„Das phonologische Inventar und seine morphonologische Klassifizierung“ (einige Bemerkungen)

ACTA LINGUISTICA LITHUANICA XLVIII (2003), 159 – 175 “The phonological inventory and its morphonological classification” (some remarks) BOHUMIL VYKYPĚL Ústav pro jazyk český AV ČR In the present paper an attempt is made to describe the Lithuanian expression plane (the phonological system) according to glossematic principles. We compare this description with the morphonological description of Lithuanian proposed by Tomáš Hoskovec. On this basis a few observations are made on the relation between the glossematic description of the expression plane of language and morphonology. In the first issue of the second volume of the journal Acta Linguistica, several reviews and notes by Louis Hjelmslev appeared, including one on Ľudovít Novák’s article “K základným otázkam štrukturálnej jazykovedy” (cf. Hjelmslev 1940–41a; Novák 1937). Ľudovít Novák, one of the “impertinent lads” of the second generation of the Prague School1, published this article in the then still (above all, thanks to Novák’s editorship) important journal Sborník Matice slovenskej and conceived it as a fundamental rethinking of the basic theses and practice of Prague structural linguistics, especially phonology. Louis Hjelmslev emphasized the importance of Novák’s article and also regretted that it had been written in Slovak, “dans une langue si peu répandue dans le monde international”. At the same time, however, Hjelmslev also added his own views to Novák’s discussion in a condensed form. The time when journal articles worth discussing appeared seems long past. Nevertheless, one can hope that the story of Novák and Hjelmslev will keep repeating itself. Recently, an article by Tomáš Hoskovec appeared in the journal Slavia (cf. Hoskovec 2002). Today’s Slavia is certainly not like the Sborník Matice slovenskej under Novák’s editorship, but Hoskovec’s article is written in Czech, that is, also in a language little spoken in the international world. Like Novák before him, Hoskovec attempts, using Lithuanian as an example, to rethink and develop a theory—namely, the functional-ge- 1 Zu den „vorlauten Buben“ (auf tschechisch „ubátí kluci“) vgl. Leška (2002: 53f.). 160 | BOHUMIL VYKYPĚL nerative description of Petr Sgall, one of the most important continuators of the Prague tradition, and, within this framework, specifically the (hitherto less considered) questions that arise here at the morphonological level. As the title of the article already indicates, its main topic is the question of what happens to the phonological inventory after its morphonological reclassification. In other words, the issue is how the phonological elements are to be reinterpreted when their roles in the construction of signifiants are taken into account. It is well known that Hjelmslev welcomed the program of morphonology (in its form at the time)—unlike that of phonology (cf. Hjelmslev 1973: 228, 234, and also below 2.4.), since morphonology, according to Hjelmslev, sought to describe the elements of a language’s expression plane in terms of linguistic form rather than substance. Although there are also considerable differences between morphonology and Hjelmslev’s description of the expression plane (above all, because morphonology, though not operating in the substance aspect of language usage as phonology does, nevertheless operates in its sign aspect, and not in language structure as Hjelmslev does; see also below 2.), morphonological description and Hjelmslev’s are nonetheless evidently closer to one another. This can also be observed in the result of the “morphonological classification” of the Lithuanian phonological inventory arrived at by Hoskovec. We therefore take this opportunity to compare morphonological description with an attempt at a glossematic description of the elements of the expression side of the linguistic sign in Lithuanian (to avoid taking sides, neither phonology nor kenematics…). This comparison concerns above all the question of the reduction of elements and some more general questions about the “object of description.” First, however, let us recall the well-known fact that glossematic theory is fragmentary and was not brought to completion in all its aspects. This is also why there are only a few, and moreover not unproblematic, descriptions of individual languages from a glossematic point of view. Nevertheless, perhaps something can be said2. 1. There is no one-to-one relationship between spoken and written Lithuanian, and they must therefore be regarded as two different languages (schemas). In what follows, “Lithuanian” will mean spoken Lithuanian. 1.1. Lithuanian has both analytical and synthetic sentences (lexemes and nexuses) and sentence groups (lexies and nexuses): it has divisible expression units that can alone constitute a catalyzed superchain (i.e. the unit immediately preceding them in the analysis), and it has extensive expression characteristics (modulation). We will not examine the category of extensive expression characteristics in detail here: the description of Lithuanian sentence intonation, as Hoskovec (2002: 269f.) also observes, remains almost a terram incognitam. 2 We will not define the glossematic terms, and refer the reader on this point to the relevant literature (cf. especially Hjelmslev 1973: 247–266; 1975). „DAS PHONOLOGISCHE INVENTAR...