ACTA LINGUISTICA LITHUANICA

LII (2005), 119–156

Thoughts on Grammaticalization (Loose, Connected Remarks)^*

BOHUMIL VYKY̌PEL

Ústav pro jazyk český AV ČR

The present paper deals with the notion of grammatical elements of language (as opposed to lexical elements). Two conceptions of grammaticality are distinguished: the one concentrating on content and the one concentrating both on content and on expression (i.e., concentrating on signs). Further, the concept of grammaticalization formulated by Christian Lehmann is compared with Prague School typology. In conclusion, a few remarks are made on the relationship between grammaticalization theory and the general linguistic method.

1. One of the topics with which general linguists frequently concern themselves nowadays seems to be so-called grammaticalization. It is well known that, in a certain sense, the programmatic work by Christian Lehmann stands at the beginning of this renewed interest Thoughts on grammaticalization. It was reissued some time ago (Lehmann 2002), and I thus take the opportunity to offer some thoughts on grammaticalization as well. Unfortunately, Lehmann’s book is a fragment (and the announced second part looks—even more interesting, particularly from the perspective of general linguistics…), yet it is indisputably the most theoretically consistent contribution to the problem and, in general, a book of wide-ranging interest and great value for learning. As for its fragmentary character, no less a figure than Louis Hjelmslev (1968: 114) wrote that nothing is more delightful for a scholar than to have before him what remains to be accomplished.

Before the actual discussion, however, let me make two further remarks (partly in anticipation):

1.1. In discussing the concept of typology, Petr Sgall (1971) divided the differences among the linguistic disciplines that call themselves typology into two main groups: those consisting in the use of a single word (namely, typology) for different concepts, and those consisting in the use of one word (namely, typology) for different interpretations of the same underlying intuitive concept. Something similar arises with the term grammaticality: the first case discussed by Sgall occurs in the difference (also vividly described by Lehmann 2002: § 2.1) between the “American” concept of grammaticality as “correctness” and the “European” one, in

^* I thank Markus Giger (Prague) for linguistic correction and inspiration regarding the content.

120 |

BOHUMIL VYKYPĚL

which grammatical entities are considered in contrast to lexical ones; the second case is found in the differences among the various interpretations of the European concept of grammaticality. In what follows, like Lehmann, I shall of course concern myself only with the second case (though at one point I unexpectedly also turn to the first).

1.2. In general, the question arises as to why one should concern oneself with grammaticality and grammaticalization at all. A provisional answer might be: because these constitute one of the instruments of linguistic description and explanation. I shall return to this later.

2. One of the problems repeatedly addressed—sometimes more, sometimes less—especially in the European linguistic tradition, is the curious difference between the so-called lexical and so-called grammatical entities of language: it is believed that some elements of language—whether (depending on the researchers’ general conception of language) signs, the contents of signs, or components of the contents of signs—can or even must be divided into these two large groups. In general, theories of grammaticality or grammaticalization can be divided into two large groups. Some researchers seek their criteria of grammaticality (grammaticalization) solely in the plane of content (however it may be construed), while others also take into account how the content elements under investigation are expressed.

2.1. As will become clear later, it is important in our context to mention that one of the linguists belonging to the first group is Petr Sgall. Sgall’s formulation of the so-called typological dominants (see most recently Sgall 2004) consists in the claim that certain elements of the so-called tectogrammatical level of functional-generative description—the grammatical elements—are expressed differently in different language types, so that the expression of grammatical elements is not relevant to their definition as grammatical. Sgall gradually expanded the part of the tectogrammatical level made up by grammatical elements. Initially, he included here only the so-called syntactic values or functors (Sgall 1979: 8, 10); later he also included the so-called morphological values or grammatemes (Sgall, Hajičová, Panevová 1986: 271); finally, he speaks of the grammatical part of the tectogrammatical level, which consists of morphological values or grammatemes, syntactic values or functors, the concept of valency, and so-called functional sentence perspective (see Sgall 1992a: 204). This grammatical part is contrasted with lexemes. Consequently, the expression of grammatical elements may take five forms, and on this basis five language types are established as constructs:

1. Expression through the ordering of phoneme sequences (morphs) that express non-grammatical functions = of the polysynthetic type;

2. Like non-grammatical functions, i.e. through phoneme sequences (morphs), but with these occurring in the form of words, that is, not being attached to the sequences expressing non-grammatical functions = of the isolating type;

3. Like non-grammatical functions, i.e. through phoneme sequences

GEDANKEN ZUR GRAMMATIKALISIERUNG

(Morphemes), which occur in the form of affixes, i.e. are attached to the successions expressing non-grammatical functions = the agglutinative type;

4. by varying (modifying) phoneme successions (morphemes) that express non-grammatical functions, where this variation takes place at the end (or, though, also at the beginning) of the successions = the inflectional type;

5. by varying (modifying) phoneme successions (morphemes) that express non-grammatical functions, where this variation takes place within the successions = the introflexive type.

As for the definition of the grammatical elements of the tectogrammatical level, it is obvious that the elements or functions designated here as grammatical correspond to one of the two cornerstones of the linguistic utterance in Vilém Mathesius’s account: the act of “linking,” of establishing relations, or the syntactic act. And Mathesius did indeed (probably mediated by Vladimír Skalička’s linguistic theory) provide an important source of inspiration for Sgall. In the ingeniously simple conception of his “functional grammar,” Mathesius famously distinguished two fundamental acts, two “pillars” through which every utterance is formed: the act of naming, or the onomatological act, and the act of linking, of establishing relations, or the syntactic act. In terms of their meaning (or content), linguistic elements are then to be divided into two groups: those that primarily serve the onomatological act, and those that primarily serve the syntactic act. It is therefore necessary first to describe how the language in question names and how it puts what is named into relations. Only afterwards can one turn to “morphology,” understood as the study of the grouping of linguistic means according to their formal kinship and thus cutting across the two parts of language description: the study of linguistic naming (onomatology) and the study of establishing relations (functional syntax). (Cf. Mathesius 1929; 1936; 1975; Daneš 2003.)

Something similar is found in Louis Hjelmslev. Initially—in his pre-glossematic period—Hjelmslev discusses the distinction between semantemes and morphemes. Semantemes are the connected elements, morphemes the connecting ones (Hjelmslev 1972: 52, 62; 1973: 85): “Det skal forstaas saaledes, at semantemerne er de elementer, der bærer rækkens meningsindhold, og morfemerne de elementer, der former dette meningsindhold paa en bestemt grammatisk maade” (Hjelmslev 1972: 52f.); “les catégories de sémantèmes expriment sans ambiguïté des catégories de notions (…) les catégories de morphèmes expriment sans ambiguïté des relations entre les notions” (Hjelmslev 1928: 210); “les morphèmes indiquent les rapports entre les sémantèmes” (Hjelmslev 1973: 87). We can see that Hjelmslev comes close to Mathesius: semantemes name, while morphemes “link” and establish relations (primarily or secondarily).

Hjelmslev’s later glossematic theory of language can be regarded in this respect as a kind of formalization. In glossematic Hjelmslev, too, we find a division of linguistic elements corresponding to the traditional division into lexical and grammatical elements. In keeping with his theory of language, Hjelmslev posits these two fundamental groups in both the content plane and the

121

122 |

B O H U M I L V Y K Y P Ě L

expression plane, and the members of these categories are formed by the potential constituents of signs, or more precisely, by expressionless elements in the content plane and contentless elements in the expression plane. Hjelmslev changed the names of the categories. The property by which these categories are defined also changed, but again and again the issue was—translated from glossematics into a language perhaps more generally intelligible—the potential (primary or secondary) capacity of some (“grammatical”) elements to connect others (“lexical”) or to place them in mutual relations. (See especially Hjelmslev 1959: 139–164; 1973: 243–254.)

In our context, it is important, at any rate, that Hjelmslev locates his criterion of grammaticality only within the respective plane: the expression of grammatical content elements is irrelevant to their definition, since it does not belong to language as a system (which is formed by the two planes), but to language use, specifically to its sign aspect.

2.2. The second group of theories of grammaticality mentioned above (§ 2) includes, among others, the other linguist who, alongside Sgall, further developed Prague typological theory, Jaroslav Popela. As is well known, Popela formulated the typological dominants differently from Sgall. The first property by which they are defined is whether the type favors grammatical signs. This is positive for the isolating, agglutinative, inflectional, and introflexive types, but negative for the polysynthetic type. The “grammatical” types are consequently subdivided according to two further properties: the cumulation of semes (content elements) in grammatical signs and the degree of fusion between the lexical and grammatical signs. There is no cumulation in the isolating and agglutinative types, whereas it occurs in the inflectional and introflexive types; fusion is minimal in the isolating type, greater in the agglutinative type, greater still in the inflectional type, and maximal in the introflexive type. (Cf. Popela 1991, 1998.)

It is thus evident that Popela’s conception of grammaticality, which he employs in defining the basic properties of language types, is a grammaticality of signs. The polysynthetic type is not non-grammatical because its content plane lacks the same or similar (grammatical) elements as the grammatical types, but because these elements in the polysynthetic type have no segmental or discrete expression: they are expressed either by word order or by semantemes in grammatical functions. Non-grammaticality thus pertains to the sign aspect of language use, to the manner of expressing certain selected content elements.

The second conception of grammatical elements, one that takes account of both content and expression, is ultimately also found (in implicit form) in the founder of Prague typology himself, Vladimír Skalička. When Skalička, for example, criticizes earlier theories of case, he accuses them, among other things, of failing to consider the fact that “cases are formations which are in permanent contact with neighbouring formations. What one language expresses by means of case, another expresses by prepositions, adverbs, derivative prefixes, number and other means” (Skalička 1950: 47).

GEDANKEN ZUR GRAMMATIKALISIERUNG | 123

Case is thus regarded as a certain form of expression that contains certain semes (content elements) and meets certain criteria with regard to its structure. Whatever fails to meet these criteria is not a case (the form of expression expresses the unit of meaning and at the same time distinguishes it from related formations that are not cases). According to the first view of grammatical elements, by contrast, case is a certain content element which—provided, of course, that it exists in the content plan of the language in question—can be expressed differently in different languages. Incidentally, Skalička is also compelled to postulate the concept of case functions, which may be “shifted to prepositions or other formations” (ibid., 48). This vacillation is also found elsewhere in Skalička. In describing the polysynthetic type, the general distinction between the basic and auxiliary linguistic elements is introduced: the former bear meaning, while the latter express the relations between the former. However, auxiliary elements exist only if an independent expression is assigned to them, and the polysynthetic type therefore has no auxiliary elements (cf. Skalička 1979: 186, 189, 195). Elsewhere, by contrast, it is also said that the polysynthetic type has a grammar (i.e. auxiliary elements), but that these are expressed by lexical elements, specifically by their order and by lexical elements occurring in grammatical functions (cf. Skalička 1955a: 16f., 18f.).

For Skalička, as for Popela, a grammatical element is thus only an element that has an expression that is more or less discrete, i.e. is a sign; moreover, Skalička also takes further properties of the expression into account in assessing a sign's grammaticality (cf. Skalička 1960a; 1962a: 125f.; 1979: 83ff.). However, there is a fundamental need to define the concept of grammatical functions, since, on the one hand, it would be impossible to speak of semantemes fulfilling grammatical functions in the polysynthetic type, while, on the other hand, it would be unclear which functions (meanings) should be examined with regard to their grammaticality.

Skalička (1965) bietet auch tatsächlich solche Kriterien zur Identifizierung von grammatischen Funktionen (Bedeutungen). Gemäß ihm haben die grammatischen Elemente (Zeichen) dreierlei Bedeutung: syntaktische, abstrakte und einreihende. Durch ihre einreihende Funktion gliedern die grammatischen Elemente das durch sie bestimmte lexikalische Element in eine Klasse ein. Auf dieser Funktion gründet sich die konstitutive Eigenschaft der grammatischen Kategorien, und zwar die Obligatorität ihrer Glieder: Jedes lexikalische Element, für das die entsprechende grammatische Kategorie relevant ist, muss mit einem Element der betreffenden Kategorie versehen werden. Die syntaktische Bedeutung oder Funktion eines grammatischen Zeichens besteht darin, dass es auf andere Benennungen verweist, dass es bestimmte lexikalische Elemente mit anderen verbindet; zugleich kann es auch den Charakter dieser Beziehung angeben. Schließlich macht die abstrakte Bedeutung ein grammatisches Zeichen zu einem Teil der Benennung: Das grammatische Zeichen modifiziert die Bedeutung des lexikalischen Elements, das es bestimmt. Sekundär kann allerdings das grammatische Element durch seine abstrakte Bedeutung auch die syntaktische Funktion erfüllen, was bei der Erscheinung der sog. Kongruenz der Fall ist. Es ist wichtig zu bemerken, dass der abstrakte Charakter der

124 |

BOHUMIL VYKYPĚL

of grammatical elements is relative, i.e. they are more abstract than semantically related lexical elements. Abstractness is understood as the capacity to encompass broader semantic fields. The difference in abstractness between a lexical and a grammatical element with the same semantic basis is therefore to be understood as gradual—in contrast to the first two properties, which are based on a privative opposition.

An instructive example of a non-formalized theory of grammaticalization, that is, one constructed not as an immanent theoretical account but for a specific purpose (namely, to explain the development of Indo-European inflection), can be found in the work of the Indo-Europeanist Adolf Erhart. The example is instructive because one can take Skalička as a starting point and arrive at Skalička again in the end, from a perspective that sheds explanatory light on the entire concept of grammaticalization. (For Erhart's conception of grammaticalization, see Vykypěl 2004a: 153–156, with relevant bibliographical references.)

