• 検索結果がありません。

Kawachi 2011 Can Ethiopian Languages be Considered as Languages in the African Linguistic Area

N/A
N/A
Protected

Academic year: 2018

シェア "Kawachi 2011 Can Ethiopian Languages be Considered as Languages in the African Linguistic Area"

Copied!
17
0
0

読み込み中.... (全文を見る)

全文

(1)

in the African Linguistic Area? The Case of Highland

East Cushitic

Kazuhiro KAWACHI

1. Introduction1

Heine & Leyew (2008) (henceforth, H & L) and Heine (2009) address the issue of whether or not Africa as a whole is a linguistic area (Greenberg 1959, 1983). They propose that there are linguistic properties that characterize (Sub-Saharan) African languages, only a few of which are found in the rest of the world. H & L demonstrate that languages in the four major African language phyla share many of these properties if they are spoken in geographically close areas. Thus, H & L’s approach shows similarities between languages that are genetically different but geographically close. However, east African languages, especially those of Ethiopia, are known to be different from other African languages in various respects (e.g., Ferguson 1970, 1976, Tosco 2000, Bisang 2006, Crass & Meyer 2008). The present study examines whether Highland East Cushitic languages (henceforth, HEC languages), especially Sidaama (Sidamo) and Kambaata, should be considered as languages in the African linguistic area proposed by H & L, and discusses the implications of H & L’s approach for these languages. HEC languages have a small number

1 I would like to express my deepest thanks to my Sidaama native speaker consultants, Abebayehu Aemero Tekleselassie, Hailu Gudura, Leggese Gudura, Yehualaeshet Aschenaki, Sileshi Workeneh, and Solomon Shaamanna. I would also like to convey my cordial gratitude to Yvonne Treis for providing very insightful comments on various portions of the present paper. I am profoundly grateful to Abebayehu Aemero Tekleselassie for his comments on Amharic, Yvonne Treis for her comments on Kambaata, and to Motomichi Wakasa for his comments on Amharic and Wolaytta. My sincere thanks also go to the audience of my presentation at the symposium, espe- cially Sasha Aikhenvald and Tom Güldemann, for their comments on my presenta- tion. The present study was partially supported by the Research Institute for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies, the Mark Diamond Research Fund, the Department of Linguistics at the University at Buffalo, and Grant-in-Aid for Scientifi c-Research Program (C) sponsored by the Japan Society for the Promotion of Science (Kaken Research Project Number: 21520431).

Abbreviations. ABL: ablative, Adj: adjective, CON: converb (connective), Dem: adnominal demonstrative, Gen: genitive noun phrase, GEN: genitive, LV: lengthened vowel, Num: numeral, Rel: relative clause

(2)

of the pan-African grammatical properties proposed by H & L, and seem to be peripheral members of the category of languages in the African linguistic area. However, if examined in detail, the properties of HEC languages are shown to be unidentical to their counterparts in other African languages. HEC languages are substantially different from languages in the African linguistic area, and therefore, it is diffi cult to include them as members. This study also points out problems with H & L’s quantitative approach in defi ning a linguistic area, including the selection of a set of properties and possible misrepresentation of the relationship between marginal members of the linguistic area.

2. Literature Review

H & L defi ne linguistic areas as follows (H & L 2008:16, 35, Heine 2009:3) (the italicized part in (1c) is added at the conclusion in H & L 2008):

(1) (a) There are a number of languages spoken in one and the same general area.

(b) The languages share a set of linguistic features whose presence can be explained with reference neither to genetic relationship, drift, universal constraints on lan- guage structure or language development, nor to chance.

(c) This set of features is not found at a comparable quantitative magnitude in langua- ges outside the area.

(d) On account of (b), the presence of these features must be the result of language contact.

They propose that there is a set of properties characteristically found in Africa as a linguistic area. According to H & L (2008:19) and Heine (2009:4), these properties are (i) “common in Africa but clearly less common elsewhere”, (ii) “found, at least to some extent, in all major geographical regions of Africa south of the Sahara”, and (iii) “found in two or more of the four African language phyla”. Specifi cally, on the basis of previous studies such as Greenberg (1959, 1983), they propose that the eleven properties in (2) are found more in African than non-African languages:

(2) [1] Labial-velar stops [2] Implosive stops

[3] Lexical and/or grammatical tones [4] ATR-based vowel harmony [5] Verbal derivational suffi xes

[6] Nominal modifi ers follow the noun.

