ACTA LINGUISTICA LITHUANICA

L (2004), 127 – 134

On Prosody in Typology

Bohumil Vykypěl

Ústav pro jazyk český AV ČR

In the present paper an attempt is made to incorporate prosody into Vladimir Skalička’s typology. Examples are taken from Lithuanian, English, and Chinese.

1. In our typological description of Lithuanian, Markus Giger and I found that, in the matter of homonymy (polyfunctionality) of grammatical formatives (phonological forms), there is a fundamental difference depending on whether prosody is included in the description or not (cf. Giger, Vykypěl 2001: 50f., 52, 71). This can be briefly demonstrated here (once again) using the declension of nouns.

At the segmental level, one finds in Lithuanian a homonymy characteristic of the inflectional type: the 67 Lithuanian morphs (morphological forms) correspond to 47 phonological forms; homonymy is thus, on average, 1: 1.43. If, however, the suprasegmental level is also included, it must be taken into account that each ending is characterized not only by its phonological form (i.e. by its segmentation or by a certain phonemic composition), but also by its inherent prosodic property, that is, by how it interacts with the stem with respect to stress. Thus, if the suprasegmental level is included and the prosodic properties of each ending are also taken into account, there is no homonymy in Lithuanian, since each ending behaves differently. Thus, at the suprasegmental level, the ending -os corresponds to the endings -os₁ and -os₂; -os₁ takes the stress offered by the stem but is itself inactive, i.e. does not attract it, and expresses the genitive singular, whereas -os₂ does not take the stress offered by the stem, does not attract it, and expresses the nominative plural. The situation is different again if the prosodic properties of the individual lexical stems are also taken into account. The stress pattern of a word form is, after all, the resultant of the prosodic properties of the stem and the ending: what matters is which stress class the lexeme belongs to; the differentiating (i.e. segmental homonymy-eliminating) potential of the ending can then be neutralized by the properties of the stem, and segmentally identical forms such as rankos (genitive singular) and rankos (nominative plural) remain homonymous (both are stressed on the stem: rankos), whereas segmentally identical forms such as galvos (genitive singular) and galvos (nominative plural) are distinguished by stress: galvōs vs. gálvos.

128 |

BOHUMIL VYKYPĚL

We have observed that an analysis restricted to the segmental level is clearly incomplete for Lithuanian and, for example, that Lithuanian and Latvian cannot be compared with respect to the homonymy of their formatives at the suprasegmental level, because they have different prosodic properties that have different morphological effects.

2. Horst Geckeler (2001: 114) encountered a similar problem in his typological description of English:

“The frequent conversion between noun and verb in English (e.g. (the) fish → to fish) is more strongly represented at the graphic level than at the phonetic level, since at the latter many word pairs differ at least suprasegmentally, i.e. through a difference in stress placement, e.g. (the) cómpound / to compóund, (the) cónsort / to consórt, (the) increase / to incréase.”

Conversion constitutes a certain lexical counterpart to the homonymy of grammatical formatives: just as the function of homonymous grammatical formatives can be identified only through their connection with a particular class of lexical formatives, so too can the function of homonymous lexical formatives involved in conversion be identified only through their connection with a particular class of grammatical formatives (or at least with a particular class of content elements that are not explicitly expressed). And just as segmentally homonymous grammatical formatives can be distinguished by their prosodic properties, so too can segmentally homonymous lexical formatives involved in conversion be distinguished by their prosodic properties: the former occurs in Lithuanian, the latter in English¹.

3. In his theory, Vladimír Skalička did not address a problem of this kind, that is, the effects of prosody in morphology. This is not surprising when one considers that Skalička’s linguistic typology was initially built on “empirical data” from “prosodically dull” languages such as Turkish, Hungarian, or Czech.

Tilman Berger (1990) attempted to create a stress typology within the framework of Prague typology. His discussion, however, presents a phonological implication of the “morphological” basic properties of the individual language types and thus concerns the question of what functions stress has in the various language types or—more generally—how the phonological system is structured in individual language types or in languages dominated by a particular type². He does not, however, address the role of prosody in the construction of signifiants and their typological

¹ Many examples of the grammatical function of prosody can already be found in Sapir (1921: 82–85).

² Skalička himself, as is well known, also devoted attention to the phonological implications of his typological theory (cf. Skalička 1967; 1979: 238–244, 307–311).

ZUR PROSODIE IN DER TYPOLOGIE

| 129

Beurteilung. Einiges zu diesem Problem hat Jiří V. Neustupný (1959) in seinen Bemerkungen zum Akzent im Japanischen und Russischen beigetragen (vgl. noch unten).

