e. g. AcI (accusativus cum infinitivo) in Latin, Greek, German, English, Spanish, French, all IE languages as far as I know of, but I suppose the feature is universal. E.g. Vidi te currere. ("I saw you[ACC] run[INF].") In the same way as the O2 is reduced in marking compared to the O1 in "ditransitive" constructions, the sentential argument is reduced in marking to indicate its subordination by using an infinitive (or less finite) verb form, e. g. "They crowned the boy king." sounds more grammatical than "They crowned the boy the king." (The latter being ambiguous as to which argument is O2.)

  3. Subordination may not be marked within the sentential argument by reduced marking in parts of the sent. arg. (e.g. infinitive of the AcI) but outside the sent. arg. by equating the unreduced sent. arg. with a complementizer (strictly speaking "the fact" is not a complementizer, since it cannot be used without "that" to link the sentential argument to the main sentence in the way that "that" can). The complementizer is an abstract word (unlike "the fact") acting as a null noun to link the O2. Correspondingly Spanish has lo for "that", and lo que for "the fact that"/"that which"/"the following, to wit"/etc. German has dass (obviously cognate to neuter article das), dialectally der/die/das was and more formally das Folgende, nämlich... (for "the fact that"/"the following, to wit"). As a non-IE example Ga has no which translates to "that", dass, lo in neuter (or inanimate) null object, demonstrative pronoun and complementizer senses.

The O2 (or "id est"-object) is identified with O1 without using a copula, but nevertheless there is identification involved. The difference between (2) and (3) is that the sentential argument is O2 in (2) and O1 in (3)*, and the difference can be used to focus rather on the content or on the action bringing about the content. E.g. I would rather use (2) "I saw you running." to emphasise that I was witness (focus on action bringing about content) and (3) "I saw that you ran."** or even more emphatic "I saw the following: You ran." (some sort of build up, probably with an hiatus before sent. arg.).

*Accordingly an abstract (or "null") noun is preferred in cases of (3) since it is inconspicuous, lacking irritating semantics by its own. It compensates the need for reduced marking of either one of the objects in the ditransitive construction. Either the complementizer or the sent. arg. has to carry reduced marking to convey the framing involved in the construction.

**Cf. arguably "I saw how you ran."


I can tell you about my language (Croatian, or BCMS): such arguments are simply subjects and they are not marked in any way. There's no dummy pronoun. But the verb still agrees with the subject: it's in the third person, singular, and the phrase has neuter agreement (the same holds if the subject is an infinitive of a verb, and in impersonal constructions).

For example:

Živcira (annoy-pres.3.sg) me (I.acc) [ da (conj.) ljudi (people.pl) još (still) čitaju (read.pres.3.pl) tu (that.fem.acc) knjigu (book.acc) ]

In the past tense it would be:

Živciralo (annoy.past.neut.sg) me je (aux.3.sg) [ da ...]

The auxiliary for the past tense is in the 3rd person: the clause is the subject. The clause doesn't change (there's no tense shift).

The interesting question is: what if a clause is an object? Again, nothing happens. The clause is unmarked.

But a clause can be an indirect object, or an object of a preposition. Then, a pronoun 'to' (actually, an adjective which functions as a pronoun as well) gets inserted before the clause and it inflects:

Razmišljam (think.pres.1.sg) o (about) tome (that.dat/loc.sg) [ da ...]

lit. 'I'm thinking about that people still read that book.'

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy