Hesperus is Bosphorus

A group blog by philosophers in and from Turkey

Knowledge is not a Propositional Attitude (at least, not in Turkish)

with 24 comments

I’m writing a paper at the moment arguing that knowledge does not entail belief.  Part of my argument is that knowing is not a propositional attitude, whereas believing is. I think there is a clear ontological distinction between facts and propositions and that what can be known are facts (and perhaps also states of affairs, and  Objects) whereas the objects of belief are propositions. The essential difference between facts and propositions is that facts are not truth apt, whereas propositions are. Amongst philosophers today the claim that knowing is not a propositional attitude is extremely idiosyncratic, however  historically something similar to the position I defend was probably the view of the majority of philosophers. In a later post I’ll give some evidence to back up this historical claim. In this post I want to point out that what I believe to be one of the strongest motivations for the claim that knowing is a propositional attitude is based on a contingent feature of English (and other Indo-European languages).

I think that one reason so many philosophers take is as almost axiomatic that knowledge is a propositional attitude is because of a contingent feature of the grammar of Indo-European languages. In these languages so-called propositional attitudes are expressed by the use of that-clauses or analogous grammatical devices. In English, for example, we distinguish between  propositional attitude verbs, which are normally expressed by a verb followed by a that-clause and an objectual attitude verbs expressed by a verb followed by a noun phrase. In English we can say ‘I see the man biting the dog’ but we say ‘I know that the man is biting the dog’. This suggests that what I see and what I know are distinct types of object, with the object of knowledge having something like the structure of a sentence. But this is a merely contingent feature of our language and we should not use this fact about English as evidence for any particular philosophical position. And I believe that this contingent linguistic fact is often illegitimately used either as evidence for, or at least strongly motivates, the position that knowledge is propositional rather than objectual.

For example, Mathew McGrath explains that, “The term ‘proposition’ has a broad use in contemporary philosophy. It is used to refer to some or all of the following: the primary bearers of truth-value, the objects of belief and other “propositional attitudes” (i.e., what is believed, doubted, etc.), the referents of that-clauses, and the meanings of sentences. And he adds a bit later., “one might doubt whether that-clauses could really refer, if reference is understood on the model of proper names. For, that-clauses are not proper names, nor are they noun phrases” [see here]. Here McGrath is clearly citing a fact about English grammar as prima facie evidence for a particular position on the nature of propositional attitudes. Similarly, Mckay and Nelson claim: “Propositional attitude reports concern the cognitive relations people bear to propositions. We typically make such reports by uttering propositional attitude reporting sentences like ‘Jill believes that Jack broke his crown’, employing a propositional attitude verb like ‘believes, ‘hopes’, and ‘knows’, followed by a clause that includes a full sentence expressing a proposition (a that-clause).” [see here] Appeals to English grammatical structures are ubiquitous in discussions of propositional attitudes.

This grammatical feature of English is not, however, standardly found in Turkish. In Turkish the (grammatical) objects of what in English are thought of as propositional attitude verbs and verbs of perception are noun-phrase not a that-clauses. In Turkish one takes what is the main verb of the equivalent English that-clause and turns it into a noun (something like an English gerund), and this verbal noun can be tensed. The rest of the information in the corresponding English that-clause then modifies this verbal noun. In Turkish ‘propositional attitude’ verbs such as ‘think’ [düşünmek], ‘know’ [bilmek] or ‘believe’ [inanmak] are transitive requiring a noun phrase in the accusative, and they work grammatically in the same way as perceptual verbs such as ‘see’ [görmek].

Thus, for example, in Turkish the translation of “I know that the cat is sitting on the mat” is “Kedinin paspasın üzerinde otur-duğ-u(n)-u biliyorum.” In this Turkish sentence, “otur” is the stem of the verb oturmak, ‘to sit’; the ‘duğ’ acts a a nominaliser forming a noun from the verb stem; the u(n) is the possessive; and the final ‘u’ is the accusative case ending making the word the gramatical object of the verb. The Turkish sentence could be re-translated back into English as “I know the cat’s it’s-sitting on the mat. The details of Turkish grammar are not particularly important here, what is important is that Turkish grammer does not, standardly, use that-clauses to express what in English are called propositional attitude verbs. In Turkish the verbs ‘to know’ and ‘to believe’ are followed by noun-phrases, not that-cluases. İf we took Turkish not English grammar as our  guide to the ontology then seeing, knowing and believing should be thought of as objectual not propositoanal.

In English saying “I saw Ayse kissing Murat yeasrterday’ sounds fine, but ‘I know Ayse’s kissing Murat yesterday” is extremely odd, but I think it better expresses the ontology of knowing than “I know that Ayse kissed Murat yesterday”. İt is clearly a very awkward and ungrammatical Englsih sentences, but as Turkish grammar shows, this akwardness is due to a contingent fact about English grammar, and not a fact about knowlege. The fact that in English the verb know is standardly followed by a that-clause should not be used as evidence for the claim that knowledge is a propositional attitude – unless, of course, you are happy doing the Philosophy of English.

UPDATE: I guess the title of this post is too strong. I should have said that knowing is not obviously a propositional attitude in Turkish. I do not want to use Turkish as providing positive evidence of my position, but merely to use it to block English speaking philosophers using contingent facts about English grammar as evidence against my position.


Written by Lucas Thorpe

February 18, 2012 at 4:56 pm

24 Responses

Subscribe to comments with RSS.