4. The question therefore arises as to how to assess the morphological effect of prosodic elements, which eliminate or reduce homonymy and thus make it problematic as a typological criterion.

One solution is to remove homonymy entirely from typological consideration as a typologically relevant phenomenon. This would also correspond to another fact that seems to make homonymy problematic, namely the need to distinguish between intraparadigmatic and extraparadigmatic homonymy (see Giger, Vykypěl 2001: 55f., 71).

The better solution, however, is a different one (and also seems more natural): homonymy is retained as a typological criterion, but at a lower point in the implicational hierarchy of type constructs (i.e. it is derived only from the basic properties of the types), while prosodic properties are incorporated directly into the types of grammatical formant structure, i.e. at the level of the basic properties of the types.

As Petr Sgall famously formulated it, the basic properties of the types consist in how grammatical functions are expressed. While lexical functions are expressed in the same way in all language types, namely by a phoneme string, i.e. by phonological forms or formants, grammatical functions are expressed in different ways: in the polysynthetic type, by the arrangement of lexical formants; in the isolating type, in the same way as lexical functions, i.e. by formants that are independent and occur as words; in the agglutinative type, in the same way as lexical formants, i.e. by formants that are dependent and occur as affixes; in the inflectional type, by a modification or alternation of lexical formants, occurring at the end of the formants; in the introflexive type, by a modification or alternation of lexical formants, occurring inside the formants (see, e.g., Sgall 1992: 204f.). These differences can also be understood as differences in the degree or level of discreteness (separability) of grammatical and lexical formants (Jaroslav Popela speaks of the degree of fusion between lexical and grammatical signs; see, e.g., Popela 1991). In the polysynthetic type, the discreteness of both grammatical and lexical formants is zero (grammatical functions concern complexes of formants, not individual formant components). In the isolating type, discreteness is absolute. In the agglutinative type, it is likewise absolute in its componential aspect, but in its linear aspect it is relativized by the affixal character of grammatical formants. The inflectional type is characterized by reduced discreteness of grammatical formants; this can be linked to Skalička’s (1938-39) concept of fluctuation in the function of the individual phonemes that make up the formant in question as its components: in the inflectional type, they express

130 |

BOHUMIL VYKYPĚL

some parts of both grammatical and lexical formants express exclusively grammatical or lexical functions (i.e. they are discrete or separable), while other parts express grammatical and lexical functions simultaneously.

At first glance, the introflexive type might be defined as one in which some parts of lexical formants express exclusively the lexical function, whereas all components of grammatical formants fulfill not only the grammatical function but also the lexical function, i.e. the discreteness of grammatical formants is zero. Such a formulation could be justified by cases such as German Vater – Väter, Bruder – Brüder, Mutter – Mütter or English foot – feet, man – men. However, among the forms of expression designated as introflexive, one can also find forms whose parts (components) may readily be regarded as discrete, i.e. as expressing only grammatical meaning: compare Old Babylonian aprus ‘I separated’, i.e. a- ‘1st person singular’, ‘u- ‘preterite’, p-r-s ‘separate’, or German du beginnst – du begannst, i.e. du & -st ‘2nd person singular’, ‘-i- ‘present’, ‘-a- ‘preterite’, beg-nn- ‘begin’. At the same time, one finds forms of expression in which there is partial fusion, as in the inflectional type (in other words, in which the boundary between the lexical and grammatical formant is difficult to determine): compare Old Babylonian aparras ‘I separate’, where a- means ‘1st person singular’ and the preterite is expressed by the two vowels a and reduplication of the second root consonant. The difference between the introflexive type and the isolating and agglutinative types, on the one hand, and the inflectional type, on the other—types with which it shares alternative discreteness and partial fusion—should therefore be formulated rather as follows: the expression of grammatical meaning may be discrete or partly fused with the expression of lexical meaning, but it always takes place within the boundaries determined by the expression of lexical meaning (this may perhaps also be considered a common denominator or connecting link with the polysynthetic type in the “typological spiral”).

Schematically, the interpenetration, or fusion, of grammatical and lexical formants can be represented approximately as follows (x, y = the individual phonological components; l = lexical function, g = grammatical function):

the polysynthetic type: (xʸl)ᵍ (xʸl)ᵍ

the isolating type: xʸl xᵍyᵍ (or (xy)ˡ (xy)ᵍ)

the agglutinative type: xʸlˣᵍyᵍ (or (xy)ˡ(xy)ᵍ)

the inflectional type: xˡyᵍxˡyᵍ

the introflexive type: xˡyᵍxˡdyˡ.³

³ Incidentally, it should be noted that the role of individual phonological components in the construction of formants provides an explanation for the correspondence that can be established between the dominance of a type in a language and that language’s writing system (see Neustupný 1997: 36f.).

