Presentation on theme: "SEMANTIC PATTERNS AND SEMANTIC PROCESSES IN THE SWADESH LIST: PRELIMINARY RESULTS A. Dybo, Moscow."— Presentation transcript:
SEMANTIC PATTERNS AND SEMANTIC PROCESSES IN THE SWADESH LIST: PRELIMINARY RESULTS A. Dybo, Moscow
The long-distance relationship research requires the capability to reconstruct Swadesh's wordlists for a protolanguage. Wordlist reconstruction is the reconstruction of language signs, that is, it requires reconstruction of the semantics, and for such entities as glottochronological wordlists, this reconstruction should rather be quite exact. On the methods of semantic reconstruction which strictly require to include full analysis of lexical microsystems see [Dybo 1996], [Dybo 2011]. Swadesh's wordlist provides a large field for experimentation here.
Some remarks to the technology of semantic reconstruction: 1. Determination of the exact meanings of the words forming the basis for reconstruction. Smaller semantic components (features) can be isolated in a lexeme's meaning, and they can be organized in a hierarchical way. Our semantic constructions are most close to the semantic model of the Meaning–Text theory. We are building semantic trees as glosses of lexemes by analysing phraseologized uses of lexemes in lexical mycrosystems for any language. 2. Word meaning analysis includes the determination of a primary meaning and of the products of semantic derivation. Historical-semantic analysis is as similar to this process as the process of phonetic reconstruction is similar to the process of building deep structures for the phonemic layer of a language. During the reconstruction, reflex units are viewed as the projections of deep (that is, protolanguage) units. In comparative-historical semantic analysis, the variation of the meaning of protolanguage word reflexes may be viewed as a kind of polysemy - that is, polysemy within a language family.
3. The criterion for discrimination between polysemy and omonymy: in polysemy, distinct meanings have a common non-elementary component which represents identical branches of glossing trees. In omonymy, the common component of the meanings either is elementary or represents non-identical branches of interpretation trees [Apresyan 1995, 184]. 4. The possibility of some universal semantic features which are conditioning the comparability of the semantics of words from distinct languages' wordlists is, on one side, an implicit axiom lying in the basis of Swadesh's method; on the other side, it is an axiom lying in the basis of the both semantic models used by the Meaning–Text theory - Apresyan-Mel'čuk's model as well as Anna Wierzbicka's semantic primitives theory.
The glossing of Swadesh's words should serve to further refinement of the methods of interviewing and Swadesh wordlist isolation, which has been started in the article [Kass. et al.]. The glossed words will allow for a stronger basis for discussing which exact features of these words are conditioning the circumstances of little transmissivity and great stability, which is essential for the understanding of experimental results in determining the degree of stability of the lexics within a wordlist [Starostin 2005]. Theoretically, the intransmissivity feature must be guaranteed by the "precultural" character of Swadesh semantics. What of the stability, the possibilities of semantic drift can be calculated for every meaning of a word, based on its ECD-interpretation [Dybo 1984], [Shaykevich, Polinskaya 1989]: a) metonymic possibilities - based on the additions and removals of semantic features within an interpretation; b) metaphoric possibilities - based on the replacements of semantic features (anim. inanim. and alike).
The 100%-stable words in the Swadesh list for the Turkic languages (marked with magenta): 20 words in the more stable half of the list ~ 6 words in the less stable half.