  1. 1. “In Turkish the (grammatical) objects of what in English are thought of as propositional attitude verbs and verbs of perception are noun-phrase not a that-clauses.”
    Technically aren’t that-clauses a sort of noun-phrase? If so, the contrast you have in mind could be between that-clauses and, say, objectival noun-phrases, or something like that.

    2. The grammatical details of Turkish superficially support your point; however, there’s a problem.
    E1 The cat’s sitting on the computer pissed me off.
    T1 Kedinin bilgisayarin uzerinde oturmasi beni kizdirdi.
    T1 translates E1. This is evidence that ‘kedinin bilgisayarin uzerinde oturmasi’ refers to the non-propositional, objectival entity, the cat’s sitting on the computer. If in the ontology presupposed by Turkish, knowledge is a relation between a person and such an entity, then we should expect the following to make sense (at least after some adjustment).
    T2 Kedinin bilgisayarin uzerinde oturmasini biliyorum.
    But T2 isn’t proper Turkish at all.

    To complicate matters, ‘kedinin bilgisayarin uzerinde oturmasi’ can sometimes be the best way to translate the that clause ‘that the cat is sitting on the computer’:

    E3 I can’t stand that the cat is sitting on the computer.
    T3 Kedinin bilgisayar uzerinde oturmasina dayanamiyorum.

    E4 It was decided that the Chair talk to the Dean about the shortage of paperclips.
    T4 Baskanin Dekan’la atas eksikligi hakkinda konusmasina karar verildi.

    What seems to be going on, I think, is that ‘oturmasi’ does both the job of English that-clauses and the job of English objectival noun-phrases. If so, the surface grammar of ‘oturdugu’ doesn’t tell us which job it’s doing when it’s next to the word for ‘know’ in Turkish.


    February 19, 2012 at 4:37 pm

    • Hi Irem,

      I don’t want to use Turkish grammar as positive evidence for my own ontology of knowing. Rather I merely want to undermine, what I take to be a common (although perhaps implicit or unconscious) motivation for the claim that knowing is a propositional attitude. English speakers have clear linguistic intuitions about when a verb should take a that-clause and when it should take a noun phrase, and, I suspect, that many English language philosophers (and those speaking other Indo-European languages which have a similar structure) take their linguistic intuitions as evidence for an important ontological distinction: seeing is (or at least can be) objectual, whereas knowing is not.

      My point is precisely that ‘oturmasi’ does both the job of English that-clauses and the job of English objectival noun-phrases so we shouldn’t appeal to the fact that English uses a that-clause after ‘know’ as evidence for the claim that knowing is a propositional attitude rather than objectual. the way English or Turkish grammar functions here should not be our guide to the ontology.

      The ontology I defend departs from Turkish grammar – as I suggest that believing is a propositional attitude, whereas knowing is objectual. The point of appealing to Turkish grammar is to stop what I take to be a very common type of argument against the position I’m defending. Namely the argument: “You’re obviously wrong. We say ‘I see the cat on the Mat’ but ‘I know that the cat is on the mat’ Clearly this grammatical difference reflects an ontological difference.” I’ve heard many people give arguments that are basically of this form.

      How would you say: “I know that the cat’s sitting on the computer pissed me off.”

      Lucas Thorpe

      February 19, 2012 at 4:59 pm

      • I see.

        “Kedinin bilgisayarin uzerinde oturmasinin beni kizdirdigini biliyorum.”


        February 19, 2012 at 5:57 pm

      • The sort of argument I am objecting to are like the following one offered by Tim Williamson in Knowledge and It’s Limits to support his claim that all evidence is propositional:

        “the kind of thing which hypotheses explain is propositional. Inference to the best explanation concerns
        why-explanations, which can be put in the form ‘… because …”, which is ungrammatical unless declarative sentences, complements for ‘that’ fill both blanks.” (p.195)

        Unless this feature of English grammar is a universal fact about all languages I’m not sure why this fact about English grammar is philosophically relevant, unless all Tim wishes to do is to provide an analysis of how certain words work in English.

        Let’s take the sentence: “The dog bit the man because the cat is sitting on the mat”

        As I understand this could be translated into Turkish in two ways:

        (1) “Kedi paspasın üzerinde oturduğu için köpek adamı ısırdı.”


        (2) “Köpek adamı ısırdı çünkü kedi paspasın üzerinde oturuyordu”

        Do you think these are both adequate translations of the English sentence?

        In the second translation the explanans is a sentence (as in English) in the first translation I think that the explanans is a noun phrase. [If any fluent Turkish speakers want to comment on this, please do!] So I guess one could translate the first sentence back into vaguely grammatical English as something like:

        (1′) “for the reason of the cat’s sitting on the mat, the dog bit the man”

        and the second sentence as:

        (2′) “The dog bit the man because the cat was sitting on the mat”

        If these are both adequate translations of the original sentence, then this suggests that grammatically one can go either way here in Turkish. In Turkish, in an explanation, the explanans can either be put in the form of a sentence or as a noun-phrase. And we can see how we can sort of do this in English by using a quite odd grammatical construction. Doing this for more complex sentences in English is probably impossible, but it can and is done in Turkish. The fact that 2′ is ‘odd’ in English however, seems irrelevant philosophically to me.

        Anyway – my point is that we shouldn’t draw any ontological conclusions about evidence just because in an explanation in English the explanans is normally a sentence. Nothing I am saying here suggests that Tim’s conclusion is false. I’m just objecting to arguments like this that appeal to contingent features of English grammar.

