Do languages with cases frequently have verbs which use particular cases for their objects?

Question

Turkish, German, Russian, Greek and Latin are examples of languages with declension. They also have lists of verbs for which the verb's single object takes a particular case apart from the "normal" object case (accusative, say). E.g., when I mention that I believe something in Turkish, I mention something in the dative case. To glance at those five languages, having verbs which can "case" their single objects different ways might appear normal for languages which use case.

Some examples:

German:

Ich sehe dich [2ps acc]
"I see you"

but

Ich helfe dir [2ps dat]
"I help you"

Turkish:

Seni [2ps acc] görüyorum
"I see you"

but

Senden [2ps abl] şüpheleniyorum
"I suspect you"

One needs to remember these case-object relations per verb. Russian seems to have quite a few verbs which "take" different cases. I wonder if all languages which use case are this way.

Searching about this tonight, I think it may be a type of "subcategorization" or "c-selection": verbs c-select particular types of complements/objects. I read half a dozen pages seemingly related pages which I think I did understand, and two or three I know I did not understand. I might be wrong in applying those two terms. I don't want to risk unwittingly asking a subtly different question by using them. I will say "casey verbs" and mean "verbs which normally don't follow the language's most common 'object case' for their complements."

Questions:

1. Those easy-to-search languages with cases above do have casey verbs. Is this representative of languages which use case?

2. If a language has several cases in addition to its nominative/accusative/ergative situation, e.g., it has dative, are there likely to be casey verbs for each of those several cases?

3. Specific examples of #2: are dative verbs common where there is a dative? ablative verbs, common? genitive verbs, common?

4. I am not sure how to phrase this question. Some English verbs seem "casey" to me as I have been thinking about this today. English doesn't have a "from" case, but some verbs are especially ready for an "incoming object": 'to defend,' 'to hear.' Casey verbs might have a "fromness," for example, built-in in this way. So my question is can they have all the "fromness" (say) built into the definition so that a verb 'to hearfrom' takes a "normal" direct object now because the declension to the "from" case would be redundant like "to hearfrom from Fred"?

Answer 1

I understand the question as follows: Do all (or a notable majority of) languages "with cases"

• make use of at least one construction for bivalent verbs where the non-subject argument is a noun phrase in a case other than the accusative, in addition to making use of the construction where the non-subject argument is a noun phrase in the accusative case (or in whatever case is normally assigned to the O argument of a transitive verb)

• have some bivalent verbs that are usually used with an accusative argument, and other bivalent verbs that are usually used with a non-accusative argument, such that for any language X with cases Nom (or Erg), A=Acc (or Abs), B, C..., it makes sense to make lists of "bivalent verbs in language X that take a noun phrase in case A as the non-subject argument" vs. "bivalent verbs in language X that take a noun phrase in case B, C... (etc.) as the non-subject argument"

(The first but not the second bullet point could be true in a language where the case of verb arguments was determined solely by syntactic rather than lexical factors; e.g. a language where any bivalent verb with a noun phrase as its complement takes a genitive argument when negated and an accusative argument otherwise.)

The question is not asking about the existence of verbs that can be used in both constructions, or how the distribution of different constructions among different verbs varies between languages.

I'm using the phrasing "non-subject argument" because, as mentioned in the comments, it can be debated whether it is correct to refer to the non-accusative noun phrases in these constructions as "(direct) objects"; that is also why I am using the phrase "bivalent verbs", since it seems to be debated whether the verb in such constructions can be called "transitive". Bivalent and divalent are synonyms meaning "possessing a valence/valency of two." That is, a bivalent/divalent verb is a verb that has two arguments. (In the cases that you are interested in, both arguments are noun phrases, but bivalent verbs can also take phrases of other categories as arguments: e.g. in the English sentence "I believe in him," "believe" is a bivalent verb with the following two arguments: the noun phrase/NP subject "I" and the prepositional phrase/PP complement "in him".)