“ 161 Lithuanian has both analytical syllables (syllabemes) and synthetic syllables: it has both units that, in catalyzed form, can alone form an uncatalyzed lexia in the expression plane, and intense expression characteristics (accents). As for the description of the system of intense expression characteristics, there appear to be two possibilities. The category can be divided into two subcategories: the first category is formed by the so-called accents, the second by the so-called intonations. The first subcategory is selected by the second (the second is syntagmatically dependent on the first), and both have two terms, which may be arbitrarily called “strong accent” (ˈ) and “weak accent” (°), or “acute intonation” (ˊ) and “circumflex intonation” (ˇ), respectively. The system is as follows (1st category ‘punctual’ and negatively oriented; 2nd category ‘rising’ and positively oriented): or—in the framework of the whole category, which is divided into two dimensions (horizontal ‘rising’ and oriented positive-negative, vertical ‘punctual’ and negatively oriented): The so-called intonations select the so-called accents, since the accents may stand alone (i.e., without intonation)—specifically, accent ' in syllables with a so-called short vowel and without l, r, n, m—but not vice versa (see also Hjelmslev 1973: 189–191). The “weak accent” (°) is intensive, since in combination with it the intonations are latent, i.e., syncretized and manifested by zero, and this is a resolvable syncretism (on this, see Hjelmslev 1973: 188, 204 and below 2.4.2.). The “circumflex intonation” (ˇ) is intensive, since in combination with it the vowels a and ə vary (see below 1.2.2.). The second possibility mentioned is not to divide the category (neither into subcategories nor into dimensions) and to posit four terms, which may be arbitrarily designated “strong accent” or “grave” (ˈ), “weak accent” (°), “acute intonation” (ˊ), and “circumflex intonation” (ˇ). The system is as follows (on the ‘rising’ dimension): 162 | BOHUMIL VYKYPĚL The coexistence of the terms of the “intonation” and “accent” categories argues against the second possibility (in other words, the second structuring of the category does not capture the syntagmatic relationship into which the two categories enter). Hjelmslev’s solution offers an alternative. Hjelmslev (1973: 197–202) eliminates the accent ´: the so-called short a and æ manifest the combination of a and æ, respectively, with ´ in the so-called stressed syllables without l, r, n, m, while the so-called short u and i manifest the combination of u and i, respectively, with ˜ in the same syllables. The system could look as follows (on the ‘rising’ dimension): β | B | γ ˜ | ´ | ° (See also below 3.) 1.2. The expression constituents (−thematic) form four categories: those selected (vowels), those selecting (consonants), those that are both selected and selecting (sonants), and those that are neither selected nor selecting. The fourth category includes expression elements that can serve as interjections, and perhaps foreign elements could also be placed here, if these may be counted as belonging to the given language structure. 1.2.1. In Lithuanian, there are two constituents that are both selected and selecting: i and u. Their vocalic varieties are manifested by the phonemes /i/ and /u/, respectively, and their consonantal varieties by /j/ and /v/ (or rather /w/). In registration, the respective varieties are reduced to a single invariant in Lithuanian on the basis of both the substance aspect and the sign aspect of language usage (for the second reduction, see tave ‘you’ vs. tau ‘to you’; gavau ‘I got’ vs. gauti ‘to get’; keturi ‘four’ vs. ketvirtas ‘the fourth’; eiti ‘to go’, ėjau ‘I went’; gijau ‘I hunted’ vs. guiti ‘to hunt’). (See also below 2.4.1.) The system of sonants is therefore as follows (on the ‘front’ dimension): α | A i | u i is intensive, since it dominates the syncretism between the so-called palatalized and non-palatalized consonants (or because the so-called non-palatalized consonants vary before i—if the palatalized consonants are reduced; see below 1.2.3.). 1.2.2. There are two selected expression constituents: a, æ. The remaining candidates for vowels can be reduced: ī ≡ ii, ū ≡ uu, ė ≡ ææ, o ≡ aa (see Hjelmslev 1973: 198–200), ǣ ≡ a variety of æ in combination with ˜ and before latent n, ā ≡ a variety of a in combination with ˜ and before latent n (see Hjelmslev 1973: 182f., 197, 201f.). The elements e and ǫ may be considered signals of foreign words and variants of æ and a, respectively. „DAS PHONOLOGISCHE INVENTAR…“ 163 All these reductions evidently depend on assuming the latency of n, which nevertheless raises problems (latent n before sibilants, but not in the future as in gerins ‘he/she/it will / they will improve’; latent n at the end of a syllable—cf. Hjelmslev 1973: 206, but not in expression forms such as ten ‘there’ or šiandien ‘today’). Without latent n, one would have to posit the following vowel inventory: a, æ, e, ö, and o and ë would have to be regarded as öō and ee, respectively. This would complicate matters because e and ö occur on their own only in loanwords, whereas as parts of identity diphthongs they also form native signs; moreover, and above all, there would be scarcely any (morphonological) grounds for such a reduction. The vowels without e and ö form the following system (along the dimension ‘front’): æ A æ a æ is intensive for the same reason as i (cf. 1.2.1.); between the vowels, however, there is syncretism in most cases. Including e and ö, the system looks as follows (horizontal ‘front’, vertical ‘mid’ or ‘close’): α A α e ö A æ a 1.2.3. Most candidates for consonant status can also be reduced. Thus the so-called palatalized consonants before u and a can be resolved into a sequence of the corresponding non-palatalized consonant and the consonantal variant of i: p’ ≡ pj, b’ ≡ bj, t’ ≡ tj, d’ ≡ dj, k’ ≡ kj, g’ ≡ gj, c’ ≡ cj, ʒ’ ≡ ʒj, č’ ≡ čj, ž’ ≡ žj, s’ ≡ sj, z’ ≡ zj, š’ ≡ šj, ž’ ≡ žj, m’ ≡ mj, n’ ≡ nj, r’ ≡ rj, l’ ≡ lj, v’ ≡ vj, f’ ≡ fj, x’ ≡ xj, h’ ≡ hj; before i and æ, as well as the other palatalized consonants, p’ etc. are variants of p etc. The so-called neutralization of the palatalization correlation (before a non-palatalized consonant and at the end of a word) should be regarded as a manifestation of the latency of j. An alternative reduction of the so-called palatalized consonants before u and a would be to resolve them into a sequence of the corresponding non-palatalized consonant and the vocalic variant of i. This reduction would be motivated by the apparent (potential) commutation between the so-called palatalized consonants and the sequence of the corresponding non-palatalized consonant with j; cf. the dialectal pjonyčia ‘drunkard’ and the variable expression forms bjaurus / biaurus ‘nasty’, pjauti / piauti ‘to cut’, pjudyti / piudyt ‘to set on’ and their derivatives (tautosyllabic Civok before u and a is manifested by /C’/ and Cikons by /Cj/). Another resolution is possible for the affricates: c ≡ ts, ʒ ≡ dz, č ≡ tš, ž ≡ dž. An alternative reduction would be the following: c’ ≡ tj, ʒ’ ≡ dj (cf. also Hjelmslev 164 | BOHUMIL VYKYPĚL 1973: 201f.); t’ and d’ are signals of loanwords or interjections and variants of t and d, respectively. The reason for resolving p’ etc. into pj or pi etc., and c etc. into ts etc., is in any case—as indicated—the absence of commutation between p’ etc. and pj or pi etc., and between c etc. and ts etc. (for this principle, cf. Hjelmslev 1943: 62f.; 1973: 217). The units f, x, h can be regarded as signals of loanwords and as variants: f involves syncretism with p, x with k, and h with zero. The resulting system can be arranged into two subcategories. 1.2.3.1. The first subcategory consists of elements that must stand immediately next to a vowel or sonorant: m, n, r, l3. Its system can be arranged either in two dimensions or in a single dimension: α A α n l A m r (horizontal ‘nasal’; vertical ‘back’) β B γ Γ n l r m (along the dimension ‘closure’) 1.2.3.2. The second subcategory has ten elements and the following system (horizontal ‘front’; vertical ‘voiced’, with syncretisms): α A β B Γ α b g z ž d A p k s š t The voiced consonants could, however, be analyzed as combinations of the corresponding voiceless consonants with h, and before another voiced consonant could be regarded as variants of the corresponding voiceless consonant. The so-called neutralization of the voicing correlation (before a voiceless consonant and 3 Here, of course, the question arises of distinguishing between the accidental and structurally conditioned absence of combinations of elements, which, however, seems solvable only axiomatically (cf. Vykypěl 2003: §2 with references). This question also arises when it is asserted that a unit, in combination with other units, must be analyzed into its components because of the absence of commutation between the unit and the combination: the absence of a combination in the relevant position in which the commutation test is carried out may also be accidental. For the same reason, in discussing reductions we do not address the question of the order of the elements into which something is reduced, since Hjelmslev’s (1973: 161, 255) rule yz = xy + yz is not necessarily valid from this perspective (cf. Fischer-Jørgensen 1979: 108). „DAS PHONOLOGISCHE INVENTAR...“ 165 at the end of a word) would be regarded as a manifestation of the latency of h. The system would then be as follows (on the ‘front’ dimension): α A β B γ Γ p k s š h t An alternative ordering of the system would be the following (horizontally ‘front’; vertically ‘stop’): β B Γ α p k t A s š h 2. We shall now try to compare the outline of the Lithuanian expression system with the morphological classification of the Lithuanian phonological inventory in some respects. 2.1. At this point it may be appropriate first to recall a few things from the history of morphonology (without, however, recounting the entire history). In the initial program of morphonology, as formulated by its founder, N. S. Trubetzkoy (1929; 1931; 1934; cf. also Skalička 1979: 77f., 83), various matters were, as is well known, included: the general task of investigating the phonological structure of (minimal) signs was divided into two areas, namely that of “bound alternations,” which are phonologically determined (i.e. by which contentless units of expression the individual signs are formed), and that of “free alternations,” which are morphologically determined (i.e. by which content elements the relevant units of expression are combined). Trubetzkoy himself, however, for whatever reasons, never completed his own program of morphonology; moreover, the status of morphonology as an independent linguistic discipline was criticized, with the claim that the tasks assigned to morphonology should instead be incorporated into phonology (“bound alternations”) and morphology (“free alternations”) (cf. Martinet 1963: 88 f.; 1965: 95; D’urovič 1967 a; 1967 b; Kurylowicz 1967; more recently cf. Patri 2002). The criticism on the second point can be summarized in the objection that there is no difference between morphonology and formal morphology4, since both investigate the structure of signifiants. Two points may be made in response. First, one can say that morphonology investigates what the INDIVIDUAL phonological elements contribute to the structure of the 4 Perhaps it is unnecessary to recall that here the study of (minimal) signs is understood as morphonology, and their signifiants as formal morphology, whereas Hjelmslev meant the description of certain content elements by morphology. 166 Bohumil Vykypěl signs, while formal morphology deals with complexes of these elements as sign components. Since, however, morphonology also describes complexes of phonological elements (for example, diphthongs or phoneme combinations that alternate with a single phoneme) and, conversely, formal morphology has to take into account the roles of individual phonemes in the formation of signs, a second possible distinction seems more important: morphonology investigates the possibilities offered by the linguistic (phonological) system for constructing signs, while formal morphology is concerned with how these possibilities are exploited, that is, how the concrete signifiants are constructed. And it is precisely on this point that we can also take up Hjelmslev’s positive assessment of morphonology, mentioned at the outset. 2.2. The difference between the possibilities offered by the system and their exploitation is, as is well known, one of the important distinguishing features for Hjelmslev between language structure and language usage (see also Vykypěl 2003: §1). The object of description of morphonology could thus provisionally be assigned to Hjelmslev’s language structure, while that of formal morphology could be assigned to language usage. Besides this point, further points of contact can be found that link morphonology with Hjelmslev’s kenematics (description of the expression plane). First, it should be recalled that the expression elements in the two disciplines are not classified according to their manifestation (phonetically or phonologically, that is, according to Trubetzkoy’s “phonological content”). The difference lies instead in the basis on which they are classified: in morphonology, according to the roles of the elements in constructing signs; in kenematics, according to the syntagmatic and paradigmatic relations into which the elements can enter, without restricting this by the framework of the sign. Nevertheless, both may be regarded as taking an internal viewpoint: kenematics operates in pure form, without taking substance (manifestation) into account; in the case of morphonology, the concern is with the sign, which is formed through the interpenetration of the two planes of language (the content plane and the expression plane), and not with manifestation, which represents the interpenetration of the planes with the world. Nonetheless, from a strictly glossematic viewpoint, it must be acknowledged that the morphonological description of the structure of signifiants—however much it is or may be restricted to possibilities and does not deal with concrete signs—must not be included in the description of language structure, because signs belong to language usage and kenematic description is directed only at contentless expression entities. The possibilities offered by the system mentioned above (2.1.) represent the syntagmatic and paradigmatic combinatorial possibilities of the elements, as well as their possibilities of manifestation, among which language usage then makes choices at various levels (in the expression plane, at the morphonological level and later at the formal-morphological level). Thus one can say that in Hjelmslev’s program for investigating the expression plane, only the following part of Trubetzkoy’s morphonology belongs primarily (that is, to the description of language structure): „DAS PHONOLOGISCHE INVENTAR…“ | 167 morphonology, namely the part devoted to the investigation of so-called bound alternations (see Vykypěl 2002: 243f.). 2.3. Although morphonology can thus be integrated into kenematics only in part, a (elective) affinity nevertheless remains between these two disciplines, evidently based on the fact that in both cases what is realizable is investigated more than what is realized. One may likewise suppose that this is also expressed in morphonology’s readiness to reduce elements. We can therefore finally return to Tomáš Hoskovec’s article. There, a reduction of the so-called palatalized phonemes in Lithuanian similar to that above (1.2.3.) was carried out: from a morphological viewpoint, the palatalized phonemes were interpreted as combinations of the respective non-palatalized consonants with a “palatalizing element” (see Hoskovec 2002: 283–291). This step was not taken arbitrarily, however, but because the consonant system was thereby “considerably simplified,” as was explicitly stated. Hoskovec also allows that the so-called voiced consonants could be reduced (see also above 1.2.3.2), but he does not make this reduction, since it would bring no simplification; in other words, from a morphonological viewpoint, he sees no use for such a modification (see Hoskovec 2002: 283). A further reduction is considered possible in the case of the long vowels /ī/ and /ū/ from a morphonological viewpoint (see Hoskovec 2002: 293f.), namely by resolving them into /ii/ and /uu/, respectively—that is, a reduction similar to that above (1.2.2.) is carried out. At the same time, however, it is stated, on the one hand, that these phonemes cannot be resolved in all cases, and, on the other, this possibility is not considered at all for the other vowels. Both points are understandable, since there are no morphonological grounds for doing so—or, in other words, such operations would not have led to any simplification from a morphonological viewpoint. Kenematic description, however, finds grounds for both, namely in the principle of generalization (on this, see Hjelmslev 1943: 63; 1975: 2): the solution by reduction, which allows the quantities ī and ū to be interpreted unambiguously in some cases and ambiguously in others, is generalized to all cases and to the other vowels as well5. This principle can also be applied to the so-called affricates. As with /ī/ and /ū/, it is also observed with affricates (in other words) that in some cases they can be interpreted morphonologically as combinations of dentals and sibilants, but in others they cannot (see Hoskovec 2002: 286–290). On the basis of the principle of generalization, however, all affricates can be resolved as above (1.2.3.). What speaks against their morphonological resolution, however, is the fact that generalizing the resolution achieves no simplification from a morphonological viewpoint. 5 It should be noted that purely phonological grounds can also be found for interpreting Lithuanian long vowels as identity diphthongs (see Trubetzkoy 1939: 172). 168 | B O H U M I L V Y K Y P Ě L 2.4. We can also take up the objection Hjelmslev raised against morphonology, and aphoristically summed up, in connection with the difference in the extent of reduction: he wrote (interestingly, once again in a note to a text by Novák) that phonology should be built on morphonology and not the other way around (cf. Hjelmslev 1940–41b; cf. also Hjelmslev 1959: 122, 127; 1973: 228, 234; 1974: 1). This means that morphonology operates on the basis of the phonological system: it takes over the phonological inventory and also the phonological method of the commutation test, i.e. manifested forms (expression magnitudes) are commuted, and this takes place within the framework of the sign (in the glossematic commutation test, by contrast, pure forms are exchanged within already established categories; for a comparison between glossematics and phonology with regard to commutation, cf. Ducrot 1967). Examples can also be drawn from the text under discussion. 2.4.1. Perhaps the clearest example is the question of whether the sounds v and u, or i and j, are independent morphonological elements or instead variants of a single element; a further question is what relation obtains between j and i, on the one hand, and the “palatalizing element” (cf. 2.3.), on the other (cf. Hoskovec 2002: 295f.). The sounds cited alternate morphonologically with one another, and their distribution is also to a large extent complementary. Nevertheless, in the case of v and u, the question posed is decided on the basis of the phonological commutation test, and v and u are explained as two different morphonological elements by exchanging manifested expression sequences, with the actual words providing the framework for the commutation test (volus vs. uolus, volą vs. uolą6). In the case of j and i, although no so-called minimal pairs are found, in the case of j and the palatalizing element it is nevertheless possible to exchange at least manifested syllables (bjaurus vs. labiau). From the glossematic point of view, however, neither v and u nor j and i are commutable, since they belong to different categories, however often they may occur, or be able to occur, in the same position within a (manifested) sign. In Lithuanian, these magnitudes can only be reduced to one each and incorporated into the category of expression elements that are both selected and selecting; this is also due to the sign aspect of Lithuanian usage, i.e. to the “morphonological alternation” of these elements (cf. above 1.2.1.). The glossematic commutation relation of linguistic elements is not equivalent to the phonological distinctive capacity of phonemes, although both are established by exchanging certain elements: in the two cases, both the exchange and the elements are different. Nevertheless, one finds 6 The question is whether the segments vo and uo, rather than va and uo, should be exchanged, since the sound written as o in the diphthong uo may perhaps be identified more readily with a. „DAS PHONOLOGISCHE INVENTAR…“ 169 one perhaps finds behind the noetic difference a common ontological object, which in the first case exists in potentia, in the second in actu7. The quasi-arbitral role of phonology is also reflected in the constant distinction between the phonological and morphonological points of view: even if there were morphonological reasons, for example, for resolving long vowels into diphthongs or for reducing v and u or j and i to a single element, this would not change their phonematic status (cf. Hoskovec 2002: 295); in other words, the morphonological reinterpretation of the phonological inventory does not retroactively alter the phonological system. 2.4.2. Further examples show that phonology, and with it morphonology, operates in language usage rather than in language structure. Length is regarded as an immanent prosodic feature of syllables because syllables can themselves change or retain their vocalic peak (cf. Hoskovec 2002: 279f.). Syllable length, however, can readily be interpreted as the presence of two specific elements in a single syllable (similarly, vocalic length can conversely be regarded as the capacity of vowels to form a group of identical vowels within a single syllable), and the identity of the syllable presupposed by the concept of immanent syllable length is given by the identity of the syllable within a sign or, more precisely, a (paradigmatically connected) group of signs8. Such an understanding of syllable identity is also presupposed by the claim that unstressed syllables, too, possess a particular intonation, which, however, manifests itself only under stress and also influences the position of stress (cf. Hoskovec 2002: 282). From the point of view of language usage, however, the relations can instead be described as follows: the so-called unstressed syllables contain the intensive expression characteristic “weak accent” (°), and the intonations become latent in connection with this “accent,” i.e. syncretized and manifested as zero (cf. above 1.1.). This syncretism of intonations can, however, be resolved, namely on the basis of the sign aspect of language usage. In other words, this means that a particular unstressed syllable, i.e. one containing the intensive expression characteristic “weak accent” (°), can be identified with a particular stressed syllable, i.e. one containing the intensive expression characteristic “strong accent” (′), on the grounds that the two syllables occupy the same position in a sign 7As regards so-called minimal pairs, Eli Fischer-Jørgensen (1979: 149) may be mentioned; she wrote that to regard minimal pairs as decisive for distinguishing between variants and invariants is to describe not the phonematic structure of the language but the contingencies of sign formation or even of the material under investigation. In describing language structure, one must always allow for “accidental gaps” in language usage, and hence also for the absence of minimal pairs. Incidentally, mention of the difference between the potential and actual system is also found in the text under discussion (cf. Hoskovec 2002: 277, 288). 8On this principle of identity, cf. also (in another context) Hoskovec’s remarks (2002: 291f., 294) on the “semantic or content-based determination” of diphthongs. 170 | Bohumil Vykypěl (have the same “sign function”), whereby the member of the intonation category contained in the “stressed” syllable is introduced into the “unstressed” syllable. The limits of variation imposed by Lithuanian language structure thus consist in this case in the fact that, in syllable groups that can form the signifiant of a word, only one syllable occurs with “strong stress” and a particular intonation, while the other syllables in this group contain “weak stress” and syncretism of intonations. How these syllables are distributed within a word or a group of paradigmatically connected words is determined by language usage, since this depends on which content elements the syllable group in question is connected with. It is also significant (or may be significant) which intonation can resolve the syncretism of intonations by being introduced. Angemerkt sei, dass allerdings auch Hjelmslev (1973: 204–209) den Synkretismus der Intonationen in sog. unbetonten Silben in einigen Fällen für auflösbar hält (d.h. in der idealen Notation enthält eine solche Silbe mit dem auflösbaren Synkretismus eine bestimmte Intonation) und diesen eben aufgrund des Zeichenaspekts des Sprachgebrauchs auflöst, wobei er die eingeführten Intonationen dann als die Distribution der anderen Prosodeme im Rahmen einer Silbengruppe beeinflussend betrachtet (die durch den Sprachbau gesetzte Variierungsgrenze ist also enger). Hier stellt sich indessen die allgemeine Frage, wie eigentlich die Hjelmslevschen auflösbaren Synkretismen aufzulösen sind (oder mit anderen Worten, wie das Ergebnis eines Synkretismus mit anderen Elementen des Sprachbaus zu identifizieren ist). Es tritt in diesem Punkte offensichtlich wieder das Problem der Unterscheidung zwischen dem formalen und dem operativen Aspekt der Beschreibung des Sprachbaus auf: Während im ersteren Aspekt vom Sprachgebrauch abzusehen ist, muss der Sprachgebrauch im operativen Aspekt der Beschreibung unumgänglich berücksichtigt werden (vgl. hierzu Vykypěl 2003: §3). Darüber, ob und inwieweit der Sprachgebrauch in den einzelnen Fällen auch tatsächlich berücksichtigt wird, soll aufgrund des Empirieprinzips entschieden werden, d.h. man muss fragen, ob die Beschreibung widerspruchsfrei, exhaustiv und einfach ist (zum Empirieprinzip vgl. Hjelmslev 1943: 14; 1975: 1). Der Unterschied zwischen der morphonologischen und der glossematischen Beschreibung scheint hier also darin zu bestehen, dass die erstere sich unbedingt nach dem Zeichenaspekt des Sprachgebrauchs richtet, während die letztere diesen nur dann berücksichtigt, wenn es für die gesamte Beschreibungsstrategie „vorteilhaft“ ist (diese muss sich dabei nach den Prinzipien und Zielen der Sprachtheorie richten; dazu vgl. Hjelmslev 1943: 13–19, 21–31). Somit scheint auch berechtigt, was Hjelmslev schrieb, nachdem er durch Synkretismusauflösung Intonationen in sog. unbetonte Silben eingeführt hatte: “One can accept these facts without thereby making the slightest concession to the current theory, which maintains that ‘the circumflex and the acute exist in an atonic (unstressed) syllable in the same way as in tonic (stressed) syllables’. Our interpretation is fundamentally different and takes linguistic facts into account in a much more realistic way.” (Hjelmslev 1973: 205). „DAS PHONOLOGISCHE INVENTAR…“ 171 The following explanation by Hoskovec would also provide eloquent evidence of the (elective) affinity between Kenematik and morphology mentioned above (2.3.): “Since we are working within a deductively constructed theory, we can decide at the outset whether to take the two aforementioned inventories [i.e. those of phonemes and prosodemes] as the elements of the phonematic-prosodematic level just as they are, or whether to modify them in some way.” (Hoskovec 2002: 268) It can be observed, however, that the criteria for such modification differ from those used by Kenematik: in the first case, they concern phonological data and data relating to the sign aspect of language usage; in the second, they concern also, and above all, the principles of the entire theory of language, which is independent of individual data. 2.5. Eli Fischer-Jørgensen (1975: 123) wrote: “What the Prague phonologists were interested in was the interplay between form and substance, the way in which the purport is formed”. From this point of view, phonology is the description of the substance aspect of language usage in the expression plane. From this perspective, morphology may be regarded as a complement to phonology and as a description of the interplay between expression form and the sign aspect of language usage—a description of how signs are formed. Thus, the two disciplines would find their place in Hjelmslev’s Organon. 