ZUR PROSODIE IN DER TYPOLOGIE

131

The situation is similar with regard to word formation. The relationship between the base word and the derived word corresponds to the grammatical function, and what is observed is the degree of fusion between the expression of this relationship and the expression of the base word: the polysynthetic type favors compounding, the isolating type the creation of entirely new words or borrowing, the agglutinative type word formation by means of affixes, the fusional type conversion, and the isolating type infixes or conversion.

The prosodic properties can then also be readily incorporated into the schema cited, by simply regarding them as one of the phonological components of the given formants and observing whether they express only lexical functions, only grammatical function, or both at the same time. Neustupný (1959) proceeds in a similar way: in his view, accent is a component of formants—just like phonemes; the agglutinative type differs from the fusional type in that, in the agglutinative type, accent belongs to the morpheme, whereas in the fusional type it belongs to the word. This means, then, that in the agglutinative type accent constitutes part of either the grammatical or the lexical formant, whereas in the fusional type it (co-)constitutes both formants.

4.1. English word formation through stress shift is to be evaluated as an introflexive feature, since the stress shift occurs within the framework determined by the lexical formant (it also follows from what was said above in §4 that pure expression through prosodemes ex definitione is introflexive). This introflexive feature is, however, combined with features of other types (for example, the isolating type in the compound or I compound, the agglutinative type in compounds or compounded); it is clear that this involves a combination of typologically different formants, rather than formants that have been partially fused.

4.1.1. An interesting combination of introflexive and polysynthetic word formation through compounding is found in Chinese: cf., for example, xiǎojiě ‘Miss’, i.e. xiǎo ‘small, young’ + jiě ‘older sister’, with the second component becoming unstressed.

At first glance, a similar combination is found in the case of “verbs” functioning as prepositions: some verbal formants occur in the function of prepositions, which is to be evaluated as a feature of the polysynthetic type, in which, according to Skalička, lexemes can also perform grammatical functions, and this nonverbal function may be accompanied by a loss of the formant’s tonal characteristic: cf. gāngbǐ zài zhuōzishang ‘The ballpoint pen is lying on the table’ vs. wǒ zai běijīng zhù ‘I live in Beijing’, i.e. zài ‘be (somewhere)’ vs. zai ‘in’.

The expression of grammatical functions through lexemes, however, differs fundamentally from the expression of grammatical functions through the arrangement of lexemes (i.e. from the basic property of the polysynthetic type) from a segmental perspective: whereas in the latter case the grammatical and lexical functions in question are completely fused in their expression, in the former case

132 |

BOHUMIL VYKPĚL

the lexical function of the given formant is de facto reduced to zero, since it is not the lexical function expressed by this formant in another position that is determined by the grammatical function expressed in the position in question, but rather a lexical function of another formant. The grammatical function and the lexical function determined by it are thus discrete in their expression. A formant in a grammatical function is identified with a formant in a lexical function only on the basis of the identity of their phonological components.

Daniel Weiss (personal communication) has even proposed distinguishing two language types here: the amorphous type, in which grammatical functions are expressed through the arrangement of lexemes, and the polysynthetic type, in which grammatical functions are expressed through the occurrence of lexemes in grammatical functions. This distinction would indeed be justified by what was said above, but I do not think it is strictly necessary. First, a good common denominator can be found for these two modes of expression, namely the role of (sentence) context, since the grammatical function of a lexical formant can be recognized or determined only by its position, just as, in the case of expressing grammatical functions through the arrangement of lexemes, the grammatical function that determines a lexical function can be recognized only by the position of the corresponding lexical formant; the role of context can also be expressed using Skalīčka’s dictum that the polysynthetic type builds its grammar on semantics. Furthermore, the aforementioned identification of a formant in a grammatical function with a formant in a lexical function, based on the identity of their phonological components, can also be interpreted to mean that the lexical function in the position in question is potential (not actual), but is nevertheless contained in the given formant; thus, in this case too, the expression consists in the absolute fusion of a grammatical and a lexical function, even if the grammatical function is not fused with the lexical function it actually determines. Finally, a general argument should be mentioned, Ockham’s razor: entia non sunt multiplicanda sine necessitate. Language types too should not be multiplied unless necessary, that is, if there is reason to combine the two modes of expression mentioned.