[7] Semantic polysemy: ‘drink/pull’ and ‘smoke’ [8] Semantic polysemy: ‘hear/see’ and ‘understand’ [9] Semantic polysemy: ‘animal’ and ‘meat’

(3)

[10] Comparative construction based on the action schema, ‘X is big defeats/surpasses/ passes Y’ or ‘X defeats/surpasses/passes Y in size’

[11] Noun ‘child’ used productively to express diminutive meaning

They state that the African languages they examine have an average of 6.8 of the eleven properties, but the languages in the other areas that they have investigated contain far fewer properties (2.6 in languages of the world minus Africa; 1.1 in Europe, 2.6 in Asia, 3.0 in Australia/Oceania, and 3.4 in the Americas; 2.3 in pidgins and creoles). They also fi nd that languages in sub-Saharan Africa average an even higher number of the properties (7.2), but those in Northeastern Africa average a much lower number (3.7). More specifi cally, according to H & L (2008:30), Western Chadic, Gur (Voltaic), and some Plateau and Guang languages have nine to ten of the properties in (2); languages in the Cameroon-Central Africa area have up to nine; and Atlantic, Mande, and Bantu languages have around six. However, languages in the Ethiopian highlands and northern (Berber) Africa have fewer than fi ve. H & L further fi nd that two-thirds of the African languages examined have six of the eleven properties, [3], [5]–[8], and [10]. About half of the languages have [11], but only a minority have [1], [2], [4], and [9]. Note that as H & L state, some properties included in the set are common outside Africa: [5] is more common in the Americas and [11] is more common in South America than in Africa, and [6] is as common in the Americas and Australia/Oceania as it is in Africa. I will return to these points later.

Thus, H & L argue that it is possible to defi ne Africa as a linguistic area using quantitative terms—specifi cally that African languages contain a suffi ciently large number of the properties listed in (2) to constitute a linguistic area, even though individual properties in the set are also found outside Africa. First, I will examine the implications of the quantitative approach for languages that have only a few of the properties in (2). Second, I ask whether it is appropriate to defi ne a linguistic area using quantitative criteria.

3. Should HEC Languages be Considered as Languages in the African Linguistic Area?

HEC languages are spoken in south-central Ethiopia and belong to the Cushitic language family, one of the language families subsumed under the Afro-Asiatic language phylum. They consist of Hadiyya, Kambaata, Sidaama, Gedeo (Darasa), and Burji. According to Hudson (2009), there are two pairs of HEC languages—(i) Hadiyya and Kambaata and (ii) Sidaama and Gedeo—which are closely related to one another, but Burji, which seems to have separated from the other four in the past, is somewhat different from the rest. Nevertheless, all the HEC languages are known to share various essential

(4)

grammatical properties and to have many lexical similarities (Hudson 1976). They have SOV word order and use suffi xes both for infl ection and derivation (and suprafi xes for case marking on nouns and adjectives). Because the languages have only recently begun to be written, there is no record of their states in the remote past. To date, they have been described by only a few researchers (e.g., Hudson 1976, Teferra 2000, Kawachi 2007, 2008, Treis 2008). The present study, which is largely based on Yvonne Treis’ data and my own, focuses on Kambaata and Sidaama, and hereafter refers to these two languages as HEC languages.

3.1. African Properties in HEC Languages

Unlike many other African languages, HEC languages have only up to two or four of the eleven properties put forth by H & L. Specifi cally, the current style of Kambaata has only [5] (verbal derivational suffi xes), although the old-fashioned style has both [5] and [7] (semantic polysemy between

‘drink’ and ‘smoke’). In addition to [5] and [7], Sidaama has [2] (implosive stop) and [10] (comparative construction based on the action schema). Yet when these properties are analyzed qualitatively, these languages do not seem as similar to other African languages as H & L’s quantitative analysis suggests. Section 3.1.1 discusses which of the eleven properties are not found in HEC languages. Section 3.1.2 discusses which of the properties are found (albeit in restricted form) and describes how their occurrence differs from that in other African languages.

3.1.1. African Properties Not Found in HEC Languages

HEC languages do not have [1], [3], [4], [6], [8], [9], or [11]. They have no labial-velar stops ([1]) and no ATR-based vowel harmony ([4]). Also, they show no semantic polysemy between ‘animal’ and ‘meat’ ([9]) (Hudson 1989). There is no general word for ‘animal’; instead, these languages use ‘wild animal, beast’ (Sidaama: moiččo, Kambaata: moočču) or ‘cattle’ (Sidaama: lalo, Kambaata: lalu). Neither of these words can mean ‘meat’, which is always expressed by another word (Sidaama and Kambaata: maala).

Furthermore, HEC languages do not use any of their nouns for ‘child’ to convey a diminutive meaning ([11]) (Sidaama: beetto ‘child’, k’aakk’o ‘baby’; Kambaata: oosuta ‘children’, c’iila(-ta) ‘baby, infant’). They also seem to show no semantic polysemy between ‘hear/see’ and ‘understand’ ([8]), though the verb for ‘see, look’ (Sidaama: la’-, Kambaata: t’uud-) can mean ‘check’ when it takes a complement clause, and the verb for ‘become visible (to an experiencer in the dative)’ (Sidaama: leell-, Kambaata: lall-) can be used for

‘become understandable (to an experiencer in the dative)’ in some contexts (Kambaata data are from Yvonne Treis 2010 and p.c.).

