Semantic Shift and Variation in Formosan Languages

Abstract

This brief study examines a number of types of semantic shift and variation in the Formosan languages. Factors behind semantic shift, such as function, shape, and geography, are discussed and exemplified, revealing much semantic variation in Formosan languages. Cognate sets are analyzed to consider their original semantics.

Headnote

This brief study examines a number of types of semantic shift and variation in the Formosan languages. Factors behind semantic shift, such as function, shape, and geography, are discussed and exemplified, revealing much semantic variation in Formosan languages. Cognate sets are analyzed to consider their original semantics.

Key words: factors, Formosan languages, semantic shift, variation

...

(ProQuest: ... denotes non-US-ASCII text omitted.)

1. Introduction

Variations in Formosan languages are found at all levels of language, including lexicon, phonol- ogy, morpho-syntax, and semantics. Phonological and lexical variations among different Formosan dialects have been reported for Rukai (Li 1977; Zeitoun 1993), Paiwan (Ho 1978), Puyuma (Ting 1978), Tsou (Li 1979), Thao (Li 1983), Bunun (Li 1988), and Atayal (Li 1998). There are a few studies concerned with morpho-syntactic variations, for example Huang's (1995) comparison of the syntactic structures of two Atayal dialects and Teng's (2011) paper on noun phrase conjunction in three Puyuma dialects.

Variations can also be found due to age (Li 1982a), sex (Li 1982b), or social class (Rau 2000) within the same speech community. Work of this type has been done more carefully for the Atayalic group; such investigation for other Formosan languages is a desideratum.

So far there has been no study of semantic variation in Formosan. This paper will examine such shifts and variations in Formosan and attempt to work out some generalizations.

2. Types of semantic shift

According to Bloomfield (1933:426-427), semantic shift may be of the following nine types: (1) narrowing, (2) widening, (3) metaphor, (4) metonymy (where meanings are near each other in space or time), (5) synecdoche (where meanings are related as whole and part), (6) hyperbole, (7) litotes, (8) degeneration, and (9) elevation. These types are generalized from his study of Indo-European languages. They may not be sufficient to account for all sorts of semantic shifts encountered in Austronesian languages. Blust (2009:322ff.) illustrates six types of semantic change in Austronesian languages: (1) interchange of prototype and category, (2) change of physical environ- ment, (3) reduced importance of the referent, (4) semantic fragmentation, (5) semantic chaining, and (6) avoidance; and Blust (2010) discusses five patterns of semantic change in Austronesian.

Instead of adopting either Bloomfield's classification of semantic shift or Blust's, I shall discuss some possible factors in semantic shift, such as function, shape, and geography, as an alternative way of analyzing it, illustrating them with examples from Formosan languages, while referring passim to instances of semantic shift taken from Bloomfield and Blust.

2.1 Function as a factor

Function is a factor in semantic shift that Bloomfield did not consider. For example, there are semantic shifts in both PAN cognates *Rumaq 'house' and *taRuqan 'hut', as attested in Formosan languages:1

Tso rmoo, Bun lumaq, Puy ruma', Paz xuma, Ami lumaq 'house'; Pai umaq 'house, grave, lair of beasts'; Sar ruma'-a 'lair or den of beasts' < PAN *Rumaq (Tsuchida 1976:232)

Tso trova, Kan taru'ane, Sar taruan-a, RukBd taovan-ane, Bun taluqan, Puy taru'an, Tha talhuqan, Paz taxuan, Ami taluqan 'hut or shelter for hunting in mountains'; Sai taLoe'aen 'house' < *taRuqan (Tsuchida 1976:169)

Saaroa reflex of the first term has undergone a semantic shift from 'house' to 'lair', while the Saisiyat reflex of the second term has undergone a semantic shift from 'hut' to 'house', as is com- monly found in many languages of the world. The terms 'house', 'hut', 'den', 'lair', and 'grave' all have the same function of providing shelter. Burials from the archaeological sites of Nankuanli and Nankuanli East on the southwestern plain of Taiwan, dating back to around 5000 years BP, show that the inhabitants at that time buried their dead within the settlement, but not inside the house (Tsang, personal communication). The semantic shift from 'house' to 'grave' must have taken place in Paiwan, where the dead were buried in a house, so that house and grave had the same function for the Paiwan natives.

Interesting semantic shifts and variations are found across Formosan languages. For example, the variant derivations of PAN *qayam mostly mean 'bird', such as zom? in Tso, alhame in Sar, aza-azame in RukBd, qazam in BunTkb, qaya-qayam in PaiBt, 'ayam in PuyPn, ayam in Paz, qayam in Amis,4 alam in Kav, and aiam in Sir, adam 'bird of omen' in Fav, but both 'aeyam in Sai and 'alame in Kan mean 'meat' (Tsuchida 1976:168). In this case, a semantic shift may have taken place in the two Formosan languages. If the original meaning is 'bird', then the shift to 'meat' is from whole to part, and the Favorlang form adam referring to a 'bird of omen' is just a semantic narrowing.

Similarly, derivations of PAN *biRaq mostly mean 'leaf in general', including RukTa, RukLa bia, PuyPn bira', Tha filhaq, SaiTa biLae', Kav biRi, and Bab bia 'leaf '; Pai viaq 'leaf used in rites (the leaf could be of various plants)'; Ami filaq 'leaf used to wrap up betel nut for chewing'; but their corresponding forms in three Rukai dialects refer to 'Alocasia', namely RukMg bia, RukTn bi'a, RukMn vi'a. The semantic shift to a special type of plant has taken place in these three Rukai dialects in the same geographical area of Maolin, but not in other dialects of Rukai elsewhere or in any other Formosan language.5 Alocasia leaves were used to wrap up meat or fish. The semantic shift had to do with the traditional way of living. The original meaning of the item might refer to the most important type of leaf that had a special function in the traditional society.

A similar example can be seen in the semantic shift from 'a certain type of bamboo' to its functional usage as a 'spear', as shown below:

PAN *buluq > Kan vulu'u , RukTa boLo , Puy buLu' , Sai boLoe' , Paz buru , Ami fuluq 'type of slender bamboo'; Pai vuluq 'spear' (Tsuchida 1976:172)

It is easy to understand the shift from 'bow' to 'arrow' or 'gun', as they are all used to shoot, as in the example below. These are all related in function.

PAN *busuR > AtaMx buh<in>ug, Tso fsuru, Kan and Sar vuuru, RukTn boso'o, Tha futulh, Sai boehoeL, Ami bucul 'bow', Bun bucul 'bow, gun', Paz buzux 'arrow' (Tsuchida 1976:128)

Another cognate semantically related to this is the verb 'to shoot (with a bow or gun)':

PAN *panaq > Kan mu-a-pana'e, Sar u-a-pana, RukBd w-a-pana, Pai panaq, Puy p<en>ana', Paz pa-pana, Kav p<m>ani, Ami mi-panaq 'shoot with a bow'; Tso pono, Bun panaq, Tha panaq, Sai panae' 'shoot with a bow or gun'

Of course, Formosan natives shot with bow and arrow originally, as shown in the usage of the verb in a majority of languages. When they acquired guns just a few hundred years ago, they applied the same term to both bow and gun, as is shown in the usage of the verb in four of the languages.

2.2 Shape as a factor

Reflexes of PAN *batuh usually refer to 'stone' in many modern languages, such as fatu in Tso, vatu in Kan, vatu'u in Sar, batu in Bun, Paz, Tao, Bab, Pap, and Hoa. However, the Thao form fatu means both 'stone' and 'testicles' due to their similar shape.6 But batu means 'testicles' while batu-nux with a suffix means 'stone' in the Mayrinax dialect of Atayal. The latter is a male form of speech in the Atayalic group; see btu-nux 'stone' in all Seediq dialects (Li 1982b).

There are two cognate forms for the word 'head' in Formosan languages: PAN *quluh > a-oLo in RukBd, quLu in Pai, uRu in Kav, ucu in Bas, and uru in Hoa 'head'; PAN *bunguh > fnguu in Tso, na-vungu in Kan, vungu'u in Sar, bungu in Bun, and vongo in Sir 'head'. The Amis form funguh means 'head, skull, forehead' (Zeng 2008:142). Blust (1999) and Wolff (2010:970) reconstruct the former as the term for 'head', while Blust reconstructs the latter as the term for 'skull', and Wolff (2010:787) as 'head, skull'. There must be a difference between the two terms, so 'skull' seems more appropriate for the latter. The cognate forms for 'brain' are PAN *punuq > pnuu in Tso, punu'u in Sar, punuq in Bun, Pai and Ami, and punu' in Puy 'brain'. However, the corresponding cognate forms punuq in Thao and punu in Paz refer to 'head', while the forms in Rukai dialects, such as punu in RukBd, mean 'forehead' (Tsuchida 1976:172). There is a semantic shift in these three languages. The relationship between 'brain' and 'head' is that of part and whole, SYNECDOCHE in Bloomfield's term. Brain and forehead are near each other in space, METONYMY in Bloomfield's term. The same cognate pu:no' in Tagalog means 'head, leader', which is a metaphorical usage of the original meaning of the term, the physical head. Dempwolff (1934-1938:120) reconstructs it as *puhun7 'trunk of tree, basis, origin'. His definition should be modified to include human 'head, brain', on the evidence of most Formosan languages. The core meaning is thus the 'base' of an animal or plant.