But if the phonological components are not completely identical—even if only at the suprasegmental level, as in the Chinese example cited with zài and zài—then these formants are indeed to be evaluated as isolating or agglutinative.

4.2. The Lithuanian endings and lexical stems cited above (§1), with their prosodic properties, constitute a feature of the fusional type: one part of the formant relates only to the grammatical function or to the lexical function (namely, the phonemic composition of the formants), while another part relates to both (namely, the prosodic

ZUR PROSODIE IN DER TYPOLOGIE

133

characteristic of the word, which represents a resultant of the interplay between the prosodic potential of the stem and that of the ending).

5. Finally, however, one further general remark must be made. It is clear that Skalička’s typology can indeed be called “word typology,” with “word” to be understood in a broader sense as the minimal combination of lexical and grammatical elements (roughly, as the ‘syntagm’ of the early Hjelmslev, not as a sign of a certain kind defined on the basis of the signifiant): the core of this typology is indeed “morphological,” if morphology is to be understood as the study of minimal combinations of lexical and grammatical elements. What matters here, however, is precisely what the “results” of this combination look like, and not what is combined. An (intro)flexive grammatical or an (intro)flexive lexical sign (a formant) could also be described (defined) as a complex of certain phonemes and a complex of certain rules for changing the other part of the minimal combination, respectively: thus, the plural in German Brüder would have a formant constituted by zero and an ability to alter the lexical formant in a certain way. The Lithuanian formant -os would be constituted by the phonemes o and s and by its ability to influence the prosodic characteristic of the lexical formants, and conversely, the Lithuanian formant galv- would be constituted by the phonemes g, a, l, and v and by the ability to influence the prosodic characteristic of, for example, the grammatical formant -os. If things were understood in this way, however, everything would be discrete (in the other sense of this word and thus also fictitious), and Skalička would be a Chomskyan generativist ante litteras⁴.

LITERATURVERZEICHNIS

BERGER, T. 1990: Zur typologischen Einordnung des russischen Akzents. Die Welt der Slaven 35, 370–380.

GECKELER, H. 2001: Englisch und Französisch: ein typologischer Vergleich nach den Kriterien von V. Skalička. In: Lingua et Linguae. Festschrift für Clemens-Peter Herbermann zum 60. Geburtstag. Hrsg. von U. H. Waßner, Aachen: Shaker, 111–124. (Bochumer Beiträge zur Semiotik, Neue Folge, 6.)

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

NEUSTUPNÝ, J. V. 1959: Accent in Japanese and Russian (A Typological Study). Archiv orientální 27, 122–142.

NEUSTUPNÝ, J. V. 1997: Typology in Prague: The legacy of Vladimir Skalička. In: B. Palek (Hrsg.): Typology: Prototypes, Item Orderings and Universals (Proceedings of LP’96). Prague: Karolinum, 29–40. (Acta Universitatis Carolinae. Philologica 1996, 3–4.)

⁴ Allerdings stellt auch die oben (§4) angeführte Segmentierung eine Segmentierung einer aktuellen und einzelnen Wortform dar, keine Segmentierung einer Verbindung zwischen der Kategorie der lexematischen und derjenigen der grammatischen Zeichen, denn aus der Sicht einer solchen potentiellen und kategorialen Verbindung besitzt jede Komponente der grammatischen Formanten im flexiven Typ auch die lexikalische Funktion und umgekehrt, da Lexeme und Grammateme hier eine untrennbare (synthetische) Einheit – das Wort – bilden (eine Gruppe lexikalischer Zeichen hat nur bestimmte grammatische Zeichen und umgekehrt, wodurch die grammatischen Zeichen auch die lexikalische Funktion mit ausdrücken und umgekehrt); somit wäre der flexive Typ vom polysynthetischen nicht unterscheidbar.

134 |

BOHUMIL VYKYPĚL

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

Sapir, E. 1921: Language. New York: Harcourt, Brace and Company.

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

Skalička, V. 1938–39: O pojem morfému. Sborník Matice slovenskej 16–17, 4–12.

Skalička, V. 1967: Die phonologische Typologie. In: Phonetica Pragensia. Hrsg. von M. Romportl & V. Skalička. Praha: Universita Karlova, 73–78. (Acta Universitatis Carolinae 1967. Philologica, 6.)

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

Bohumil Vykypěl

Gauta 2004 04 06

Ústav pro jazyk český AV ČR

etymologické oddělení

Veveří 97

CZ-60200 Brno

([email protected])