Examples of the groups of similar time depth: Baltic: 23 ~ 19 Slavic: 32 ~ 24
1. All Kass. et al.: "1. He has cut down all the trees 2. He has killed all animals 3. All these men are brothers The plural all (= Latin omnis). Not to be confused with each, every (cf. context 3). If possible, should be separated from the singular all (Latin totus, in contexts like: All of the water in the lake/pool became frozen/evaporated; He has eaten all the meat), which we recommend to exclude from the list." Usage structure in languages
Semantic derivation in Turkic: "Being" > omnis or syncr. – 5 cases "Collected, taken" > omnis or syncr. – 6 cases "Covered" > omnis or syncr. – 1 case "Remained" > omnis or syncr. – 1 case "Evenly, exactly" > omnis – 2 cases "raised" ("up to the brim") > syncr. – 1 case "Pressed" ("chock-full") > syncr. – 1 case "through" > syncr. – 1 case "Healthy" > totus – 1 case "Finished" > totus or syncr. – 4 cases "Filled" > totus or syncr. – 1 case Borrowings: omnis 9, syncr. 3, totus 0
Primary words: 1.1: all (totus): PT*bütüm Derived from PT *büt- 'to be finished, completed'. || PAlt mut ʽ i 'to be finished/ to assemble' > PNMong *möči- 'hardly, barely' || PTM *mute-w- 'to finish, to fulfill' (Caus) || PKor *mòt- 'to assemble in flocks' > mòt 'all' (omnis).|| PJap *muta 'together with'. 1.2: all (omnis): PT *bār Semantical derivate from PT *bār 'exists, is available; presence' || PA *bā́ra 'to have, to receive' > PMong *bari- 'to take, grasp', PTM *bara-n 'many', PKor *pə ̄ r- 'to receive'. 1.3: all: Оnly Old Тurkic qamug 'totus & omnis'. PA *kаma 'to unite, together': PMong *kam 'together', *kami- 'to join', PTung *kamu- 'to collect, to join', PJap *kama-pa- 'to arrange'. 1.4: all (весь): Only Old Тurkic *qop 'totus & omnis'. PA *k ʽ opa-: PMong *kow 'totus & omnis', PNTung *(х)up- 'omnis', PKor *kòp- 'to redouble', PJap *kúpá-pa- 'to add '.
For the reflexes of *bütüm the meaning remains almost unchanged in all branches of Turkic languages, except Oghuz, where an expansion to quantifier uses occurs; the latter may be related to the complete exclusion of the use of primary quantifier word *bār in this range of meanings. Proto-Turkic quantifier meaning for *bār reflexes can be reconstructed completely reliably (it shows everywhere, including the most distant branches). Its use with collective nouns, it seems, can be reconstructed for PCT on the same basis. As for the use with abstract nouns, it can either be understood as CT or as an innovation, independently appearing in three different areals – Yakut, Central Asiatic in late times and Karaim; the latter being less probable. Further syncretization of the meaning is certainly area-dependent – in Yakut and Central Asiatic (Kazakh, Karakalpak and Uzbek) areas. Innovative words that draw a more or less clear distinction between the two meanings' types under consideration appear mostly in areas where syncretization of old words' semantics occurs (see Scheme 1, Scheme 2).
2. Ashes It's reasonable for this clause in Swadesh list to differentiate the following meanings: 'cinder' – shapeless black remains of burned down fuel in a fire or a fireplace 'coals' - shaped black remains of burned down fuel in a fire or a fireplace 'ashes' – very fine light gray powder or thin coating, which can be found amongst these remains; also very fine light gray powder, formed by smoking or burning of small objects. 'embers' - glowing hot shaped remains of burned down fuel in a fire or a fireplace 'soot' – black coating, appearing on surfaces in containers or rooms as a result of burning, which falls down in flocks, reaching sufficient thickness.
Kass.et al.: "1. The campfire has left only ashes. 2. The wind scattered the ashes. 3. He scooped up a handful of ashes. Ashes as the basic result of combustion of wood, grass, dry dung (not after cremation, as in Russian прах). Not to be confused with special words for tobacco ashes, volcanic ashes, or embers (hot ashes/coals)" In the Russian text, however, зола was chosen, corresponding rather to cinder. The choice is not evident - for Slavic lists пепел was taken. Evidently, the semantic structure of cinder, зола is simpler than that of пепел. For the Turkic material the situation here is relatively simple.
Semantic derivation in Turkic: to heat, to burn >> cinder: 1 spark > cinder: 1 embers > cinder: 1 to heat, to burn >> ashes: 2 litter, dirt >> ashes: 3 to strike fire >> spark: 1 to extinguish >> spark: 1 dry dung >> spark: 1 to burn >> soot, fume: 1 smoke >> soot, fume: 1 Borrowings: "ashes" 1, "spark" 1.