2.6. So far, morph(on)ology has been discussed predominantly from the glossematic point of view. However, “(mor)phonological objections” may likewise be raised against Kenematik. The most fundamental of these, as is well known, concerns the operative side of description (which we also touched upon above in 2.4.2.). We have tried to describe this in other texts and have also proposed a kind of compromise between glossematics and phonology, namely, the diversification of the phonological system according to the syntagmatic relations of its elements (cf. Vykypěl 2002; 2003). On the one hand, this solution can obviously be compared with the “morphonological classification of the phonological inventory” in the essay under discussion; on the other—and this seems to us even more interesting for the future—it also recalls an almost forgotten alternative version of classical Prague phonology, namely that of VILÉM MATHESIUS. We are convinced that an important task for phonological research is to analyze Mathesius’ conception of phonology and its influence9. 9 Für einige Literatur zu Mathesius’ Phonologie vgl. Vykypěl (2002: 241). – Nebenbei sei gesagt, dass weder die syntagmatische Strukturierung des Ausdruckssystems mit der immanenten (formalen) noch die paradigmatische mit der transzendentalen (substantiellen) Strukturierung gleichgesetzt werden darf, wie vielleicht einige Formulierungen in zahlreichen Texten von Oldřich Leška eben auch zu Mathesius’ Phonologie (und auch Hoskovec 2002: 275 an einer Stelle) suggerieren könnten: Die paradigmatische Strukturierung kann entweder extensional (formal) oder intensional (substantiell) sein, und ebenso kann die syntagmatische Strukturierung entweder formal sein und die bloßen Abhängigkeiten berücksichtigen, oder aber substanziell und die Reihenfolge und Kombination der Ausdruckselemente als Manifestation ihrer syntagmatischen Beziehungen untersuchen. 172 | Bohumil Vykypěl A more general objection that may be raised against glossematic description, or more precisely against the radical reduction in the number of elements, is the claim that it brings no simplification (it is significant here that this may be considered an argument internal to the theory, since simplicity forms part of Hjelmslev’s empirical principle; see also Vykypěl 2003: §2). As Henning Spang-Hanssen (1949) has already shown, simplicity is a relative notion: reducing the number of elements, as Hjelmslev preferred, can complicate the rules for combining and/or manifesting the elements. We could also observe this above (2.3.), albeit in negative form: reducing voiced consonants or affricates would bring no simplification in the morphological aspect and was therefore not carried out. 3. In conclusion, we return to the question of how the category of intensive expression characteristics is structured. It was mentioned above (1.1.) that Hjelmslev divides this category neither into subcategories nor into dimensions. He nevertheless claims that it is impossible to divide the category of intensive expression characteristics (accents) and that of extensive expression characteristics (modulations) not only in Lithuanian but in all languages (whether into subcategories or dimensions). Two objections may, however, be raised against this view. First, it should be noted that Hjelmslev (1973: 196) substantiates his claim that the categories of accents and modulations never contain further subcategories or dimensions empirically in the usual sense, i.e. on the basis of incomplete induction, which is not permissible within glossematic theory10. The second objection is that the parallelism of the planes, one of the most important points of glossematic theory, would not hold if the categories of expression characteristics could not be divided into subcategories or dimensions, since the categories of content characteristics (morphemes) are divided into subcategories, and these into dimensions (cf. Hjelmslev 1959: 152–164). The category of intensive expression characteristics (accents) could be compared with that of number-gender among the intensive content characteristics. From a panchronic point of view, the “basic meaning” of the number-gender category, namely ‘consistency’ (cf. Hjelmslev 1959: 157; 1972: 90–103), would provide the closest basis for comparison. The third dimension of this category is ‘massive–punctual’ and could be compared with the “accents” that we distinguished above (1.1.) along the ‘punctual’ dimension. The second dimension of the number-gender category is ‘expansion–concentration’ and could be compared with the “intonations” distinguished above (1.1.) along the ‘rising’ dimension. Parallels may also be found in the extensional structuring of the categories. The syncretism of intonations in connection with the 10 It is clear that the text concerning Lithuanian prosodemes was written before the “deductive turn” in the development of glossematic theory. „DAS PHONOLOGISCHE INVENTAR…“ 173 The “weak accent” in Lithuanian can, for example, be compared with the syncretism of genders in the plural in German. As regards the number of members and the extensional structuring, it would be better to compare the Latvian category of intense expression characteristics with the German