(5)

HEC languages also differ from other African languages in their word order. Nominal modifi ers precede rather than follow the noun ([6]). With the exception of the GenN order, the orders of the noun and its various types of modifi ers are not correlated to OV order (Dryer 1988, 1991, 1992, 2007), and thus they are discussed here as indications of some of the differences between HEC languages and languages of the African linguistic area (the following information is taken from Dryer in Haspelmath et al. 2005:Chapters 87, 88, 89, 96, and 97)2. The AdjN, DemN, and NumN orders found in HEC languages are also common in languages of southwest Europe, the Middle East, and many parts of Asia and the Americas, but uncommon in other parts of Africa. The AdjN order is common among OV languages only in Eurasia but is rarely found in OV languages in other parts of Africa. OV languages in Asia tend to have the RelN order, whereas those in Africa tend to have the NRel order. Thus, in regard to the orders of the noun and its modifi ers, HEC languages differ from central members of the African linguistic area.

Furthermore, HEC languages do not have a lexical or grammatical tone ([3]). They are often described as stress languages in which one syllable of each word is stressed (e.g., Hudson 1976, Teferra 2000). In fact, however, they use pitch to indicate the prominence of a vowel segment or syllable. Sidaama is a pitch-accent language in which the penultimate vowel segment of the citation form of a noun or an adjective has high pitch irrespective of the vowel length. (A high pitch on the penultimate vowel segment is not marked in this paper, though it could be represented as in míne ‘house’, beétto

‘child’, and šiíma ‘smart’, for example.) Thus, the location of a high pitch in a noun or adjective is predictable and unassociated with vowel duration. Pitch has grammatical signifi cance in that the accusative or genitive case-marking suprafi x is realized as a high pitch on the fi nal vowel segment of an unmodifi ed noun or an adjective, as seen in (4) and (5). Note that a modifi ed noun is not marked with the suprafi x and as a whole has a fl at pitch (e.g., beetto in ise šiimá beetto la’-’-ino. ‘She saw the small child’; min-i in (5)).

(4) ise beettó la’-’-ino.

3SG.F.NOM child.ACC see-3SG.F-PERF.3

‘She saw the child.’

(5) beet-ú min-i šiima=ho. child-GEN.M house-NOM.M small=PRED.M

‘The boy’s house is small.’

2 Note that although there is a correlation between VO and NRel order, there is no such relation between OV and the order of relative clause and noun (Dryer 1988, 1991, 1992, 2007).

(6)

Also, according to Treis (2008), Kambaata is a language in which every word has one syllable that is prominent in terms of both pitch and loudness, and the location of the accented syllable of a noun shifts depending on its case.

3.1.2. African Properties Found in HEC languages

At most, two of the proposed African properties, verbal derivational suffi xes ([5]) and semantic polysemy between ‘drink’ and ‘smoke’ ([7]), are found in Kambaata, and these as well as two additional properties, the implosive stop ([2]) and a comparative construction based on the action schema ([10]), are found in Sidaama. However, as H & L state, [5] and [7] are widespread in languages of the world, and [2] is not a common property found in African languages, though [10] is common in Africa but not outside Africa. Moreover, [2] and [10] in Sidaama are qualitatively different from their counterparts in other African languages, and [7] is only found in restricted expressions. This section examines these four properties.

HEC languages have a set of verbal derivational suffi xes (Hudson 1976): the suffi x that verbalizes nouns, the suffi x that verbalizes adjectives, the causative and double-causative suffi xes, the reciprocal suffi x (at least in Sidaama), the passive voice suffi x, and the middle voice suffi x.

HEC languages may idiomatically use the verb for ‘drink’ in the sense of ‘smoke’ (Sidaama: ɡayyá aɡ- [cigarette.ACC drink-] ‘smoke a cigarette’). However, in Kambaata, this use is considered a little old-fashioned and is usually found in the speech of older generations or speakers in the countryside (Yvonne Treis p.c.). Both languages have a verb for ‘smoke’ wiliišš-, which is used in a wider range of contexts.

An implosive stop occurs in Sidaama (and also Gedeo and Burji), but not in Kambaata (or Hadiyya) (Hudson 1976). Normally, a language with only one implosive has /ɓ/, but Sidaama is unusual in that its only implosive is the dental /ɗ/ (Greenberg 1970). It is unclear whether or not the presence of anɗɗ implosive in Sidaama is due to language contact. The implosive /ɗ/ may be aɗɗ tradition of HEC languages. Hudson (1976) hypothesizes that /ɗ/ historicallyɗɗ turned into the glottal stop in Kambaata (and Hadiyya). Also, Lowland dialects of Sidaama, which have been infl uenced by other languages and are thus considered to be less authentic than Highland ones, have replaced /ɗ/ with /t’/ ɗɗ (e.g., Highland: baɗɗeessa,ɗɗ Lowland: batt’eessa ‘rainy season’; Highland: ɗiw-an-t-ino,

ɗɗ Lowland: t’iw-an-t-ino [cause.sickness-PASS-3SG.F-PERF.3]

‘She is/got sick’).

Sidaama has comparative constructions that use the action schema ‘X is big defeats/surpasses/passes Y’/‘X defeats/surpasses/passes Y in size’. However, these are not the only comparative constructions available. According

(7)

to Heine (1994) and Leyew & Heine (2003), some Ethiopian languages use only the action schema or the source schema for comparative constructions, but others use both, perhaps as the result of language contact due to the location of Ethiopia in an overlapping zone between Asia and Africa. Sidaama has not only comparative constructions that use either the action schema or the source schema, but also those that use both simultaneously. When instances of an action or a state/state change are compared in a sentence with an action or state-change verb predicate, constructions with roor- ‘exceed’, those with the ablative suffi x -nni for a noun phrase for the standard of comparison, and those with both are all possible, as in (6a)–(6f).