"Fume, fliying ashes": South Siberian *pïr, *pïrïm PT *bur-uk 'dust, smoke' || PA *bŏ́ru 'dust, smoke' > PMo *bur-gi- / *bür-gi- ' to rise (of dust, smoke)' || PTM *bure-ki-n 'dust, flying snow' || PKor *pr-m 'wind'. "Coal": PT *kömür || PA *kume 'black' > PKor *kə ̄ m- 'to be black', Old Koguryo *kămul 'black'. "Embers": 1) PT *kř || PA *k ʽ ū́ŕkV > PTM *xurkï 'soot'. 2) PCT *čōg || No etymology PT "cinder" < PA "to warm, to burn" ? CT "ashes" < PA "dirt" PT "soot, fume" < PA "soot, ashes" PT "embers" < PA "soot" ~ "embers"
3. Bark Kass. et al.: "1. He tore some bark off the tree. 2. A piece of bark came loose from the tree. The neutral word for bark, applicable to as many different tree species as possible". This clause belongs as well to the encyclopedically loaded lexics. As Swadesh word it's not a very good one being geographically restricted. In the Turkic languages of South Siberia, for example, we saw the following: only names of certain trees' bark are used, due to their practical importance: birch bark and larch bark; bark in general is named with the same word that stands for animal and human skin (*aγač tere-si), and informants are leaning towards scientific, botanical use of this word, not everyday use. Moreover, an opposition bark as a part of a plant ~ bark as a production material = stripped bark (parallel to tree ~ timber) is observed.
Semantic derivation in Turkic: bark < bark, skin 2; bark < peel 2; bark < leaf 1 (and in the same region foliage < bark 2) bark < to strip 4 bark < cambium < hard 1 Borrowings: 1
Primary words: 3.1: bark: PT *Kāpuk 'bark, shell' || PA *k ʽ ā́p ʽ à 'bark; skin; leaf' > PNMo *kawda-sun 'bark'; PTM *xabda(-nsa) 'leaf'. PKor *kàph- / *kə̀ph- 'bark, skin'. PJap *kapa 'skin, bark'. 3.5: bark: CT *Kās 'bark, peel' < PSam *käsa 'bark' bark: CТ *qa/āz < PT *Ka/āř 'кора' || PA *kéŕà 'skin, bark, shell', PTung. *xere- 'to peel, to skin'; PJap *kárà 'shell'. 3.8: bark: PT *Kaδiř 'bark, stripped bark' || PA *kádí(-rV) 'to scrape; any instrument for skin tanning': PMong *kederge 'scraper', PSTung *xargan ' stock for fish and game skins processing', PJap *káintúr- 'scrape'. "Birch bark": PCT *Tōz || PWA *t ŕu 'birch bark, birch bark container'; PNMo *duru-sun 'birch bark, bast' || PTM *duri 'cradle made of birch bark'. Only PT *Kāpuk 'bark, shell' and PT *Kaδiř 'bark, stripped bark; > scales' evidently claim for PT meaning bark of tree. *Kās, apparently borrowed from Samodic, and *Kāz, which evidently goes back to the PA name for bark (although maybe *Kāpuk as well) are CT words. It still appears impossible to reconstruct clear semantic differences between all these names; only *Kaδiř is, most probably, stripped bark.
4. Belly Kass. et al.: "1. While fighting, he punched him in the belly. 2. Wrap this belt around your belly. Part of the human body located directly above the pelvis. Not to be confused with various terms that denote internal organs (stomach, intestines) or semantically/stylistically marked words (paunch)" Certainly in this glossing two sememes are merged: a) the name of the "exterior" (according to Wierzbicka and Arutyunova), that is, visible without destroying the organism, body part (the front part of the trunk between chest and groin) and b) the name of the "interior" body part - the container of viscera: stomach, intestines, liver, spleen, kidneys, urinary bladder. In principle, distinct words can be used to signify the two sememes in a language (like in Tatar: корсак 'exterior belly' (diagnostic contexts "he has big belly", "hit in the belly" or "fasten the buckle of one's belt on his belly"), карын 'interior belly' ("my belly hurts", "my belly is empty")).