number-gender category: Gender/“intonation” German | β | B | γ; German | F. | M. | N.; Latvian | ~ | ˙ | ^; Number/“accent” | α | A; Pl. | Sg.; ° | ˈ The syncretism relations would also be parallel: German Latvian German: β | B | γ; Syncretism | Pl. | α | °; F. | M. | N. | Sg. | A | ˈ; Latvian: β | B | γ; Syncretism; ~ | ˙ | ^ Wenn man mit Hjelmslev (1959: 248) im Deutschen in der Numerus-Genus-Kategorie nur mit den Elementen ‘Femininum’, ‘Maskulinum’, ‘Neutrum’ und ‘Plural’ (bzw. auch ‘Kollektiv’) rechnet und die zweite oben (1.1.) vorgeschlagene Strukturierung der litauischen intensen Ausdruckscharakteristiken annimmt, so kann man ein paralleles System aufstellen: German | β | B | γ | Γ | (Γ₂); German | F. | M. | N. | Pl. | (Coll.); Lithuanian | ~ | ˙ | ° | ˈ | (virtual) Hjelmslev’s structuring of the category of intense expression characteristic (cf. above 1.1.) also permits a comparison with the category of intense content characteristics, number-gender, for example with that in French: French | β | B | γ; French | F. | M. | Pl.; Lithuanian | ~ | ˙ | ° Finally, let us point to another striking connection between the category of intense expression characteristics and the content category of number-gen- 174 | BOHUMIL VYKYPĚL Reference was made to the number. As is well known, Latvian has an “intonation” that is absent in Lithuanian, namely the so-called broken intonation (ˆ). This would correspond to Latvian also having, in its content plane, a content element of the number-gender category that is not found in Lithuanian, namely ‘personality’ (cf. vinš, vina vs. tas, tā). For lovers of areal linguistics and linguistic areas, it is certainly interesting here that comparable circumstances are found in the neighboring Germanic languages: cf. the Swedish intonations and Swedish han, hon ‘he, she (personal)’ vs. den, det ‘he/she, it (non-personal)’, both of which are absent in German. LITERATURVERZEICHNIS Ducrot, O. 1967: La commutation en glossématique et en phonologie. Word 23, 101–121. D’urović, L. 1967a: Diskussionsbeitrag zu Kurylowicz 1967. In: Hamm 1967, 170–171. D’urović, L. 1967b: Das Problem der Morphonologie. In: To honor Roman Jakobson. Essays on the Occasion of his Seventieth Birthday 11 October 1966. I. The Hague–Paris: Mouton, 556–568. (Janua Linguarum. Series Maior, 31.) Fischer-Jørgensen, E. 1975: Trends in Phonological Theory, Copenhagen: Akademisk forlag. Fischer-Jørgensen, E. 1979: 25 Years’ Phonological Comments, München: Fink. (Internationale Bibliothek für allgemeine Linguistik, 31.) Hamm, J., Hrsg., 1967: Phonologie der Gegenwart. Vorträge und Diskussionen anläßlich der Internationalen Phonologie-Tagung in Wien 30. VIII – 3. IX. 1966, Graz–Wien–Köln: Hermann Böhlaus Nachf. (Wiener Slavistisches Jahrbuch. Ergänzungsband, 6.) Hjelmslev, L. 1940–41a: Rezension von Novák 1937. Acta Linguistica 2, 64–65. Hjelmslev, L. 1940–41b: Rezension von L’. Novák: Quelques remarques sur le système phonologique du hongrois. Acta Linguistica 2, 67. Hjelmslev, L. 1943: Omkring sprogteoriens grundlæggelse, København: Munksgaard. Hjelmslev, L. 1959: Essais linguistiques, Copenhague: Nordisk Sprog- og Kulturforlag. (Travaux du Cercle linguistique de Copenhague, 12.) 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.) Hjelmslev, L. 1974: Aufsätze zur Sprachwissenschaft, Stuttgart: Ernst Klett Verlag. Hjelmslev, L. 1975: Résumé of a Theory of Language. Hrsg. von F. J. Whitfield, Copenhague: Nordisk Sprog- og Kulturforlag. (Travaux du Cercle linguistique de Copenhague, 16.) Hoskovec, T. 2002: Fonologický inventář a jeho morfonologické třídění. Obecná metodologická rozvaha nad konkrétním materiálem jazyka litevského. Slavia 71, 267–300. Kurylowicz, J. 1967: Phonologie und Morphonologie. In: Hamm 1967, 158–169. Leška, O. 2002: We have finished reading Vachek’s Prolegomena. Travaux du Cercle linguistique de Prague, n.s. 4, 52–54. Martinet, A. 1963: Grundzüge der Allgemeinen Sprachwissenschaft, Stuttgart: W. Kohlhammer Verlag. (Urban-Bücher, 69.) Martinet, A. 1965: La linguistique synchronique (études et recherches), Paris: Presses universitaires de France. (Le Linguistique, 1.) Novák, L’. 1937: K základným otázkam štrukturálnej jazykovedy. Sborník Matice slovenskej 15, 3–23. Patri, S. 2002: La méthode de Troubetzkoy morphonologue. Cahiers Ferdinand de Saussure 55, 63–83. Skalička, V. 1979: Typologische Studien. Hrsg. von P. Hartmann, Braunschweig–Wiesbaden: Vieweg. „DAS PHONOLOGISCHE INVENTAR....“ 175 SPANG-HANSSEN, H. 1949: On the Simplicity of Descriptions. Travaux du Cercle linguistique de Copenhague 5, 61–70. TRUBETZKOY, N. S. 1929: Sur la „morphonologie“. Travaux du Cercle linguistique de Prague 1, 85–88. TRUBETZKOY, N. S. 1931: Gedanken über Morphologie. Travaux du Cercle linguistique de Prague 4, 160–163. TRUBETZKOY, N. S. 1934: Das morphonologische System der russischen Sprache, Prague: Jednota československých matematiků a fysiků & Otto Harrassowitz. (Travaux du Cercle linguistique de Prague, 5/2.) TRUBETZKOY, N. S. 1939: Grundzüge der Phonologie, Prague. (Travaux du Cercle linguistique de Prague, 7.) VYKYPĚL, B. 2002: Fragen der Beschreibung des Ausdrucksplans (mit Beispielen aus dem Obersorbischen). In: R. Blankenhorn, S. Dönninghaus, R. Marzari, Hrsg., Beiträge der Europäischen Slavistischen Linguistik (POLYSLAV) 5, München: Verlag Otto Sagner, 236–250. (Die Welt der Slaven. Sammelbände/Сборники, 15.) VYKYPĚL, B. 2003: Vokale und Konsonanten als Selbst- und Mitlaute (mit Beispielen aus dem Obersorbischen). Sborník prací filosofické fakulty brněnské university A 51 (im Druck). Bohumil Vykypěl Ústav pro jazyk český AV ČR etymologické oddělení Veveří 97 CZ-60200 Brno ([email protected]) Gauta 2003 05 02