(6) (a) The verb roor- ‘exceed’ is the main verb, the action or state is expressed with a noun phrase, and the noun phrase for the standard of comparison is in the accusa- tive or ablative case.

(b) The action or state-change verb is the main verb, the verb roor- ‘exceed’ is a subor- dinate verb, and the noun phrase for the standard of comparison is in the accusative or ablative case.

(c) The action or state-change verb is the main verb and is modifi ed by the adverb roore ‘exceedingly’ (derived from roor-), and the noun phrase for the standard of comparison is in the accusative or ablative case.

(d) Neither schema is used. The action or state-change verb is the main verb, and the noun ale ‘aboveness’ is modifi ed by the genitive form of the noun phrase for the standard of comparison.

(e) The noun phrase for the standard of comparison is in the ablative case, the action or state-change verb is the main verb, and the noun ale ‘aboveness’ is adverbially used to modify the verb.

(f) The noun phrase for the standard of comparison is in the ablative case, and the action or state-change verb is the main verb.

The constructions in (6a)–(6f) are arranged in the order of traditionality or authenticity. In each of (6a)–(6c), the noun phrase for the standard of comparison can be in either the accusative or ablative case, but the use of accusative case is considered more traditional and authentic. (6a) with the standard of comparison in the accusative case seems to be considered the most traditional and authentic Sidaama construction and is also the most common; (6f) is the least traditional and is used only informally. The set of comparative constructions in (6a)–(6f) are illustrated below in (7a)–(7f), in that order, using the state-change verb busul- ‘become smart’. The perfective form of busul- can be used to express the current state of being smart as well as the already completed state change of becoming smart. Action verbs are used in the same way as state-change verbs—for example, in order to mean that

(8)

she runs better (faster) than him, busule-te-nni in (7a) is replaced by dod-a- te-nni [run-INF-GEN.F-LOC], where the infi nitive form of the action verb is followed by the genitive and locative suffi xes. Similarly, busul-t-anno in (7b)–(7f) is replaced by the imperfective form of the action verb ‘running’, dod-d-anno [run-3SG.F-IMPERF.3]. Note that the suffi x -nni is locative when attached to some locational nouns, and ablative when attached to other nouns or pronouns in the genitive case.

(7) (a) ise isó/isí-nni-i busule-te-nni

3SG.F.NOM 3SG.M.ACC/3SG.M.GEN-ABL-LV smartness-GEN.F-LOC roor-t-anno.

exceed-3SG.F-IMPERF.3

(b) ise isó/isí-nni-i roor-t-e

3SG.F.NOM 3SG.M.ACC/3SG.M.GEN-ABL-LV exceed-3SG.F-CON busul-t-anno.

become.smart-3SG.F-IMPERF.3

(c) ise isó/isí-nni-i roore 3SG.F.NOM 3SG.M.ACC/3SG.M.GEN-ABL-LV exceedingly busul-t-anno.

become.smart-3SG.F-IMPERF.3

(d) ise isí ale-e-nni busul-t-anno.

3SG.F.NOM 3SG.M.GEN aboveness-LV-LOC become.smart-3SG.F-IMPERF.3 (e) ise isí-nni-i ale-e

3SG.F.NOM 3SG.M.GEN-ABL-LV aboveness-LV busul-t-anno.

become.smart-3SG.F-IMPERF.3

(f) ise isí-nni-i busul-t-anno.

3SG.F.NOM 3SG.M.GEN-ABL-LV become.smart-3SG.F-IMPERF.3 ‘She is/will become smarter than him.’

When a state is expressed by a predicative adjective, counterparts of only one of the subconstructions of (6c) (with the ablative case standard) and the constructions in (6d)–(6f), which are each shown below as (6c’)–(6f’), can be used:

(6’) (c) The predicative adjective is modifi ed by the adverb roore ‘exceedingly’ (derived from roor-), and the noun phrase for the standard is in the ablative case (not the accusative case).

(d) Neither schema is used. The noun for ‘aboveness’ is modifi ed by the genitive noun phrase for the standard of comparison.

(9)

(e) The noun phrase for the standard is in the ablative case, and the noun for ‘above- ness’ is adverbially used to modify the predicative adjective.

(f) The noun phrase for the standard is in the ablative case.

To say ‘She is smarter than him’ using the adjective busule ‘smart’ in the predicate, busul-t-anno in (7c)-(7f) is replaced by busule=te [smart=PRED.F]. (Note that in the counterpart of (7c), isó [3SG.M.ACC] cannot be used; isí-nni-i [3SG.M.GEN-ABL-LV] must be used instead.)

Thus, the constructions that use one of the schemas and those that use both coexist in Sidaama. It is not known how these constructions have developed historically. Construction (6f), which uses only the source schema, may have been recently introduced from neighboring languages.