        Lucas Thorpe

        February 19, 2012 at 11:32 pm

  2. Irem: So I guess the best translation of this would be something like: “I know my being pissed of with my sitting on the computer cat”.

    Lucas Thorpe

    February 19, 2012 at 6:20 pm

    • Hi Lucas, no this last definitely doesn’t work; such reverse translations are of course impossible because of word order even if incrementally we can do them to get a sense of how a language works—the closest rendering of the last would be, “I know my being pissed of with the cat’s on-computer-sitting.” or better yet, “I know my being pissed off with the cat-on-computer-sittingness” As you can imagine, just as you have to bring the “I know” at the end of the Turkish sentence to the beginning of the English sentence no matter what, you’d have to reverse word order “bigisayarin uzerinde” into “on computer” no matter what.
      I think Irem’s point is important–the distinction between “oturmasini” and “oturdugunu”. I think it’s conceptually important that “oturmasi” comes form the infinitive and “oturdugu” is more *like* a gerund, but in fact propositional in tense and other conditions, such as: “I know its sitting/ its having-sat, its going-to-be-sitting somewhere/ somehow/ with someone.” Contrast A and B below:

      A1. Konusmayi biliyorum (I know how to talk); –I know “to talk” [irrelevant: konusma can also be a regular noun, “a talk”, “the talk”, but that’s not an option for all verbs]
      2. By contrast: usage-wise “Yuzme biliyorum” (I know how to swim); –I know “to swim”. no -(y)i used here
      3. Konusmak istiyorum/ isterim (I want/ would like to talk). –I want “to talk”
      4. Konusmayi cok istiyorum/ isterim. (I really/ very much want “to talk”) -(y)i is added in some cases where the verbal noun is thus qualified.

      B1. Konustugumu biliyorum (I know that I’m talking; I know that I talked;conjugation is not definite here)
      B2. Ne hakkinda konustugumu biliyorum (I know what I talked about)
      B3. Nasil konustugumu biliyorum (I know how I talked)
      B4. Kiminle konustugumu biliyorum. (I know with whom I talked)
      B4. Konusmakta oldugumu biliyorum (I know that I am talking)–emphasizes present tense
      B6. Konusmus oldugumu biliyorum (I know that I talked–emphasizes past tense
      So to me these possiblities are far more propositional than “object/ noun”.

      Let\s go back to your example. The cat’s “to sit” (oturmasi) is a fixed thing or frame that “pisses me off.” The cat’s “oturdugu” is a proposition that “I know”, a proposition that is itself modifiable as to time/ tense, & conditions/when-what-with whom-how


      February 22, 2012 at 2:23 am

  3. One more thing. You can actually modify the “–mak/ –masi” infinitive/ verbal noun form too as to tense and conditions. “Kedinin dun butun gun bilgisiyarin uzerinde oturmus olmasi beni kizdirdi” (I was pissed off that the cat sat on the computer yesterday all day/ The cat’s having sat on the computer yesterday all day pissed me off”. You can have a future version too, “oturacak olmasi”… But they all convey a *fixed* state, which (did, will) have consequences.
    Perhaps so far we have considered mutually exclusive uses–but equal simultaneous validity can shed some light:
    1. “Onu gordugune cok sevindim”
    “Onu gormene cok sevindim” [possibly colloquial]
    Both mean: (I’m so glad that you saw him/her/it)
    I would actually be more inclined to use the first sentence for a phenomenon, an event. Such as, “I’m very glad that you saw your friend.” It somehow answers the question, “Why am I so glad?” I would be more inclined to use the second for an object, a moment, a fixed frame–such as “I’m so glad that you saw the mistake/ the coincidence/the lost object” It somehow answers the question, “What made me so glad?”,
    2. “Geldigin iyi oldu”
    “Gelmen iyi oldu”
    (It was good that you came) Almost interchangeable to me… and yet the first implies the unexpected, a happenstance; the second implies it was somewhat expected, perhaps even necessary.
    3. “Geldiginize cok sevindim”
    “Gelmenize cok sevindim”
    (I am very glad that you came) Almost interchangeable to me; the first sounds more proper, and the above distinction might somewhat apply.

    Essentially, that’s where the distinction seems to be:
    “Kedinin oturmasi beni kizdirdi”–“THe cat’s sitting [that the cat sat] pissed me off” –cannot use “oturdugu” here. A given fixed situation causes a certain outcome.
    “Kedinin oturdugunu gordum/ duydum/ hissettim/ biliyorum” (I saw/ heard/ felt/ know that the cat is sitting) Cannot use “oturmasi” with verbs of this meaning. It again seems to me that *the senses (in terms of perception, that is, indeterminate), as well as knowledge are propositional, as of a dynamic, unfolding event in the “oturdugunu” form, and not a given object as of the infinitive/fixed verbal noun form, “oturmasini”.


    February 22, 2012 at 4:03 am

    • Hi Burcu,

      Thanks for the replies. Very helpful.

      Here are a few comments (and I’ll try and make them in a way that can be understood by non Turkish speakers. For English speakers I guess one big take-home here is that it Turkish there are far more ways of transforming a verb into a noun than in English – and this makes it a lot easier to construct verbal noun-phrases.

      (1) I think your translations are better than mine. The point of the reverse translation was to make the point (and a wierd-sounding but sort of understandable English sentence) that the gramatical structure in Turkish is something like: “I know my being pissed of with [blah blah blah]”, and I wasn’t too concerned with the grammatical structure of [blah blah blah].