Erhart's formulation of the properties of grammatical elements is similar to Skalička's (1965). Grammemes are dependent and always bound to particular lexemes (i.e. are potentially classifying). Grammemes are substantially simpler in meaning than lexemes (i.e. they are more abstract) and may also be entirely formal, i.e. designate only a syntactic relation (i.e. they may be syntactic). The fact that the inventories of grammemes are usually closed is also connected with their simplicity. According to Erhart, the change of lexical elements into grammatical ones, i.e. grammaticalization, has three phases: 1. loss of independence (the element begins to bind itself to another element), which may be accompanied by simplification of the element's semantics; 2. generalization, i.e. broadening and obligatorification; 3. complete formalization, i.e. the element has only a syntactic meaning.

These properties, which concern the content plan or can at least be related to it, can also be connected with properties concerning the expression of grammatical elements (signs). The abstract character of grammatical elements (to which syntacticity can also be assigned) favors the limited extent of their means of expression. This limited extent, in turn, favors both the cumulation of grammemes within a single sign and the synonymy and homonymy of grammatical signs. The obligatory character favors equipollent expression, i.e. the absence of so-called zero formatives, including so-called fixed word order, and also the bound expression of grammatical elements, i.e. expression by affixes or endings rather than by words. Congruence also belongs here: when congruence is more prevalent in a language, this means, among other things, that every word is obligatorily accompanied by certain forms of expression (endings), including words that do not necessarily need them. The semantically abstract character and the limited extent of the form of expression of grammatical elements can be subsumed under the common concept of simplicity, and another property of grammatical elements posited by Erhart can also be associated with simplicity, namely “uniformity of formation” of the forms of expression, i.e. these should be formed only by

GEDANKEN ZUR GRAMMATIKALISIERUNG | 125

one type of sign (by words, affixes, endings, or introflexion, but not in a mixture).

On the basis of the properties of the form of expression, Erhart even posits three phases leading to complete grammaticality, in which the meanings under observation—corresponding to the meanings of certain lexical elements—are expressed in different ways: 1. the cryptogrammatical phase (expression through word order or freely used words); 2. the pregrammatical phase (word order and affixes); 3. the semigrammatical phase (affixes, but also zero formatives; no congruence or cumulation). It is obvious that these phases can be compared with Skalička's language types: the cryptogrammatical phase with the polysynthetic or isolating type, and the pregrammatical and semigrammatical phases with the agglutinating type. In any case, one can observe that Erhart's properties of grammatical elements strikingly resemble those of the inflecting type (in Skalička's typology). One may therefore also agree with Tomáš Hoskovec (2001: 12):

“Erhart counters the aporias of Indo-European reconstructions with a distinctive typological apparatus. Its hierarchical conception is provocative, placing the Indo-European languages at the top of the grammaticalization pyramid and thus, in effect, of the intellectualization of semantic relations in language. Perhaps we can forestall misunderstandings if we bear in mind that we view the world's languages through a magnifying glass that brings everything into focus on the Indo-European languages, which are closest to us and which we know best: we always view the world from somewhere.”

It must be said, however, in fairness that even within an understanding of grammaticality or grammaticalization that locates its criteria in both the content and expression plans, i.e. examines the grammaticality of signs, criteria can be established that are not tied to the concrete character of the signifiant and are therefore correlatable with Prague typology. There are two such criteria, and they have already been mentioned. The first is uniformity in the formation of forms of expression: it makes no difference here whether grammatical functions are expressed by words, affixes, endings, or introflexion; only the expression of the individual members of the category in question must be the same. The second criterion is the absence of zero formatives: no member of a grammatical category should be expressed by a zero formative; in other words, the individual expressions of the members of a grammatical category should be equipollent to one another, not privative (on this, see also Skalička 1960a and Vyky̌pel 2005: 210).

3. I can finally return to Christian Lehmann and his thoughts on grammaticalization¹.

¹ Here I will essentially confine myself to this book by Lehmann, although, of course, his other writings on grammaticalization are also important and illustrate what I have said so far and

126 |

B O H U M I L V Y K Y P Ě L

3.1. It is obvious that Lehmann’s grammaticalization belongs to the second of the two groups mentioned above (§ 2), since what is to be investigated with regard to grammaticalization is signs (cf. Lehmann 2002: VII, 109f.). It also follows that in this case, too, one must proceed as with the other grammaticalization theories in this group: first, the relevant content elements must be established, and only then can the signs into which these content elements enter be examined with respect to their degree of grammaticality². It is important that in the second step both sides of the linguistic sign be taken into account, and thus both the meaning and the expression of the content elements in question be included among the grammaticalization parameters (cf. Lehmann 2002: ch. 4). In Hjelmslev’s terms, certain content elements are selected from the structure of language, and their fate in language use is traced, both in its substance aspect (the meaning of the content elements that form the signifié of a sign) and in its sign aspect (the expression of the content elements within a sign).

At first glance, the first step—the establishment of those content elements that determine which signs are examined with regard to grammaticalization and which are not—might not seem indispensable: one simply applies the semantic criteria and the expression criteria to the signs of the language and determines which signs meet them, and to what extent, and which do not. There are, however, several reasons for the opposite approach.

First, there is the point already hinted at above (§ 2.2.): in the polysynthetic type, or in languages with a clearly dominant polysynthetic type, it would be impossible or difficult to distinguish between grammar and lexicon. For example, it would not be clear when a sign functions only as a verb and thus does not enter the grammaticalization channel, and when it also functions as an adposition and is thus at the beginning of grammaticalization (Lehmann 2002: 94; cf. also Haspelmath 1998: 328f.). Everything would potentially be grammatical or, in Erhart’s words, “cryptogrammatical”; but if something is cryptic, one cannot see it (and if one cannot see something, one can hardly speak of it; and what one cannot speak of, one must remain silent about, as Wittgenstein would have it…).

A further argument for the conviction that one must first determine which content categories are relevant to investigating, with respect to grammaticalization, the signs whose components they help form is provided by the concepts of renovation and innovation (Lehmann 2002: § 2.4.). Renovation presupposes the existence of a shared content category, whose members enter into signs located at different points on the grammaticalization scale. The starting point of in-

what else I want to say. For the main subject of this contribution—the attempt to provisionally integrate grammaticalization theory into a broader theoretical framework—it should, however, suffice to refer to this major work.

² I understand signs bilaterally, as an inseparable connection between elements of the expression plane and the content plane (the elements of the two planes of language, as two parts of the structure of language, are potential components of linguistic signs, which belong to language use). Therefore, one can say that a content element (or, of course, an expression element) enters into a linguistic sign by forming (part of) it. (See also below § 4.3., Remark I.)

Gedanken zur Grammatikalisierung

| 127

Innovation, in turn, constitutes the emergence of a grammatical content category (or, more precisely, of a content category whose members then co-constitute the signs followed with respect to grammaticalization), and one must therefore have a criterion by which to decide on the emergence of the category and its separation from the other content categories: It is true that “Latin had no grammatical category which corresponded to the French articles, so that nothing has been renovated by these” (Lehmann 2002: 18), for Latin had only a corresponding lexical category; the question nevertheless arises of how a lexical category such as the Latin demonstrative pronouns and a grammatical one such as the French article are to be distinguished. If such a question did not seem relevant or important, it would also be unclear what one is talking about, and why, when lexical and grammatical categories are discussed.

The need for a content link between the variously grammaticalized signs is also addressed by Lehmann (2002: 111f., 150), when he speaks of the requirement of functional similarity between the signs being compared or of their common functional basis. It should be added that this applies both intralinguistically, when comparing signs of one language with respect to grammaticalization, and cross-linguistically.

3.2. It is also generally true that comparison can be made in two ways (cf. Vykypěl 2005: 195, 286): either on the basis of an extralinguistic axiomatics, Hjelmslev’s substance, or on the basis of an intralinguistic axiomatics, i.e. a general, “panchronic” theory. The general theory that serves as the necessary point of departure in the case of grammaticalization is a theory of content. This can also be expressed by saying that the grammaticalization theory, a conception belonging to the second of the groups mentioned above (§ 2), presupposes one of the grammaticality theories from the first of the two groups above. This presupposed theory may be Sgall’s functional-generative description with its tectogrammatical level, Mathesius’ functional linguistics with its two acts, or Hjelmslev’s linguistic theory with its plane of content.

The need for a general theory of content can be seen in several examples.

3.2.1. One may, for example, think that the clarity of relations within a category, which appears in the grammaticalization parameter of paradigmaticity (Lehmann 2002: 118), fundamentally depends on the clarity of its description (to quote again from the Hjelmslev primer, “if a thing looks complicated, it is chiefly because it is looked upon in a complicated way”; Hjelmslev 1973: 120). The same applies to the homogeneity of the category as “a certain amount of similarity among its members and of regularity in their differences” (Lehmann 2002: 120): for example, with Hjelmslev (1959: 157), one may find clear basic meanings even where they have traditionally not been seen³.

³ Incidentally, the notion of basic meaning also touches on another problem. The difference between basic meaning and total meaning can certainly be interpreted (with Lehmann 2002: 114) as a difference between two designations and understood metaphorically:

128 |

B O H U M I L V Y K Y P Ě L

3.2.2. Another point is paradigmatic variability (Lehmann 2002: § 4.2.3.): the more grammaticalized an element is, the less the freedom to choose among the members of a category depends on the speaker’s naming intention. This, however, should not be assumed without further ado. First, two kinds of relations must be distinguished: those between categories and those between elements. As Hjelmslev has already shown, for example, there can be interdependence between the category of number and that of case, but an individual member of the number category and an individual member of the case category can be freely combined with one another (cf. Vykypěl 2005: 69f.). From the point of view of language structure (linguistic structure), one may assume that relations regulated (obligatory) to varying degrees exist only between categories, whereas the regulation of combinations between elements from different categories belongs to language use. Language use can make some of these combinations between elements more or less obligatory and exclude others. From the point of view of language structure, however, this rests on the properties of the elements determined by the internal structure of the respective categories or—if these properties are manifested—on their semantics (in the plane of content). Thus, phenomena such as government or valency consist in one content element being “required” by another, but this “requirement,” as a relation between elements (not between categories), may be regarded as established by a convention of language use: combinations between content elements that do not meet the requirements of government or valency are considered “nonsensical,” i.e. as forming, or being capable of forming, no sense or part of the (current) world (the world as meaning is shaped in various ways by the elements of language and their various combinations)⁴.

It is claimed of the Latin preposition ad: “Ad can only have a complement in the accusative. While this would be expectable on semantic grounds in sentence as E72.a [Caesar legatos ad Hannibalem misit], it would be inappropriate in b [Caesar ad Cannas pugnam commisit] if the case of an NP dependent on a preposition had an independent semantic function” (Lehmann 2002: 82). However, the combination of ad with the accusative in the second sentence can be considered “inappropriate” only from the point of view of the conventions of another language, in which the connection between the equivalent of ad from the first sentence and the equivalent of the accusative from the second sentence is considered “nonsensical.” (Its counterpart in the plane of expression, incidentally, is the so-called “unusual” phoneme combinations: they are usually unusual from the point of view of another phonological system.)⁵

Basic meaning is somewhat more concrete (basic-), total meaning somewhat more abstract (total-). However, one should perhaps also not forget that these concepts belong to two different theoretical systems, namely Hjelmslev’s and Jakobson’s, and are therefore not, strictly speaking, directly comparable (cf. Vykypěl 2005: 223–225).

4 For a semantic definition of the phenomenon of government, see also Isačenko (1948), from a different point of view.

5 Here one might find a point of contact between “American” and “European” grammaticality. “American” grammaticality corresponds approximately to the correctness of linguistic constructions. This correctness may be associated with the usualness of a kind

GEDANKEN ZUR GRAMMATIKALISIERUNG

3.2.3. It seems that here we encounter an echo of Jespersen’s squinting grammar. A similar impression arises elsewhere, too. The combination of inter and facere in interficere is supposed to yield an irregular meaning, namely ‘kill’ (Lehmann 2002: 88). However, the meaning is irregular (or “nonsensical”) only from the standpoint of another language convention, in which the combination of the approximate equivalents of inter and facere forms no sense: for example, there is no German word zwischenmachen. interficere might, however, seem less irregular from the standpoint of Czech, for example, which has the verb oddělat ‘interficere’, formed from od ‘off’, which is also a preposition, and dělat ‘facere’. Similarly, it is not clear why, for example, the so-called serial verbs form complex lexical units that may be discontinuous, as in Efik dá dí in the sentence dá ìkwá dí! ‘bring a knife!’ (Lehmann 2002: 94). Hopefully, this is not because dá dí can be translated into English with a single verb⁶. One presumably does not distinguish a relational noun from a transitive verb in polysynthetic languages by translating the former into a Western European language with a noun and the latter with a verb (Lehmann 2002: 95).

3.2.4. The potential irregularity in question is allegedly a characteristic of word formation in comparison with grammar. Another characteristic is said to be the fact that word-forming elements do not attach to all lexemes, but display different degrees of productivity, and that they attach to different lexemes with different results (Lehmann 2002: 89). The latter is obviously another aspect of irregularity: lexical meaning is not, or is difficult to be, predicted on the basis of word-forming meaning or word-forming structure—at least from a certain point of view. One may note, however, that this can also occur with some elements usually regarded as grammatical (cf., for example, the Ancient Greek genera verbi!).

The former (the different degrees of productivity) is, in turn, a case of neutralization. In general, the question of the obligatoriness of an element, which stands in an inverse relationship to the degree of its paradigmatic variability and also in a direct relationship to the degree of its integration into a paradigm, can be correlated with the question of neutralization. The fact that the lower the degree of contextual restrictions on the occurrence of an element, the more grammatical it is (cf. Lehmann 2002: 10f., 125f.), means that an element is grammatical to the extent that it does not enter into syntagmatic combinations that cause its syncretism with another element of its category (for the concept

and manner of shaping the world, and this correctness seems most discernible or conspicuous precisely in elements that are grammatical in the “European” sense. One may argue that the two views consider roughly the same thing, but from different standpoints: the “American” from the operational one (the identification of relations in the language system, namely, by identifying possible combinations of elements), the “European” from the formal one (the definition of elements).

