Science of language in the language of science Linguistics in the paradigm of empirical sciences Dorota Zielińska Kraków, Poland DorotaZielinska@yahoo.co.uk Plan Part One: Methodology: 3-stages; Part Two: Major descriptive frameworks Part Three: An application Part One The three stages of scientific research 1.1 Description (depends on the purpose) 1.2 Search for patterns (phenomenological laws) 1.3 Search for explanatory laws (principles) implying the patterns observed Models and theories According to Bunge (1972), theories describe general assumptions concerning a certain type of behaviour. They are not testable per se. Theories are tested indirectly by applying them to models, which are testable. Models, however, include approximating conditions and may also depend on additional laws. Part Two: Basic descriptive frameworks 2.1 2.2 2.3 Rductionism (ontological and epistemic) (Holism) Relationalism 2.4 Systemism cum Emergentism (ontological, but not epistemic reductionism) Bunge, (2003) Relational description “All that we can know in nature are relations and in the final analysis any kind of knowledge is just the knowledge of relations.” I.Z. Tsekhmistro ”Thus, for instance, when investigating the characteristics of elementary particles, all we can really establish is how they interact with the surroundings – the measuring device. In other words, we may know their states only in relation to the measuring device and with the precision limited by possible experimental set-ups.” Relational methodology in linguistics “We know the word by the company it keeps”. Firth (1934) Wittgenstein, in turn, brought to our attention the relational character of meaning contained, e.g., in kin terms, and in the names of chess pieces (not to mention pronouns). Still in a precomputational era, Harris (1988) proposed to define the meaning of a lexeme by syntagmatic relations it enters. RM in Quantitative corpus linguistics Corpus linguists study wide ranging collocations, contextual properties of thematically organized items, correlations between specific linguistic items and their position in texts, properties of linguistic items acquired in social contexts. As recent studies on priming in corpus linguistics indicate, virtually all items have specific environments, which they favour. Reversely, these environments identify the meanings of the respective items. This fact has already been exploited, for instance, in spell checker software. Systemism cum emergentism System: components combine and interact as a whole with a new environment, fulfilling a new function (an interaction not active for the components). This new interaction influences the components. Atoms in water/ ice a bird in a tree, could you open the window? Material system description Bunge (2003): First, defining the mechanism which has created and which supports the observed behaviour of the given material system or the process in a given environment - its structure and components. Second, defining the relevant environment, which will restrict possible behaviour of the system considered. Bunge (2003: 20) defines a mechanism as “a set of processes in a system, such that they bring about or prevent some change – either the emergence of a property or another process – in the system as a whole.” Part Three: an application 3.1 Search for patterns: The ordering of adjectives in English AAN noun phrases (adjective adjective noun) opinion, size, shape, age, colour, nationality, material a pretty rectangular table, a comfortable wooden bed Gradable (descriptive) categorizing English, Polish, Hungarian, Chinese, German, Italian (in a reversed order) Experimental results Wulf (2003) The independence from comparison index (IndComp) - the ratio of the number of occurrences of a given adjective in non-comparative degree to the number of all occurrences of the given adjective in a given corpus. The mean IndComp values for adjective1 and adjective2 in her study differ highly significantly (p < .001). Wulf’s Semantic closeness of an adjective. Zipf’s privilege of occurrence, Behaghel’s Law . 3.2 Search for principles: Linguistics as an empirical science ”Languages do not develop or evolve by themselves and there are no mechanisms of linguistic changes, in particular evolutionary forces. Only concrete things, such as people can develop and evolve. And, of course, as they develop or evolve, they modify, introduce, jettison linguistic expressions. The history of mathematics is parallel: mathematicians do come up with new mathematical ideas, which are adopted or rejected by the mathematical community, but mathematics does not evolve by itself.” Bunge (2003: 62) Language is a bio-psycho-social phenomenon which can be presented within a systemism cum emergentism framework Language depends on the history of the contingencies of its creation. Only statistical relations can be postulated. Analogy: defining the relationship between the height of a child and that of his/her parents. Besides, we have access only to a sample of „language”. Statistical character of linguistic principles Objectivity of measurement Degree of being gradable (evaluating) = the number of tokens in comparative and superlative forms/the number of all tokens Degree of being categorizing in Polish = the number of post modifying uses of adjectives / the number of both postmodifying and premodifying uses gradable adj. + descriptive adj. + categorizing adj. + noun The basic mechanisms (theory) of language use and formation Mechanisms forming form-reference correlations in the brain have been selected for by evolution. They lead to automatic local optimisation (regularization) of the semiotic system. They are approximately reductionist and bio-cognitive in nature. Non-reductionist (systemic) bio-psycho-social mechanisms driven by the natural selection like processes (depending on the function of language) (Altmann 1978) lead to global optimisation of the semiotic system (reflected, e.g., by implicational universals). These changes have been selected for because they allowed a given community to dominate due to communicating more effectively. A Model of AAN ordering Principle: The ordering observed makes language more efficient Processing categorizing adjectives first, and gradable ones second increases the precision of the information conveyed. długi drewniany most (a long wooden bridge) vs. długi kamienny most (a long steel bridge) When categorizing adjectives are used first in A2A1N phrases, the resultant encoded content of categorizing adjectives is more precise, while the encoded content of evaluating adjectives is not affected. a big red bird, a red big bird* With such an ordering, processing effort is also decreased Experimental results: Zielinska(2007) The division of the colour category into a category of intuitively highly relative colour terms, such as light, pale, vivid, dark and intuitively less relative ones such as red, blue, yellow, results in the category ‘relative colour’ being statistically more likely to precede other semantic categories than the category ‘non-relative colour’. Similarly, the subdivision of a given category of adjectives containing the information about age into a subcategory of the adjectives more and less categorizing, respectively, e.g. into {pre-war, renaissance, baroque, etc.} and such ones as {20year old, centenial, etc.}, results in more categorizing subcategory following other selected categories statistically more frequently than the other subset of the category ‘Age’ does. Future research on the order of adjectives Checking the order „Opinion_size /// Shape_Age_Colour/// Origin_Material in several languages Operationalizing categoriability and gradability In Polish: Categoriability = the number of Postpositional vs. prepositional uses = the number of separated postpositional usage to all uses (separated and not) Ciepły barszcz czerwony, czerwony barszcz ciepły Marta ma: długie, ciemne włosy: włosy długie, ciemne,ciemne włosy długie, długie włosy ciemne. Bibliography Gabriel Altmann 1978 in: Glottometrics Mario Bunge 1973 Matter, Model, Methodology 2003, Emergence and convergence Stephane Wulf. 2003. Multifactorial Analysis of the order of adjectives (Tom Grice) Thank you Relational approach in physics Such reconstruction led to the development of quantum mechanics, and a similar reasoning underlies also the philosophy of Einstein’s general and particular theories of relativity. It does not follow however, that quantum world does not exist independently of the measurements as was assumed by Copenhagen interpretation of quantum mechanics, but merely that our knowledge of it depends on the ways we can interact with it.