dr MIESZKO TAŁASIEWICZ Institute of Philosophy, Warsaw University, Poland m.talasiewicz@uw.edu.pl Simple syntax and substantial pragmatics comments on slides ABSTRACT The talk is a part of a program of defending categorial grammar (CG) in its simplest form, established by Ajdukiewicz and refined by Geach [against (1) different types of grammar & (2) sophisticated and complicated modern versions of CG]. In Szklarska I would present the summary of what is available within this version of CG and argue that we can have quite simple categorial account of syntax for quite complicated quantifier phrases (like: A man who always agrees with whoever he is talking to never tells only the truth – an example from M. Schenner’s last year talk in Szklarska + some other examples) – provided we are ready to admit quite extensive pragmatic, informal reasoning. NOTE! NUMBERS BEFORE PARAGRAPHS CORRESPOND TO THE NUMBERS OF SLIDES PART ONE: General introduction Semiotics: theory of language • Ch. Morris: syntax, semantics, pragmatics – different sets of semiotic functions of an expression • N. Chomsky: syntax semantics • Now: „semiotic holism” – syntax, semantics and pragmatics are aspects of overall semiotic value of an expression; they work always together dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 3 3. Traditionally, after Charles Morris (1930’s), semiotics was divided into three main areas in which different relations and functions concerning expressions: syntax, semantics and pragmatics. Accordingly there are three theories of language to develop: syntactic theory, semantic theory and – if that’d be possible at all – pragmatic theory. Noam Chomsky in (1950’s – 1960’s) introduced an order to the areas: he claimed syntactic theory the fundamental one, giving input for semantic theory; pragmatics – as (temporarily perhaps) unsusceptible for strict, formal treatment – would be set out of concern. 1 Nowadays – at least since 1980’s (due to emergence of formal pragmatics in the 1970’s) we can witness the growing awareness of what I’d call ‘semiotic holism’: the view that the ‘meaning’ – or better to say ‘overall semiotic value’1 – of an expression is a product of factors belonging to all the three areas simultaneously. It remains true that syntax enables semantic interpretation of an expression and that this interpretation is an input for pragmatic theory, saying what communicative purpose that expression could possibly serve, uttered in such and such conditions. However, this is only half of the story, because we are aware now of the relations the other way round: context can decide what is the meaning of an expression, and this meaning often happens to be a necessary input for establishing syntactic structure of the expression. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Drawbacks of semiotic holism: temptation for a ‘total theory’ • Multitude of competing theories • Difficulties in comparing theories, which are often (practically) incommensurable • Every theory sooner or later proves empirically inadequate • Regress in general understanding of language: there is no view from outside the battlefield dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 4 4. Semiotic holism – together with Chomskyan idea that theory of syntax should be testable against empirical evidence – gives rise to a strong temptation to formulate a ‘total’ theory that would explain all semiotical aspects of given expressions together, as they together really work. Such a theory, if one could develop such a theory reasonably fine, would give us a thorough and sound understanding of language. Alas, this seems a goal far too hard to achieve now and in predictable future. The result is then quite the reverse: the multitude of competing theories – each of them is intended as ‘total’ and is capable of explaining this or that phenomenon but proves inadequate very soon outside the initial scope – theories that are hardly comparable (even if they are not incommensurable in principle, they often are in fact, as the task of comparing them becomes itself a difficult academic enterprise). In this state of affairs we loose even the general understanding of what language is and how it – roughly – works. Every ‘theory’ gives us its own perspective on the whole matter, but it gives us no reason why to believe rather in it than in its competitor. Even if we have grounds for assessing relative success of a theory in explaining this or that phenomenon, it does not shift for justification of general claims. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Illocutions and perlocutions (as well as other communicative goals) are included in the ‘meaning’ which ceases to be a purely semantic notion. 