Souborné dílo Vladimíra Skaličky - 3. díl (1964-1994), Dodatky, Bibliografie

This is the last volume of Vladimír Skalička's Comprehensive Work. Vladimír Skalička was not only the first professor of general linguistics and a founder of the Finno-Ugric field of study at the Charles University Faculty of Arts, but also an important Czech and Slavic scholar. The first two volumes containing Skalička's works from 1931-1963 were published in 2004. The third volume contains texts from 1964-1994 and Skalička's complete bibliography.

The last volume of Skalička's collected works contains his writings from 1964 to 1994 including also book reviews and small contributions to a Czech encyclopedy. The last article "Praguian typology of languages" was written together with Petr Sgall who systematized Skalička's doctrine and published the article in 1994 (Skalička died 1991). The editors who performed a heroic work by collecting, translating and ordering 1304 pages of text, added Skalička 's complete bibliography using different sources. Three registers (Names, Objects, Languages) yield an excellent guide through Skalička's work. Though the three volumes are now available in Czech, the interested reader can find the majority of the articles in world languages from which they have been translated.
The first most conspicuous feature of all his works is their shortness. He did not like many words, he said what was necessary and passed to the next idea. This was both his speaking and writing custom.
The second important feature is Skalička's place in the history of linguistics. He is known as typologist but his typology is rather of a synergetic kind, distinguishing him even from present-day typologists. Zipf and Skalička can be considered predecessors of language synergetics. Zipf as quantifier, discoverer of quite new dimensions of language, Skalička as typologist with enormous knowledge of languages seeing everywhere interrelations between language properties. As a matter of fact, he can be called the pope of linguistic interrelations.
Though in more than 300 publications of the three volumes one can find articles on language development, syntax, glottochronology, hyposyntax, sociolinguistics, history of linguistics, morphology, phonemics, style, translation, book reviews, encyclopaedic articles etc., his main interest concentrated to typology and we shall restrict ourselves to this dominant feature of his work.
Though one finds in his work allusions to very modern concepts like redundancy, compensation (in a self-regulating systems we call it functional equivalents, e.g. morphology and syntax are mutual compensations), necessity, probability, possibility, probabilistic relations, etc. he did not use any mathematics. Nevertheless, he gave excellent hints at the study of relations at all levels of language. Skalička defines the tasks of typology as follows: (1) Study of phenomena, (2) study of interrelations, (3) quantification of interrelations (p. 981) which are probabilistic (p. 986). As a matter of fact, he himself came at the second stage and developed it better than any typologist before him, stage three is the initial object of modern synergetic linguistics and is performed with mathematical means.
However, looking retrospectively at his work we see a lot of difficulties connected with the quantification of his interrelations. As a matter of fact, phenomena themselves, e.g. properties, must be quantified, and then interrelations can be ascertained almost automatically.
There are still several other levels above his requirements, namely the theoretical derivation of the interrelations, i.e. their embedding in a theory and at last their explanation by means of laws (subsumption under laws). Peter Sgall asks the essential question "why some feature combine with one another?" (p. 1205). Skalička tried to give some reason or causes for it but the realisation of his program is confronted with enormous problems. Though in typology and language theory we cannot avoid his results and a time must come when everything he said must be "quantified", considering some features that can be found throughout his work we are rather desperate. Let us mention only some of them: root length; word length; distinguishing word classes; word complexity; conversion; number of affixes; extent of derivation; number of synsemantics; affix length; building of compounds; number of preposition and postpositions; homonymy of affixes; synonymy of affixes; fixedness of word order; inflection; internal inflexion; number of clauses; number of endings in the word; morpheme discontinuity; existence of infinitives, participles, verbal nouns; number of vowels and consonants; extent of congruency; differentiation of root and auxiliary elements; differentiation of inflection and derivation; sentence markedness; vowel harmony; existence of suppletivism; article formation; possessivity; extent of declination, which is only a modest extract from the diversity of features. Some of them can easily be quantified, e.g. word length, but the majority of them is a source of enormous problems even from the qualitative point of view. We know, for example, what is word order but what should be quantified? Quantification is not a classification, e.g. in word order types like SVO, OVS etc. it is at least an ordering, scaling etc. How to quantify the "strength of word order" or the "extent of conversion" (theoretical or observed)?