However, Kambaata, which uses the source schema for comparative constructions, does not seem to have a comparative construction with the action schema (Yvonne Treis p.c.). There is no Kambaata verb cognate to the Sidaama roor-. Although Kambaata does have the verb abbis- ‘exceed’, this verb cannot be used to form a comparative construction3. Admittedly, it is possible that the Kambaata comparative construction using the action schema is accidentally absent from Treis’ data; however, even if this were the case, it would be much less common than the predominant construction, which uses the source schema.

As it has been shown, Kambaata has at most two of the eleven properties put forth by H & L as characteristics of languages in the African linguistic area: [5] (verbal derivational suffi xes) and, in slightly old-fashioned speech, [7] (the use of the verb for ‘drink’ for ‘smoke’). In addition to these two properties, Sidaama has two properties, [2] (implosive stop) and [10] (comparative construction based on the action schema), but they differ qualitatively from their counterparts in other African languages. Therefore, HEC languages are not good candidates for inclusion in the African linguistic area. In fact, if African languages are characterized by the properties in (2), HEC languages may be considered outside the African linguistic area.

3.2. Challenges to the Quantitative Approach to Defi ning Linguistic Areas As shown in section 3.1, two of the HEC languages, Kambaata and Sidaama, have only a few of the properties that H & L propose are pan-African. In one view, the further the languages are located from the center of the region, the fewer are the properties that they possess of their linguistic area, and Ethiopian languages exhibit a small number of properties because they

3 Nevertheless, its converb form can be used as a degree adverbial for ‘much’, for example, optionally in a comparative sentence (Yvonne Treis p.c.).

(10)

are peripheral members of the African linguistic area (H & L 2008:30). Alternatively, these languages may be excluded from the African linguistic area because of the qualitative differences of the properties that they have compared to those that central members of the area have, as well as the small number of pan-African properties that they contain.

However, the number of properties found in Kambaata (one or possibly two) and that found in Sidaama (four) do not say anything about whether they should be included or excluded from the African linguistic area. It is problematic to delineate a linguistic area using a quantitative method: any number, if used as a criterion for the inclusion of a language in the given linguistic area, would be strictly arbitrary. Moreover, the size of the linguistic area would grow or shrink on the basis of the number of properties in the set and the minimum number needed for inclusion. The more properties are combined to form a set that is used to characterize a linguistic area, the smaller the area becomes (if languages must have a large number of the properties in order to be considered members of the area). The area grows wider if languages with a small number of the specifi ed properties are included. In the latter case, because of a part of the defi nition of a linguistic area, namely, (1a) (“There are a number of languages spoken in one and the same general area”), as long as languages are spoken in (a contiguous part of) the same continent, all of them could be regarded as languages in the same linguistic area, or an even larger area could be regarded as a linguistic area. Using this method, it is diffi cult to determine where the African linguistic area ends, unless it can be shown that languages in adjacent surrounding regions (for example, in the Middle East and South Europe) have signifi cantly fewer of the specifi ed properties (or none at all).

H & L seem to be aware of the limits to this methodological approach. They write that isopleth mapping conducted by means of their quantitative approach to linguistic areas is discouraging in two respects (H & L 2008:31). First, one can approximate the structure of a linguistic area even without adopting a quantitative approach. Second, the quantitative analysis does not contribute much to the reconstruction of linguistic history or history in general.

The quantitative approach faces other challenges, which I will explore in the subsequent sections. Section 3.2.1 discusses the problem of how to determine which properties should be included in the set that defi nes a given linguistic area. Section 3.2.2 brings up the issue of how to treat marginal members of an area. When all these issues are taken into account, HEC languages are shown to have little in common with the languages in the African linguistic area.

(11)

3.2.1. Selection of a Set of Properties Found in a Linguistic Area

H & L state that what characterizes Africa as a linguistic area is not any individual property in (2), but “the combinations of these properties” (H & L 2008:35). However, the question that arises is how miscellaneous properties can be combined to form a characterization of the area. In fact, many of the eleven properties are not characteristic of African languages at all. This point has been noted by H & L and is raised again below along with related issues that pertain to languages that have only a few of the properties that characterize their linguistic area.

Property [2] is not common in Africa, and some of the properties listed in set (2), specifi cally, [5], [6], [7], [8], [9], and [11], seem to be common in other parts of the world, as H & L admit. In fact, only some of the properties, such as [1], [3], [4], and [10], may be genuine areal properties of Africa.

Verbal derivational suffi xes ([5]) are by no means restricted to Africa; they are quite common all over the world. The causative suffi x, for example, is widespread (Song 1996). Hence, as H & L themselves note, it is not clear how the existence of verbal derivational suffi xes can be regarded as an African areal property (or, for that matter, can serve as a criterion for any linguistic area).

The placement of nominal modifi ers after the noun ([6]) is likewise found in many parts of the world (Dryer 1988, 1991, 1992, 2007). The NRel, NGen, and NAdj orders extend into Europe and Asia. Only the NNum and NDem orders seem to be areal properties of Africa: although they are common both in and outside of Africa, they are not commonly found in regions immediately contiguous to Africa.