      (2) I think that the hard thing to get in English is that Turkish verbal nouns can be tensed. In English we can capture some Turkish constructions using the Gerund. In the sort of constructions I am interested in, it seems to me that in everyday speech the past tense is used to express both the past and the future, but there is a separate future marker. So, for example, here are three sentences and their translations. I’ll capitalise the tense marker in the verbal noun:

      (a) I know that the dog bit the man.
      Köpeğin adamı ısırDIǦını biliyorum.

      (b) I know that the dog is biting the man.
      Köpeğin adamı ısırDIǦını biliyorum.

      (c) I know that the dog will bite the man.
      Köpeğin adamı ısırACAǦını biliyorum.

      So – as far as i understand how Turkish works (and My Turkish is not great – so please correct me if I’m wrong here), the DIǦ suffix is ambiguous between past and present. But it is common to find such ambiguity. In English we can use the present tense to talk about the past, for example. The disambiguation here is normally provided by context – but there are other simple ways to disambiguate – by, for example adding an adverb.
      The ACAǦ suffix is a future marked. [The suffixes I am talking about here are actually in fact DIK and ACAK, but the K is softened to a Ǧ when followed by a vowel. But I’m trying to make this a comprehensible as possible for those who don’t speak Turkish.]

      It is extremely difficult to translate (c) back into English.

      Perhaps one could do it adverbially and say: “I know the dog’s biting the man in the future” or perhaps more literally: “I know the dog’s will-be-biting the man”. I guess that one can do something similar with the past: “I know the dog’s having-bitten the man”. The important point I want to make is that the ODDNESS of these sentences in English is just a contingent fact about English grammar and we should not draw any philosophical consequences from this oddness.

      The reason I think this is important philosophically is because there are many that-clauses in English that are almost impossible to translate into English noun phrases – e.g. that-clauses in the past or future tense.
      And I think this plays an important role in the Intuitions of English speaking philosophers. But the fact that we cannot do it in English does not mean it cannot be done.

      My friend Ilhan is working on constructing a language (influenced by the way turkish works) that is just like English – except that it contains no sentences. The logical connectives are fun to play with in this project. I think that showing that such a language can be constructed has important philosophical implications.

      (3) I’d like to understand better the distinction between “oturmasini” and “oturdugunu”. Perhaps a linguist could help us out here. i think it’s an interesting distinction.

      Lucas Thorpe

      February 22, 2012 at 12:03 pm

  4. I’ve been trying to clarify my position to a friend on Facebook. Here’s how I explained my position if anyone is interested:

    I think that knowing is objectual, whereas believing is a mental attitude directed towards a proposition. I don’t think that one attitude entails the other – although I think that for humans knowing normally causes believing. My position does not rule out the possibility that we are so constituted that knowing necessarily causes believing – although: (1) such a necessary relation would not be entailment and (2) I think this is empirically false.

    One thought is that believing is a more complicated mental state/attitude than knowing. I think that although knowing is not propositional, this does not imply that it is non-conceptual. I have a sort of adverbial understanding of the role of concepts in, say, perceiving (which I take to be a way of knowing). So concepts can be thought of in terms of our way of grasping objects. So I think I can make a distinction between, say, seeing a house and seeing something AS a house. The second requires that one possess the concept ‘house’. BUT although seeing something AS a house involves concepts this does not mean that the object the activity of seeing is a proposition.

    I think that believing does not just require that we possess concepts, but that we also have the capacity to represent our concepts. So the thought is that believing involves some sort of meta-cognition. SO I think that there might be some animals that are capable of knowing, but not capable of believing. And I think that it might be the case that young human infants start to know before they are capable of forming beliefs.

    I think that it is (at least metaphysically) possible that there are creatures capable of knowing which are not capable of forming beliefs. Such creatures would could know a’s being b without forming any attitude to the proposition ‘a is b’ – perhaps because they were unable to have any attitudes towards propositions.

    Lucas Thorpe

    February 22, 2012 at 12:11 pm

  5. hI again Lucas–
    A couple of notes:
    I do find your general philosophical argument persuasive. I am more intrigued with the wish to find linguistic evidence for that, however, since it’s not only bare grammatical constructions but their connotations too that go into the equation–though of course there must be many languages that offer so many other perspectives we’re not even able to take into consideration. One might have to take note of other inflections too. Even in English, for example, there are differences in objectual/ propositional *meaning* when we say:
    1–I saw that the woman kissed a man.
    2–I saw the woman kissing a man.
    3–I saw a woman (who was) kissing a man.
    Not only does the indefinite/ definite change the emphasis on what part of the sentence might count as “objectaul”,”propositional” (or, in context, valuable information) but the choice of construction itself can imply things. There are circumstances in which one would feel more compelled to say (1) versus (2). The second does imply the sighting of a state, situation, action, but the first, the “that” clause, carries greater strength as a self-conscious claim. (2) sounds like a more casual statement to me than (1), and more objectual than propositional.
    Let’s consider the same thing with “knowing”, though of course the words work differently
    1. I know you. I know this city.
    2. I know that you are engaged in philosophy. I know that traffic is a huge problem in this city.
    Bracketing how one can “evaluate” the assertions themselves (what does it mean to know something, to know someone etc), this is one way in which we can gauge the objectual versus propositional nature of these sentences. The purely objectual would resist attributes but require particularity. (If I say “I know the spotted cat”, my assertion is not about the cat’s being spotted; I’m saying that only to specify which particular cat,and my assertion remains: “I know that cat”). SO I suppose what I mean with (1) is more like, “I know who you are. I know what this city is like.” Object’s attributes cannot be included. Based on this one could argue that a sentence about knowing *any* attribute would go from the purely objectual (1) toward the propositional (2).