In your German example sentences "Ich sehe dich" and "Ich helfe dir", "sehe" and "helfe" are bivalent verbs that take as their non-subject arguments noun phrases of two different cases (accusative and dative respectively). (The verb sehen can also be used intransitively, but that is not relevant to your question.)

I don't know; it might depend on the case inventory of the language and on how we define "case" and "language with cases"

I think it may depend on what you mean by "case". I don't think it is a very clear term, and it is sometimes used to refer to a morphological category but sometimes to a syntactic category.

There seem to be a lot of different things that can potentially be encoded by nominal inflection, and not all inflected forms can necessarily be used as the complement to a verb.

One example, even though it isn't a typical or central example of a "case", is the English -'(s) "genitive" construction. As far as I know, no finite verb in English takes an argument marked with the -'(s) genitive: this marking is used only in other grammatical contexts (mainly on noun phrases used as determiners; also on noun modifiers in the "classifying genitive" construction, on certain arguments of non-finite verb forms in certain contexts, and arguably on the complement of the preposition of in the "a friend of mine" construction).

The English -'(s) construction is difficult to analyze morphologically, which is part of the reason why there is argument about whether it is a case. But if it is considered a case, then you might be able to say that English is a language where a noun phrase used as the complement of a bivalent verb can only be in one case (the one traditionally referred to as the "accusative case", although some modern linguists seem to be reluctant to apply that traditional categorization to the modern English case system*). (For the purposes of this argument, we need to ignore the use of forms like "I" (i.e. forms traditionally called "nominative") in contexts like "It is I" or "They already told my parents and I"; some linguists argue that these uses of "I" do not actually belong to the core syntax of English, but instead are used and found acceptable by speakers of standard English for some other reason.)

I don't know of a clearer counterexample against your examples of Turkish, German, Russian, Greek and Latin, but I am not familiar with the variety of case systems used across the world. Like you, I have the impression that languages with cases frequently have bivalent verbs that select arguments of a particular case other than the main case used for objects.

---

*For example, in some theories, "nominative" is redefined or reinterpreted as the absence of case; based on this, Omer Preminger says in "Case in 2017: some thoughts" that "insofar as English has anything you’d want to call ‘nominative’ [...] it’s [...] the thing we’ve been calling ‘accusative’ or ‘objective’ case" and describes what is traditionally called the "nominative" in English as instead being a case "assigned by T⁰ under closest-c-command" (page 29).

Answer 2

Yes, they do - in general, it’s always language-specific. There may be morphosyntactic alternations (e.g. two cases alternate depending on the governing verb’s polarity) and lexico-semantic alternations. The latter are called “non-canonical objects” (at least in Turkish). It also depends on the theory you adopt, there are direct objects, indirect objects, thematic objects... However, in most languages non-canonical objects can’t be passivised so it really depends on how broad your definition of ‘object’ is. For example, your example “ich sehe dich“ can be directly passivised (“du wirst gesehen“) whereas in the case of “ich helfe dir” the undergoer would remain in the dative: “dir wird geholfen”. I don’t speak Turkish but in Russian it’s no different.

Answer 3

Atamiri's answer provided the key term: non-canonical marking. This question is easier to investigate via that term. Googling that, I found this ebook (of a 2001 book): Non-Canonical Marking of Subjects and Objects (Typological Studies in Language) . That book helped me answer my questions substantially. Thanks Atamiri and everybody who joined in guiding the question. @ewawe, if you mentioned non-canonical subject cases because they interest you, then I can tell you that this book discusses them more than it discusses non-canonical objects.

One overarching note: the book's introduction breaks verbs into semantic types (such as perception verbs) and subtypes. It discusses how verbs of these types are those that receive non-canonical subject/agent/object treatment.

1. Many languages have a "canonical" way of marking a verb's main noun phrase argument, such as accusative and absolutive cases. However, well-known languages with canonical object marking of this sort (German, Turkish, Russian, Latin, Greek, Finnish: search for example "german verbs that take dative") have groups of verbs the main/single objects of which "take" other cases. ewawe's answer provides one sophisticated definition of this idea, while keelan has made some specific examples which Nico helpfully requested easy to find in the question. When particular verbs "take" their main arguments in other cases, this may be regarded as one type of "non-canonical" marking. The ebook linked above discusses it in case-marking languages found around the world.