⁶ Vgl. auch Haspelmath (1998: 332) zu Lehmann: „The main reason why [japanisch – B.V.] ue-ni […] is taken as one constituent seems to be that semantically it corresponds to the single constituent on in English.“

129

130 |

BOHUMIL VYKPĚL

der Neutralisation im Inhaltsplan vgl. auch Trnka 1982: 149–155, 311–314, 330–332, 336–339, 356–360).

Several types of neutralization can, however, be distinguished in this context. The conventional neutralization is that brought about by a syntagmatic combination between two grammatical categories and resulting in the neutralization of a contrast between two or more members of one of the categories (e.g. case neutralization in the plural). A total syncretism between the members of a content category, however, may also be considered neutralization when it is brought about by a syntagmatic combination of the content category with a lexeme or a lexical category and expressed by zero. Categories affected by such neutralization are usually called word-forming. Because of the zero expression, however, neutralization and the absence of the content category are difficult to distinguish (one may also speak here, in distributionalist terms, of defective distribution). Yet another type of neutralization is likewise based on a syntagmatic combination of a content category with a lexeme or a lexical category, but the syncretism is not expressed by zero. This applies, for example, to number with personal names. A personal name combined with the plural may, however, also be considered a “nonsensical” combination, and such a combination—like the “incorrect classification” of a word with classifiers discussed by Lehmann (2002: 54)—can be used “for jocular or derogatory effects,” since the speaker is always free to choose among the members of an obligatory category. The obligatoriness of a category means the obligatoriness of that category, not of its members.

As regards the role of word-forming elements in grammaticalization (which seems to be questioned; cf. Lehmann 2002: 89ff.), it should be noted that it is entirely conceivable that word-forming formatives may be integrated into a unified paradigm and thus grammaticalized. Helena Kurzová (1993) has demonstrated this, with general implications, using Greek and Latin as examples. She speaks of the paradigmatization of word-forming elements, that is, their integration into a paradigm. (For a general discussion of the difference between word formation and grammar and the potential transition of one into the other, see also Dokulil 1968, 1994.)

3.2.5. The problem of distinguishing between word formation and grammar is in fact the problem of the paradigmatic definition of the word: When is something merely another form of a single word, and when is it already a new word? Or, in other words, when does a set of paradigmatically related expression forms constitute a single whole, a word, and when, by contrast, does it constitute two or more such wholes? The solution is indeed simple: the condition that a series of signs with certain—grammatical—content elements not be repeated in the set in question is (implicitly) posited here for the set to count as a single word. Clearly, then, the distinction between word formation and grammar presupposes a prior definition of grammatical elements in the content plane.

3.2.6. Structural obligatoriness would thus, according to what was mentioned above (§§ 3.2.2., 3.2.4.), represent the interdependence between categories. The actual problem that

GEDANKEN ZUR GRAMMATIKALISIERUNG | 131

What arises here, however, is the (indicated) operational question of how, in the case of zero stress, to distinguish between the zero formant of a constituent and the absence of the category. Here we arrive at a general question of language description, or of the description of content, which is not based on expression.

Another general problem, connected with the foregoing, emerges in the notion of “structural scope,” or syntagmatic weight (Lehmann 2002: § 4.3.1.), namely the question of what the description of the content plane should be based on: should one proceed from the analysis of signs or from the analysis of the relations among content elements (meanings)? The parameter of syntagmatic weight is based on the extent of the unit determined or co-formed by the grammatical element. It is not, however, immediately clear how to establish what is determined by the respective grammatical element. Should one take as one’s starting point the sign structure of the sentence and observe where, with which sign, the sign expressing the element in question occurs? Or should one proceed from the pure content plane, or the (Sgallian) level of sentence meaning, and assign no decisive role to signs? For example, the sign for number often occurs with the sign for the verbal lexeme, but it does not determine the verbal lexeme; rather, it determines the verbal person (with respect to its consistency). Similarly, Hjelmslev (1959: 156, 165–191) famously argued and demonstrated in detail (using the so-called nominal sentence as an example) that the so-called verbal grammatical content elements always determine the whole sentence, even when they are usually expressed by signs that are more or less closely connected with the sign for verbal lexemes.

A reverse case is represented by constructions such as English all good old men’s works. Hjelmslev (1959: 199–210) regards all four “lexical” content elements, expressed by the forms of expression all, good, old and men, as determined by the “grammatical” content category of case. The only difference is that with all, good and old there is syncretism of all the members of the case category, expressed by the position of the corresponding lexical signs, the “fixed word order” (i.e. a segmental zero), whereas with men the genitive is present and expressed by the sign -s. Thus, from the viewpoint of the content plane, the English construction is interpreted in the same way as its Latin equivalent opera virorum omnium bonorum veterum, in which so-called agreement comes into play (see also below § 3.3.2.). Apposition, as the fixed position of a sign containing a “lexical,” determined content element, and agreement, as the repetition of a sign containing a “grammatical,” determining content element, express the same thing: a dependency relation between content elements, which may itself be regarded as a content element.

Two phenomena that differ at the level of signs, agreement and apposition, are interpreted in the same way in the content plane, namely as a kind of dependency relation. In the case of agreement and apposition, Sgall and Hjelmslev thus also agree (see Vykypěl 2005: 236–239): phenomena that differ at one level, on the morphematic level or in the sign aspect of language use (are expressed differently), are interpreted by them as identical at the tectogrammatical level or in the content plane of language structure (where Hjelmslev sees dependent repeated

132 |

BOHUMIL VYKYPĚL

content elements or syncretisms of members of content categories, Sgall sees functors of the tectogrammatical level).

Hjelmslev goes further, however, and argues that with all, good and old in the English construction mentioned above there are also syncretisms of all the members of the content category of gender and of that of number (and that they too are expressed by “fixed word order”). The syncretism of numbers is then resolved with men, and the plural is in this case expressed by introflexion, whereas the syncretism of genders in English is resolved only in connection with third-person personal pronouns—which, however, is precisely the reason for positing the content category of gender in English, since otherwise total syncretism would be equivalent to the absence of the category.

I do not wish to claim that these specific observations about English are readily acceptable; rather, I wish only to sketchily show that the parameters of grammaticalization cannot be regarded as informative for language description from a certain other point of view.

3.2.7. The example of the resolution of gender syncretism in connection with third-person personal pronouns in English also teaches us something else: one and the same content element of one and the same content system can be expressed differently in different combinations. Thus, in English the “nominative” (subjective) and the “accusative” (objective) are expressed by “fixed word order” in combination with so-called nouns, but by introflexion or suppletion in combination with personal pronouns. One would say that the respective expressions are grammaticalized to different degrees. But can signs with the same content element be grammaticalized differently? And are we in fact dealing here with the same content element?

3.2.8. Another aspect of the question (posed above in § 3.2.6.) of how an analysis of content should be constructed concerns the degree of relevance of the formants (signifiants) of the content elements being described. For example, the question arises as to why the Latin prepositions causā and gratiā are supposed to be ablatives of causa and gratia (Lehmann 2002: 70). The identification of the content elements contained in the prepositions with the ablatives is made implicitly on the basis of the identity of their formants.

In the same place, a criterion is also given for distinguishing between a form of a noun and a preposition: when a noun form becomes a preposition, it can no longer be expanded by an attributive modifier. One may initially observe that this presupposes a change in the position of the content element in question within the structure of the content plane, as well as a non-gradual opposition between lexical and grammatical elements in content (see above § 3.1. and below § 3.4.): the change consists in the fact that the element can now determine, but can no longer be determined, or that it can be dependent, but not independent; that is, it becomes a converted morpheme in Hjelmslev’s sense (1959: 192–198). Besides prepositions, Hjelmslev counts personal pronouns, for example, among converted morphemes, and thus I come to the second point I wish to make. Personal pronouns, like prepositions, are grammatical content elements, or Hjelmslevian morphemes—in this case, converted morphemes—and they

GEDANKEN ZUR GRAMMATIKALISIERUNG

133

therefore do not ordinarily admit expansion by an attribute. In unusual cases, however, they can be developed by an attribute: the eternal I or the eternal before (over whose semantics, for example, some linguists argue eternally). Are I and me and before and before the same if they have the same formant, as was the case with the prepositions and ablatives causā and gratiā? (On the question of identifying content elements in general, see Vykypěl 2005: 87–90, 133–135, 344–346, 349–353.)

3.2.9. Finally, the whole of chapter 3 of Lehmann (2002) can be viewed as a collection of examples of which parts of the content plan should be examined with regard to the construction of the signs in which they occur. Here one finds conventional grammatical content elements such as mood (o.c., § 3.1.3.), tense and aspect (o.c., § 3.1.4.), diathesis (o.c., § 3.1.5.), article and person (o.c., §§ 3.2.1., 3.2.2.2.), number (o.c., §§ 3.3.1.1.-2.) or case (o.c., § 3.4.), but also less conventional ones such as emphasis and negation (o.c., §§ 3.1.5., 3.2.2.3.). For the determination of “grammatical” content elements within the framework of Hjelmslev’s conception, see Hjelmslev (1959: 152–164) and Diderichsen (1966: 148–168); for the enumeration of functions and grammatemes of the tectogrammatical level of Sgall’s functional-generative description, see Sgall, Hajičová, Panevová (1986: 154–174) and Panevová (1994)⁷.

In general, however, no particularly strict or clear criteria are offered for delimiting specific domains in which grammaticalization occurs. It must be admitted, though, that the same applies to definitions of the components of the tectogrammatical level, where one finds fairly simple lists of elements. Thus, incidentally, one returns to the question of defining grammatical elements in contrast to lexical ones, or of defining the domain of grammatical elements in contrast to that of lexical elements (§§ 2.2., 3.1.). The question may also arise whether every lexical element can become grammatical or be subject to grammaticalization. On the one hand, it seems that this is not the case, since the fate of a category is “largely predetermined” once its meaning and syntactic function are known (Lehmann 2002: 22). This would mean that some categories are more prone to grammaticalization than others. Alongside this, however, it is asserted (ibid.) that grammaticalization is omnipresent and not specific to any particular area of language.

3.2.10. Despite these problems, allow me nevertheless to make a few further remarks on grammaticalization domains.

Interessant ist der Fall der Kopula (Lehmann 2002: § 3.1.2.). Die Kopula sei „a word which turns a nominal into a predicate“ (Lehmann 2002: 23). Diese Funktion muss indessen nicht unbedingt ein Wort erfüllen, wie die türkischen prädikativen Suffixe zeigen, und sie kann sogar durch die feste Wortfolge ausgedrückt werden, wie wiederum die rei-

⁷ In this connection, it may also be casually noted that the distinction between relational and absolute grammatical elements (Lehmann 2002: 115) can be compared with the distinction in the framework of Mathesius’s theory between elements that primarily serve the syntactic act (case, person) and those that primarily serve the onomatological act (number, tense). Elements such as the article can primarily serve both acts (cf. the generic meaning of the article and its role in the so-called functional sentence perspective).

134 |

B O H U M I L V Y K Y P Ě L

… nominal sentences of the type found, for example, in Hungarian (see also Giger, Vykypěl 2001: 72f.). This means, on the one hand, that the copula can also be regarded as a content element, namely one belonging to the category of Hjelmslev’s extensive, that is, roughly verbal, “grammatical” content elements; and, on the other hand, that the expression of this function may vary, for example depending on which language type predominates in the language in question. What is important here is that the point of departure must be a definition of this copula function or this copula content element, from which one can then proceed to observe the signs in which it occurs with regard to their grammaticalization or typological diversity.

Noch interessanter scheint die Nominalisierung und Attribution (Lehmann 2002: §§ 3.3.2.–3.; vgl. auch Lehmann Ms.a: §§ 5.–6.). Über die mögliche Interpretation der Attribution und der damit eventuell verbundenen Kongruenz war schon oben (§ 3.2.6.) die Rede und vgl. noch unten (§ 3.3.2.). Was die sog. Nominalisierung angeht, so lässt sich aus der Hjelmslevschen Perspektive behaupten, dass in der Erscheinung dieser Dependenz zwischen zwei oder mehreren „verbalen Konzepten“ ein Inhaltselement namens Konnektiv auftritt, welches eben diese zwei Konzepte als komplexe Inhaltselemente oder Komplexe von Inhaltselementen verbindet (vgl. Hjelmslev 1959: 165–191, Vykypěl 2005: 76f., 232f.). Der Ausdruck dieses Inhaltselementes kann verschieden sein. 1° Es kann sich nur um die „feste Wortfolge“ handeln, d.h. eine bestimmte Aneinanderreihung von „Haupt-“ und „Nebensätzen“, aus der dann die gegenseitige Beziehung der Sätze (d.h. die Tatsache, welcher von welchem dependent ist) hervorgeht. 2° Es kann ein für diese Funktion nichtspezialisiertes Wort auftreten, welches in anderen Kontexten eine lexikalische Funktion erfüllt. 3° Es kann ein für diese Funktion spezialisiertes Wort auftreten. 4° Schließlich kann auch ein Affix auftreten. 1° kommt beispielsweise im Chinesischen vor, in dem die Temporal- und Konditionalnebensätze vor dem Hauptsatz stehen wie allgemein das Determinans vor dem Determinatum, die Objektnebensätze dagegen nach dem Hauptsatz wie allgemein das Objekt nach dem Subjekt und Prädikat, ohne dass in beiden Fällen der Nebensatz durch ein Zeichen bestimmt zu werden braucht. 2° kommt beispielsweise im Vietnamesischen vor, in dem Finalnebensätze durch die Wörter để oder cho determiniert werden, welche auch ‘lassen, stellen’ resp. ‘geben’ bedeuten können. 3° ist der Fall der anständigen Konjunktionen und 4° begegnet man etwa bei allen sog. Partizipien, Infinitiven, Gerundien etc. 1° und 2° lässt sich im Rahmen der Prager Typologie als polysynthetisch, 3° als isolativ und 4° als agglutinativ bewerten. Für den flektierenden Sprachtyp würde man mit dem Ausdruck durch eine Endung rechnen, Skalička (1979: 173–179) behauptet jedoch, der flektierende Typ löse das Problem des zweiten Verbs (welches dem Problem der Nominalisierung entspricht) durch reich entwickelte Nebensätze. An einer anderen Stelle haben wir, Markus Giger und ich, nichtsdestotrotz zu zeigen versucht, dass diese Behauptung – wie andere sog. syntaktische Eigenschaften im Rahmen von Skaličkas Typologie – nicht problemlos ist (vgl. Giger, Vykypěl 2001: 72–79).