2.3 Geography as a factor

A semantic shift can be drastic. A case in point is PAN *lahud > PuyPn LauD 'east', Paz rahut 'west', PaiBt Lauz 'south'; and similarly, PAN *Daya > Paz daya 'east', PuyPn Daya, Kav zaya 'west'. The semantic reconstruction for the former is 'lowland', while that for the latter is 'inland'. It is clear that the geographical location of each of the Formosan languages determines their seman- tic shifts. The Central Mountains run from north to south in Taiwan. Pazih is spoken to the west, Puyuma and Kavalan to the east, and Paiwan to the south of the Central Mountains, after the dispersal of the Formosan ethnic groups. See Blust's (2009:323) terminology, 'change of physical environment'. Most Formosan languages in the different subgroups still retain the original meanings:

Ata maka-raya, Sed daya, a-m-a-a-cala, Ruk Daza, Bun daza, Pai i-zaya, Tha i-tana-saya, Sai raya 'upland, uphill'; Paz daya 'east'; PuyPn Daya, Kav zaya 'west' < *Daya (Tsuchida 1976:240),

Ata mak-ahu', Kan 'ama-laucu, Ruk LaoDo, Tha i-tana-raus, Sai Laehoer 'lowland, downhill'; Tso mua-rovcu 'blow downhill'; Sar tala-la-laucu 'look down'; PuyPn LauD 'east'; Paz rahut 'west'; PaiBt Lauz 'south' < *lahud (Tsuchida 1976:181)

2.4 Great semantic variation

Many examples of great semantic variation can be found in the Formosan cognates. For example,

Ruk Damar 'moon', Pai zama-n 'torch', Ami lamal (< A l-/r-), Sir lamag, Kav Ramaz (< M *D/*R), Bas namaD (< M *D/*R and < A n/m) 'fire, light' < *DamaR (Tsuchida 1976:154)

The general meaning for the cognate above is 'source of light', which is probably the original meaning. Notice that all four languages in the 'East-Formosan' subgroup (Blust 1999) share the same meaning 'fire, light'.

Tso hicu, Sar 'ilhicu (< A i/a), BunTkb qanitu 'god, ghost, evil spirit'; RukMg alicu 'octopus tree, which is considered sacred' < *qaNiCu (Tsuchida 1976:166)

It is hard to imagine that a type of tree would be related to an evil spirit. One must know, like Tsuchida, that the octopus tree is considered sacred in the Maga dialect of Rukai.

Kan tucu'u, Sar tusu-a, RukBd wa-toDo, Bun tuduq-an, Pai t<em>uzuq, Tha tusuq-in, Sai t<om>oroe', Ami mi-tuluq 'to leak from the roof'; Pai tuzuq, Thao tusuq, Ami tuluq 'drop of water' < PAN *tuDuq (Tsuchida 1976:154)

Tso mou-su-suhcu, Sar kira-a-ua-ulhusu 'move forward step by step', Bun mu-sunu' 'next' < *suNuD (Tsuchida 1976:155)

Kan cuvu'u, Sar cuvu'u, Pai cuvuq 'bamboo shoot'; RukMg s-cubu, RukMn 'a-cuvu 'treetop'; Pai c<em>uvuq 'to sprout' < PAN *Cubuq (Tsuchida 1976:167)

What is the original meaning of the item above? The meanings in modern languages have to do with a 'sprout' or 'new growth'. Tsuchida (1976:314) glossed it as 'grow' and Wolff (2010:788) as 'grow, shoot'.

PAN *bajaq > Kan taa-vala'e , Ami ma-vanaq , Tha ma-fazaq , Paz ma-baza , Sir mou-vana 'know, be able, understand'; RukBd baga 'language, word'; Pai ki-badaq 'inquire, request'; Sai bazae' 'listen' (Tsuchida 1976:224)

The nouns 'language, word' are closely related to the verbs 'to know, be able, inquire'. However, it does not seem to be so obvious that it should be related to the verb 'to listen', but not to the verb 'to talk'. Both Dempwolff (1934-1938:18) and Wolff (2010:746) give the original meaning 'inform' for this cognate.

PAN *pajey > Ata pagay, Sed payay, Tso pai, Bun paz, Ruk pagay, Pai paday, Tha pazay, Kav panay, Pap pada 'rice plant, unhusked rice'; Ami panay 'rice plant'; Sai pazay 'rice plant, cooked rice'; Paz pazay, Kan pai 'glutinous rice'

Sai pazay refers to 'rice plant' and 'cooked rice' rather than 'unhusked rice', the meaning in most other Formosan languages. In addition to the inherited sense, the reflex of *pajey in Saisiyat has come to mean 'cooked rice', which is a striking semantic innovation. Moreover, both Paz pazay and Kan pai refer to a special type of rice, namely 'glutinous rice'. This is a case of semantic narrowing.