1 2 Linguistics and Philosophy 25 (2002) ‘There is no doubt that the 25 years since the launching of Linguistics and Philosophy have witnessed an explosion in our understanding of linguistics semantics. There is, however, one area in which we have arguably made little progress – indeed I wish to suggest here that we have perhaps gone backwards. And this concerns the fundamental question of overall organization and architecture of the grammar – in particular, how the systems of syntax and semantics work (or don’t work) together’ Pauline Jacobson, The (Dis)organisation of the Grammar: 25 Years dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 5 Handbook of Logic and Language, 1997 ‘In the 80s, ‘‘frameworks’’ started appearing trying to change and monopolize part of the research agenda, and authors felt the need to present their ideas more forcefully as ‘theories’ with appealing names, forming schools and proselytizing. Part of this may be symptomatic for a young emerging area trying to establish itself, a phenomenon well–documented in fields like linguistics and computer science. This trend toward separatism and rivaling research agendas, though it may have had positive effects in stimulating foundational discussions, has hampered communication, and generated much fortuitous competition’ Johan van Benthem, Alice ter Meulen, Editors’ preface dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 6 5–6. Its not only my opinion. These are two generalizing comments of prominent philosophers of language, which support the picture I have presented. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Back to Morris • I recognize the holism as a matter of fact: semiotic value of an expression just is multiaspectual and usually one cannot (easily) disjoin the aspects ‘in the flesh of language’ • I reject the holism in theory of language: one can and should disjoin the aspects conceptually. Let us have simple and powerful theory of syntax and idealised semantics, motivated by general logical and philosophical intuitions; and let us appoint pragmatics to be the battlefield: the area for competing theories, providing the empirical interface (on the one hand it makes syntax and semantics more of philosophical interest and pragmatics, on the other hand, the most interesting field for empirical linguistics) dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 7 7. Whether one is allowed to do so, is a complicated methodological question. I cannot discuss it here. Roughly, I would refer to something like ‘the hard core’ of a theory and to the 3 notion of conceptual frame. It is a matter of fact that contemporary methodology of empirical sciences allows in general that parts of theories are not directly testable (they are assessed on the grounds of fertility or plausibility etc. – cf. my book Tałasiewicz 2000). –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Back to Ajdukiewicz • We have good and extremely simple syntactic framework, grounded in philosophical reflections of Husserl and logical considerations of Leśniewski. It is categorial grammar of Ajdukiewicz • The task is to clean up the philosophical issues about categorial grammar (with the definition of syntactic/semantic category in the first place and including of necessary semantic component) and delimit the field left for pragmatics dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 8 8. The ideas behind Ajdukiewicz’s grammar had been providing quite a constant picture of syntax for around 70 years, form the early writings of Husserl up to the 1970’s, when they became ‘corrupted’ in order to get in touch with the requirement of empirical testing (which was presumably triggered by the competition with generative grammar). The other reason for ‘corruption’ however – by ‘corruption’ I mean theoretical amendments, like the idea of category shifting, which makes CG a little bit more flexible to match the real speech, but much more complicated and obscure – so the other reason for these amendments was that CG really needed some philosophical refinement and correction – but rather towards further simplifying, not complicating. The main points needing correction are the definition of syntactic/semantic (depends on terminology) category and the capability to answer a semantic question of how can we get syntactic information from primitive ostensive procedure. These philosophical analyses are the main topic of my book presently being prepared (Tałasiewicz 2006). Below I will give some of the results. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– PART TWO: The Core of Categorial Grammar Functoriality Principle One… to rule them all, One… to find them One… to bring them all and in the darkness bind them J.R.R.Tolkien, The Lord of the Rings dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 10 4 Functoriality Principle - rephrased • In every syntactically well-formed expression there is exactly one part (a functor) which takes all other parts as its arguments and – binding them (in the darkness?) - yields the whole expression – The parts of an expression are not on a par among themselves; there is always one superior to others, one ‘to rule them’ (contrary e.g. to generative grammar) – A functor binds its arguments iff the functor and the arguments belong to proper syntactic/semantic categories dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 11 10–11. It is Functoriality Principle (FP) what distinguishes categorial grammars (CG), not the notion of category itself. Categories one can find anywhere, as well in generative grammars (GG). There is terminological mess about whether the categories in categorial grammar are ‘syntactic’ or ‘semantic’ ones. No wonder – there are three levels of FP: syntactic (in given sentence there is exactly one syntactic operator); semantic (in the position of operator can stay only an expression that belongs to the category of functor); and ontological: functors denote functions that take arguments from the designata of the arguments of the functor and give values in the set of designata of the compound expression yielded by the functor. There are good reasons to call the categories in CG ‘semantic’. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Definition of semantic category postulates • Interchangeability • Primitive categories: ‘sentence’ (s) and ‘name’ (n) • Atomic principle: given the categories of functor and its arguments we have the category of the compound; given the category of the compound and the categories of the arguments of the functor in this compound, we have the category of the functor. dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 12 12. Original definition of semantic category goes as follows (Ajdukiewicz 1935/1967: 223): ‘The word or expression A, taken in sense x, and the word or expression B, taken in sense y, belong to the same semantic category if and only if there is a sentence SA, in which A occurs with meaning x, and which has the property that if SA is transformed into SB upon replacing A by B (with meaning y), while retaining exactly the same meaning of the other words in sentence SA and the same syntax of this sentence, the resultant SB will also be a sentence’ [slightly revised translation –MT]. 5 This definition has been rightly criticized on the grounds that its definiens features terms with unclear meaning and that any attempts to define these terms without contradicting our intuitions require references to the concept of… semantic category. The definition of semantic category is thus in danger of falling victim to the vicious circle. One line of amendments required for CG (mentioned before) is to replace this definition with a protodefinition by postulates, as a preliminary and intuitive guideline to the denotative definition by recursion. My proposition is following: Let us have Interchangeability Principle not as a connotative definition but as a postulate; there being no thread of vicious circle then. It says that two expressions mutually interchangeable ceteris paribus in a sentence salva syntactic well–formedness belong to the same semantic category and vice versa. IP is taken as a defining principle and cannot be confronted with any empirical evidence. It is pretty obvious that it results in discriminating the notion of syntactic well–formedness from the notion of acceptability by native speakers. I must and I do admit that there are syntactically (according to syntactic theory) well–formed expressions that are ungrammatical according to traditional grammar or inacceptable for the competent user of a given language. The division into primitive (or basic) and derived (or functor) categories refers to Husserl’s notion of propositional and nominal kind of intentionality. Only sentences and names can have a meaning on their own. I reject the idea of having one or three or whatever number of primitive categories. There are two of them. Fullstop. Atomic Principle (AP) is popular in the first half of its formulation; I insist on the second half (sometimes called ‘reverse’ AP). In language acquisition – on early stages there is sole ostensive procedure involved – we are given only primitive categories of some simple and some compound expressions. These compounds are defined ostensively as wholes: we know their meaning and category without knowing their structure, which is to be discovered (by the ‘reverse’ AP). –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Rules of syntactic analysis: summary of the CG philosophy 1. The rule of primitive categories: the arguments of the functor in an expression belonging to primitive category belong as well to primitive categories (not necessarily the same) 2. The rule of superfunctors: the arguments of a superfunctor belong to the same category as the functor yielded by the superfunctor or to a primitive category 3. General guideline: at first try to get as big parts as possible according to points 1 & 2 dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 13 13. The rules summarize what is doable in analysis on the above grounds (and what is indeed being done). In an expression of a basic category we are given initially the components belonging to basic categories: we can recognize names and sentences (on ostensive grounds). All the rest – is the functor (possibly compound itself). In a compound functor there must be a so called superfunctor which yields the given functor out of… well, what it could possibly be? Names and sentences, of course (as in n/n//n/n///s), but not only. Given the principle of interchangeability we can ask if there is a part of the compound functor under analysis, which (the part) can play the same role as the whole functor (can be interchanged with it salva 6 whatever there is to be sa(l)ved). If there is such a part – this part has the same category as the whole functor (say: f); the rest is the superfunctor of the category f/f. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Simple example John | n | passionately | s/nn | s/nn//s/nn | loves | |s/nn | Mary (s) | n | dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 14 14. In ‘John passionately loves Mary’, which is a sentence, we can recognize names: ‘John’ and ‘Mary’. The rest is the functor – compound, in this case: ‘passionately loves’. This functor makes a sentence out of two names, so it belongs to the predicate category s/nn. In this functor we cannot find any names or sentences. So we look for a part of it which would be interchageable with the whole functor. We find ‘loves’ (wherever one can put ‘passionately loves’ one can put mere ‘loves’ as well). It belongs to the category s/nn, of course. So the superfunctor is ‘passionately’ which makes s/nn out of s/nn – therefore it itself belongs to the category s/nn//s/nn. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Simple example 2 John loved Mary but Mary didn’t love John | s ||s/ss|| s | | n || s/nn|| n | | n || s/nn || n | |s/nn//s/nn|| s/nn | NOT: John loved Mary but Mary didn’t love John | n || ? || n || ? || n || ?? || n | dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 15 15. It shows that we really need to look for as big parts as possible. Sometimes we can recognize small parts belonging to basic categories but it would leave so many incoherent parts aside that it would become impossible to make a single functor out of these remainders. –––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– 7 Selected particular guidelines for semantic interpretation of syntactic analysis • General names are variables with the scope of variability (x: xP); individual names are constants; generic names could be both • Adverbials are superfunctors over predicates; if required they are compound and contain a nominal (usu: temporal) variable that can be quantified over – that amounts to what is sometimes called ‘quantification over situations’ • Quantifiers have always the category s/s; they stand at the beginning of the sentence and are ordered – normally – according to the order of variables being quantified. Often quantifiers are tacit; the user has to quess which quantifier is applicable – on pragmatic grounds dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 16 16. This point requires substantial comment – there is around 20 pages in my book on this. The point is – roughly – that we do not need any substantial paraphrases: names are names (not predicates nor quantifiers), the order of quantifiers (which are often implicit) is normally2 given by the order of variable names quantified over etc. What is worth stressing, I am not very fond of quantifying over situations (although I do believe, of course, that sentences denote situations). What is quantified over in such cases is an adverbial parameter, usually temporal (‘always’). This parameter I would represent as a variable index at predicate. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– PART THREE: A hard example A man who always agrees with whoever he is talking to never tells only the truth • We have four explicit quantifiers here and – as we will see – three more implicit. One of the latter is an implicit general quantifier for the name ‘a man who always agrees with… talking to’. • So we have: [Every]a[never]b[only]c{[A man who…talking to]a[tell]b[truth]c} | s/s | s | | s/s | s | | s/s | s | | n |s/nn| n | dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 2 19 It means that a change in this order is possible, but requires special justification. 8 A man who always agrees with whoever he is talking to (n) |……...|n/s|…………………… s ………………………….| here we have a sentence with another implicit quantifier, this time existential, binding the name ‘A man’: [there is]d[always]e[whoever]f{[A man]d[agrees with]e[one (that) he is talking to]f} | s/s | s | | s/s | s | | s/s | s | | n | s/nn | n | |……| n/s |………s……….| There we have another sentence with another implicit existential quantifier binding the name ‘(some)one’: [there is]g{[he][is talking to][someone]g} | s/s | s | | n | s/nn | n | dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 20 19–20. The example is taken from last year’s presentation on the Workshop by Matthias Schenner (Syntactic partioning revisited), where it is ascribed to Barbara Partee (no analysis of it is given in Schenner’s paper, except that it is suggested there are three quantifiers in it, whereas I would count seven). Indexes at quantifiers are just a technical device to help to remember where the quantifiers are applicable. Implicit quantifiers are in italics. Some quantifiers are not to be analyzed away in the first order. It is so because there are names made out of sentences (by nominalizing functors like ‘who’, ‘that’ etc.). Quantifiers from these sentences do not appear in the sentence in which the name is a part. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Semantics (s) x IS TALKING TO (heconst, x) he is talking to someone (n) x: IS TALKING TO (he, x) - nominalizing functor ‘that’ Such a (person) to whom he is talking (s) ytx AGREES WITHt [y: yman, x: IS TALKING TO (y, x)] There is such a man that he always agrees with anybody he is talking to (n) y:tx AGREES WITHt [y: yman, x: IS TALKING TO (y, x)] - nominalizing functor ‘who’ Such a man that he always agrees with whoever he is talking to; or A man who always agrees with whoever he is talking to dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 21 9 y ~t’ z ~ TELLt’ [y: tx AGREES WITHt (…), z: ztruth] note: ‘only S is P’ means ‘every not-S is not P’ or y t’ z TELLt’ [y: tx AGREES WITHt (…), z: ztruth] every man who agrees with whoever he is talking to always tells something untrue or … never tells only truth dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 22 21–22. Once we’ve got the syntactic atoms and the structure, we can build up the semantic interpretation of the whole, starting from bottom – up. So, at the beginning we have a simple situation that certain person (referred to by the pronoun ‘he’)3 is talking to someone. Then, by the nominalizing functor ‘that’ we get the general name of such persons to whom this ‘he’ is talking to. And we store this aside, for a while. In the meantime we consider also quite simple situation in which a man agrees with someone. Then we substitute for the indefinite ‘someone’ the name we have built a moment ago and stored; we get situation in which a man agrees with someone he is talking to. (And if we are smart enough we notice here, that we got the reference for ‘he’ – it is indeed a variable, but it is quantified over somewhere else). Now we notice a general quantifier for the interlocutors, so we understand, that our ‘a man’ agrees with everyone (=whoever) he is talking to. And we begin to wonder, what it means, that he (‘a man’) always agrees with whoever he is talking to. For a moment we entertain the idea that there is a man who at every moment of time is talking to somebody and always agrees with whoever he is talking to, but soon we get serious and assume that ‘always’ in this context means ‘any time he talks to somebody’ (purely pragmatic reasoning). Then, by the nominalizing functor ‘who’ we get a general name of such persons who always agree with whoever they are talking to – and store it for a moment. In the meantime we consider third relatively simple situation in which somebody is never telling only truth. We notice that it means that he always tells something untrue (when ‘always’ means of course ‘at every moment in which he tells something at all’). To this simple situation we apply the name we have just built and stored and we get finally the interpretation saying that every man who always agrees with whoever he is talking to never tells only the truth. ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– In the bottom we do not know exactly who it is; it is ‘a’ person for us, but it shall not be represented by a variable: the pronoun informs us that when we proceed in semantic interpretation, we will find identifying information somewhere further. 3 10 Conclusion • We have analysed a very complicated sentence with seven quantifiers involved with very simple means of basic Categorial Grammar • Doing so, we have been saying things quite obvious, but nevertheless hard to be formulated formally. This is the task for pragmatics: make a theory for these obvious but evasive things (they are evasive because they depend on our overall knowledge of the world, which is vast and partially tacit – this undermines every account of it) • Let be syntax simple as it is dr Mieszko Tałasiewicz, Institute of Philosophy, Warsaw University 23 23. Not only pragmatic theory is needed to formalize the story told above about building semantic representation. There is as well a room for linguistic analysis of traditional grammar. Grammar plays different roles; one of them is to give information for logical syntax (e.g. what is the order of arguments of the functor etc.), but there are other roles; for instance to embody the lexicalized knowledge of the world (e.g. which objects are countable, or living etc.) It is a task for linguistics to establish in general terms what information gives traditional grammar of a given language about the logical structure of the expressions of this language (it could be different in different languages). We have seen that in particular cases it should be simple and easy to know what is what in an expression – thanks to general knowledge of the language one is a competent speaker of, and to general knowledge about the world one is a competent inhabitant of. But a general theoretic approach would be welcome (an example of such a work: a classification of predicates according to tense and aspect in Polish by Labenz (2004). ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– Here are some more analyses: A. Only one class was so bad that no student passed the exam (von Fintel 1995 after I.Heim) – source of example, not analysis, of course. syntax: Only one class was so bad that no student passed the exam | s/s | s | | n |s/nn| n | |….