In any case, we see here again (as in the case of the copula) that the stages of grammaticalization can be correlated approximately with the types of the Prague typology. This, however, brings me to the subject of the following section.

GEDANKEN ZUR GRAMMATIKALISIERUNG

135

REMARK: Before that, however, on this occasion I will allow myself to add a brief typological digression.

As soon as the dependency of “sentences” receives a more or less discrete expression, “fixed word order” (the order of morphemes) can be freed up to express the actual articulation (the functional sentence perspective). The same applies to the constituents of the sentence themselves. Many examples can be cited. When a Chinese causal subordinate clause is marked by yīnwèi ‘because’, it can also follow the main clause, especially if it does not represent the topic. In Indonesian, a word serving the function of the indirect object must occur after the verb and before the direct object; if, however, it is determined by a preposition, it can also occur after the direct object. The same is well known to hold for the indirect object in English. An example from English is given by Sgall (1988: 465): Mary moved from Chicago to Philadelphia vs. Mary moved to Philadelphia from Chicago. These phenomena may be taken as an occasion to reflect on the typological evaluation of the expression of actual articulation (the functional sentence perspective).

As mentioned above (§ 2.1.), Sgall counts actual sentence articulation, together with functors and grammatemes, as part of the grammatical component of the tectogrammatical level, whose expression is consequently postulated as the dominant feature or basic property of the five language types. Certain ambiguities arise here⁸. Thus, expression through variation of lexical morphemes, that is, through endings, is the dominant feature of the inflectional type, whereas actual articulation is expressed through the ordering of lexical morphemes, which is, however, regarded as the dominant feature of the polysynthetic type. The examples from Chinese, Indonesian, and English cited above would mean that when functors or grammatemes (here, case and connective functions) are expressed through the ordering of morphemes, actual articulation must be expressed differently, for example through intonation or certain constructions; whereas, if functors or grammatemes are expressed by their own morphemes, actual articulation is expressed through the ordering of morphemes. One may assume that there is a kind of compensatory relationship between the expression of functors and grammatemes (“proper grammar”), on the one hand, and that of actual articulation, on the other⁹. The question, however, is whether this compensation belongs to the constructs of the types or whether it is a matter of individual languages, which, as is well known, are regarded in Prague typology as hierarchically organized combinations of properties of different types. One may refer here, for example, to the famous Japanese topicalization suffix -wa, found in (allegedly) dominantly agglutinative Japanese: here, then, we have the same (in this case, agglutinative) solution for the domain of grammatemes and functors and for that of—

⁸ Zur aktuellen Gliederung vom typologischen Gesichtspunkt aus vgl. auch Sgall (1982) und Sgall, Hajčová, Panevová (1986: 241–244).

⁹ For the concept of compensation in the deduction of properties of language types, see Skalička (1966: 27).

136 |

BOHUMIL VYKYPĚL

…with the current articulation. Conversely, one might believe that the expression of the current articulation through the order of morphemes in predominantly inflecting languages constitutes a feature of the polysynthetic type. This brings us back to the twofold theoretical uncertainty of Prague typology: How are the types actually to be constructed, and how is the typological dominant of a language to be established? (See also Vykypěl 2005: 197–203.) A possible solution to the second question is to hierarchize the language system according to its relevance for determining the typological dominant of an individual language: The most important would be, for example, the expression of functors and grammatemes, or that of functors followed by that of grammatemes; next would come the expression of current articulation; further on would be the areas of naming (word formation), semantics, syntax, and phonology.

3.3. The theory of grammaticalization, i.e. a particular assessment of certain signs, thus falls into two parts. In its “content aspect,” it corresponds to a theory of content (see above § 3.2.). In its “sign aspect,” it can in turn be compared with Hjelmslev’s description of language usage (in both its substance and sign aspects; see above § 3.1.), with Mathesius’ morphology, or with Sgall’s description of the morphemic level. Most directly, however, it can be compared with Prague typology, which can itself also be viewed as a description of types of Hjelmslevian language usage, of Mathesiusian morphology, or of the connection between Sgall’s tectogrammatical and morphemic levels.

First, two general properties shared by grammaticalization and Prague typology can be noted (and Markus Giger has drawn attention to them). If, provisionally and approximately, individual phases of grammaticalization are equated with the Prague types, two things become apparent: the fact that signs with different degrees of grammaticalization occur in one and the same language corresponds to the Prague thesis that an individual language is a combination of properties of several language types; and the fact that the parameters of grammaticalization vary in degree among individual signs corresponds to the possibility of combining properties of several language types within a single sign.

Moreover, concrete examples can be given of the correlation between phases of grammaticalization and the Prague language types. The correspondences are numerous (see also Sgall 1992a: 205f., 1993a: 323, 1999: 31; Giger 2002: 46); I will limit myself to just a few of them¹⁰.

3.3.1. For example, highly grammaticalized cases (Lehmann 2002: 76f.) exhibit features of the inflecting type: their formatives cumulate several functions, are shorter, enter into agreement, undergo fusion, and are accompanied by stem alternation.

¹⁰ For typology (though predominantly its pre-structural precursor to Prague typology) from the perspective of grammaticalization theory, see Lehmann (1985; 1986; 1987).

GEDANKEN ZUR GRAMMATIKALISIERUNG | 137

Perhaps the most striking correlation is in the area of syntagmatic cohesion, or bondedness (Lehmann 2002: § 4.3.2.). The stages of fusion between the lexical stem and the grammatical element (p. 132) correspond exactly to those of Prague typology; the indistinct boundary between morphemes and the cumulation of functions in morphemes as features of advanced grammaticalization (p. 137) correspond to properties of the inflecting type. 3.3.2. The section on bondedness also discusses coordination syntagms (Lehmann 2002: 133f., 135f.). Coordination syntagms contain content elements that are equal in status relative to one another. When they are determined by another content element, the latter can either be expressed repeatedly with each of the content elements it determines or be expressed only once. The first mode of expression is considered a feature of greater grammaticalization of the determining sign. Repetition of a sign is also the essence of what is called agreement (see Skalička 1937): by repeating a sign that contains a particular content element, the dependence of one lexical content element on another is expressed. Agreement as the expression of such a dependency relation occurs in the inflecting (or introflecting) type, whereas in the other types the relation is expressed through apposition, i.e. fixed word order. In addition to what was already mentioned above (§ 3.2.6.), namely that two phenomena that differ at the level of signs—agreement and apposition—can be interpreted in the same way in the content plane, it becomes clear here that the expression of the relevant content relation is pertinent to both typology and grammaticalization: agreement and repetition of a sign are inflectional and more grammaticalized, while apposition and non-repetition of a sign are non-inflectional and less grammaticalized. Once again, we see development toward the inflecting type as a strengthening of grammaticalization.

3.3.3. The phenomena discussed (§§ 3.2.6., 3.2.10., 3.3.2.)—attribution, predication, nominalization, or agreement—can be described as particular syntagmatic relations in the content plane, or as reflections of these relations, or as content elements through which these relations are formed. The “lexical” content elements thus connected to one another form different groups depending on the syntagmatic relations into which they enter (or the “grammatical” content elements by which they are determined). These groups are conventionally called parts of speech¹¹, and they can be differentiated to varying degrees and distinguished from one another in expression¹². A strong differentiation

¹¹ Zum Begriff der Wortarten vgl. Trnka (1982: 315, 327f.) (vgl. auch die inspirierenden Bemerkungen beim frühen Hjelmslev 1928: Kap. V).

¹² This distinction concerns whether the expression of the corresponding grammatical and lexical content elements differs or is the same. Thus, for example, in Chinese and Vietnamese, “nouns” on the one hand and “verbs” and “adjectives” on the other are distinguished in that the former, unlike the latter, occur in the predicate with the copula; that is, unlike the second group, the first requires explicit expression of the copula function, and there is therefore a difference in the expression of the grammatical content element ‘copula’; however, hardly any other differences are found, and the “verbs” and

138 |

BOHUMIL VYKYPĚL

Word classes are regarded in Prague typology as a property of the inflecting type, whereas in the other language types word classes are distinguished less sharply (cf., e.g., Skalička 1979: 87, 117, 190, 201, 336, 338–341). Now, the high degree of differentiation among word classes attributed to the inflecting type is found in grammaticalization theory as a reflection of a high degree of grammaticalization: the more clearly the word classes are distinguished, the more grammaticalized they are (cf. Lehmann Ms.a).

3.3.4. In addition to the homogeneity of a paradigm (cf. § 3.2.1.), an opposite property is also postulated as another feature of a high degree of grammaticalization, namely allomorphy (and suppletion) (Lehmann 2002: 122). This may seem somewhat surprising, since homogeneity represents regularity, whereas allomorphy is an irregularity; its explanation, however, lies in the correlation between high degrees of grammaticalization and the inflecting type, and also in the state of affairs in ancient Indo-European languages.

3.3.5. Another example of the similarity between the highest degrees of grammaticalization and the state of affairs in ancient Indo-European languages may be the peak of grammaticalization of number (Lehmann 2002: 52): the number of members in the category is reduced, number formatives become obligatory and fused, number agreement arises, originally semantically distinct formatives become allomorphs, and stem suppletion also occurs. These are all properties of the inflecting type. A similar description of the fully grammaticalized category of number, which “coincidentally” corresponds to the state of affairs in ancient Indo-European languages, is found in Erhart (1970: 69f.).

3.3.6. A subtler and interesting example is the case of the so-called fixed word order. In the last section of his fascinating book, Lehmann (2002: § 4.4.4.) resolves one final intriguing question: the so-called fixed word order can occur both at the beginning and at the end of grammaticalization. From the perspective of Prague typology, I would like to make the following observations:

As in grammaticalization, the so-called fixed word order also occurs in two places in Prague typology, namely in the polysynthetic and isolating types. It should, however, be emphasized that there is a substantial difference between these two fixed word orders. The first, polysynthetic fixed word order, which stands at the beginning of the grammaticalization scale, can be described as semantic and compared with Skalička’s implicit or lexical syntax: as is well known, Skalička (1960b) distinguishes several “special forms of syntax,” and one of these forms is implicit or lexical syntax, in which the relations between

“adjectives” are not distinguished. In Hungarian, nominal possessive suffixes and verbal personal suffixes are partly identical, and in Turkish the plural suffix is the same for nouns and for verbs in the third person; conversely, in these languages the same lexical nominal expression form may be either dependent (i.e. adjectival) or independent (i.e. substantival). Differences in the corresponding expression forms are also frequent, however. By contrast, in languages such as Czech or Lithuanian, for example, the expression forms of both dependent nouns and the grammemes that determine them, as well as those of independent nouns and their grammemes, are usually different, so that the so-called adjectives and the so-called substantives are clearly distinguished in expression.

GEDANKEN ZUR GRAMMATIKALISIERUNG

139

words (lexemes or signs) are given by the meaning of the words themselves¹³. Relations between words are indeed expressed through their order, through the position of words within the sentence, but they are not determined by it; rather, they are determined by the semantics of the words. By contrast, the other, isolating fixed word order, which stands at the end of the grammaticalization scale, can be described as “formal”: relations between words are expressed and determined by their order.

The difference between the two kinds of fixed word order can also be formulated as follows: in the second (isolating) case, every change in the order of words also changes the sentence meaning, the relation between the words. In contrast, in the first (polysynthetic) case, a change in the order of words does not necessarily change the sentence meaning, the relation between the words: if the relation between the words within the sentence, and thus the sentence meaning, is clear from the semantics (the meaning) of the words themselves, then it is not necessarily necessary to leave their order unchanged. This accords with Skalička’s general maxim that the polysynthetic type seeks to build its grammar on semantics.

Skalička himself gives examples from predominantly polysynthetic Chinese of freedom of word order when the relations between words are clear on the basis of their meaning (Skalička 1946: 400f., 409; abridged in Skalička 1979: 192). An example from equally strongly polysynthetic Vietnamese may be added. In Vietnamese, the future of an action can be expressed by placing the relevant sign for a time expression before or after the subject, but still before the predicate: Chiều hôm nay tôi mua từ điển ‘This afternoon I will buy a dictionary’. The past of an action, by contrast, is expressed by placing the sign at the end of the sentence: Tôi mua quyển sách trưa hôm nay ‘I bought the book at noon today’. But when the time expression is clear from the meaning of the words, the signs are not bound by these rules: Chiều hôm qua tôi đi xem chiếu bóng ‘Yesterday afternoon I went to the cinema’ and Tôi đi xem chiếu bóng chiều hôm qua ‘I went to the cinema yesterday afternoon’ (cf. Vũ 1998: 64).