Kan ma-'aini, Sar ma-ailhi 'salt', PuyKl 'asil, AmiSa qacil 'salty'; cf. Tongan m-ahi 'sour' < PAN *qasiN (Tsuchida 1976:128)

The words 'salty' and 'sour', referring to different tastes, are derived from the word 'salt' in the example above.

Sed dara, Kan cara'e, Sar cara'e, PuyKl zarah, Fav tagga 'blood'; Sar m-uru-cara'e, Pai djaq, Ami lalaq 'menstrual flow' < *daRaq

AtaSq ramu', Pai djamuq, Sai ramo', Paz damu 'blood'; Kav zamu 'menstruation'; Bun damuq, Tha samuq 'dew'; < *damuq (Wolff 2010:814)

There are four terms for four different types of blood in some Formosan languages: (1) the blood circulating in the body, (2) the blood coming out of the body when wounded, (3) menstrual blood, and (4) the blood from a nose bleed. The Formosan languages that have these distinctions and retain the cognate forms of *daRaq all point to the sense of 'the blood coming out of the body' or 'menstrual blood or flow', so the original meaning is probably 'the blood coming out of the body' (Tsuchida, personal communication).

It seems hard to explain how 'dew' is related to 'blood' to establish the cognate above. Blust (2009:326-327) gives several examples of lexical innovations meaning 'blood' derived from words meaning 'sap' or 'juice' in Malayo-Polynesian languages. He suggests that each of these innovations are 'avoidance of term' because 'blood is a sign of danger'. If that is the case, then the original meaning of the Formosan cognate *damuq could be 'dew', with the meanings of 'blood' and 'menstruation' being innovations.

The form papak means 'ear' in most Atayal dialects, but it refers to 'leg' in the Palngawan dialect of Atayal. It is not clear whether the identical form is related or accidental. If related, it will be a problem to account for the semantic shift, although both terms refer to body parts.

2.5 Noun and verb with the same semantic field

A noun and a verb may share the same or similar semantic field in related forms. For example, AtaMx wagi', RukBu vai, RukMg a-vee, Bun vali, Sir wagi 'sun'; Puy wari 'day'; Ami wali 'east'; Tso m-vore, Kan pa-ari, Sar pa-ari, RukMg o-p-vee, RukMn to-pa-'ai (< M), PaiKu v<n>ayvay, AmiTa pa-wali 'to dry in the sun' < *waRih (Tsuchida 1976:144). The Puyuma and Amis cognate forms show a natural semantic shift from 'sun' to 'day' (for each sunrise and sunset) or 'east' (where the sun rises). Note that some verb forms take the prefix pa- or p- 'causative' in some of the languages, as cited above.

Similarly, Tso me-ah'o 'to give birth', ng-ho'? 'pregnant', Sai 'al-'alak 'young', RukBd valake, Pai alak, Puy alak, Tha az-azak 'child' < *waNak (Tsuchida 1976:146). All the meanings have to do with a 'baby'.

Still another example is the fact that the noun 'tongue' is related to the verb 'to lick': Sar s<um>a-silae, RukTa diLa, Pai dj<em>ilaq, PuyKl L<em>izah (< M), Tha s<m>iraq, Sai riLae' 'to lick'; Hoa la-rila, Sir da-dila, Tag di:la' 'tongue' < *Dilaq (Tsuchida 1976:154).

The same cognate forms may refer to 'farm' or 'to work on a farm': AtaMx quma-qumah, Kan 'uuma, Sar umu-uma, RukBd oma-oma, Bun quma, Pai quma, Puy 'uma, Sai 'oem-'oemaeh, 'farm, dry field'; Paz umamah 'wet field'; AtaSq qumah, Tso mo-mo 'to work, to till, to cultivate the field' < PAN *qumah (Tsuchida 1976:133).

Still another example is PFN *?aSik > Bun 'asik , Ami sa-'asik 'broom'; Bun ma-'asik , Kan m-aru-asiki, Ami mi-'asik 'to sweep' (Tsuchida 1976:183).

What part of speech is the original meaning in each of the cognates above? Nouns are usually the base forms, while their corresponding verbs are derived with an affix.

2.6 Variations across different languages or dialects

Semantic variations among different languages may be greater than what we would expect, especially in the case of verbs. For example, Sar um-ali-a-valee, RukBd tu-a-bale, Paz pa-baret 'to answer'; Pai v<n>alet 'to oppose someone, to talk or strike back', ki-valet 'to take revenge' (Ferrell 1982:335), Bun mim-baas (< A -a/u), Sai Su-baLeh 'to revenge'; Sar m-utu-a-valee 'to echo' < *bales (Tsuchida 1976:129). The core meaning is 'to do something in return'.