Therefore, non-canonical marking is probably pretty common among in case-marking languages the world over; the well-known examples are probably not misleading.

We do not know from this book that all case-marking languages do use non-canonical marking, though. Introducing the non-canonical marking concept on page 3, it states "some languages" do this.

2. So if a language has non-canonical object marking using cases as well as several cases like dative, ablative, etc., might it employ each of its additional cases for different verbs' non-canonical verb marking? I was satisfied by finding, from three different language families, one example that it can happen and two examples that non-canonical marking can be more restricted. The Icelandic chapter analyzes each of Icelandic's four cases as taking an object position sometimes. On the other hand, Finnish has verbs which do "take" single arguments in several other Finnish cases (satisfying my own purpose), but the chapter authors find that most of them do so in ways that do not meet the book's definition of 'non-canonical objects.' A paper on non-canonical objects in Turkish specifies that "the set of verbs which simply take non-canonical objects... can have either ablative or dative objects": a subset of Turkish's cases.

3a. Are dative non-canonical objects common in languages which both distinguish a dative and which do non-canonical objects? I don't know if it's common in all languages which distinguish a dative. The well-known languages above use dative objects non-canonically. So do Icelandic and Japanese (both chapters in the book), as well as Hungarian. In the book, Imbabura Quechua and Amele have a dative cases, and appear not to use them the way the others do.

3b. Ablative--same question. The book's Finnish chapter has verbs which take ablative objects; it mentions Mordvin has a very few such. Turkish, Russian (links in the question) and Latin have verbs like this. Japanese has an ablative particle and does not seem to use it in this way.

3c. Genitive--same question. The book considers on page 42 that Genitive (with acc. and dat.) is among the most common non-canonical object cases. "Especially Slavic" languages do this (p57). Icelandic does this, too. Googling "latin verbs "take the genitive"" returns an affirmative; so, too, to some fossils from German. From the above sources, Bengali, Imbabura Quechua, Japanese, Finnish, and Turkish are examples of languages with genitive cases not "taken" by particular sets of verbs.

3d. The Russian link above lists verbs which "take" the instrumental, as well. Finnish does partitive, elative and allative, that I recall from scanning that chapter.

4. If someone were interested in this question, I could recommend looking at the book's classes of verbs which typically get involved with non-canonical objects, then finding how same work in the languages which I have mentioned don't do the non-canonical objects. My curiosity is satiated for the moment.

Further reading:

Here's a tidbit in a lengthy quote from John Roberts, in his chapter on the Amele language, "Whaley (1997) following Gerdts (1992) suggests that languages can be classified as to whether they are direct-object-centred or indirect-object-centred languages. Direct object-centred languages tend to have very little case marking and have verb agreement only with subjects or only with subjects and direct objects. They also tend to permit recipients to appear as direct objects. English is cited as a good example of a direct-object-centred language. In contrast indirect-object-centred languages tend to have robust case-marking systems and to allow verb agreement with indirect objects. Moreover, they often permit patients, which are typically treated as direct objects in direct-object-centred languages, to be encoded as indirect objects. Choctaw is cited as an example of such a language."

Answer 4

I don't know if I get your question right. If so, a distinction between intra-linguistic phenomena (within one specific language) vs cross-linguistic phenomena is in order. In the former, the same verb might license different cases depending on the meaning. For example, in German "glauben" ('believe') requires the accusative when referring to facts ('I believe this story'), the dative when the referent is a person ('I believe you') and the preposition "an" + accusative in the sense of 'I believe in your talent'. From a cross-linguistic perspective, instead, there are differences inasmuch as some languages encode verbs denoting beneficiaries with the dative, as in German with "helfen" ('help'), whereas others don't and rather encode them as patients via accusative (see the same verb in Italian and French, where the distinction between the two cases shows up pronominally).