3.3.7. In connection with the case of two kinds of fixed word order, allow me to make a few further observations.

The most striking—or most banal—observation is that here we see an instructive example of Gabelentz’s typological spiral: the beginning and the end—here, the so-called fixed word order—are similar, but not identical.

It can further be observed that the claim that isolating fixed word order represents the end of grammaticalization seems hardly to be theoretically grounded; rather, it appears to be empirical, based on the development of Western European languages after the loss of inflection (the empirical standpoint here thus resembles the historical one, i.e. a potentially contingent one). The same applies to the loss or the ma-

¹³ Here, incidentally, another maxim of Skalička (1962a) may also be mentioned, namely that syntax is anthropocentric (and morphology superfluous), which applies especially to implicit or lexical syntax.

140 |

BOHUMIL VYKYPĚL

…ximal reduction of grammatical signs, i.e. their reduction to zero, which is regarded as the highest stage of grammaticalization (e.g. Lehmann 2002: 12, 57f., 64, 121, 137): This reduction (immediately preceding the formation of the isolating fixed word order) followed, in Western European languages, the phase of grammaticalization designated as the penultimate one, which corresponds to the maximization of the inflecting type.

In general, however, the claim that reduction to zero represents the end or the maximum of grammaticalization cannot readily be accepted either, for nothing is nothing and not something, i.e. it is no end, either. Rather, the end of grammaticalization is that “penultimate stage” (e.g. Lehmann 2002: 77) which corresponds to the inflecting or, even more so, the introflexive type (cf. Giger, Vykypěl 2001: 79f.).

3.3.8. One may ask what the account given in §§ 3.3.1–7 means. It may also mean that the criteria of expression in grammaticalization theory are based on a squinting grammar that looks askance at inflecting Latin and its isolating West European daughter languages. To put it more sharply, what is strongly grammaticalized is what looks roughly like what Latin grammar calls grammatical, and what is even more strongly grammaticalized is what it became in West European languages. One is led back to Hoskovec’s words about Erhart, cited above (§ 2.2). Recalling what was said above (§ 3.2), the two comparative axiomatic systems mentioned there—the extralinguistic one (i.e. practically general phonetics for the expression plane and, for the content plane, metaphysics in traditional linguistics, a kind of logic among Chomskyans, or so-called cognitive linguistics in the latest fashion in linguistics) and the intralinguistic one—can be called explicit bases of comparison. In addition, there may be an implicit and therefore, in the Hjelmslevian manner, a priori basis for comparison and description, a squinting grammar¹⁴.

3.4. The correlation between high stages of grammaticalization and the properties of the inflecting type was based primarily on phenomena concerning the expression side of linguistic signs (and further examples could also be cited). This is, moreover, understandable, since the core of Prague typology consists in observing which forms are assigned to the functions found in languages. Lehmann (2002: 109f.) rightly emphasizes, however, that grammaticalization parameters concern both sides of the linguistic sign, even if it is sometimes difficult to distinguish them. If it were possible to establish a grammaticalization scale on the basis of content or semantic parameters as well, it would be impossible to correlate it with Prague typology, since this typology assumes precisely that there is no such scale in the content plane (at the tectogrammatical level), but only elements that are either “lexical” or “grammatical,” and that in different language types the lexical elements are expressed in the same way, while the grammatical ones are expressed differently.

At this point, therefore, the grammaticalization parameters that concern content (semantics) to a greater or lesser extent remain to be discussed.

¹⁴ For an interesting attempt to establish a basis for cross-linguistic comparison of content, a “functional linguistics,” see Lehmann (2004a, 2004b). – On the semantic basis for a synchronic comparison of languages, see also Sgall (1978) and Haspelmath (2004: 572f.).

GEDANKEN ZUR GRAMMATIKALISIERUNG | 141

3.4.1. The parameter of variability may be set aside in this context. The two types of variability, paradigmatic and syntagmatic (Lehmann 2002: §§ 4.2.3., 4.3.3.), presuppose a non-gradual change in the content plane. In the case of syntagmatic variability, a new or different form of relations between certain content elements is reflected by new relations between the expressions associated with them. Paradigmatic variability is determined as a whole by relations in the content plane, not by signs (cf. also above § 3.2.2.).

Im Parameter der paradigmatischen Kohäsion (Lehmann 2002: § 4.2.2.) lassen sich semantische resp. inhaltliche Aspekte einer Grammatikalisierung besser identifizieren. Einen von ihnen stellt der geringe Umfang des Paradigmas (oder der Kategorie) dar. Allerdings kann man hier zweierlei anmerken: Einerseits scheint dies wieder mit dem Zustand in den hochflektierenden altindogermanischen Sprachen korrelierbar (vgl. auch oben § 3.3.5.), andererseits stellt sich grundsätzlich die Frage, was zu einem Paradigma (einer Kategorie) gehört (vgl. auch Vykypěl 2005: 215–223). Man kann beispielsweise sehr gut ein Kriterium aufstellen, nach dem Präpositionen und die affixalen Kasus in eine einzige Kategorie eingegliedert werden. Mit anderen Worten, auch dieser Parameter setzt eine Inhaltsbeschreibung voraus, durch die folglich der graduelle Charakter im Verhältnis zwischen den betreffenden Kategorien bezweifelt werden kann (wie im Falle von Kasus und Präpositionen). Ein weiterer semantischer Aspekt der paradigmatischen Kohäsion, die einfachen Beziehungen innerhalb der Kategorie, wurde schon oben (§ 3.2.1.) behandelt.

Lehmann (2002: § 4.3.2., S. 138–140) macht auch auf den semantischen Aspekt der syntagmatischen Kohäsion (der Gebundenheit) aufmerksam¹⁵. Diese Voraussetzung ist sehr interessant (u.a. auch durch alle ihren Implikationen, die Lehmann a.a.O. behandelt). Sie scheint uns jedoch wieder vor das Problem einer Theorie des Inhalts zu stellen. Es stellt sich hier zunächst die Frage, was woran gebunden ist, d.h. wodurch was determiniert wird, und welche Abhängigkeit gemeint ist (denn beispielsweise auch zwischen Teilen eines Kompositums, d.h. im Bereich des Lexikons kann Abhängigkeit bestehen). Es stellt sich auch die Frage, wie die angeblich geringeren und größeren Abhängigkeiten zu unterscheiden sind. Man stößt an die allgemeine Unsicherheit von Kriterien in der Semantik und den Kontinuumcharakter des Sprachgebrauchs. Das bedeutet auch, dass man hier wieder kaum natürliche graduelle Gegensätze finden kann, sondern nur durch die Form involvierte diskrete Gegensätze.

Das syntagmatische Gewicht (Lehmann 2002: § 4.3.1.) entspricht der Frage, was wodurch determiniert wird. Wie oben (§§ 3.2.6., 3.3.2.) angedeutet, muss vom Gesichtspunkt des Inhaltsplans kein Unterschied zwischen Konstruktionen bestehen, die ein einziges determinierendes Zeichen enthalten, und jenen, in denen mehrere solche Zeichen enthalten sind. Somit bestehen auch hier kaum graduelle inhaltliche oder semantische Unterschiede.

¹⁵ It should be pointed out that the semantic (content-related) dependence of a grammatical element on a lexical one is also found among Erhart’s constitutive properties of grammatical elements (cf. above § 2.2.), and that, moreover, it is possible to discern it in the medieval Modistae as well, in their modi significandi.

142 |

BOHUMIL VYKPĚL

The semantic and formal aspects seem best distinguished in the parameter of paradigmatic weight (Lehmann 2002: § 4.2.1.): A sign’s greater grammaticalization is associated with its greater desemanticization. This criterion is also encountered in Skalička’s definition of grammatical meanings (cf. above § 2.2.). Some reservations may, however, be expressed about it.

First, as is well known, semantic description generally presents a fairly substantial problem, and one is often confronted with vagueness or with difficulty in grasping concepts, or with the intervention of extralinguistic or at least metalinguistic factors. Semantics can also (from a Hjelmslevian perspective) be regarded as the manifestation of relationally defined content elements, i.e. as the interpenetration of language structure and the world, and thus as irrelevant to the definition of elements and their categories. From a gnoseological or operational point of view, semantics or semantic description can also be understood as a “translation” into another language. (Cf. also above §§ 3.2.3.-4. and Miloš Dokulil’s (1978) distinction between word-formational meaning, i.e. structurally determined meaning, and lexical meaning, i.e. meaning determined by usage.)¹⁶

Semantics can therefore be considered something that, on the one hand, is irrelevant to the definition of content elements or is a consequence of this definition (the position of the elements in the structure of the plane of content) (cf. also Hjelmslev 1959: 192–198), and that, on the other hand, can be determined in terms of its abstractness or concreteness only extralinguistically or on the basis of usage, and hardly within the framework of language structure (the structure of the plane of content) itself. The gradual nature of the semantic opposition between related lexical and grammatical elements is given extralinguistically. Ultimately, what has been said is also connected with the continuous nature of phenomena in this domain, which Skalička (1959, 1979: 297–299) also discussed when he described the fundamental semantic relationship between certain lexical and certain grammatical elements and spoke of the transposition of the former into the latter. Hjelmslev considered precisely this fact to be the (operational) reason why grammatical elements cannot be defined semantically (cf. Vykypěl 2005: 244f.). Here one also returns to the question mentioned above (§ 3.1.): when is a sign to be examined with respect to its grammaticalization, and when is it not yet?

3.4.2. Nevertheless, it should be mentioned that in Skalička (1955b, 1960a) one finds an assessment of his language types in terms of their grammaticality and abstractness, or more precisely, of the grammaticality and abstractness they produce¹⁷, which can thus be compared with

¹⁶Incidentally, Lehmann (2002: 114) also answers the question of how one recognizes that something has fewer semantic features by referring to a metalanguage (it is described by fewer metalinguistic sentences). – The general problems of semantics can also be demonstrated by a single problem, namely that of grading cases (case functions) according to their grammatical character (Lehmann 2002: 84f. and 3.4.2.2.): the semantic criteria seem somewhat vague, and, ultimately, the measure of grammaticalization in content is tied to that in expression (Lehmann 2002: 85, 100).

¹⁷ Vgl. auch Skalička (1949).

GEDANKEN ZUR GRAMMATIKALISIERUNG | 143

the semantic aspect of the grammaticalization parameter of paradigmatic weight can be compared with. This can also be called Skalička’s own interpretation of grammaticalization in his language typology. What is interesting here is that, unlike the expression criteria, Skalička’s typology does not correspond to the grammaticalization scale.

According to Skalička (ibid.), language types can be grouped in terms of their grammaticality into approximately two binary oppositions: the polysynthetic type stands opposed to the isolating type, and the agglutinating type to the inflecting type.

The polysynthetic type builds its grammar (Mathesius’ relating) on semantics (it has no specialized grammatical signs) and is thus semantic and agrammatical, whereas the isolating type builds its grammar on formal elements (with the status of words) and is thus formal and grammatical. Similarly, in forming words, lexemes (Mathesius’ onomatology), the polysynthetic type is highly descriptive, i.e. its lexemes are strongly motivated (the main procedure of polysynthetic word formation is compounding), whereas the isolating type in this domain is highly etiquetting, i.e. its lexemes are strongly unmotivated and arbitrary (the isolating type favors the creation of entirely new words, e.g. abbreviations, as well as borrowing)¹⁸. (On the difference in semanticity between the polysynthetic and isolating types, cf. also the discussion of the so-called fixed word order above in § 3.3.6.)

The agglutinating type produces a large number of affixes and uses them to solve problems of both grammar and onomatology. These two properties consequently lead, on the one hand, to the possibility of reducing the asymmetry of the linguistic sign (the relation between form and meaning within the sign becomes clearer) and, on the other hand, to obscuring the relation between grammar and onomatology (word formation). Both also entail that the agglutinating language type is less grammatical and abstract, and more agrammatical and descriptive (with motivated complex signs), though not to the same extent as the polysynthetic type. The inflecting (and introflexive) type, by contrast, reduces the number of affixes until they become endings. This strengthens the asymmetry of the linguistic sign (synonymy and homonymy, the cumulation of functions in morphemes). In the inflecting type, grammar and onomatology have at their disposal complicated sets (paradigms) of signs with no or little motivation and with a strongly obscured relation between form and meaning. The inflecting language type is thus more grammatical and abstract, but not as abstract and unmotivated as the isolating type, since inflectional formatives do, formally (in expression), form more coherent, albeit complicated, systems.

The degree of abstractness in Skalička’s language types can be represented schematically as follows:

¹⁸For the distinction between descriptive and etiquetting designations, which goes back to Mathesius, and for these German terms, cf. Daneš (2003: 37).

144 |

BOHUMIL VYKÝPĚL

The criteria of expression, by contrast, create a straight line (which may, where appropriate, turn into Gabelentz’s spiral):

or:

For a straight line, the types must be ordered differently in the case of semantic criteria

or:

It is not without interest that in Skalička’s famous typological triangle (1935: 81), too, the polysynthetic and the isolating types are found as two extreme points:¹⁹.