A semantic variation may have to do with body parts that are closely related, for example Tsou t'ornga 'chest': Kan takeranga 'ribs'.

Semantic variations are also found between different dialects in the same language, such as in Rukai: RukBd ta-korapange, RukMg t-korpang?, RukTn ta-koapange 'frog'; RukTa ta-ko-ra- rapange, RukMn korapange 'toad'. One may say that frog and toad are of the same species, and similar in shape, but each of these Rukai dialects has different terms for frog and toad. Semantic change must have taken place in these dialects.

3. The original meaning

It is not always easy to determine the original meaning of a proto-form. For instance, reflexes of the cognate set of *waNiS in some Formosan languages refer to 'tooth', 'tusk', or 'wild pig': Tso hisi, Kan anisi, Sar alhii, RukBd valise, Pai aljis, Puy wali, AmiSa wazis 'tooth'; Bun vanis, Sai waliS, Paz walis 'tusk'; Bun vanis, Tha wazish, Sai waliS-an 'wild pig' (Tsuchida 1976:147). The Bunun form refers to both 'tusk (of a wild pig)' and 'wild pig'. The original meaning of *waNiS was probably 'tusk of a wild pig'. It was then extended to cover the wild pig in a few languages, a case of synecdoche. In many other languages, it was extended to the general term for 'tooth', a semantic widening, but with the shift being in a different direction. See *(n)ipen, the PAN cognate generally reconstructed for 'tooth', and Blust (1996), who discusses the semantic history of PAN *waNiS and *waNis-an at some length.

Similarly, reflexes for the same cognate form *qamiS have very different meanings in modern languages: AtaMx qamis-an, Sed mis-an, Kan 'amis-ane, Sar 'amis-an, Sai 'aemiS-an, Paz 'amis-an 'winter'; Bun qamis-an, Puy 'ami 'year'; Puy 'ami, Paz amis-an, Kav imis, AmiSa qamis-an, Bab amis-an 'north'. How are the meanings of 'winter', 'north', and 'year' related, and what is the original meaning? The related cognate forms ami(h)-an in several Philippine languages refer to 'north wind' (Tsuchida 1976:160). There is north wind in the winter time, and also during the New Year period. The original meaning is probably 'north wind', as suggested by Tsuchida (1976:317).

If we know the cultural traditions, it will help us determine the original meaning of a cognate. For example, AtaMx m-atas, Sed m-atas, Tso t<m>opos-?, Bun ma-patas, Tha m-atash 'to write'; RuBd wa-pacase 'to write, to draw'; Tso ta-tpos-a, Kan tapase, Sar taa-tapa-a 'pattern, design'; RukMg ptas?, RukMn u-paca'e 'to embroider' < *pataS (Tsuchida 1976:151). The Formosan aborigines had no writing system, embroidery, or pattern to design at an earlier stage, but they have practiced tattooing for thousands of years. Consequently, the original meaning of the cognate must have been 'to tattoo'.

Many Formosan languages have the same form for 'hand' and 'five'; only Rukai and Bunun have a slight difference in form: Ruk Lima 'five' versus aLima 'hand', Bun ima 'hand' versus hima 'five':

*lima > Ruk a-Lima, Bun ima, Pai lima, Puy Lima, Paz rima, Tha rima, Paz rima, Kav rima, Tao rima, Bab rima, Pap rima, Hoa lima 'hand'

*lima > AtaMx ima-gal, Sed rima, Tso rimo, Kan lima, Sar ku-lima, RukBu Lima, Bun hima, Pai lima, Puy Lima, Tha rima, Kav rima, Ami lima, Pap rima, Hoa lima, Sir ri-rima 'five'

What is the original meaning of this cognate set? Is it 'hand' or 'five' or both 'hand' and 'five'? Intuitively we may think that 'five' is derived from 'hand'. Dempwolff reconstructs *lima' 'hand, five', while Wolff (2010:952) reconstructs *lima 'five' versus *qalima 'hand', on the basis of Rukai aLima and reflexes in a few Philippine languages, such as Cebuano alima, Surigao alima and Kalamian kalima' 'hand'. If Wolff is correct, then the body part 'hand' is derived from 'five' by adding a prefix *qa-, attested in Rukai and a few Philippine languages. Note that k in Kalamian is a regular reflex of *q (Ogawa 1940).