Properties [7], [8], [9], and [11] are also common outside Africa. Semantic polysemy between ‘drink/pull’ and ‘smoke’ ([7]) has been reported in other languages (e.g., in languages of New Guinea; see Aikhenvald 2009:102–106). Semantic polysemy between ‘hear/see’ and ‘understand’ ([8]) and between

‘animal’ and ‘meat’ ([9]) in non-African languages are also described in Vanhove (2008) and Boyeldieu (2008), respectively. Furthermore, the productive use of the noun for ‘child’ to convey a diminutive meaning ([11]) does not appear to be unique to African languages at all. In fact, Jurafsky (1996), in a cross- linguistic study of diminutive markers, fi nds that they often come from the word for ‘child’. He hypothesizes that the notion of ‘child’ is at the center of the category of diminutive meanings. Hence, these six properties are not exclusive to Africa even though they can be found in many African languages.

When a language does not have many of the properties that are said to characterize a linguistic area, it is necessary to examine the individual language properties in order to determine if the language should be included in the area. In examining the case of HEC languages, we see that verbal derivational suffi xes ([5]) and semantic polysemy between ‘drink’ and ‘smoke’ ([7]) are common

(12)

outside Africa, and the occurrence of an implosive stop ([2]) is not common even in Africa. Thus, it seems that the only African property found in HEC languages is the Sidaama comparative construction based on the action schema ([10]). When this is taken into consideration, HEC languages appear to have very little in common with central members of the African linguistic area.

Even if a given property is common both in and outside Africa, it can still be considered an African property as long as it is not prevalent everywhere and is not common in any region that immediately neighbors Africa. However, H & L’s descriptions do not show that this is the case for the properties in (2).

Finally, the use of a set of properties to defi ne a linguistic area is problematic in other ways. In the quantitative approach, the properties that characterize the languages in a given linguistic area are each assumed to occur independently rather than in correlation with other properties—and to be equally important. However, it is obviously diffi cult to determine whether or not individual properties play roles of the same importance in characterizing a linguistic area. For example, the use of a lexeme for a certain sense (e.g., [7]: whether or not the verb for ‘drink/pull’ is used to mean ‘smoke’ in a language) does not seem to be as important as a property that pertains to the grammatical system as a whole (e.g., [5]: the use of verbal derivational suffi xes). There is also the problem of whether each property can be treated as a binary feature. 3.2.2. Peripheral Members of Languages in a Linguistic Area

Another challenge to the use of the quantitative approach in defi ning a linguistic area is the matter of how to treat peripheral members. As H & L (2008:31–34) demonstrate for languages in northern Nigeria, the quantitative approach is useful for showing that languages of different genetic affi liations spoken in neighboring regions share many properties if they are central members of the linguistic area. However, unless a detailed qualitative analysis is also conducted, this approach has the potential to give only a limited amount of understanding of the relationships between peripheral members of a linguistic area and the differences between peripheral members and languages outside the area. In the following discussion, it is assumed that Ethiopian languages, including HEC languages, are peripheral members of the African linguistic area.

Sidaama and Kambaata share the same genetic origin, and are spoken in very close (though not contiguous) regions, and are thus very similar to each other in various respects. For example, they show 62% lexical similarity (Bender 1971). However, as shown in section 3.1, they have (at most) two and four out of the eleven properties in (2), respectively. Four is within the range of the number of the properties that H & L expect languages in the Ethiopian highlands to have (“fewer than fi ve”); however, two is probably not. It is lower

(13)

than the average number of properties found in languages outside Africa (2.6), in Asia (2.6), and in the Americas (2.3). Only one of the properties is found in present-day Kambaata, and this is about the same as the average number in Europe (1.1), the area with the lowest average examined by H & L. The difference between the number of African properties found in Sidaama (four) versus Kambaata (one or two) is a difference of either two or three, a gap that is great relative to these small numbers. A variance of two or three in the number of African properties contained by central members is not as signifi cant given that these numbers are larger to begin with; the difference is more meaningful for Sidaama and Kambaata. However, given the data, it does not make sense to say that Sidaama is a member of the African linguistic area but Kambaata is not. Thus, a purely quantitative analysis can misrepresent the relationship between peripheral members.

Sidaama and Kambaata differ in as many as two or three of the properties. The use of the verb for ‘drink’ to mean ‘smoke’ is found in both Sidaama and slightly old-fashioned Kambaata, but not in current Kambaata. Also, Kambaata might have had but then lost the dental implosive, which is still present in Sidaama. Therefore, the only acute difference between the two may be the use of the action schema in comparative constructions.

Data from other Ethiopian languages that also have only small numbers of the properties in (2) are useful to illustrate this. As shown in Table 1, Amharic, an Ethio-Semitic language widely spoken as the lingua franca in Ethiopia, has at most one of the eleven properties in (2) (Abebayehu Aemero Tekleselassie p.c., Motomichi Wakasa p.c.), and Wolaytta, an Omotic language spoken immediately to the south of the Kambaata-Xambaaro zone and immediately to the west of the Sidaama zone, has only two or three of the properties (Motomichi Wakasa p.c.). Both languages are distantly related to and geographically close to HEC languages.4

Table 1. African properties in Sidaama, Kambaata, Amharic, and Wolaytta Sidaama Kambaata Amharic Wolaytta

[2] Implosive stops yes no no no

[3] Lexical and/or grammatical tones no no no yes

[5] Verbal derivational suffi xes yes yes no yes

[7] Semantic polysemy: ‘drink/pull’ and

‘smoke’

yes no/yes* no/yes* no/yes*

[10] Comparative construction based on the action schema

yes no no no

*: in (slightly) old-fashioned use

4 Note that there are researchers who treat as an implosive the consonant that Wakasa (2008) regards as a voiced glottalized dental stop, which he argues involves no ingressive airstream (Wakasa 2008:62-63).

(14)

In Amharic, the use of the verbs for ‘drink’ and ‘pull’ to mean ‘smoke’ was common in the past but does not occur in current usage (especially in formal speech). If we exclude this use from the analysis, Amharic has none of the properties in (2). Similarly, in the past, Wolaytta seems to have had a similar usage for the verb ‘drink’ to mean ‘smoke’, but it does not seem to be used currently (Wakasa p.c. states that the usage may accidentally be missing from his data). Wolaytta has a lexical tone (also a grammatical tone in the archaic form) as well as derivational suffi xes such as the causative suffi x and the passive suffi x (Wakasa 2008).

If the four languages are compared on the basis of the properties in (2), Kambaata looks more similar to Amharic and Wolaytta than to Sidaama. It differs from Amharic only in the presence of [5], and from Wolaytta only in the absence of [3]. However, the differences between Kambaata and Sidaama are overrepresented, and those between Kambaata on the one hand and Amharic and Wolaytta on the other are underrepresented here. Sidaama and Kambaata are so closely related and similar to each other that it would be diffi cult to convincingly claim that Kambaata’s similarities to Amharic and Wolaytta, which are due to language contact, are greater than its similarities to Sidaama, which are attributed to a shared genetic affi liation.

4. Conclusion

This study has shown that Kambaata and Sidaama have only one or two and four, respectively, of the eleven properties that H & L propose as properties characteristic of languages in the African linguistic area. These languages seem to be much less like the central members of languages in the African linguistic area than these numbers suggest. The present study has also pointed out the limitations of the quantitative approach to defi ning linguistic areas, namely, those involving the selection of a set of properties in a linguistic area and the treatment of peripheral members in a linguistic area. If these are also considered, HEC languages are even more different from languages in the African linguistic area.5

5 There are myths that HEC speakers originated from an area northeast of the African continent. There are grammatical properties that HEC languages (and other Ethiopian languages as well) have in common with languages in southern Europe and southwestern Asia, as is clear from many of the maps provided by Haspelmath et al. (2005). A pos- sible future research project would be to investigate and compare grammatical properties found in languages in these regions.

(15)

References

Aikhenvald, A. Y. 2009. “‘Eating’, ‘drinking’ and ‘smoking’: A generic verb and its semantics in Manambu.” The Linguistics of Eating and Drinking. Newman, John (ed.). John Benjamins 2009, 91-108.

Bender, M. L. 1971. The languages of Ethiopia: A new lexicostatistic clas- sifi cation and some problems of diffusion. Anthropological Linguistics, 13.5: 165-288.

Bisang, W. 2006. “Linguistic areas, language contact and typology: Some im- plications from the case of Ethiopia as a linguistic area.” Linguistic Areas: Convergence in Historical and Typological Perspective. Matras, Y., A. McMahon, and N. Vincent. (eds.). Hampshire: Palgrave MacMillan. Boyeldieu, P. 2008. “From semantic change to polysemy: The cases of ‘meat/

animal’ and ‘drink’.” From Polysemy to Semantic Change: Towards a Typology of Lexical Semantic Associations. Vanhove, M. John Benjamins 2008, 303-339.

Crass, J., and R. Meyer. 2008. “Ethiopia.” A Linguistic Geography of Africa. Heine, B., and D. Nurse (eds.) . Cambridge: Cambridge University Press 2008, 228-250.

Dryer, M. S. 1988. “Object-verb order and adjective-noun order: Dispelling a myth.” Lingua, 74, 185-217.

Dryer, M. S. 1991. “SVO languages and the OV/VO typology.” Journal of Linguistics, 27, 443-482.

Dryer, M. S. 1992. “The Greenbergian word order correlations.” Language, 68, 81-138.

Dryer, M. S. 2007. Word order. In T. Shopen (ed.) Clause Structure, Language Typology and Syntactic Description, Vol. 1, Second Edition. Cambridge University Press 2007, 61-131.

Ferguson, C. A. 1970. “The Ethiopian language area.” Journal of Ethiopian Studies, 8.2, 67-80.

Ferguson, C. A. 1976. “The Ethiopian language area.” Language in Ethiopia. Bender, M. L., J. D. Bowen, R. L. Cooper, and C. A. Ferguson. (eds.). London: Oxford University Press 1976, 63-76.