¹⁹

It is likewise not without interest that the cited alternative arrangement of language types could be compared with the phases of lexicalization as formulated by Lehmann (2005: § 3.5): “Recalling the techniques of expressing a semantic component in linguistic structure […], we can say that lexicalization leads from (lexical-)syntactic expression via derivational to lexical expression of a concept […].” The main word-formation process in the polysynthetic type is compounding, in the agglutinating type affixal derivation, and in the inflecting type morphological

GEDANKEN ZUR GRAMMATIKALISIERUNG | 145

4. Now I would like to venture some GENERAL REMARKS, which may even be related to the question of linguistic method.

4.1. In the preface to his treatise, Lehmann (2002: VII) reveals:

„I must also warn the reader that I have great conceptual difficulties with the present subject, and I will leave many questions open. The problem is not so much an empirical one: there are sufficient analyzed data, and the empirical phenomena in themselves appear to be reasonably clear. What is highly unclear is how the phenomena are to be interpreted, classified and related to each other.“

One may note that the interpretation of empirical data is always highly uncertain if one is not a realist and reality thus appears to one as a continuum, for the interpretation of which one needs a theory that requires nomina. But even if one does not consider oneself a nominalist, the need for an interpretive theory nevertheless arises in the case of so-called grammatical entities, since these have scarcely any boundaries of their own in empirical data. Skalička (1962b: 211) captured this with the metaphor of the twofold stylization of the world by language (cf. Vykypěl 2005: 192–194): grammatical or morphological categories form the second stage of this stylization and therefore reflect the world only indirectly²⁰. Hjelmslev (1943: § 5) famously resolves the dilemma with his thesis of the twofold, realistic and arealistic character of linguistic theory: linguistic theory is indeed based on certain data of experience, on empirical evidence, but at the same time it frees itself from them²¹. From another perspective, this can be expressed by saying that linguistic theory

conversion, and in the isolating type, borrowing or the creation of entirely new words or words connected only conceptually (cf. also Vykypěl 2004b: 131). (On the shift from compounding to affixal derivation as an expression of grammaticalization, cf. Haspelmath 1992: 70–74.)

20 Skalička’s twofold stylization, which I have attempted to describe in greater detail in the work cited above, could be compared in an interesting way with Lehmann’s (1993: 320f.) distinction between the domain of universal concepts and that of the language-specific combination of these concepts. But perhaps that is a topic for another time. – Incidentally, it may be noted that, in his more recent clarification of the difference between lexicon and grammar (and correspondingly between lexicalization and grammaticalization), Lehmann seems to have come somewhat closer to Mathesius’ distinction between naming and establishing relations (cf. Lehmann 2002a, 2005: § 3.5.). This, however, would be another topic…

21 From an operational point of view, the arealistic character can also be compared with a principle according to which one does not necessarily take into account the facts of language use, i.e. in the content plane, the expression form of signs and the semantics of content elements. As for the

146 |

BOHUMIL VYKYPĚL

It should be deductive: the starting point should be a defining classification, a simple, fixed, and intersubjectively controllable system, from which one proceeds to the components of the class; that is, one moves from the closed, simple system of linguistic categories and their elements to the question of how this system is realized in individual languages. Lehmann (2002: 108) also notes that, in the investigation of grammatical entities, the deductive procedure is more advantageous than the inductive one, and that Chapter 4 establishes a class that specifies the parameters of grammaticalization.

4.2. One might, however, also ask the question (already mentioned above in § 1.2) of what purpose is served by isolating the domain of grammatical elements and grammaticalization. Without a general theory, the topics of grammaticalization do not seem to be amenable to meaningful or relatively definitive solutions. When they are incorporated into a theory, they seem in fact to disappear: I have tried to show that grammaticalization presupposes a theory of content, on the one hand, and coincides with Prague typology, on the other²².

Moreover, Prague typology seems to possess considerably greater explanatory power. In his talk Grammatikalisierung und Sprachtypologie at the Prague Linguistic Circle on 6. 1. 2003, whose written version has unfortunately still not appeared, Markus Giger demonstrated several points of agreement between grammaticalization theory and Prague typology, some of which were mentioned above (§ 3.3.). He also raised a question that grammaticalization theory does not answer: why are grammaticalization processes carried through to completion in some languages, while in others they are halted at a certain stage? He suggested a solution: this depends on which language type is dominant in the language in question (see also Giger 2002: 46f.). If, for example, the polysynthetic type is dominant, grammaticalization processes are considerably weaker than when the inflectional type is dominant²³.

In seinem Vorwort beschreibt Lehmann (2002: VII) die Sprachtheorie, die für die Erfassung des Problems der Grammatikalisierung geeignet wäre:

With regard to signs, the impossibility or difficulty of a consistently “realistic” procedure has also been seen, for example, above (§§ 3.2.4., 3.2.6., 3.3.2., 3.3.7.), in questions such as how to decide whether, in certain cases, no member of a grammatical category (an Inhaltskategorie) is present, or whether what is involved is a syncretism of all the members of that category, expressed by zero or at least by a segmental zero.

²² It should be expressly emphasized that this claim is not intended to mean that grammaticalization is nonsense, as the “antigrammaticalists”—against whom Lehmann (2005) rightly polemicizes—probably believe. On the contrary, research on grammaticalization reveals many valuable and interesting, sometimes even fascinating, facts. The point here, rather, is to clarify perspectives or to foster mutual understanding.

²³ The Prague theory also proves illuminating with regard to the general question of explaining language change, when language is viewed as a system of signs, as it is in grammaticalization theory. Lehmann (2002: 3f.) mentions two factors in language development, the drive toward ease and the drive toward clarity, which are found in Georg von der Gabelentz and, in modified form, in Antoine Meillet (see also Has pelmath 1998: 318–322). These can also be found among Prague linguists, placed on a semiological-functional basis (see Vykypěl 2005: 28–31, 256–258, with references).

GEDANKEN ZUR GRAMMATIKALISIERUNG | 147

„The theory of language which is to account for the systematicity, goal-directedness and dynamism inherent to grammaticalization must be structural, functional and operational in nature. It is essentially the theory of Wilhelm von Humboldt (1836), which has been elaborated in more recent times by Eugenio Coseriu (1974) and Hansjakob Seiler (1978). This theory has never been made fully explicit; but will become transparent through all of the present treatment, and an attempt to make it more explicit will be presented in the last chapter.“

Unfortunately, the final chapter was left out (see, however, Lehmann 1993 and also Lehmann 2005). An adherent of the Prague School cannot, however, fail to mention that the language theory sought here is precisely that of the Prague School, which also seems conceptually clearer, on the one hand, and more thoroughly considered and thematically more diverse, on the other. One can only regret that the Prague School is, strangely, still so little known (despite all efforts to promote it, particularly on the part of Josef Vachek; see, among others, Vachek 1964, 1966, 1976, 1983, 2002, 2003 and Hladký 1997, Ehlers 1999; see also Trnka 1982, Luelsdorff 1994, Luelsdorff, Panevová, Sgall 1994, Leška 1995, 1996, 1998, 1999, Sgall 1992b, 1993b, 1997, Leška, Nekvapil, Šoltys 1993, Winner 1998, Schooneveld 1996, Nekula 2003). Anyone who feels that explicitness and/or operationality are lacking in Prague language theory can also turn to Petr Sgall’s functional-generative description (see Sgall 1984, Sgall, Hajičová, Panevová 1986, Panevová 1994, Sgall 1995, Hajičová, Panevová 1996).

4.3. One purpose of grammaticalization theory, which was asked about above (§§ 1.2., 4.2.), is implicitly offered by Lehmann (2002: 108f.): grammaticalization reflects the freedom with which linguistic signs are used; that is, the more grammaticalized a sign is, the less freely it can be used (the “measurement” of grammaticalization may represent a measurement of the sign’s autonomy or of the freedom of its use) (see also Lehmann 2005: § 6.1.). A substantively corresponding view, though not tied to the concept of grammaticalization, was formulated by Skalička (1948) and later by Jakobson (also cited by Lehmann 2002: 147) (see Vykypěl 2005: 129–131): the more extensive a sign-unit is, the more freely it is formed.

It can, however, be argued with good reason that the formation of all signs is always both free and bound, or determined, from the point of view of the language system (considered, for instance, as Hjelmslevian language structure) as well as from the point of view of the speaker’s intention or from an extralinguistic point of view:

From the point of view of language structure, the formation of signs is determined in that the (paradigmatic and syntagmatic) relations between the expression and content categories of language structure, as well as the structure of these categories, must be respected in their formation (see also above § 3.2.2. on so-called government). From the point of view of language structure, the formation of signs is free, by contrast, because elements or categories between which there is no interdependence or dependence, and which are at the same time combinable, can be combined freely at the speaker’s discretion.

From an extralinguistic point of view, the formation of signs is determined by what the speaker wants to say, since this must consequently be expressed only in a particular

148 |

BOHUMIL VYKYPĚL

in a particular way and in no other way. It is free, in turn, because the speaker can say what they want, including things that seem nonsensical, i.e. things that do not seem to shape any part of the world (this can happen through an unusual combination of lexemes or through a disallowed government; the former may be called poetry and is usually praised, while the latter may be called an ungrammatical construction and is usually not praised…).

This was intended to say, somewhat ponderously, that the question of freedom in language, and the question of what determines that freedom, is fundamentally connected with what one understands by language. What Hjelmslev would call language use, i.e. the application of the language system within a speech community, determined on the one hand by the structure of the language system and, on the other, possibly regulated in various ways by further conventions, Lehmann (2002: 108) has aptly and ingeniously condensed into the term language:

„Language is an activity which consists in the creation of interpersonally available meanings, i.e. signs. This activity can be more free or more regulated; accordingly, the ways in which the signs are formed will either depend more on the actual decision of the language user or more on the social conventions laid down in the grammar.“

The fundamental difference between the language system and language use, the intention underlying Hjelmslev’s distinction between these two concepts, is that the former is something fixed, certain, invariable, and determinable, whereas the latter is something unfixed, uncertain, variable, and indeterminate; consequently, the immanent object of linguistics as an autonomous modern science is supposed to be the language system, while language use can be studied only in connection with other sciences. One may suppose that a similar intention was also at work for Ferdinand de Saussure when he distinguished langue from parole; he simply did not restrict the domain of langue as much as Hjelmslev restricted that of the language system, and included, for example, signs within it. (See Vykypěl 2005: 19f.)

The difference between Hjelmslev’s or de Saussure’s conception of language and the conception represented, for example, by Lehmann can also be understood as the difference between an entity-based and a process-based conception of language. This distinction is also expressed by the two possible terms for the underlying intuitive object of the present treatise: grammaticality and grammaticalization.

From the perspective of what might be called general methodology, the difference under discussion could also be viewed as one between the epistemological and ontological standpoints. The epistemological perspective asks, for example, what and how one sees, and how one can formulate what one has seen; the ontological perspective gives less consideration to questions of seeing and instead tends to register and describe immediate data.

At the end of his overview of the history of grammaticalization research, Lehmann (2002: 7) writes:

GEDANKEN ZUR GRAMMATIKALISIERUNG | 149

„One tradition is conspicuously absent from this picture, namely that of structural linguistics, from de Saussure to our day. This is by no means an accident: whereas historical linguistics and typology have been concerned, from their beginning, with processes and continuous phenomena and thus could easily accommodate grammaticalization as a process which creates such phenomena, structural linguistics has tended to favour a static view of language and clear-cut binary distinctions.“

One may first note here that a problem comparable to the one addressed by grammaticalization theory (see above § 2.1.) also occurs within the structuralist tradition, but is construed there in more entity-based terms. The Prague typology can offer a process-based conception of this problem if its types are interpreted as processes—as stages of grammaticalization (see above §§ 2.2., 3.3.).

Two points should be made about the general characterization of structural linguistics. On the one hand, the tendency to favor “a static view of language and clear-cut binary distinctions” should be understood as an expression of the fundamental epistemological standpoint: an effort to organize empirical data in order to make one’s way among them and gain clarity about what one is actually talking about. On the other hand, the characterization quoted applies primarily to conceptions of structural linguistics such as those of Hjelmslev or de Saussure (and even then not without qualification!), but by no means to the functional structuralism of the Prague School (as already mentioned above in § 4.2.). Bohumil Trnka had already demonstrated this vividly in his programmatic texts from the 1940s (see Trnka 1982: 32–60), and it is reflected in such watchwords as the potentiality of linguistic phenomena (Mathesius 1911), the non-consistent character of language (Skalička 1935), the complexity of the units of language (Skalička 1957), dynamic synchrony (Vachek 1983a), or the freedom of language (Sgall 2002)²⁴. The most appropriate term for the linguistics of the Prague School is certainly functional-structural linguistics (see also Skalička 1947–48: 142 = 1965: 154): to a great extent, it combines the structural and communicative approaches to language, Hjelmslevian theoretical rigor and the tradition’s practical perspective, the static synchrony of theory and the dynamic practice of development, and the epistemological and ontological standpoints in their deepest phenomenological

²⁴ Similarly, when it is claimed that “structuralism has taught us that linguistic analysis starts from expressions and their structures, because this is the only way the linguist can access the meaning or function of the form” (Lehmann Ms.b: § 5.4.4., p. 16), the question arises as to which structuralism is meant. One might mention here, for example, the famous passage in Hjelmslev (1943: 68) on commutation in the plane of content, which criticizes the demand that the point of departure should in principle be taken in the plane of expression; one might mention Mathesius’ “functional linguistics” (see above § 2.1.), on which Skalička bases his typology, in which languages are viewed as “different solutions to the same problems” (see Skalička 1935: 62; Leška 1995: 14f.; Sgall 1971: 83 = 1984: 31; Sgall, Hajičová, Panevová 1986: 270); or, finally, one might point to Sgall’s functional-generative description (see above § 4.2.), which takes the level of sentence meaning (the tectogrammatical level) as its point of departure and proceeds from there “downward” to the phonetic level.

150 |

Bohumil Vykypěl

inspiration; in the Prague School, one finds the clarity of theory combined with the lack of clarity of practice, of “reality” as an empirical continuum.