    I think this distinction about connotation also holds about the apparent “past tense” sense of “Oturdugunu” (versus “oturmasini”). Let me explain this some other way. As I said earlier, if my emphasis were on the tense, I could easiily distinguish between past, present, and future, with the same “-dugunu” construct in various conjugations:
    “Oturmus oldugunu biliyorum” (past)
    “Oturmakta oldugunu biliyorum” (present)
    “Oturacagini biliyorum/ Oturmus olacagini biliyorum) (future). Of course, in the first two, “oldugunu” is a helping verb and continues the “dugunu” construct, which seems, at first glance, to be past in construct, and past/present ambiguous in tense. But its *meaning* is solely about the accomplishment of the act. That is, the “dugunu” construct is not about time, “its sat-ness yesterday” “its sitting-ness now”, but about an accomplished act bridging the past and the present: “its having-satness.” I think this last English rendering quite captures the “meaning” of the construct “dugunu”–which corresponds not to “did” but to “having done”.


    February 22, 2012 at 3:59 pm

  6. “This grammatical feature of English is not, however, standardly found in Turkish. In Turkish the (grammatical) objects of what in English are thought of as propositional attitude verbs and verbs of perception are noun-phrase not a that-clauses.”

    In Turkish, there are other grammatical devices for belief and knowledge reports that exactly mimic the ‘that-clauses’ in English.
    For example,

    (A) I know[/believe] that the cat is sitting on the mat.
    (A’) Kedinin masanın üstünde oturduğunU[/oturduğunA] biliyorum[/inanıyorum].
    (A”) Biliyorum[/inanıyorum] ki kedi paspasın üstünde oturuyor.

    Here, in (A”), the word ‘ki’ exactly plays the role of the word ‘that’ plays in (A).
    Although the sentence (A’) employs the most common way of reporting a belief or a knowledge claim, (A”) is neither ungrammatical nor an archaic expression. It is quiet standart.


    February 23, 2012 at 8:10 pm

  7. Hi Lucas,

    Many thanks for such an exciting entry.

    1) The ontological (or generic) distinction between facts/states of affairs and propositions is far from clear (at least according to the opponents of truthmaker ontologies). Facts or states of affairs do not seem to be truth-apt, but they surely could be viewed as existence- or actuality-apt entities. And some philosophers, following Bradley, (early) Russell and (early) Wittgenstein, see something stronger than a similarity between a proposition’s coming out true and a state of affairs’ coming into existence, namely, IDENTITY (so, being true, they argue, is being identical with an existent or actual state of affairs, that is, a fact). Thus, the difference between the object of a belief and the object of a ‘knowing-state’ is analysed away as a pure modal difference. I guess this might have some bearing on the wider question you’re dealing with in your paper.

    2) Although it’s obvious that Turkish inclines towards nominalisation in constructing the (grammatical) objects of ‘propositional-attitude verbs’, it also uses another tool to serve the same purpose, as you well know, which is quite similar to the Indo-European ‘that’ (or ‘dass’, or ‘que’, or ‘quod’…): the ‘ki’-clause (earlier form: ‘kim’). So instead of ‘Kedinin beyaz olduğuna inanıyorum’ (‘Cat’s-it’s-being-white believe I’), a native speaker of Turkish could quite naturally go with ‘İnanıyorum ki kedi beyaz(dır)’ (Believe I that cat white is). ‘Ki’ is probably a loan from Persian – but then Persian itself is an Indo-European language! Nevertheless, since ‘ki’-sentences (which leave the embedded sentence as it is) sound quite natural to the native speaker and are interchangeable with the corresponding nominalisations (in most of the contexts), it is highly probable that ‘-dığı’/’-ması’ constructions are still considered ‘nominalised’ propositions (and yet still ‘propositions’) by the native speaker, and not ‘nomina’ proper. Moreover, the merely syntactical difference between a noun-phrase and a ‘that’-clause (or ‘ki’-clause) by no means guarantees that there is a corresponding generic difference in the realm of reference. Turkish is almost infinitely capable in nominalisation (see the suffix ‘-lik’), but I haven’t had the chance yet to meet a fellow native speaker who is committed to ‘objects’ of the form ‘zamanından önce gelmiş olmamızdan kaynaklanıyor olamazlık’. 🙂



    Arman Besler

    February 24, 2012 at 2:04 am

  8. Hi!

    “(3) I’d like to understand better the distinction between “oturmasini” and “oturdugunu”.”

    In Turkish, there are three non-finite verbal forms: 1) verbal nouns, or ‘isim-fiiller’, 2) participles, or ‘sıfat-fiiller’, and 3) adverbials, or ‘bağ-fiiller’. Approaching the problem from this milieu, “oturmasını” is a verbal noun, which is constructed by means of ‘-me’ suffix: oku-ma. Verbal nouns constructed in this way can be used as a component of noun compounds in both primary and secondary positions. For example, kedilerin otur-ma-sı, or otur-ma-n-ın faydalar-ı. Here, both -sı and -ı in secondary positions are the same suffix which is necessary to form a noun compound. -ın in the second compound is genitive ending.


    S1: Kedinin bilgisayarın üzerinde otumasına dayanamıyorum.