| n |….…………..n/n………………….| |n/n//s| s | |s/s| s | | n | s/nn | n | semantics: ~x PASSED (x: xstudent, the exam) [‘the exam’ is an individual name] : ~x PASSED (x: xstudent, the exam) [‘colon phrase’ belonging to category n/n; a colon in standard notation represents functor ‘such (or so) … that’] bad [general name in third order of analysis – not quantified] 11 bad: ~x PASSED (x: xstudent, the exam) [compound name of something bad so that no student passed the exam; second name in second order] y: yclass [first name in second order] 1y WAS [y: yclass, bad: ~x PASSED (x: xstudent, the exam)] Only one class was bad so that no student passed the exam. What is interesting in this analysis is that it does not say the students must belong to the class in question. We can restrict the scope of the variable x to students of this class if the context suggests so – on the pragmatic grounds. However, context not always suggests such restriction. Consider a case in which we decide about a certain whole (e.g. a transport of fruits) on the basis of testing a sample. We can perfectly well say that given box of apples was so bad that no apple would be allowed for sale – meaning no apple from the whole transport at all: the whole transport would be rejected. The restrictions on scope are not in the syntax here. B. Every linguist knows two languages. (Chomsky (?), somewhere). It’s been told that this sentence is ambiguous according to the two possible orders of implicit quantifiers. a) x 2y KNOWS (x: x linguist, y: ylanguage) b) 2y x KNOWS (x: x linguist, y: ylanguage) It would be a sort of counterexample to my claim that the order of names gives the order of quantifiers. I would say in defence of myself that standard meaning is only (a). Meaning (b) is a misinterpretation of conceivable interpretation (b’): (b’) x KNOWS (x: x linguist, twolanguages) where twolanguages is an individual name for, say, English and French together. (Exactly as ‘three musketeers’ is the name for Atos, Portos and Aramis) C. When Kim visits her parents, she often takes the train. (von Fintel 1995) The word ‘when’ is a functor correlating two situations denoted by two sentences; thus it is a connective belonging to category s/ss, which semantics is close to implication. The only temporal quantifier is the word ‘often’. We can tell this seeing that the order of ‘when’ and ‘often’ can be easily changed without change in meaning (‘Kim often takes the train when she visits her parents’). What shall be preserved is the order of ‘when’ and ‘Kim visits her parents’ – because the change of this order would reverse the order of arguments of implication: ‘When Kim takes the train, she often visits her parents’ means something completely different. So: syntax Often when Kim visits her parents, she takes the train. | s/s | s | | s/ss | s | s | | n | s/nn | n || n | s/nn | n | |n/n| n | semantics 12 OFTENt [VISITSt (Kim, her parents) ––> TAKESt (Kim, the train)] All the difficulty is hidden in discussion about how to interpret the quantifier ‘often’. In this context it is reasonably to suggest that it should be understood in a following way: the whole compound sentence is true iff the sentence ‘Kim takes the train’ is true in many (most) times at which the sentence ‘Kim visits her parents’ is true. (One can call it quantification over situations, but the situations are represented here by their temporal index). SELECTED BIBLIOGRAPHY Ajdukiewicz, Kazimierz (1935/1967). Syntactic Connexion. In Polish Logic 1920-1939, Storrs McCall (ed.), 207-231. Oxford 1967: Oxford University Press. [First published in Studia Philosophica 1 (1935) in German: Die Syntaktische Konnexität.] - (1960/1978). Syntactical Connections between Constituents of Declarative Sentences (1960). In The Scientific World-Perspective and Other Essays, 1931-1963. Edited and with an Introduction by Jerzy Giedymin, p. 269-281. Synthese Library Vol. 108. Dordrecht/Boston 1978: D. Reidel. [First published in Studia Filozoficzne 6(21)/1960, p. 73–88.] Husserl, Edmund (1901/????), Logical Investigations, [Logische Untersuchungen, 1901] von Fintel, Kai (1995/1997), „A minimal theory of adverbial quantification”, [in:] B. Partee, H. Kamp (eds.), Context dependence in the analysis of linguistic meaning: Proceedings of the workshops in Prague, February 1995, Bad Teinach, May 1995, IMS Stuttgart Working Papers, Stuttgart 1997, p. 153-193 [internet version 04/03/2005: http://web.mit.edu/fintel/www/minimal.pdf] Labenz, Piotr (2004), Event–calculus semantics of Polish aspect, master’s thesis, University of Amsterdam, http://www.illc.uva.nl/Publications/reportlist.php?Series=MoL]. Schenner, Mathias (2005), “Syntactic partioning revisited”, paper on “VI Szklarska Poreba Workshop”. Tałasiewicz, Mieszko (2000), Pojęcie racjonalności nauk empirycznych [The Concept of Rationality of Empirical Sciences], Wydawnictwo WFiS UW, Warszawa – (2006) [in preparation], Filozofia składni [Philosophy of Syntax] 13