Consider the following cognate set:

*qenay > Kan 'enai 'earth, ground', 'ena-'enay 'soil'; Kav mra-nay 'earth, soil'; Ruk enay 'sand'; Puy 'enay 'water' (Wolff 2010:962)

Reflexes of modern Formosan languages show these four possible meanings: 'earth', 'soil', 'sand' and 'water' for this cognate. What is the original meaning? The cognate *DaReq is generally reconstructed for the word 'earth' and *DaNum for the word 'water'. The semantic shift to 'water', only in Puyuma, probably took place later. So the choice is between 'soil' and 'sand' for the original meaning. Two Formosan languages (Kanakanavu and Kavalan) have the meaning 'soil', while one Formosan language (Rukai) and Malayo-Polynesian languages have the meaning 'sand'. Kanakanavu, Kavalan, Rukai, and Malayo-Polynesian belong to four primary subgroups (Blust 1999), so the weight is equal on either side. Blust reconstructs it as 'sand', a meaning widespread in Malayo-Polynesian languages. However, Wolff (2010:962) prefers 'soil', saying that 'the meaning of Kanakanavu and Kavalan "soil" is probably older and the meaning "sand" comes into being later.'

4. Summary and conclusion

I have examined some interesting examples of semantic shift and variation as manifested in Formosan languages. Some of them can be explained by Bloomfield's types of semantic shift, such as narrowing, widening, metaphor, metonymy (meanings related in space or time), and synecdoche (meanings related as whole and part). However, the others do not fall into any of his categories. I have offered a few other explanations, including function, shape, and geography, as factors for the semantic shift in Formosan languages. But these are only factors in, not types of, semantic shift.

To summarize, I have examined some Formosan cognates in which various factors have played a part in the semantic shift, including

Function as a factor:

*Rumaq 'house' > 'house, grave, lair'

*taRuqan 'hut' > 'hut, house'

*qayam 'bird' > 'bird, meat'

*biRaq 'leaf' > 'leaf (in general), leaf used in rites, betel leaf, Alocasia'

*buluq 'type of bamboo' > 'type of slender bamboo, spear'

*busuR 'bow' > 'bow, arrow, gun'

*panaq 'shoot' > 'shoot with a bow, shoot with a bow or gun'

Shape as a factor:

*batuh 'stone' > 'stone, testicles'

Geography as a factor:

*Daya 'inland' > 'east, west, south'

*lahud 'lowland' > 'east, west'

Great semantic variation:

*DamaR 'light' > 'moon, torch, fire, light'

*qaNiCu 'soul' > 'god, ghost, type of plant considered holy'

*tuDuq 'leak' > 'leak from the roof, drop of water'

*Cubuq 'new growth' > 'bamboo shoot, treetop, to sprout'

*bajaq 'know' > 'know, be able, understand; language, word; inquire; listen'

*pajey 'rice' > 'rice plant, unhusked rice; rice plant, cooked rice; glutinous rice'

*damuq 'dew' > 'dew; blood, menstruation'

In addition, I have also examined examples of nouns and verbs sharing the same semantic field:

*waRih 'sun' > 'sun, day, east, dry in the sun'

*waNak 'child' > 'child, young, pregnant, give birth'

*qumah 'farm' > 'farm, work on a farm'

*?aSik 'broom' > 'broom, sweep'

I have also discussed problems of deciding the original meaning of a cognate, such as:

*waNiS 'tusk of a wild pig' > 'tooth, tusk, wild pig'

*qamiS 'north wind' > 'winter, year, north'

*pataS 'tattoo' > 'write, design, embroider'

*lima 'five' > 'five, hand'

It is not always easy to determine the original meaning of a proto-form. It is even harder to suggest what semantic shift may have taken place. As a matter of fact, some semantic variations can be so great they seem to go beyond what could be imagined.

Semantic shift and variation are an interesting aspect of language that requires more careful study.

Footnote

* An earlier version of this paper was presented at the First World Congress of Taiwan Studies, Taipei, April 26-28, 2012. I am grateful to Shigeru Tsuchida for his valuable suggestions and new additions to the language data, and to Elizabeth Zeitoun, Hui-chuan J. Huang, Dorinda Tsai-hsiu Liu, and two anonymous reviewers for their suggestions for improvement.