    S2: Kedinin miyavlaması beni rahatsız etmiyor.

    Here, I think, the noun compound, “kedi-nin … oturma-sı” took the dative ending, i.e. -(y)a.

    On the other hand, “otur-duğ-u-n-u” is a participle which took the accusative ending -u. -dık x4 is the suffix by means of which these kind of participles are formed. I think that -u is 3rd person sg. possessive ending. When the participles take possessive endings, they can be declined (according to cases). It is also important to note that this participle is “substantialized”.


    (a) I know that the dog bit the man.
    Köpeğin adamı (dün) ısırDIǦını biliyorum.

    (b) I know that the dog is biting the man.
    Köpeğin adamı (şimdi) ısırDIǦını biliyorum.

    This “ambiguity” resembles the one that is present in the ancient greek infinitives. However, in ancient greek the disambiguation is accomplished by means of the distinction between time and aspect. That is to say, present indicative and aorist indicative infinitives in ancient greek signifies only aspect, not the time. The time is “contextual”. May be the problem can be approached from this perspective. We can say that (a) is in punctual aspect, but (b) in progressive. Ι’m not sure that it works, though.


    P.S: It must be clear by now, but let me tell anyway that I’m not a linguist. 🙂


    February 24, 2012 at 2:49 am

  9. hi Adil, Arman and Ramin,

    Thanks for these responses. They are all helpful. I’m too tired to respond adequately to some of the more complex point – but when (if?) I find a bit of free time I’ll try and respond in more detail.

    Adil and Arman: As an English person learning Turkish, of course I know the “ki” construction – this is the only construction I’m really capable of using. And it the first construction that (at least European) foreigners are taught when when we learn how to construct more complex sentences. This is why I said the ‘standard’ construction. But three quick points:

    (1) I’m not using Turkish grammar as evidence for my own position. Rather, I’m appealing to Turkish grammar in an attempt to block English speakers from using a particular fact about English grammar as evidence FOR their position. The type of argument I want to block goes something like this:

    “Although one can artificially convert that-clauses into noun phrases, such noun-phrases are ungrammatical. And the ungrammatically [is that even a word?] of such phrases (in English) is evidence for the claim that knowing is not objectual.”

    Most of papers I have read on “propositional attitudes” (in English) I have read I seem at some point to implicitly make an argument like this. So the the fact that such noun-phrases in turkish are grammatically fine in Turkish is enough for my purposes. Because the main point of my appeal to Turkish grammar is that the ungrammaticality of such noun-phrases is merely a contingent fact of English grammar.

    (2) I do think that “Kedinin masanın üstünde oturduğunu biliyorum” is the ‘standard’ translation of “I know that the cat is/was sitting on the mat” rather than “Biliyorum ki kedi paspasın üstünde oturuyor”
    For me an interesting question is: are there any claims that you can only express by using the ‘ki’ construction.

    (3) I’ve heard it said that the ‘ki’ construction is a fairly recent introduction into Turkish. this claim is base on a vague memory of a fairly long ( probably drunken) discussion with a well known Turkish linguist.
    So it would be interesting to know if the ‘ki’ construction is found in Osmanlı (Ottoman Turkish), or in, say 19th century village Turkish. If there are any Turkish linguists reading this I’d be interested in some feedback on this. (Arman: when you talk of “the earlier form: ‘kim’” how far back to you trace this? – can you give a reference for this?)

    I don’t thank that Points (2) and (3) are really important for my argument [although I’d be interested to hear if and why you disagree]. I’m just interested.

    Arman: I just vaguely gestured towards the distinction I want to make between facts\states of affairs\objects on the one hand and propositions on the other. I actually an a reductionist about propositions. I think one can reduce all talk of propositions to talk about concepts. And so knowing is a mental attitude directed toward facts (etc) whereas the object of belief is (or at least involves) a concept. But one can only fight one battle at a time. So I don’t want to get into a discussion of my weird view about propositions in the paper I’m writing. The basic idea here is that one can, in principle, convert all sentences into noun phrases – and such noun phrases better express the objects of propositional attitudes than sentences. So what one believes is [something like]: “the instantiation of [the concept] ‘the cat’s being on the mat'” what one knows is: the cats being on the mat. And i think that one can have a mental attitude (knowing) directed towards the cat’s being on the mat without having any mental attitude at all directed towards the concept “the cat;s being on the mat”. So ultimately I don;t really want to explain this in terms of a distinction between propositions and facts but in terms of a distinction concepts and objects. But for the time being I’ll stick with talking about propositions: one has to fight one war at a time.

    The object of knowledge (which in my view not a propositional attitude) is what is. The object of belief (which is, on my view IS a propositional attitude) is a purported representation of what is. I suspect that, historically, this was probably the mainstream view.

    Ramin: Thanks for that. it was helpful

    Lucas Thorpe

    February 24, 2012 at 5:30 am

  10. Lucas,

    Thanks for your reply.