Greenberg, J. H. 1959. “Africa as a linguistic area.” Continuity and Change in African Cultures. W. R. Bascom and M. J. Herskovitz (eds.). Chicago: Chicago University Press, 15-27.

Greenberg, J. H. 1983. “Some areal characteristics of African languages.” Current Approaches to African Linguistics, Vol. 1. Dihoff, I. R. (ed.). Dordrecht: Foris 1983, 3-21.

Haspelmath, M., M. S. Dryer, D. Gil, and B. Comrie. (eds.) 2005. The World Atlas of Language Structures. Oxford: Oxford University Press.

(16)

Heine, B. 1994. “Areal infl uence on grammaticalization.” Language Contact and Language Confl ict. Pütz, M. (ed.). John Benjamins 1994, 55-68. Heine, B. 2009. “Africa as a linguistic area.” Concise Encyclopedia of

Languages of the World. Keith, B., and S. Ogilvie (eds.). Oxford: Elsevier 2009, 3-7.

Heine, B., and Z. Leyew. 2008. “Is Africa a linguistic area?” A Linguistic Geography of Africa. Heine, B., and D. Nurse (eds.). Cambridge: Cambidge University Press 2008, 15-35.

Hudson, G. 1976. “Highland East Cushitic.” The Non-Semitic Languages of Ethiopia. Bender, M. L. (ed.). East Lansing, MI: African Studies Center, Michigan State University 1976, 232-277.

Hudson, G. 1989. Highland East Cushitic Dictionary. Hamburg: Buske. Hudson, G. 2009. “Highland East Cushitic languages.” Concise Encyclopedia

of Languages of the World. Brown, K. and S. Ogilvie. (eds.). Oxford: Elsevier, 2009, 488-492.

Jurafsky, D. 1966. “Universal tendencies in the semantics of the diminutive.” Language, 72.3, 533-578.

Kawachi, K. 2007. A Grammar of Sidaama (Sidamo), a Cushitic Language of Ethiopia. Ph.D. Dissertation. University at Buffalo, the State University of New York.

Kawachi, K. 2008. “Middle voice and refl exive in Sidaama.” Proceedings of the 40th Annual Meeting of the Chicago Linguistic Society (2004), Volume 2: Panels. Adams, N., A. Cooper, F. Parrill, and T. Wier (eds.). Chicago, IL: Chicago Linguistic Society. 119-134.

Leyew, Z., and B. Heine. 2003. “Comparative constructions in Africa: an areal dimension.” Annual Publication in African Linguistics 1: 47-68.

Song, J. J. 1996. Causatives and Causation: a Universal-Typological Perspective. New York: Longman.

Teferra, A. 2000. A Grammar of Sidaama. Ph.D. Dissertation. Hebrew University of Jerusalem.

Tosco, M. 2000. “Is there an ‘Ethiopian language area’?” Anthropological Linguistics 42.3: 329-365.

Treis, Y. 2008. A Grammar of Kambaata, Part I: Phonology, Nominal Morphology and Non-verbal Predication. Köln: Rüdiger Köppe Verlag. Treis Y. 2010. Perception verbs and taste adjectives in Kambaata and beyond.

Perception of the Invisible. Religion, Historical Semantics and the Role of Perceptive Verbs (SUGIA – Sprache und Geschichte in Afrika, 21). Storch, A. (ed.). Cologne: Köppe, 2010, 313-346

(17)

Vanhove, M. 2008. “Semantic associations between sensory modalities, pre- hension and mental perceptions.” From Polysemy to Semantic Change: Towards a Typology of Lexical Semantic Associations. Vanhove, M. John Benjamins 2008, 341-370.

Wakasa, M. 2008. A Descriptive Study of the Modern Wolaytta Language. Ph.D. Dissertation. The University of Tokyo. (available at: http://www.world- lang.osaka-u.ac.jp/user/liccosec/africa/AF_Wakasa_DrDissertation.pdf)

Table 1. African properties in Sidaama, Kambaata, Amharic, and Wolaytta

参照

関連したドキュメント

These abstract machines are inspired by Girard’s Geometry of Interaction, and model program execution as dynamic rewriting of graph representation of a pro- gram, guided and

Lemma4.1.. This is not true if f is not positively homogeneous as the following example shows.. Let f be positively homogeneous. We shall give an example later to show that

Lomadze, On the number of representations of numbers by positive quadratic forms with six variables.. (Russian)

For the multiparameter regular variation associated with the convergence of the Gaussian high risk scenarios we need the full symmetry group G , which includes the rotations around

All (4 × 4) rank one solutions of the Yang equation with rational vacuum curve with ordinary double point are gauge equivalent to the Cherednik solution.. The Cherednik and the

We show that a discrete fixed point theorem of Eilenberg is equivalent to the restriction of the contraction principle to the class of non-Archimedean bounded metric spaces.. We

Instead an elementary random occurrence will be denoted by the variable (though unpredictable) element x of the (now Cartesian) sample space, and a general random variable will

Thanks to this correspondence, formula (2.4) can be read as a relation between area of bargraphs and the number of palindromic bargraphs. In fact, since the area of a bargraph..