Remark I: In this connection, I would also like to point out that the structural perspective represented by the Prague School, as well as by other European structuralist movements, should not be confused with the perspective of expression: structure does not resemble expression (of a meaning or function), as was presumably the case in American descriptivism and as Lehmann (2004a: 149–154; 2004b: § 2.2.) seems to assume, but structure is the linguistic system itself, through which the world is shaped, observed, and understood, and by which this shaping is approximately determined (for the concept of structure, see Benveniste 1962). And this system has both its expression side (or “formal” side in Prague) and its content side (or “functional” side in Prague). Here, for example, consider only Sgall’s repeatedly stated distinction in his functional-generative description between linguistically structured meaning and cognitive content, in which he even invokes de Saussure and Hjelmslev’s distinction between the form and substance of content (see Sgall, Hajicová, Panevová 1986: 13; Vykypěl 2005: 285). The aforementioned double attribute “functional-structural” of the Prague School expresses the fact that the Prague School regards linguistic structure and its description as fundamentally determined by the communicative task of language, which consequently also means that it assumes a fundamentally reciprocal relationship between language and the world (see also Vachek 1975 = 1976: 91–103; Vykypěl 2005: 272ff.).

Remark II: If a further topic were to be added to those mentioned above (note 20) for discussion, one could also mention here Lehmann’s plea for the relevance and necessity of an empirical linguistics that is more oriented toward linguistic data and engages with it (from both theoretical and practical perspectives), among other things in view of the extinction of languages. Lehmann (2004c: § 4.1.) interestingly distinguishes roughly three ways of doing linguistics: the logical (which, according to him, has in recent times been represented above all by the various strands of structuralism), the hermeneutic, and the empirical, with only the last making an adequate treatment of data possible. One might, however, note that it is precisely in the Prague School that all three of these aspects are found in close balance and interconnected. The “hermeneutic” study of language can be seen above all in the developed literary theory of the Prague School, which is largely linguistically grounded and represented by names such as Jan Mukařovský, Felix Vodička, and Miroslav Červenka. The “empirical” study, with all its careful reflection, about which Lehmann (2004c) writes, is perhaps most clearly expressed in the collection of material on Czech dialects (i.e. also “endangered languages”) since the 1960s: on the one hand, this project fostered natural and spontaneous collaboration between “traditional” and structuralist linguists; on the other hand, the work was carried out to a large extent on the basis formulated by Prague linguistic theory (see

Gedanken zur Grammatikalisierung

151

the introduction to the first volume of the Czech Linguistic Atlas – Balhar, Jančák 1992; a similar situation is also found in the case of the Sorbian Linguistic Atlas, see SSA 1965–96). From another point of view, however, it is of course not at all clear what empirical research actually is, but more on that another time (see my scattered remarks on this in Vykypěl 2005: 317–332).

Furthermore, two aspects of the matter should be pointed out. The demand (Lehmann’s demand) for the existence and recording of data independently of the researcher accords well with the neopositivist (logical positivist) requirement that statements be intersubjectively verifiable (which, as is well known, was later developed above all by W. V. Quine), and neopositivism was, as is well known, an important source of inspiration for Hjelmslev. A serious, albeit fragmentary, consideration of what data are and how they should be handled is found even at the heart of “extreme structuralism,” namely in Hans Jørgen Uldall (1959).

Remark III: The process-ontological conception of language mentioned above (§ 4.3.) also corresponds to the conception of explanation in linguistic description presupposed by grammaticalization theory. This explanation is essentially diachronic or historical (a diachronic description is one whose object is two or more systems, or parts of systems, in terms of their interconnectedness as potential temporal substitutes; a historical description deals with substitutions that have actually occurred). A system can be understood diachronically or historically only if it is transformed into a preceding system or—viewed from another perspective—if its elements are (re)structured according to the structure of the preceding system (with some exaggeration, for example, Neogrammarian linguistics can be regarded as a squinting grammar that casts longing glances toward the proto-language). Something similar is found in grammaticalization theory when a phenomenon is understood through “tentative degrammaticalization”:

„Human beings […] want to make sense out of formal structure. Most of us will have made the experience that a grammatical feature of some language was totally obscure to them until they discovered that it was a grammaticalized variant of something which had more semantic content and therefore was understood from start. […] Many people have gone through the experience that prior knowledge of Latin can be quite helpful in the learning of a Romance language, while the reverse sequence of learning does not provide any systematic aid (apart from cognate vocabulary, that is). This is another example of the phenomenon that laymen just as linguists understand grammatical structures in terms of less grammaticalized structures. This example is particularly interesting for the theory of language, as it would appear to provide an argument for the psychological reality of the presence of diachrony in synchrony.“ (Lehmann 1993: 331, 332f.)

The presence of diachrony in synchrony (and of synchrony in diachrony), or the synchronic dynamism of language, is of course only one of the possible instruments of linguistic explication and, moreover, for describing language as a whole or the position of an element or phenomenon

152 |

Bohumil Vykypěl

…tion in the language system, i.e. for their definition, is of no decisive importance; on the contrary, this definition, and thus the description of the configuration of systems, constitutes the prerequisite for dynamic-synchronic, diachronic, or historical explication, since it discovers structural tensions as potentialities for change. (For an instructive discussion of the historical character of the explanation implied by grammaticalization theory, see also Haspelmath 1992: esp. 72, 74, 81.)

5. In conclusion, I even venture to claim that the innermost, probably initially unintended, meaning of grammaticalization theory is a polemic with structural linguistics, or a defense of pre-structural linguistics—a linguistics based on the deep, everyday human experience that the world is complicated, or at least seems so to us. I believe that structuralists, if there are any left in the world, should study grammaticalization theory closely, in their own interest. The sketches above are nothing more than preliminary remarks.

Generally speaking, there are in principle two ways out of the feeling that reality is an empirical continuum of phenomena, a theatrum mundi: one is that of the medieval nominalists and their magnificent interpretive theories; the other is that of the medieval realists and their humble endeavor to understand the world, without claiming that it will ever be understood. The first way threatens to become meaningless, i.e. to shape no part of the world; the second, in turn, may become pointless. Perhaps there is also a compromise: an approximate, humble understanding of the world through an approximate, magnificent theory. At present, I feel that the Prague School comes closest to this in linguistics. Perhaps this is more than mere patriotism.

LITERATURVERZEICHNIS

Balhar, J., Jančák, P., Hrsg., 1992: Český jazykový atlas 1, Praha: Academia.

Benveniste, É. 1962: « Structure » en linguistique. In: É. Benveniste: Problèmes de linguistique générale. Paris: Gallard, 91–98.

Daneš, F. 2003: Vilém Mathesius’ Konzeption der funktionalen Linguistik. In: Nekula 2003, 31–47.

Diderichsen, P. 1966: Helhed og struktur. Udvalgte sprogvidenskabelige afhandlinger. Selected Linguistic Papers with Detailed English Summaries, København: G. E. C. Gads Forlag.

Dokulil, M. 1968: Zur Frage der Stelle der Wortbildung im Sprachsystem. Slovo a slovesnost 29, 9–16. [Reprint in Panevová, Skoumalová 1997, 125–133.]

Dokulil, M. 1978: K otázce predikability lexikálního významu slovotvorně motivovaného slova. Slovo a slovesnost 39, 244–251. [Reprint in Obsah – výraz – význam. Výbor z lingvistického díla Miloše Dokulila I. Hrsg. von O. Uličný, Praha: Filozofická fakulta Univerzity Karlovy 1997, 41–50.]

Dokulil, M. 1994: The Prague School’s Theoretical and Methodological Contribution to “Word Formation” (Derivatology). In: Luelsdorff 1994, 123–161. [Reprint in: Panevová, Skoumalová 1997, 179–211.]

Ehlers, K.-H. 1999: Josef Vacheks letzter Beitrag zur Geschichte der Prager Schule. Beiträge zur Geschichte der Sprachwissenschaft 9, 288–294.

Erhart, A. 1970: Studien zur indoeuropäischen Morphologie, Brno: Universita J. E. Purkyně. (Opera Universitatis Purkynianae Brunensis. Facultas philosophica, 148.)

Gedanken zur Grammatikalisierung

153

Giger, M. 2002: Aktuálne otázky jazykovej typológie na príklade slovanských jazykov. In: P. Žigo, Hrsg., BraSlav 1. Zborník z medzinárodnej slavistickej konferencie, konanej na Filozofickej fakulte Univerzity Komenského v Bratislave 14.–15. novembra 2002, Bratislava, 43–50.

Giger, M., Vykypěl, B. 2001: Die Typologie des Litauischen und des Lettischen (mit einem allgemeinen Ausblick zu den Perspektiven der Prager Typologie). Sborník prací filosofické fakulty brněnské university A 49, 45–86.

Hajicová, E., Panevová, J. 1996: Functional generative description and its Praguian roots. The Prague Bulletin of Mathematical Linguistics 65–66, 21–28.

Haspelmath, M. 1992: Grammaticalization theory and heads in morphology. In: M. Aronoff, Hrsg., Morphology now, Albany: State University of New York Press, 69–82, 194–198. [auch zum Downloaden unter http://email.eva.mpg.de/~haspelmt/1992ghd.pdf]

Haspelmath, M. 1998: Does grammaticalization need reanalysis? Studies in Language 22, 315–351. [auch zum Downloaden unter http://email.eva.mpg.de/~haspelmt/1998gre.pdf]

Haspelmath, M. 2004: Does linguistic explanation presuppose linguistic description? Studies in Language 28, 554–579. [auch zum Downloaden unter http://email.eva.mpg.de/~haspelmt/ExDesc.pdf]

Hjelmslev, L. 1928: Principes de grammaire générale, København: Bianco Lunos Bogtrykkeri. (Det Kgl. Danske Videnskabernes Selskab. Historisk-filologiske Meddelelser, 16/1.)

Hjelmslev, L. 1943: Omkring sprogteoriens grundlæggelse, København: Ejnar Munksgaard.

Hjelmslev, L. 1959: Essais linguistiques, Copenhague: Nordisk Sprog- og Kulturforlag. (Travaux du Cercle linguistique de Copenhague, 12.)

Hjelmslev, L. 1968: Die Sprache. Übers. von O. Werner, Darmstadt: Wissenschaftliche Buchgesellschaft.

Hjelmslev, L. 1972: Sprogsystem og sprogforandring, Copenhague: Nordisk Sprog- og Kulturforlag. (Travaux du Cercle linguistique de Copenhague, 15.)

Hjelmslev, L. 1973: Essais linguistiques II, Copenhague: Nordisk Sprog- og Kulturforlag. (Travaux du Cercle linguistique de Copenhague, 14.)

Hladký, J. 1997: Bibliography of Professor Josef Vachek’s works. Brno Studies in English 23 (Sborník prací filosofické fakulty brněnské university, S 3.), 15–42. [auch zum Downloaden unter http://www.phil.muni.cz/angl/bse/archive/BSE_23_02.pdf]

Hoskovec, T. 2001: Prooemium ad Grammaticum. In: O. Šefčík & B. Vykypěl, Hrsg., Grammaticus. Studia linguistica Adolfo Erharto quinque et septuagenario oblata, Brno: Masarykova univerzita, 11–13.

Isačenko, A. V. 1948: Morphologie, syntaxe et phraséologie. Cahiers Ferdinand de Saussure 7, 17–32.

Kurzová, H. 1993: From Indo-European to Latin: The Evolution of a Morphosyntactic Type, Amsterdam–Philadelphia: John Benjamins. (Current Issues in Linguistic Theory, 104.)

Lehmann, Ch. 1985: The role of grammaticalization in linguistic typology. In: H. Seiler & G. Brettschneider, Hrsg., Language Invariants and Mental Operations. International Interdisciplinary Conference, held at Gummersbach/Cologne, Germany, September 18–23, 1983, Tübingen (Language Universals Series, 5), 41–52.

Lehmann, Ch. 1986: Grammaticalization and linguistic typology. General Linguistics 26, 3–21.

Lehmann, Ch. 1987: Sprachwandel und Typologie. In: N. Boretzky, W. Enninger & Th. Stolz, Hrsg., Beiträge zum 3. Essener Kolloquium über Sprachwandel und seine bestimmenden Faktoren vom 30. 9. – 2. 10. 1987 an der Universität Essen. Bochum (Bochum–Essener Beiträge zur Sprachwandelforschung, 4), 201–225.

Lehmann, Ch. 1993: Theoretical implications of grammaticalization phenomena. In: W. A. Foley, Hrsg., The Role of Theory in Language Description, Berlin–New York: Mouton de Gruyter (Trends in Linguistics. Studies and Monographs, 69), 315–340.

Lehmann, Ch. 2002: Thoughts on grammaticalization. 2nd, revised edition, Erfurt: Seminar für Sprachwissenschaft der Universität. (Arbeitsapiere des Seminars für Sprachwissenschaft der Universität Erfurt, 9.) [auch zum Downloaden unter http://www.uni-erfurt.de/sprachwissenschaft/ASSidUE/ASSidUE09.pdf]

Lehmann, Ch. 2002a: New reflection on grammaticalization and lexicalization. In: I. Wischer & G. Diewald, Hrsg., New reflections on grammaticalization, Amsterdam–Philadelphia: John Benjamins (Typological Studies in Languages, 49), 1–18. [auch zum Downloaden unter http://www.uni-erfurt.de/sprachwissenschaft/personal/Lehmann/CL_Publ/New_reflection.pdf]

154 |

BOHUMIL VYKYPĚL

LEHMANN, CH. 2004a: Funktionale Grammatikographie. In: W. Premper, Hrsg., Dimensionen und Kontinua. Beiträge zu Hansjakob Seilers Universalienforschung, Bochum: Universitätsverlag Dr. N. Brockmeyer (Diversitas linguarum, 4), 147–165.