The "exterior" word must correspond more with the criterion of "intransmissivity", because it's less linked with the "illnesses and medicine" conceptual area which is fatal for the transmissivity of the names of body parts. It's clear which of these words corresponds with the criterion of stability if we accept that stability is linked to the number of potential single-step metonymic shifts - for the "interior" belly it's all the types of entrails contained within it and probably also its spiritual "content", the inner world of a human being or the nutritional content - food (so, at least 8); for the exterior belly there are only 3 possibilities: the body part which is higher, that is, chest; the body part on the same height level, only not in front but in side, that is, side ~ hip; and the hyperonymic meaning "the whole trunk". Hence, the name of the "exterior" belly is a more "Swadeshian" word. But, as quite often the two sememes discussed above are syncretically expressed with a single word, and a diachronic shift between syncretical and distinct expression is often imperceptible, we consider the expression of both meaning types below. Also, we must pay attention to the meaning "animal belly", that is, "barrel" (often linked to stylistically lowered uses in relation to human).
PT *Krïm 'exterior belly'. NB: the development of "interior" semantics reveal the regular synchronic polysemy 'reservoir' > 'contents', so it can be independent in all the cases. PT *ič 'interior belly' > syncr. in the North-Eastern area PCT *kurgsak 'stomach' (remaining in the periphery) > Karluk-Kypchak 'interior belly' > syncr. The inverted opposition of 'interior' and 'exterior' belly in the Tatar-Bashkir-Nogai area needs more detailed investigation on the material of the medieval texts now difficult of access.
5. Big Kass. et al.: "1. This person has a big nose, that one has a small nose. 2. A big stone (leaf, tree, pool, etc...). An adjective or verb that may be applied to different objects. The contexts are the same as in No.78 small." As in the theoretical semantics was shown, the meanings like"big" shall be glossed with the help of the feature "norm" or "standart". "A big house" is "a house bigger than standart", and "a big apple" is "an apple bigger then standart". Other features that can be included in the word meaning are different features characteristic for the object of atributivisation. So, the size attributes may differ by the animated - inanimated character of the object (English small vs little). Normally the words exist that define the "bigness" of an object by individual dimensions ("high" – vertical, "long" - horisontal). If it is an object characterised by one marked dimension, the usage with the neutralisation of the feature is possible (high tree = big tree), what leads to a kind of polysemy and to a possibility of semantic shifts. Theoretically, a shift in both directions is possible (big > high and high > big ).
Here the most neutral word is required, that is with the maximal list of objects. Such a word must be highly frequent, what principially could provide "impenetrability". But its stability shall be worse as for any more concrete word, while for any word neutralising characteristic feature it is just one semantic shift to the neutral "big", and for the neutral "big" the narrowing of the meaning can go gaining any of 4 dimensional features (high, long, wide, deep) and moreover any other kind of feature. A type of meaning near to "big" is the lexical function 'Magn', related to the objects characterised by any kind of features, and not only by physical size. Here also senmantic shifts are possible (X > size and size > X), that are metaphorical (as Russian большой ученый). The second type is regular for Turkic languages; the examples for the first type are Shor poγda ("sacred" > "big") ot Tatar зур ("strong" > "big").
For CT the Swadesh word must be *ulug. But it cannot be reconstructed as PT "big": in Old Bulgar we have only one usage within a title. *Ülken receives the Swadesh meaning within a coherent isogloss. *Bedük is the PCT word for "high". "Swadesh" usage for *idrik 'coarse' increases spontaneously in two areas. We have two candidates for PT "big": PCT *ulug 'big' (Bulgarian 'Magn') and Bulgarian *bn- 'big' (in other Turkic languages 'adult'). For *ulug Altaic parallels speak for common meaning "Magn" (quantity and quality); for *bn- - rather for "many" ("quality"). So the choice is ambivalent.
6. Bird Kass. et al.: "1. Something is moving in the bushes, I cannot tell if it is a bird or an animal. 2. Birds lay eggs, animals and people bear children. 3. There is a bird flying on high, I cannot tell what kind. Some languages make a contrast between small bird / large bird or have a special word for large (predatory) bird. The contexts have been specially chosen so as to assist in revealing the more neutral term." So, in a language a common word for "bird" may lack. In Tutkic material the Kirgiz system contains three words: "predatory/ hunting bird", "small bird" and "game (edible) bird". From the other part, the cases exist when a word for "bird" means in reality any flying animal. The second case offers no difficulty; itn the first case evidently we shall analyse both (or all the 3) words.