1 Abbreviations of language and dialect names used in this paper are: Ami, Amis; AmiTa, Tavalung dialect of Amis; AmiSa, Sakizaya dialect of Amis; Ata, Atayal; AtaMx, Mayrinax dialect of Atayal; AtaSq, Squliq dia- lect of Atayal; Bab, Babuza; Bas, Basay; Bun, Bunun; BunTkb, Takbanuaz dialect of Bunun; Fav, Favorlang; Hoa, Hoanya; Kan, Kanakanavu; Kav, Kavalan; Pai, Paiwan; PaiBt, Butanglu dialect of Paiwan; PaiKu, Kulalau dialect of Paiwan; Pap, Papora; Paz, Pazih; Puy, Puyuma; PuyPn, Pinam dialect of Puyuma; Ruk, Rukai; RukBd, Budai dialect of Rukai; RukLa, Labuan dialect of Rukai; RukMg, Maga dialect of Rukai; RukMn, Mantauran dialect of Rukai; RukTa, Tanan dialect of Rukai; RukTn, Tona dialect of Rukai; Sai, Saisi- yat; SaiTa, Ta'ay dialect of Saisiyat; Sar, Saaroa; Sed, Seediq; Sir, Siraya; Tao, Taokas; Tha, Thao; Tso, Tsou.

2 Most of the cognate sets in this paper are taken from Tsuchida (1976). I have updated them by adding a few new cognate forms that Tsuchida missed or left out in the early 1970s, and by correcting a few errors, as based on my own field notes, or consulting dictionaries compiled by Egerod (1980, 1999), Ferrell (1982), Fey (1986), Li & Tsuchida (2001, 2006), Nihira (1988), Ogawa (2006), Tsuchida et al. (1991), and Zeng (2008).

3 I generally adopt Romanized writing systems for Formosan languages except for the high back unrounded vowel ? in Tsou and the high and mid-central vowels in the Maga dialect of Rukai. In addition, the symbol q is used to stand for the pharyngealized stop in Amis, and for the glottal stop in Philippine languages.

4 The form qayam also means 'chicken' in some dialects of Amis, such as Fataqan; also see Javanese ayam 'chicken'. Reflexes of the same cognate refer to 'dog' in a few Philippine languages, such as qay?m in Bilaan (Reid 1971:70).

5 The meaning of the reconstructed form is usually assumed to be 'Alocasia sp., the giant taro' (Blust, ACD, Wolff 2010:768). However, since all or most Austronesian major subgroups are found in Taiwan (Blust 1999; Ross 2009), Formosan evidence carries more weight than extra-Formosan, even though Malayo-Polynesian languages have a much wider geographical distribution. It is much more likely that the original meaning is 'leaf' rather than 'Alocasia'. Incidentally, Paz biarax 'Alocasia', cited in Wolff (2010:768), is not a cognate form.

6 Cf. the French word tête 'head', which is historically derived from Latin testa 'pot' due to their similar shape. Similarly, the English word 'vagina' is derived from the Latin word vagina 'sheath' due to similar shape or function (Heine 1997:132).

7 There is a metathesis of *n and *q > *h in Malayo-Polynesian languages.

8 Following Tsuchida's (1976) convention, < A indicates assimilation and < M metathesis.

References

References

Bloomfield, Leonard. 1933. Language. New York: Holt, Rinehart and Winston.

Blust, Robert. 1996. Notes on the semantics of proto-Austronesian *-an 'locative'. Voice in Austrone- sian, ed. by Marian Klamer, 1-11. Jakarta: Badan Penyelenggara Seri NUSA, Universitas Katolik Indonesia Atma Jaya.

Blust, Robert. 1999. Subgrouping, circularity and extinction: some issues in Austronesian compara- tive linguistics. Selected Papers from the Eighth International Conference on Austronesian Linguistics, ed. by Elizabeth Zeitoun & Paul Jen-kuei Li, 31-94. Taipei: Institute of Linguistics, Academia Sinica.

Blust, Robert. 2009. The Austronesian Languages. Canberra: Pacific Linguistics.

Blust, Robert. 2010. Five patterns of semantic change in Austronesian languages. A Journey through Austronesian and Papuan Linguistic and Cultural Space: Papers in Honour of Andrew Pawley, ed. by John Bowden, Nikolaus P. Himmelmann & Malcolm Ross, 525-546. Canberra: Pacific Linguistics.

Dempwolff, Otto. 1934-1938. Vergleichende Lautlehre des Austronesischen Wortschatzes. Berlin: D. Reimer.

Egerod, Søren. 1980. Atayal-English Dictionary. London: Curzon Press.

Egerod, Søren. 1999. Atayal-English Dictionary (2nd edition), ed. by Jens Østergaard Petersen. Copenhagen: The Royal Danish Academy of Sciences and Letters.

Ferrell, Raleigh. 1982. Paiwan Dictionary. Canberra: Pacific Linguistics.

Fey, Virginia A. 1986. Amis Dictionary. Taipei: The Bible Society.

Heine, Bernd. 1997. Cognitive Foundations of Grammar. New York: Oxford University Press.

Ho, Dah-an. 1978. Wuzhong Paiwan fangyan de chubu bijiao [A preliminary comparative study of five Paiwan dialects]. Bulletin of the Institute of History and Philology Academia Sinica 49.4:565-681.

Huang, Lillian M. 1995. The syntactic structure of Wulai and Mayrinax Atayal: a comparison. Bulletin of National Taiwan Normal University 40:261-294.

Li, Paul Jen-kuei. 1977. The internal relationships of Rukai. Bulletin of the Institute of History and Philology Academia Sinica 48.1:1-92.

Li, Paul Jen-kuei. 1979. Variations in the Tsou dialects. Bulletin of the Institute of History and Philology Academia Sinica 50.2:273-300.

Li, Paul Jen-kuei. 1982a. Linguistic variations of different age groups in the Atayalic dialects. Tsing Hua Journal of Chinese Studies, New Series 14.1-2:167-191.

Li, Paul Jen-kuei. 1982b. Male and female forms of speech in the Atayalic group. Bulletin of the Institute of History and Philology Academia Sinica 53.2:265-304.

Li, Paul Jen-kuei. 1983. Notes on Thao dialects. Bulletin of the Department of Anthropology National Taiwan University 43:48-50.

Li, Paul Jen-kuei. 1988. A comparative study of Bunun dialects. Bulletin of the Institute of History and Philology Academia Sinica 59.2:479-508.

Li, Paul Jen-kuei. 1998. The dialectal differences in Atayal in I-Lan. Selected Papers from the Second International Symposium on Languages in Taiwan, ed. by Shuanfan Huang, 49-78. Taipei: Crane.

Li, Paul Jen-kuei, & Shigeru Tsuchida. 2001. Pazih Dictionary. Taipei: Institute of Linguistics, Academia Sinica.

Li, Paul Jen-kuei, & Shigeru Tsuchida. 2006. Kavalan Dictionary. Taipei: Institute of Linguistics, Academia Sinica.

Nihira, Yoshiro. 1988. A Bunun Vocabulary: A Language of Formosa (3 edition, enlarged). Tokyo (Privately published).

Ogawa, Naoyoshi. 1940. Karamian-go to Agotaya-go [The Calamian and Agotaya languages]. Ando Masatsugu Kyoju Kanreki Shukuga Kinen Ronbunshu [Collection of Theses and Essays in Commemoration of the 60th Anniversary of Professor Masatsugu Ando], 1215-1228. Tokyo: Sanseido.

Ogawa, Naoyoshi. 2006. A Comparative Vocabulary of Formosan Languages and Dialects, ed. by Paul Jen-kuei Li & Masayuki Toyoshima. Tokyo: Research Institute for Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies.

Rau, D. Victoria. 2000. Phonological variation and sound change in Atayal. Oceanic Linguistics 39.1:144-156.

Reid, Lawrence A. 1971. Philippine Minor Languages: Word Lists and Phonologies. Honolulu: University of Hawai'i Press.

Ross, Malcolm. 2009. Proto Austronesian verbal morphology: a reappraisal. Austronesian Historical Linguistics and Culture History: A Festschrift for Robert Blust, ed. by K. Alexander Adelaar & Andrew Pawley, 295-326. Canberra: Pacific Linguistics.

Teng, Stacy Fang-ching. 2011. Noun phrase conjunction in three Puyuma dialects. Language and Linguistics 12.1:171-204.

Ting, Pang-Hsin. 1978. Gu Beinanyu de nice [Reconstruction of proto-Puyuma phonology]. Bulletin of the Institute of History and Philology Academia Sinica 49.3:321-392.

Tsuchida, Shigeru. 1976. Reconstruction of Proto-Tsouic Phonology. Tokyo: Institute for the Study of Languages and Cultures of Asia and Africa, Tokyo University of Foreign Studies.

Tsuchida, Shigeru, Yukihiro Yamada, & Tsunekazu Moriguchi. 1991. Linguistic Materials of the Formosan Sinicized Populations I: Siraya and Basai. Tokyo: University of Tokyo.

Wolff, John U. 2010. Proto-Austronesian Phonology with Glossary. Ithaca: Southeast Asia Program Publications.

Zeitoun, Elizabeth. 1993. Degrees of grammaticalization in the Rukai dialects: synchronic and diachronic considerations. Paper presented at the 2nd International Conference on Chinese Languages, June 23-25, 1993. Paris, France.

Zeng, Siqi. 2008. Taiwan Amisyu Changyong Cizhui yu Dongci Cigen Shouce [A Handbook of Productive Affixes and Verb Roots in Amis]. Beijing: Nationalities Press.

[Received 10 June 2013; revised 4 December 2013; accepted 19 December 2013]

Institute of Linguistics

Academia Sinica

130, Sec. 2, Academia Road

Nankang, Taipei 115, Taiwan

[email protected]