LEHMANN, CH. 2004b: Documentation of grammar. In: O. Sakiyama, F. Endo, H. Watanabe & F. Sasama, Hrsg., Lectures on Endangered Languages 4. From Kyoto Conference 2001, Osaka: Osaka Gakuin University (Endangered Languages of the Pacific Rim Publication Series, C-004.), 61–74. [auch zum Downloaden unter http://www.uni-erfurt.de/sprachwissenschaft/personal/lehmann/CL_Pub/Documentation_of_grammar.pdf]

LEHMANN, CH. 2004c: Data in linguistics. The Linguistic Review 21, 275–310. [auch zum Downloaden unter http://www.uni-erfurt.de/sprachwissenschaft/personal/lehmann/CL_Pub/Data_in_linguistics.pdf]

LEHMANN, CH. 2005: Theory and method in grammaticalization. In: G. Diewald, Hrsg., Grammatikalisierung, Berlin: de Gruyter (Zeitschrift für Germanistische Linguistik, Themenheft) (im Druck). [auch zum Downloaden unter http://www.uni-erfurt.de/sprachwissenschaft/personal/lehmann/CL_Pub/Theory&method_in_grammaticalization.pdf]

LEHMANN, CH. MS.A: Grammatikalisierung und Wortarten. http://www.uni-erfurt.de/sprachwissenschaft/personal/lehmann/CL_Pub/Grammatikalisierung&Wortarten.pdf

LEHMANN, CH. MS.B: Motivation in language. Attempt at a systematization. http://www.uni-erfurt.de/sprachwissenschaft/personal/lehmann/CL_Pub/Motivation_in_language.pdf

LEŠKA, O. 1995: Prague School teachings of the classical period and beyond. Travaux du Cercle linguistique de Prague (n.s.) 1, 3–22.

LEŠKA, O. 1996: The meaning of Meaning in the Prague School. In: B. H. Partee & P. Sgall, Hrsg., Discourse and Meaning: Papers in honor of Eva Hajicová, Amsterdam–Philadelphia: John Benjamins, 127–140.

LEŠKA, O. 1998: Prague Circle Linguistics. Linguistica Pragensia 8, 57–72.

LEŠKA, O. 1999: Prague School Linguistics: Unity in Diversity. Travaux du Cercle linguistique de Prague (n.s.) 3, 3–14.

LEŠKA, O., NEKVAPIL, J., ŠOLTYS, O. 1993: Ferdinand de Saussure and the Prague Linguistic Circle. In: J. Chloupek & J. Nekvapil, Hrsg., Studies in Functional Stylistics, Amsterdam–Philadelphia: John Benjamins (Linguistic & Literary Studies in Eastern Europe. 36), 9–50.

LUELSDORFF, PH. A., Hrsg., 1994: The Prague School of Structural and Functional Linguistics, Amsterdam–Philadelphia: John Benjamins. (Linguistic & Literary Studies in Eastern Europe, 41.)

LUELSDORFF, PH. A., PANEVOVÁ, J., SGALL, P., Hrsg., 1994: Praguiana 1945–1990, Amsterdam–Philadelphia: John Benjamins. (Linguistic & Literary Studies in Eastern Europe, 40.)

MATHESIUS, V. 1911: On the Potentiality of the Phenomena of Language. In: Vachek 1964, 1–32 & In: Vachek 1983, 3–43.

MATHESIUS, V. 1929: Functional Linguistics. In: Vachek 1983, 121–142.

MATHESIUS, V. 1936: On Some Problems of the Systematic Analysis of Grammar. In: Vachek 1964, 306–319.

MATHESIUS, V. 1975: A Functional Analysis of Present Day English on a General Linguistic Basis. Hrsg. von J. Vachek. The Hague–Paris: Mouton (Janua linguarum. Series Practica, 208) & Prague: Academia.

NEKULA, M., Hrsg., 2003: Prager Strukturalismus. Methodologische Grundlagen / Prague Structuralism. Methodological Fundamentals, Heidelberg: Winter. (Slavica. Monographien, Hand-, Lehr- und Wörterbücher, 3.)

PANEVOVÁ, J. 1994: Valency Frames and the Meaning of the Sentence. In: Luelsdorff 1994, 223–243.

PANEVOVÁ, J., SKOUMALOVÁ, Z., Hrsg., 1997: Obsah – výraz – význam. Výbor z lingvistického díla Miloše Dokulila II, Praha: Filozofická fakulta Univerzity Karlovy.

POPELA, J. 1991: Prospects of V. Skalička’s Linguistic Typology. In: B. Palek, P. Janota, Hrsg., Proceedings of LP’90. Prague. Karolinum (Acta Universitatis Carolinae, 1992. Philologica, 3–4.), 237–243.

POPELA, J. 1998: Ein Beitrag zur allgemeinen Sprachtypologie und zur Typologie des Deutschen. Germanistica Pragensia 13 (Acta Universitatis Carolinae, 1996. Philologica, 5.), 75–91.

SCHOONEVELD, C. H. VAN 1996: What has the Prague School Given to Linguistics and to Science? Travaux du Cercle linguistique de Prague (n.s.) 2, 3–58.

GEDANKEN ZUR GRAMMATIKALISIERUNG

155

Sgall, P. 1971: On the Notion “Type of Language”. Travaux linguistiques de Prague 4, 75–87. [Reprint in: Sgall 1984, 19–37.]

Sgall, P. 1978: Synchronní srovnávání jazyků a jeho sémantické východisko. Slovo a slovesnost 39, 25–32.

Sgall, P. 1979: Die Sprachtypologie V. Skaličkas. In: Skalička 1979, 1–20.

Sgall, P. 1982: Zur Typologie der Thema-Rhema-Gliederung. In: G. Jäger, R. Heinisch, Hrsg., Studien zum Tschechischen, Slowakischen und Deutschen aus vergleichender Sicht, Leipzig: Karl-Marx-Universität, 173–185.

Sgall, P., Hrsg., 1984: Contributions to Functional Syntax, Semantics, and Language Comprehension, Amsterdam: John Benjamins (Linguistic and Literary Studies in Eastern Europe, 16.) & Prague: Academia.

Sgall, P. 1988: Natürlichkeit, Syntax und Typologie. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 41, 463–470.

Sgall, P. 1992: Valenz und Typologie. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 45, 200–206.

Sgall, P. 1992b: Classical Structuralism and Present-Day Praguian Linguistics. In: H.-H. Lieb, Hrsg., Prospects for a New Structuralism, Amsterdam–Philadelphia: John Benjamins, 75–90. (Amsterdam Studies in the Theory and History of Linguistic Science, 96.)

Sgall, P. 1993a: Skaličkas Sprachtypologie und ihre Fortsetzung. Sprachtypologie und Universalienforschung 46, 318–329.

Sgall, P. 1993b: The Czech Tradition. In: J. Jacobs, A. von Stechow, W. Sternefeld, T. Vennemann, Hrsg., Syntax: Ein internationales Handbuch zeitgenössischer Forschung. 1. Halbband, Berlin–New York: Walter de Gruyter, 349–368. (Handbücher zur Sprach- und Kommunikationswissenschaft, 9/1.)

Sgall, P. 1995: Formal and Computational Linguistics in Prague. Travaux du Cercle linguistique de Prague (n.s.) 1, 23–35.

Sgall, P. 1997: Functionalism in Czech Linguistics and in the World. Linguistica Pragensia 7, 64–81.

Sgall, P. 1999: Types of Languages and Probabilistic Implication Laws. Travaux du Cercle linguistique de Prague (n.s.) 3, 25–34.

Sgall, P. 2002: Freedom of Language: Its nature, its sources, and its consequences. Travaux du Cercle linguistique de Prague (n.s.) 4, 309–329.

Sgall, P. 2004: Types of Languages and the Simple Pattern of the Core of Language. In: P. van Sterkenburg, Hrsg., Linguistics Today – Facing a Greater Challenge, Amsterdam–Philadelphia: John Benjamins, 243–265.

Sgall, P., Hajicová, E., Panevová, J. 1986: The Meaning of the Sentence in its Semantic and Pragmatic Aspects. Dordrecht: Reidel & Prague: Academia.

Skalička, V. 1935: Zur ungarischen Grammatik. In: Skalička 1979, 59–125.

Skalička, V. 1937: Bemerkungen zur Kongruenz. Sborník Matice slovenskej 15, 35–38.

Skalička, V. 1946: Sur la typologie de la langue chinoise parlée. Archiv orientální 15, 386–412. [Verkürzte deutsche Übersetzung: Über die Typologie des gesprochenen Chinesisch. In: Skalička 1979, 180–197.]

Skalička, V. 1947–48: Kodaňský strukturalismus a „pražská škola“. Slovo a slovesnost 10, 135–142. [Verkürzte russische Übersetzung: Копенгагенский структурализм и «Пражская школа». In: В. А. Звегинцев, ред., История языкознания XIX–XX веков в очерках и извлечениях II. 3. Aufl. Moskau 1965, 147–154.]

Skalička, V. 1948: The Need for a Linguistics of “la parole”. In: Vachek 1964, 375–390.

Skalička, V. 1949: On Analogy and Anomaly. In: Luelsdorff, Panevová, Sgall 1994, 147–170.

Skalička, V. 1950: On Case Theory. In: Luelsdorff, Panevová, Sgall 1994, 45–70.

Skalička, V. 1955a: Sur les langues polysynthétiques. Archiv orientální 23, 10–28.

Skalička, V. 1955b: O konkretnosti a abstraktnosti při tvoření slov. Universitas Carolina, Philologica 1, 75–84.

Skalička, V. 1957: Komplexnost jazykových jednotek. Universitas Carolina, Philologica 3, 15–25.

Skalička, V. 1959: Über die Transposition. Slavica Pragensia 1 (Acta Universitatis Carolinae, 1959. Philologica, Supplementum.), 45–52.

156 |

BOHUMIL VYKYPĚL

Skalička, V. 1960a: Agrammatismus und Grammatikismus. Slavica Pragensia 2 (Acta Universitatis Carolinae, 1960. Philologica, 2), 3–6.

Skalička, V. 1960b: Über die besonderen Formen der Syntax. In: C. Bosák, Hrsg.: Rusko-české studie, Praha: Státní pedagogické nakladatelství (Sborník Vysoké školy pedagogické v Praze, Jazyk a literatura, 2), 37–42.

Skalička, V. 1962a: Das Wesen der Morphologie und der Syntax. Slavica Pragensia 4 (Acta Universitatis Carolinae, 1962. Philologica, 3.), 123–127.

Skalička, V. 1962b: Typologie a konfrontační lingvistika. Československá rusistika 7, 210–212.

Skalička, V. 1965: Über die Bedeutung der grammatischen Elemente. Slavica Pragensia 7, 9–11. (Acta Universitatis Carolinae 1965. Philologica, 1.)

Skalička, V. 1966: К вопросу о типологии. Вопросы языкознания 15 (4), 22–30.

Skalička, V. 1979: Typologische Studien. Hrsg. von P. Hartmann, Braunschweig–Wiesbaden: Vieweg. (Schriften zur Linguistik, 11.)

SSA 1965–96 = Sorbischer Sprachatlas 1–15. Bearb. von H. Faßke, H. Jentsch und S. Michalk, Bautzen: Domowina.

Trnka, B. 1982: Selected Papers in Structural Linguistics. Hrsg. von V. Fried, Berlin–New York–Amsterdam: Mouton. (Janua linguarum. Series Maior, 88.)

Uldall, H. J. 1959: On the Preparation of a Text. Archivum linguisticum 11, 1–17.

Vachek, J., Hrsg., 1964: A Prague School Reader in Linguistics, Bloomington: Indiana University Press.

Vachek, J. 1966: The Linguistic School of Prague, Bloomington–London: Indiana University Press.

Vachek, J. 1975: Zum Zusammenspiel von internen und externen Faktoren bei der Sprachentwicklung. In: D. Cherubim, Hrsg., Sprachwandel. Reader zur diachronischen Sprachwissenschaft, Berlin – New York: Walter de Gruyter, 190–207.

Vachek, J. 1976: Selected Writings in English and General Linguistics, The Hague–Paris: Mouton (Janua linguarum. Series Maior, 92) & Prague: Academia.

Vachek, J., Hrsg., 1983: Praguīana: Some Basic and Less Known Aspects of the Prague Linguistic School, Amsterdam: John Benjamins (Linguistic & Literary Studies in Eastern Europe, 12.) & Praha: Academia.

Vachek, J. 1983a: Remarks on the Dynamism of the System of Language. In: Vachek 1983, 241–254.

Vachek, J. 2002: Prolegomena to the History of the Prague School of Linguistics. Travaux du Cercle linguistique de Prague (n.s.) 4, 3–81.

Vachek, J. 2003: Dictionary of the Prague School of Linguistics. Hrsg. von L. Dušková, Amsterdam–Philadelphia: John Benjamins. (Studies in Functional and Structural Linguistics, 50.)

Vũ, Duy-Tú 1998: Lehrbuch der vietnamesischen Sprache, Hamburg: Buske.

Vykypěl, B. 2004a: Adolf Erhart (31. Mai 1926 – 11. August 2003). Sborník prací filosofické fakulty brněnské university A 52, 146–158.

Vykypěl, B. 2004b: Zur Prosodie in der Typologie. Acta linguistica Lithuanica 50, 127–134.

Vykypěl, B. 2005: Glossematikstudien. Unzeitgemäße Betrachtungen zu Louis Hjelmslevs Sprachtheorie, Hamburg: Verlag Dr. Kovač. (Philologia – Sprachwissenschaftliche Forschungsergebnisse, 66.)

Winner, T. G. 1998: Prague Functionalism. In: R. Posner, K. Robering, T. A. Sebeok, Hrsg., Semiotik. Ein Handbuch zu den zeichentheoretischen Grundlagen von Natur und Kultur. 2. Teilband, Berlin–New York: Walter de Gruyter, 2248–2255. (Handbücher zur Sprach- und Kommunikationswissenschaft, 13/2.)

Bohumil Vykypěl

Gauta 2005 06 01

Ústav pro jazyk český AV ČR

etymologické oddělení

Veveří 97

CZ-60200 Brno

([email protected])