7. Bite, кусать. Kass. et al.: "1. The dog bit him. 2. He bit his opponent in a fight. Not to be confused with to bite off (while eating) or to gnaw. Said of animals or people (not a separate word like to sting used for insects, snakes, stinging plants etc.)." The glosses for Russian кусать in MAS are: "to take, to clench one's teeth; to wound smb. clenching teeth". "to separate bits from smth. clenching teeth" "to wound smb. sticking a sting" > "to sting (about nettle, wind)" Apparently, the Swadesh meaning is "to wound smb. clenching teeth". This meaning belongs to the group of predicates of physical influence, the soubgroup of predicates of destruction. Such predicates are defined by the types of instrument and object; the instrument and the (animated) object of this verb provide "precultural" character of the word (and such instrument as sting is geographically restricted). But the semantic shifts for the predicates of destruction are multiple - the metaphoric ones are innumerable, the metonymic ones are provided by the shift of types of object, of instrument and of the grade of destruction.
Our meaning can be expressed syncretic with all of enumerated meanings, and such verb we shall take for a Swadesh one, if there is no verb meaning strictly "to wound smb. clenching teeth". Nearest meanings: "to gnaw" = to clench an object with teeth repeatedly, with the expected result of destruction of this object. "to chew" = to clench an object with teeth repeatedly, with the expected result of grinding of this object.. Semantic derivations: 'bite' > 'eat' 3 'bite' > 'assault' 3 'bite' > 'gnaw' 3 'bite' > 'chew' 3 'bite' > 'sting' 2 'chew' > 'gnaw' 1 'snap' > 'bite' 1 'snatch by mouth' > 'bite' 1 'to break' > 'bite' 2 'to mow' > 'to sting' 1 'to crack (nuts)' > 'to sting' 1 Borrowings: 1
Primary words: 7.1: bite: PCT *ïsïr- || PA *ìsú 'to break, to crush': PTM *(x)ise- 'to break, to beat ' || PKor * ɨ s ɨ r- 'to break, to crush' || PJap *ùsú 'mortar'. 7.2: bite: CT * dīš-lA-, a derivate from PCT *dīš 'tooth', lexical function Oper. The most common meaning is 'to take by teeth' > 'to chew', 'to eat' and 'to bite'. It seems that this is the Proto-Common Turkic meaning, and not many separate derivatives made by productive model: another productive Oper - *dīš-A- - reveals such meanings as "to shed teeth", "to dent" and "to whet a toothed instrument". 7.3: bite: CT *Kap- PMo *kab-či-, *kab-la- 'to clasp, to squeese' || PTung *xap-ki- 'to strangle'. 7.4: bite: PT *dāla- 'to bite, to tear by teeth' || PWA *tā́lV > PMo *dol[u] ɣ a- 'to lick'', PTung *dala- 'to lap, swill; to eat (about animals)'. 7.6: bite: Chuv. sïrt- < PT *jïrt- 'to tear to pieces', an intensive from the PT *jīr- / *jy ̄ r- 'tosplit, to break, to tear'. "to gnaw": PCT *gEmür-|| PA *kĕ̀ma 'to gnaw, bite', PMo *kemeli- 'to gnaw > to bite', PJap *kàm- 'to bite'. "to chew": PCT *čAj-na-|| PWA *čă ǯ V 'cheek, cheekbone' > PMo * ǯ a ǯ i-'to chew; part of cheek' || PTM * ǯ a ǯ i- 'cheekbone'.
The geography of reflexes of *ïsïr- (the periphery of CT) let us presume that it is the PCT "bite". It cannot be drawn to the PT layer, but it is interesting that in consideration of the Altaic parallels it is the same semantic derivate as the Chuvash word: "bite" < "break". The CT derivate *tīš-le- is the realization of the lexical function Oper from "tooth", so a more common semantic than "bite" shall be reconstructed - "to take, to clench one's teeth". The PT *Kap- 'to grasp, to take by mouth' shows the shift to "bite" as a entral innovation. The PT *dala- 'to bite, to tear by teeth', apparently is the most probable candidate to the PT "to bite"; The Altaic parallels speak for the common meaning "to eat (about animals)".
Conclusion The adduced exemples show the possibilities of formal semantic analysis of Swadesh words meaning. Only such method permits to accomplish a reconstruction of Swadesh list for a protolanguage. We can see by the way that as in any structured sphere of language, in historical semantics many processes disturbing the tree- shaped scheme of language developments work.