Skalička considers a language type a non-existing theoretical construct, a set (system) of mutually cooperating phenomena (p. 985) realized in a real language to a certain extent. Each language has components of all types. Hence, type is not a class of languages and typology is not classification. Today, we would say that a real language is a kind of steady state in which a function of the vector of all (quantified) properties results in a constant (in different terminology: an attractor). This recalls the first steps of G.K. Zipf in the domain of word frequency distribu-tions. If a feature changes its value, the functional equivalents must change their own values in order to restore the equilibrium. Zipfcame to control cycles of maximally four features, Köhlers system containing also the requirements and mathematical relations contains about 20 features, Skalička's system without any quantification contains about 100 features. It would be lost of energy trying to enlarge his system, the most important task would be to incorporate his features step by step in Köhlers system. Even the results of classical typologists like Finck, Steinthal, Misteli, Lewy, Sapir etc. and of all new ones should be embedded in Kóhleťs system.
One sees that the fulfilling of this task is not only the highest aim of typology but at the same time a step towards language theory. Language theory does not consist of rules or sentence structures but of stochastic laws whose possible existence has been hinted at by Skalička. He assumed a development in this direction and spoke of the possibility of a deductive typology (p. 1024) which is nothing else but language theory.
Unfortunately, this way will not be simple. Taking two properties whose interrelation will be corroborated in several languages we necessarily meet a language in which it does not hold. Does it mean that the preliminary theory is false? No, it means that a part of the variance (deviations) must be captured by a third variable. But which is the third variable under say 100 possible ones? One cannot check all, either one has luck or one has ingenuous ideas. Skalička showed us a direction but not a way. Besides, no single linguist is competent in so many language as are necessary for corroborating a theory. He himself brought examples from 368 languages and language families, an enormous achievement for a linguist, but examples do not corroborate anything. A systematic theoretical work accompanied by empirical work of specialists in individual languages is necessary in order to accomplish his program. No future typology can be built without his results unless it remains a torso. Thus a translation of at least his basic works in English would be a very meritorious deed.
Influenced by the typology of his predecessors, Skalička kept the old names of types like inflective, agglutinating, polysynthetic, isolating and introflective but did not consider them as classes but rather as a set of cooperating properties. A language does not belong to a type, in almost all languages almost all types are present. The three volumes are full of properties of individual types, here we present only P. Sgall's summary (p. 1270f) of the polysynthetic type (we found more than 30 properties in the work): (a) Some lexical words are used also as auxiliary words and there are no other grammatical means in the morphology; (b) no distinguishing of word classes; (c) word building is performed by composition; (d) there are no affixes and endings, hence they cannot be in opposition; (e) the phonemic form of grammatical morphemes is similar to that of lexical words, hence they can be homonymous; (f) the lack of grammatical means is associated with the grammaticalized word order; (g) compounds fulfil also the functions which in other types are fulfilled by dependent clauses. One easily sees that these properties are extreme cases. In individual languages they are realized "to a certain extent". And this "extent" is exactly what is searched for in synergetic linguistics. Thus starting from Skalička's enormous number of hypotheses about language structure, one could step by step set up a very ample theory of language. We hope that the edition of his collected work in Czech will stimulate his fellow countrymen to further develop the most promising and the most vital theory that ever arose in typology. But here we see an analogous inscription to that at the entrance of Dante's hell: "Give up every hope if you enter without mathematics!"

I-II Journal of Quantitative Linguistics 14 (1), 2007, str. 100-102, Reviewed by L. Hřebíček and G. Altmann.