    You’re absolutely right about the ‘standards’. The ‘ki’-construction can never completely replace the ‘-dığı/-ması’ construction, and I’m sure one can find many contexts where one is available while the other is (pragmatically) not. Like other loans from Persian, the ‘ki’ form leaves a more poetic taste on the audience, and serves a different pragmatic purpose (or conveys a different pragmatic meaning) in most of the cases. Since the ur-source is Persian, I cannot agree with the linguist you mentioned – it must have ‘infiltrated’ Altaic Turkish much earlier, like when Seljuk Turks ruled Persia (or maybe a little later). Ottoman Divans are full of ‘ki’-constructions (along with the phonetic variant ‘kim’, which must not be confused with the singular interrogative pronoun ‘kim’, which, in turn, in Ottoman Turkish is frequently used in conjunction with ‘her’, resulting in ‘her kim’, translating ‘he/she who’). Anyway, I find both forms in the works of Yunus Emre, a late 13th–early 14th century Anatolian mysticist. An example of the rarer one:

    Dilerim kim bana feryad iresin
    Güç olmuş kişiye sen dad iresin

    (‘Dilerim kim’ = ‘I wish that’)

    There is also the much more common ‘diyor ki’ (s/he says that), whose earlier form is ‘aydır ki’. Again from Yunus: ‘Kur’an aydır ki “Vetteku” gene aydır ki “Tezrauu” ‘. I have to admit, however, that ‘ki’ has several functions in formal and informal Turkish, and forming propositional attitude phrases might be the most uncommon among those.

    As for the other issue, I suppose I get your point. It somehow reminded me of Leibniz’s approach to propositions in ‘Generales Inquisitiones’ (where he suggests that every linguistic complexity can be reduced, ultimately, to a concept/term).



    Arman Besler

    February 25, 2012 at 12:02 am

    • Hi Arman,

      Sorry for taking so long to reply. I’ve been really busy here. but I appreciate your comments. They are all very interesting.

      I also thought that the ‘ki’ construction was Persian – and this thought led me to believe that there was not such a construction in Altaic Turkish – I think this was the main point of the the half remembered conversation with the linguist. I guess I just mis-remembered when it entered Turkish. I wonder whether there was a similar construction in Altaic Turkish? Ilhan (Inan) and I plan to have lunch with a linguist who works on these issues. Hopefully we’ll find out some more.

      The stuff on concepts was very rough and unthoughtout – but you’re right my suggestion is something along the lines of Leibniz’s account. Although I haven’t thought deeply about Leibniz for a while. [When I started my phd the plan was to write on Leibniz, Kant and Hegel – obviously over ambitious – and I worked through a lot of Leibniz then.] Do you have a pdf of his ‘Generales Inquisitiones’? I guess it would be a good idea for me to have a look at it. I think he makes a similar suggestion elsewhere too – but I’d have to have a proper look through the texts. Most of my books are in storage at the moment – our offices are being renovated. So I guess this will have to wait. In the paper I’m working on I’m going to just assume that there are propositions and that belief is a propositional attitude. Ultimately I suspect all talk of propositions can be reduced to talk of terms, and so I suspect that don’t really need propositions in my ontology of mind – but I guess I can leave this aside for now.

      Anyway – thanks for your helpful comments.

      Lucas Thorpe

      March 5, 2012 at 10:50 pm

  11. Hi Lucas,

    No need to be sorry.

    Unfortunately I don’t have an electronic copy of the text. Just let me remind you that a quite accurate English translation of it can be found in the following edition:


    Please let us all know about the progress you make on the issue. Thank you for sharing your ideas.

    All the best,


    Arman Besler

    March 7, 2012 at 10:47 pm

    • Hi Arman,
      Just got a copy from Interlibrary loan. I’m looking forward to reading it.

      Lucas Thorpe

      March 8, 2012 at 9:25 pm

  12. On the title, ‘Knowledge is not a Propositional Attitude (at least, not in Turkish)’: is this not strictly speaking nonsense (at least on ordinary readings of ‘in’)?

    Knowledge itself is a propositional attitude if it consists in an agent taking an attitude of some sort to a proposition enjoying certain properties (e.g. truth). Cases of knowing are cases of propositional attitude if they consist in agents taking an attitude to a proposition enjoying certain properties (e.g. truth). If that’s roughly right, then what could it mean to say that knowledge is a propositional attitude in some language?

    By ‘in’, do you really mean something like ‘going by the surface features of’? Or do you really mean to say something about the concept of knowledge, or something of the kind?

    Pardon me if this seems pedantic.

    Tristan Haze

    March 13, 2012 at 8:27 am

    • Hi Tristan,

      As should be apparent from my two recent posts I’m inclined to agree with your observation that the parenthetical remark in my title is strictly speaking nonsense. It was vaguely meant as a slight parody of the way in which many epistemologists argue. (and was also meant to be provocative).

      The main point of this post is to criticise those (many) philosophers who use how the the word ‘know’ functions IN English as evidence for philosophical conclusions about knowledge. I am not opposed in principle to appealing to grammatical features of all languages as defeasible evidence for a particular philosophical position (see my post here).The point I was trying to make is that the grammatical feature that many Anglophone philosophers appeal to is a contingent feature of English. And so, I think does not provide us with any evidence about knowledge, but only tells us about how the word ‘know’ functions IN English (and other Indo-European languages).

      I also think that in English we don’t have to interpret the grammar of ‘knows’ in terms of a propositional attitude. It depends if we think that the very is ‘knows’ or ‘knows that’. If we think that the verb is ‘knows’ then grammatically the object of this verb is not a sentence but a that-clause. And one might interpret that-clauses as non-phrases. if one thinks that the basic attitude is ‘knowing-that’ (and I think that philosophers who distinguish between ‘knowing-how’ and ‘knowing-that’ tend to do this] then the object of this attitude is grammatically a sentence, and so the sort of thing that appears to be truth-apt.

      If one thinks that the the primary goal of epistemology is conceptual analysis, one might think that different cultures have different concepts of knowledge (reflected in their languages). I don’t buy this. I’m an epistemic realist – so I think the task of epistemology is not to analyse our concept of knowledge, but to understand what knowledge is. (as I argue here).

      So my point is that in discussion the question of whether knowing is a propositional attitude or not we should NOT appeal to grammar (of either English or Turkish). But, many many philosophers do (either implicitly or explicitly) appeal to English grammar when arguing that knowing is a propositional attitude. The main point of this post is that they should stop doing this.

      Lucas Thorpe

      March 13, 2012 at 1:38 pm

      • Many thanks for the clarifications.

        Tristan Haze

        March 14, 2012 at 8:33 am

      • Hi Lucas,

        I certainly agree that there is much to be learned from a cross-linguistic study of knowledge ascription, but I don’t think you’ve isolated a difference between Indo-European languages and Turkish. Factive verbs can take infinitival complements (marked for aspect) in German and Spanish as well (see Ana Perez-Leroux and Petra Schulz’s 1999 paper “The role of tense and aspect in the acquisition of factivity” for some useful discussion). Although the relevant grammatical constructions are different, it seems that across the board these verbs embed clauses presupposed to be true (where nonfactive mental state verbs carry no such presupposition), and I think this distinction is what really matters for Williamson. As far as I am aware the factive/nonfactive contrast is a feature of all natural languages.

        Strictly speaking it might be better if he’d present the manner in which this distinction is expressed in English as one illustration of the phenomenon of interest, rather than risk having his readership interpret him as saying that the English-language devices are universal. But I don’t see anything in his argument that commits him to saying that the particular grammatical features he alludes to are themselves universal.

        Jennifer Nagel

        March 14, 2012 at 11:51 am

        • Dear Jennifer,

          Thanks for that. I didn’t know the paper by Perez-Leroux and Schulz. I had a quick look and when I have time I’ll look at it properly – and perhaps have more to say in reply to you. (I also added a link to the paper in your comment – in case anyone else was interested in looking at it).

          One claim I found particularly interesting was that: “According to Abbeduto & Rosenberg (1985), children before the age of four interpret all verbs Cforget, rementber, klio~v, believe and think) as factive, that is they interpret the complement as true, independent of sentential negation and type of verb. This suggests that young children only pay attention to the embedded clause and interpret it as if it was a
          main clause.” I’m not sure what to make of this. I wonder if one finds the same effect in Turkish.

          Part of my motivation is that I think that knowing is developmentally prior to believing – that one develops the capacity to know before one develops the capacity to believe. The thought here is that believing involves some capacity for meta-representation whereas knowing does not. So I think that knowing does (or at least can) involve the exercise of conceptual capacities – but believing involves not just the exercise of such capacities but the capacity to represent these concepts. So I’m very interested in the developmental literature (although not at all an expert on it).

          On Williamson: I agree that there is nothing in his argument that commits him to saying that the particular grammatical features he alludes to are universal. But I do think that his appeal to the way England works does (at least implicitly) play a justificatory role in his argument. I don’t think my appeal to Turkish in any way blocks his conclusion – I just would like him (and other philosophers) not to implicitly use contingent facts about how English works as evidence for their position.

          You write: “Although the relevant grammatical constructions are different, it seems that across the board these verbs embed clauses presupposed to be true (where nonfactive mental state verbs carry no such presupposition)” – I’m not sure that this is right. What do you mean by an embedded “clause”? In the standard Turkish construction [and please – any native Turkish speakers, correct me if I’m wrong here] ‘know’ ‘see’ ‘believe’ ‘think’ ‘imagine’ work grammatically in the same way. The complement of these verbs is a noun phrase, with the verbal noun being in the accusative case – with the verbal-noun being something like an English gerund, but which can be tensed, have a person etc. And, grammatically, there is no clause embedded in the sentence which standing alone can be true or false (unless of course, like Ilhan, one wants to reduce truth to reference). Now of course – one can interpret this Turkish noun phrase as the Turkish way of expressing a proposition (or the same propositional content as the corresponding English sentence). But the surface grammar of Turkish does not force us to do this – we could equally well interpret an English that-clause as expressing something like a Turkish noun-phrase – something that is not truth-apt.

          Having said this: In Turkish one can talk of “wrongly knowing” and it might be the case that “know wrongly” functions grammatically differently from “know”. (Murat and Nurbay talk a bit about this in a paper I mentioned in another post) – which might give some support to your claim.

          Anyway – I guess my main point here is that on this issue perhaps grammar is not our best guide as different languages do things quite differently. But people working on knowledge in English (and on “propositional attitudes” in general) incessantly appeal to English grammar.

          Lucas Thorpe

          March 14, 2012 at 2:58 pm

  13. I may be missing something but if your aim here is merely to diffuse arguments from (contingent) grammatical features of English against your claim that knowing is not a propositional attitude, then why not appeal to following example and the like that one can easily find in English:

    A) She knows the distance between the moon and the earth.

    The object of knowledge in A is clearly not a proposition. A is not an instance of Know-How. Is there any reason to render A as irrelevant linguistic evidence? Whatever that reason is, it cannot be, on pain of circularity, that the object of A is not a proposition.

    If A is a legitimate instance of knowing, which I don’t see why not, then the fact that knowing is mostly propositional in English should not be used to refute your claim. There are non-propositional knowings in English. So I don’t think you even need the Turkish case to diffuse the arguments from English grammar against your position.

    nurbay irmak

    March 15, 2012 at 8:06 pm

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: