When “what” is “where”: A linguistic analysis of landscape terms, place names and body part terms in Marquesan (Oceanic, French Polynesia)

Available online at www.sciencedirect.com

Language Sciences 30 (2008) 200–226 www.elsevier.com/locate/langsci

When ‘‘what’’ is ‘‘where’’: A linguistic analysis of landscape terms, place names and body part terms in Marquesan (Oceanic, French Polynesia) Gabriele H. Cablitz

*

Institut fu¨r Allgemeine und Vergleichende Sprachwissenschaft, der Universita¨t Kiel, Ohlshausenstr. 40, 24098 Kiel

Dedicated to Godfrey Pitt

Abstract This article describes how the topography of the Marquesas Islands is reflected in the Marquesan language (East Polynesian, Oceanic, Austronesian) and how speakers of Marquesan use their linguistic resources to orient themselves in their environment. It addresses the semantic and lexical relationships (taxonomic, antonymic/oppositional, partonymic, etc.) between landscape terms, place names and body part terms, and how linguistic resources are mapped onto the environment. Particular focus is placed on morphosyntactic aspects of landscape terms. Landscape features form an unusual ontological category because their status as ‘being objects’ or ‘being places’ is unclear [Lyons, J., 1977. Semantics. Cambridge University Press, Cambridge; Smith, B., Mark, D.M., 1998. Ontology and geographic kinds. In: Poiker, T.K., Chrisman, N. (Eds.), Proceedings of the 8th International Symposium on Spatial Data Handling (SDH’98), pp. 308–320; Smith, B., Mark, D.M., 2001. Geographical categories: an ontological investigation. International Journal of Geographical Information Science 15 (7), 591–612]. Marquesan is interesting in this respect because speakers are sensitive to a ‘‘what’’ and a ‘‘where’’ category in their language when talking about spatial relations. Classification of landscape terms is examined in this context. Furthermore, the article compares landscape terms with body part terms because studies of grammaticalisation of locatives also deal with the issue of when and how a body part term (a ‘‘what’’) becomes a marker of location (a ‘‘where’’). As will be shown, the grammaticalisation of body part terms to locatives reflects a similar kind of ambiguity as observed in Marquesan landscape terms. Ó 2006 Elsevier Ltd. All rights reserved. *

Tel.: +49 4318802669; fax: +49 4318807405. E-mail address: [email protected]

0388-0001/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.langsci.2006.12.004

G.H. Cablitz / Language Sciences 30 (2008) 200–226

201

Keywords: Marquesan; Spatial language and conceptualisation; Ontological status; Morphosyntax; Semantic categorisation; Linguistic anthropology

1. Introduction 1.1. General outline of article After discussing some basic aspects of the notions ‘object’ and ‘place’, an introduction into the topography of the Marquesas Islands, the genetic affiliation and some typological characteristics of the Marquesan languages will be given. This introduction is followed by two major parts. In the first part a discussion and overview of Marquesan landscape terms and place names are undertaken in showing that the naming of places is often related to the labels of geographical entities or landscape features. It will be further discussed in which way the common definition of proper names of places (cf. Lyons, 1977; Lerner and Zimmermann, 1991) is problematic. The second part of the article focuses on the linguistic marking and morphosyntactic construction types of landscape terms in relation to other location-denoting nominals such as place names and local nouns. Despite the fact that there is a relation of semantic labelling between place names and landscape terms in Marquesan, this section discusses the ontological status of landscape terms showing that we are dealing with an ontological ‘‘in-between’’ category. As Marquesans are sensitive to a ‘‘what’’- and ‘‘where’’-category in their language when talking about spatial relations, the linguistic distinction will be a basis of analysis of how landscape features are perceived and conceptualised. Finally, the analysis of the ontological status of landscape terms is compared to body part terms which have become grammaticalised markers of location in showing that the border between a ‘‘what’’- and ‘‘where’’-category is blurred in a similar way as with landscape terms. Morphosyntactic, syntactic and semantic evidence in constructions with body part and landscape terms will reveal what is ‘‘what’’ and what is ‘‘where’’ in Marquesan.1

1.2. Some preliminaries: categorical distinction between ‘‘what’’ and ‘‘where’’ Landau and Jackendoff (1993) claim that all languages distinguish between a ‘‘what’’and ‘‘where’’-category. Lyons (1977) distinguishes between so-called ‘first-order entities’ and ‘places/locations’, arguing that there is a categorical difference between the two. According to Lyons the physical world contains discrete, three-dimensional and self-moving entities such as humans and animals as well as entities which are not self-moving, but manipulable as in the case of objects. Lyons calls them first-order entities and views them as ontologically different from locations (1977, p. 693).2 Although many researchers assume that there is a basic distinction between first-order entities and places, the nature 1

For example, prepositional marking, modification and construction types. For a more detailed analysis in the Pacific context, see Cablitz (2005, pp. 163–204, 2006, pp. 309–331). 2 Note that Lyons does not include plants in his definition of ‘first-order entities’ although there is no good argument for treating them differently from any other three-dimensional objects.

202

G.H. Cablitz / Language Sciences 30 (2008) 200–226

of first-order entities and places is rarely discussed in detail. So what is the basic distinction between the two? The most significant difference between the two categories is that of perceptual boundedness. First-order entities such as cars, tables and chairs have clear perceptual boundaries, whereas places such as the inland region of a Marquesan valley have no visually perceived boundaries. Places such as towns, countries or terrains can be bounded by law or agreement, but these boundaries are rarely physically/visually marked along their complete extensions. Similarly, locations in small-scale reference, such as the ‘above’ or ‘under’ region of a table do not have boundaries which can be perceived. Miller and JohnsonLaird (1976, p. 59) define these locations as concepts rather than as percepts because their extensions are often vague and can vary depending on different contexts and conditions of usage (e.g. the wood worm in the cupboard vs. the clothes in the cupboard). Although criteria such as moveability and material existence typically define first-order entities, these criteria can also be true for locations. Locations can be of a material nature consisting of earth, mountains, lakes, meadows and the like. They can also be moved: the ‘above’ region of a table changes its location once the table is moved. First-order entities can also be considered as one class of places: due to their material nature they always occupy some location in space at any point in time. The locations which are occupied by objects can be of a temporary nature if objects are moveable. According to Klein (1991) an object (or first-order entity) has to be distinguished from its eigenort, i.e. the proper location it occupies by its material boundaries at a certain point in time. Klein therefore classifies objects or first-order entities as a category which is distinct from locations. Eigeno¨rter shall be considered as a different class of locations than towns, countries or any other landscape region. 1.3. The topography of the Marquesas Islands The Marquesas Islands are characterised by high volcanic mountains and rugged steep cliffs on which one typically finds black lava, basalt and red tufa. They are in many ways quite different from other volcanic islands in the South Pacific as they typically do not have a coral reef formations around the islands due to the cold Humboldt Stream coming from the Antarctic. The mountainous topography of the islands, on which most valleys are deep and narrow and not easily accessible by land as well as by sea, is thought to have caused great isolation and a clanish and aggressive culture promoting warfare even among neighbouring valleys and tribes of one and the same island (Handy, 1923; Kellum-Ottino, 1971; Hughes and Fischer, 1998; Thomas, 1990). The Marquesas Islands are situated in the Pacific Ocean in the southern hemisphere between 7°50 0 and 10°35 0 latitude and 138°25 0 and 140°50 0 longitude west of Greenwich (GB) (Hughes and Fischer, 1998, p. ix). They are one of the five archipelagos of French Polynesia, a POM (= pays d 0 outre mer) of France with a relative political autonomous status. Tahiti, which is the political, economical, educational and administrative capital of French Polynesia, is situated about 1400 km southwest of the Marquesan archipelago. Geographically, linguistically and, to some degree, culturally two distinct groups are distinguished in the Marquesan archipelago: the northwest and the southeast Marquesas (Thomas, 1990, p. 1). The Marquesan archipelago has twelve islands of which six are inhabited: Nuku Hiva, 0Ua Pou, 0Ua Huka in the North Marquesas, and Hiva 0Oa, Tahuata and Fatu Iva in the South Marquesas (see Fig. 1).

G.H. Cablitz / Language Sciences 30 (2008) 200–226

203

Fig. 1. The Marquesas Islands.

The highest mountain (Temetiu) is situated on Hiva 0Oa, reaching 1276 m. In particular, the southern islands are characterised by high central mountain ridges and they have, in contrast to the northern islands, a humid climate and a rich and lush green vegetation. Although the northern islands are much more arid, all inhabitated islands are abundant in tropical fruits and vegetation in general. The months of July to September are the coolest months and characterised by heavy rainfall. Most of the year the south-eastern trade winds (Alize´e) are dominant except for the months between December and March where south-western winds prevail bringing cyclones to eastern Polynesia. Some years are dominated by droughts, whereas others are dominated by heavy rainfall. It is generally assumed that the first settlers of the Marquesas belonged to the Lapita cultural complex originating in Southeast Asia or western Melanesia (Thomas, 1990, p. 6; Ottino, 1997, p. 7). According to archaeological estimates the first settlers arrived 200 BC ± 150 from Tonga or Samoa (Ottino, 1997, p. 7). Despite the fact that Polynesians are well-known as skilled seafarers, the contact between Marquesans and other eastern Polynesian populations is said to have been sporadic (Thomas, 1990, p. 9). First European contact was made by Alvaro de Mendan˜a y Castro in 1595, but it took another 200 years before European contact became more intense. Today about 8000 people live on the Marquesas. Around 5% are of non-indigenous origin (Blanchard, 1997).3 All villages and most dwellings are located along the coasts. Despite a drastic change in lifestyle in the past 40 years many Marquesan households are basically still self-sufficient: they go fishing and hunting and do some gardening (taro, maniok, yam, banana).4 Copra, the former major cash crop, has been largely replaced by noni.5 3

Nowadays, a large Marquesan community of around 14,000 people lives on Tahiti, having migrated there for reasons of work opportunities and further education. 4 French Polynesia and the Marquesas have become more and more westernised, a development which was greatly accelerated by the installation of the Centre d’Expe´rimentation du Pacifique (CEP), the atomic testing ground on Moruroa (Tuamotu) in the 1960s. 5 Indian mulberry (Morinda citrofolia).

204

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Some fruits (lemons, grapefruits and mangos) are exported to the local market in Tahiti. A number of Marquesans produce and sell handicrafts (wood-, bone- and stone-carvings, tapa (bark cloth), oils, etc.). Traditionally, the Marquesan material culture was based on plants. There was the occasional use of animal materials – in particular from marine animals – but metals were unknown before the arrival of Europeans. While people of the Lapita cultural complex were pottery makers, the tradition of pottery-making probably ceased shortly after the settlement of the Marquesas Islands. 1.4. The Marquesan languages and their genetic affiliation The Marquesan languages belong to the Eastern Polynesian subgroup within the Eastern Oceanic branch of the Austronesian language family (Pawley, 1966; Green, 1966; Marck, 1996). Although the literature generally divides the Marquesan vernaculars into two distinct groups, North Marquesan ( 0 eo 0 enana) and South Marquesan ( 0 eo 0 enata) (Green, 1966; Hughes and Fischer, 1998, see also Ethnologue), each island vernacular has distinct lexical, phonological and partly morphological characteristics. The Marquesan languages are most closely related to Hawai’ian and Mangarevan, which all form the Marquesic subgroup which is distinct from the Tahitic subgroup consisting of Tahitian, Tuamotuan, Rarotongan and New Zealand Maori (Pawley, 1966; Green, 1966). Southeast Marquesan (S-MQA) is spoken on the islands Hiva 0Oa, Tahuata and Fatu Iva, and Northwest Marquesan (N-MQA) is spoken on Nuku Hiva, 0Ua Pou and 0Ua Huka. Marquesan is an accusative case-marking language. It distinguishes between inalienable and alienable possessive constructions. It has an inclusive versus exclusive first person pronoun. There is no class of adjectives and no verb–noun distinction, information structuring is characterised by clause chaining or complex nominalisations and tense is not obligatorily marked on the lexical head of verbal phrases. Marquesan is characterised by frequent passivisation of transitive verbals and by a high number of noun phrase ellipses (in particular of subject noun phrases). For the purpose of the present article it should be mentioned that polysemy is a very characteristic feature of Marquesan – and Polynesian languages in general (cf. Broschart, 1991; Vonen, 2000). A number of researchers of Polynesian languages point out that word classification (Broschart, 1991; Vonen, 2000) – in particular that of nouns and verbs – is problematic as they cannot be precategorised on the lexical level. As for the verb–noun distinction it is the constructional, or rather syntactic, phrase level (i.e. the nominal or verbal phrase), with its different formal or morphosyntactic properties, which allows us to classify a word as a verbal or a nominal. Meaning, more than anything else, determines the morphosyntactic properties of a phrase (Cablitz, 2006).

2. Landscape terms and place names in Marquesan: an overview This section gives an overview of the nominal labels for features and entities of the environment such as mountain, sea, lake, river, etc. as well as their structural and semantic properties. The landscape terms discussed in this article are generally used in all six Marquesan dialects with only slight lexical differences between the vernaculars. The dialectal differences are indicated in the table below.

G.H. Cablitz / Language Sciences 30 (2008) 200–226

205

Geographical entities are also often referred to as ‘landmarks’ and their generic labels as ‘landmark terms’ in the literature and in my own work (Cablitz, 2006). For the purpose of conformity with other contributions to this special issue, I here use ‘geographical entity’/‘landscape feature’ and ‘landscape term’ respectively. Place names are described with regard to their relationship with landscape terms as well as their importance in the Marquesan culture. Furthermore, the etymology and nature of these descriptive labels are discussed by analysing patterns of naming of places (cf. Senft, 2008). The term ‘place name’ is used for the class of proper nouns naming locations of the environment. ‘Landscape terms’ label phenomena of the environment. Unconventionally, I also treat here labels for celestial phenomena such as stars as landscape terms (see further below).

2.1. Marquesan landscape terms This subsection lists the different vernacular landscape terms and their English glosses according to the different geographical domains of the Marquesas (e.g. SEA- vs. LANDarea, water and small land features); it also includes man-made features (e.g. houses, ceremonial sites, agricultural areas). Celestial phenomena like ‘sky’, ‘cloud’ and ‘star’ are included here by virtue of their significance as reference points in navigation. The geographical zones are divided in the following way. The sea and the island are the most dominant geographical domains in the Marquesan archipelago, represented in the table as ‘sea region’ and ‘mountain area and large land features’. The coastal area also represents a separate category as it can neither be clearly defined as sea- nor as land-area. I also make a distinction between ‘mountain area and large land features’ (category 3) and ‘small land features’ (category 4). The category ‘mountain area and large land features’ lists mountain features and large landscape features such as ‘valley’ or ‘island’, whereas the fourth category only lists small landscape features such as pu 0 u 0 u ‘rock’ or ‘pointed stone’. These small landscape features are in some ways perceptually salient and sometimes serve as orientational points of reference for Marquesans, e.g. when navigating to fish banks (see Table 1). Most of the landscape terms are simplex forms such as tai ‘sea’, ava ‘passage’ or tahuna ‘gravel beach’. Some landscape terms have a complex form or structure of which tai nui ‘ocean’ (lit. ‘sea big’), topa 0 ia vai ‘waterfalls’ (lit. ‘falling water’), henua ataha ‘uninhabitated land’ (lit. ‘land to walk’), 0  ua vai ‘waterhole’ (lit. ‘hole water’) are clearly transparent compounds. Other landscape terms seem to be complex forms (e.g. ka 0 avai ‘river, valley, village’ (
206

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Table 1 Marquesan landscape terms Vernacular label of landscape term 1. Sea region tai maoanaa tai nui (almost obsolete) a u toka take aoma 0 ama

English gloss ‘sea’ ‘far out at sea’ ‘sea current’ ‘fish bank’ ‘sea-bottom’ ‘sea-surface’

2. Coastal region ava haka (UP)b/ha 0 a (NH, UP)/hana (south) a0 ıac (NH)/akau (south) puka (north)/puna (south)d ko utu (north)/ 0 o utu (south) papa uta  one tahuna tahatai  opata mata 0 ae (north)/oho (south) motu

‘sea/reef passage’ ‘bay’ ‘reef’ ‘coral’ ‘rocky coastline’ ‘rocky coast, lava stone formations’ ‘ashore’ ‘sand, (sandy) beach’ ‘gravel, gravel beach’ ‘shoreline, beach’ ‘cliff (by the sea)’ ‘cape’ ‘rock in the sea, islet’

3. Mountain area and large land features pına 0 i tuaivi mouka (north)/mouna (south) kapua (north)/ 0 apua (south) ivi pukeı (north)/pukeo (south) tua meho 0 a/ho 0 a henua ataha (north)/fenua ataha (south) uta vao ka 0 avai henua (north)/fenua (south) vahi (north)/taha (south) motu

‘cliff (inland area)’ ‘mountain, mountain at the side of a valley’ ‘inland mountain’ ‘mountain crater’ ‘top of mountain range, hill’ ‘mountain peak’ ‘complete mountain range’ ‘plateau’ ‘uninhabited land’ ‘inland area’ ‘deep inland area’ ‘valley’ ‘land, terrain, country’ ‘place, terrain, location’ ‘island’

4. Small land features ana ava keho papa ke 0 a ho 0 oho 0 o pu 0 u 0 u ma 0 a 0 eita

‘cave’ ‘mountain passage’ ‘basalt rock’ ‘lava stone rock’ ‘stone’ ‘assemblage of stones’ ‘large pointed stone’ ‘bush, grass, uncultivated growth of plants’ (continued on next page)

G.H. Cablitz / Language Sciences 30 (2008) 200–226

207

Table 1 (continued) Vernacular label of landscape term

English gloss

5. Fresh water features ka 0 avai vai puna 0  ua vai topa 0 ia vai

‘river’ ‘fresh water’ ‘fresh water source’ ‘water hole’ ‘waterfalls’

6. Sky features 0 aki (north)/ 0 ani (south) 0 oumati ao kohu mano 0 aiatae paepaeohinaf ao hatu anuanua hetu (north)/fetu (south) me 0 ama (UP, NH)/mahina (all other vernaculars)

‘sky’ ‘sun’ ‘cloud’ ‘cloud covering mountain’ ‘cloud formation in form of a shark’ ‘fuzzy cloud formation’ ‘black sky’ ‘rainbow’ ‘star’ ‘moon’

6. Man-made features ka 0 avai ha 0 e (north)/fa 0 e (south) tohua paepae (all vernaculars)/upe (‘Ua Pou, Hiva ‘Oa) me 0 ae p apua va 0 anui uap u pakatea

‘village’ ‘house’ ‘public assemblage place, floor (house)’ ‘house stone foundation’ ‘former religious ceremonial place’ ‘garden, enclosure, parc’ ‘road, route’ ‘small path, route’ ‘quay’

7. Beyond the local environment hiva aoma 0 ama

‘outside world, in the far lands’ ‘world, planet earth’

a Moana is a Tahitian loanword. It has largely replaced the Marquesan label for ‘far out at sea’, tai nui (lit. ‘sea big’). b UP is an acronym for the 0Ua Pou dialect; likewise NH stands for Nuku Hiva dialect. c Some very protected bays have coral reefs. d ‘North’ refers to all northern vernaculars (Nuku Hiva, 0Ua Pou, 0Ua Huka) and ‘south’ refers to all southern vernaculars (Hiva 0Oa, Tahuata, Fatu Iva). e In the cloud formation the ‘head’ of the shark is pointed seawards; when this cloud formation appears, it is believed that a lot of fish can be caught. f Paepaeohina literally means ‘the house foundation of Hina’; Hina is a goddess in the Polynesian mythology; in Marquesan there are a number of natural phenomena which contain the name Hina (e.g. mama 0 oohina, a fern variety, lit. ‘standing up of Hina’).

best translated as ‘far out at sea’. The English terms ‘sea’ and ‘ocean’ correspond more closely to MQR tai. A number of landscape terms are clearly polysemous like ka 0 avai ‘1. river, 2. valley, 3. village’, aoma 0 ama ‘1. sea-surface, 2. world’, henua/fenua ‘1. land, terrain, 2. country (with cultural, geographical or political borders)’, uta ‘1. ashore, 2. inland’, and ava ‘1. reef passage, 2. mountain passage’. The analysis of the morphosyntactic properties of polysemous landscape terms sheds light on how speakers categorise landscape terms, in particular with regard to their ontological status (see further in Section 3).

208

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Local nouns which belong to a small closed class of lexemes exclusively used for the description of spatial relations can also denote geographical entities or geophysical phenomena. Local nouns are nominals since they only occur as lexical heads of locative noun phrase constructions:6 e.g., ma[prep] 0 uka[local noun]o[poss] te[art] tapu[table] ‘on top of the table’ (for more details, cf. Cablitz, 2006, pp. 331–397). The local nouns 0 uka ‘up’ and 0 a 0 o ‘down’ are used to refer to local and equatorial sea currents and then mean ‘upstream’ ( 0 uka) and ‘downstream’ ( 0 a 0 o). 0 Uka ‘upstream’ is used when people go against the sea current, and 0 a 0 o ‘downstream’ is used when one goes with the sea current flow. The local nouns 0 oto, lit. ‘inside’, and vaho, lit. ‘outside’, are also used to refer to geographical entities. 0 Oto can denote the ‘bay’, whereas vaho denotes the part of the sea outside a bay. The meaning of vaho therefore only partly overlaps with moana, namely in the meaning of ‘far out at sea’. This metaphorical usage of 0 oto and vaho indicates that Marquesans view the relationship between island and sea as a relationship of containment ( 0 oto ‘bay = inside ) part of island’) vs. non-containment (vaho ‘sea = outside ) not part of island’). 0 Oto ‘bay’, vaho ‘sea (outside bay)’, 0 uka ‘upstream 0 and 0 a 0 o ‘downstream’ are used especially for navigation on sea, e.g., when localising fish banks or indicating travelling directions (cf. Lavonde`s, 1983; Cablitz, 2006, pp. 510–519). 2.2. Marquesan place names Most Marquesan place names are semantically transparent and derived from definite descriptions. One can reconstruct their descriptive meanings, which provides a convenient basis of classification of Marquesan place names. In the following I will briefly describe these classes by listing them in the tables below with a few representative examples.7 The classification is based on the lexical components of the respective place names and their combinatorial patterns. The classes of place names can consist of e.g. landscape or body part terms, names of plants or animals, names of mythological figures, or of combinations of these lexical components. Table 2 shows place names with simple definite descriptions or names. Table 3 shows the combinatorial options of the listed sources of place names in Table 2. Some sources of Table 2 combine with semantic modifiers (e.g. verbals, common full words, stative verbals, proper names of persons). Some complex place names in Table 3 are formally nominalisations of verbal clauses and locative constructions. A number of descriptive labels of place names contain the general article te (e.g. Teho 0 oho 0 o ‘the assemblage of stones’, Teputu 0 i ‘the disobedience’, Te 0 ehi ‘the coconut’). The article te is an integral part of the place name since proper names (place names and personal names) do not take any articles (Cablitz, 2006, pp. 71–77). It is not predictable when a place name contains te. For instance, place names whose descriptive labels consist of plant names or body part terms occur with the article te as well as without it (e.g. plant names: Tepua ‘the flower’ and Fa 0 aku 0 a ‘red hibiscus’; body part terms: Te 0 uma ‘the breast’, Haha ‘mouth’).

6

Marquesan locative constructions are similar to English locative constructions of the type on top of (the table) or in front of (the door), containing a preposition, a local noun as lexical head and a possessive construction. 7 The list of place names is by no means exhaustive. Examples are taken from Laporte (1995) and my own fieldwork data.

G.H. Cablitz / Language Sciences 30 (2008) 200–226

209

Table 2 Simple definite descriptions of Marquesan place names Lexical component

Vernacular form of place name (+island location)

English gloss of descriptive label

1. Landscape terms

Tepapa (Tahuata) Tepuna (Hiva 0Oa)

‘the lava rock’ the fresh water source’

2. Plant names

Tenoni (0Ua Huka) Tetumuotehutu (Nuku Hiva)

the Indian mulberry’a ‘the barringtonia tree’ (lit. ‘the trunk of the barringtonia’)

3. Animal names

Kio 0 e (0Ua Huka) Fafa 0 ua (Hiva 0Oa)

‘rat’ ‘manta ray’

4. Names for objects or things

Ke 0 atu 0 ipopoi (Hiva 0Oa) Popoma(Hiva 0Oa)

‘popoib pounder’ ‘fermented breadfruit ball’

5. Body part terms

Te 0 uma (Hiva 0Oa) Kopukiki (0Ua Huka) Te 0 ae (Fatu Iva)

‘the breast’ ‘backside, ass’c (lit. ‘side red’) ‘the forehead’

a

Citrifolia morinda. Popoi is a traditional Marquesan dish made of fermented and freshly cooked breadfruit, the paste mashed with a pounder. c Kopukiki is considered bad vulgar language. b

With respect to the complex forms of place names in Table 3, landscape terms frequently combine with all the other lexical components shown in Table 2. Landscape terms combine e.g. with plant names (Hanamı0 o ‘bay (of) pink wood’), animal names (Papatoetoe ‘lava rock (of) black crabs’, Motumanu ‘bird’s island’) or words expressing events or states (Teavahinena 0 o ‘the passage (of) falling in love’, Hakaha 0 a ‘bay (of) rage’). Structurally these complex forms of place names look like descriptive noun phrase constructions in which verbals or nominals modify a nucleus of a phrase.8 Whereas the nucleus – the first element in a complex form of a place name – refers to an individual entity (e.g. a geographical entity), the second, modifying nominal or verbal expresses a quality of the modified entity. For instance, Hanamı0 o (lit. ‘bay rosewood’) is the name of a bay at Fatu Iva which was characterised by having a lot of rosewood trees.9 Most place names containing the name of a mythological figure are structurally noun phrases with a possessive construction: (1) Te = ava = o = tupa ART = passage = Poss = Tupa (=name of mythological figure)10

8

This is according to a pattern typical of Polynesian languages (cf. Mosel and Hovdhaugen, 1992; Bauer, 1997; Elbert and Pukui, 1979). 9 Mı’o ‘rosewood’ (Thespesia populnea). 10 Abbreviations used: AGS = agens, ART = article, bp = body part, CAUS = causative prefix, Dem = demonstrative, DIR = directional particle, DL = dual, DO = direct object preposition, indef = indefinite, LD = locational-directional preposition, loc = locative, NUM = numeral particle, OBL = oblique preposition, Perf = perfective, PL = plural, POSS = possessive preposition, PREP = preposition, RED = reduplication, TAM = tense-aspect-mood particle.

210

Table 3 Complex forms of Marquesan place names and their combinatorial patterns Vernacular form of place name (+island location)

English gloss of descriptive label

1 Landscape term + landscape term

Motupapa (0Ua Pou) Motuivi (Hiva 0Oa)

‘islet of lava rock’ ‘hilly islet’ (lit. ‘islet hill’)

2 Landscape term + stative verbal

Teava 0 i 0 i (Tahuata)

‘strong current reef passage’ (lit. ‘the reef passage strong’) ‘long terrain’ (lit. ‘terrain long’)

Taha 0 oa (Hiva 0Oa) 3 Landscape term + animal name

Motumanu (Hiva 0Oa) Teavahonu (Tahuata)

a‘bird’s island’ (lit. ‘island bird’) ‘the turtle reef passage’ (lit. ‘the reef passage turtle’

4 Landscape term + plant name

Hanamı0 o (Fatu Iva) Hanahutu (Fatu Iva)

pink wood bay’ (lit. ‘bay pink wood’a) ‘barringtonia tree bay’ (lit. ‘bay barringtonia’)

5 Landscape term + mythological figure

Teohootupa (Hiva 0Oa) Anaotiu (Hiva 0Oa)

‘the cape of Tupa’ ‘cave of Tiu’

6 Body part term + animal name or landscape term

Hahamano (Tahuata) Matafenuab (Hiva 0Oa)

‘big shark’s mouth’ (lit. ‘mouth big shark’) ‘land’s end’ (lit. ‘eye land’)

7a Description of events (nominalisation)

Tetopatiatohu (Hiva 0Oa) (Art + V + N)

‘the revealing of truth’ (lit. ‘the falling of the truth’) ‘the teeth breaking’

Tenihofati (Fatu Iva) (Art + N + V) 7b Description of events (verbal/nominalised verbal + mythological figure)

Tutaekena (Hiva 0Oa) (V + PN) Temininaohina (Tahuata) (Art + V = Nom + Poss + PN)

‘Kena defecates’ ‘the urinating of Hina’

8 Locative

Teunaotemanu (Hiva 0Oa) 0 Otopuhi (Fatu Iva)

‘above the bird’ ‘moray’s hole’ (lit. ‘inside moray’)

a b

Thespesia populnea. Matafenua is the most south-easterly point of the Marquesan archipelago.

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Lexical components

G.H. Cablitz / Language Sciences 30 (2008) 200–226

211

Fig. 2. Photograph depicting the place of Temiminaohina, ‘the urinating of Hina’, (Tahuata, Motupu). According to the legends and anecdotes of the goddess Hina, the extreme splitting of the rock – as shown in the photo – is explained by the fact that Hina’s urine was so strong that it made the rock split apart.

Place names containing the name of a mythological figure can also contain words denoting events (Tutaekena ‘Kena defecates’) and may describe extraordinary geological phenomena of the environment (Temiminaohina ‘the urinating of Hina’, see Fig. 2). A number of place names contain descriptive labels of landscape features, the most frequently used being motu ‘island, islet’, ana ‘cave’, haka/hana/ha 0 a ‘bay’, mata 0 ae/oho ‘cape’, papa ‘lava rock’, vai ‘water’, ava ‘reef or mountain passage’ and 0 ua ‘water hole’. Landscape terms such as motu ‘island, islet’, haka/hana/ha 0 a ‘bay’, mata 0 ae/oho ‘cape’ and ava ‘reef or mountain passage’ are geographically salient places with which the inhabitants daily interact in one way or the other. Geographical entities such as mata 0 ae/oho ‘cape’ or motu ‘island, islet’ are, for example, important points of reference when locating fish banks on the ocean by creating an angle with these salient geographical entities. Ana ‘cave’ and 0  ua ‘(water) hole’ are culturally important places. Marquesans buried the mortal remains of their ancestors and their personal belongings in caves on steep almost inaccessible mountain ranges for protection. Most of these caves were set under a taboo to enter them. Even today these taboos persist. Marquesans believe that if these taboos are disrespected people will be punished by ancestral spirits. This can be manifested in illnesses or even result in death. Likewise, 0  ua ‘hole, (water) pool’ is also of cultural importance because in former Marquesan society particular members of the noble classes had their own private pools in the river. It was strictly taboo for other people to use these pools. In general one can say that place names containing landscape terms actually do refer to the entity denoted by their descriptive landscape label. This aspect is interesting because it is disputed in the literature (cf. Lyons, 1977; Lerner and Zimmermann, 1991) whether

212

G.H. Cablitz / Language Sciences 30 (2008) 200–226

proper names, including place names, have a meaning or not. Lyons (1977, p. 219), for instance, argues that proper names have reference, but not sense. According to Lyons the semantic function of proper names is to ‘‘identify their referents, not by describing them in terms of some relevant property or properties which the name denotes, but by utilizing the unique unequivocally and arbitrary association which holds between the name and its bearer’’ (Lyons (117 p. 214; my italics, G.C.). I agree that place names, like other proper names, are semantically unequivocal and have direct referentiality, namely that the relationship between the place name and the place is referentially unique in a precisely defined context.11 Lyons’ definition, however, is problematic because Marquesan place names containing landscape terms do indeed identify their referents by describing a property which the name denotes, namely the geographic entity described by the landscape term. The place name Teanaotetua ‘the cave of Tetua’ still refers to a cave – and not any other entity – in a similar way as kind-denoting nominals such as dog or house denote an entity of the class of dogs or houses. In fact a definite description such as Peter’s dog is comparable to Teanaotetua ‘the cave of Tetua’ as it is referentially specific and unique, but most researchers would agree that Peter’s dog cannot be characterised as being a proper name like Teanaotetua. The distinction between proper names and kind-denoting nominals in Marquesan is not always clearcut and whether they are classified as a kind-denoting nominal or a proper name depends on how it is used. If Tetua would possess several caves to which one would generally refer as ‘the cave of Tetua’ it would no longer be a proper name. Likewise, if Peter’s dog would have no other name and would always be called Peter’s dog then Peter’s dog would be a proper name, and so forth. The simple descriptive labelling is a very characteristic feature of naming locations or places in Marquesan. Names of places and persons are extremely important in the Marquesan culture – like elsewhere – regardless of whether we are dealing with names for salient geographical entities or just a small rock where an important event took place. Names of places and persons represent a link to their history and past, and most importantly to their ancestors. In the Marquesan society names and name-giving as well as name-exchanging were traditionally important social and political tools of interaction. If a chief would give his name to an outsider of his tribe, this person had the same rights as the chief including the right to ask for sexual favours from the chief’s wife. Name-giving was regulated by strict taboos. In former times the grandparents would decide which name a newborn offspring would be allowed to carry. The grandparents would transmit the right to the parents to pass on a particular name of a family ancestor to the child; but once the newborn was given the name, the parents had to ‘‘return the right’’ of giving that particular name to the grandparents. Names gave social as well as land rights. Some children were named according to the names of land, thus place names, they would inherit at a later stage in life. Finally, it is interesting to note that some Marquesan place names also represent a symbolic/metaphorical system. The naming of the Marquesas Islands is rooted in their creation myth (Dumond-Fillon, 1996). Each of the six inhabited islands of the Marquesas is named after a part of a Marquesan house or refer to the creation myth in some way.

11 Note that one place name can actually refer to more than one location (e.g. the two Frankfurts in Germany). However, in a precisely defined context they are referentially unique, often by adding a certain characteristic becoming part of the place name itself (e.g. Frankfurt am Main).

G.H. Cablitz / Language Sciences 30 (2008) 200–226

213

Oatea and Atanua, the first man and woman to arrive on the Marquesas Islands, decided to build a house by starting with two pillars (0Ua Pou, lit. ‘two pillars’), continuing with the rafter (Hiva 0Oa, lit. ‘side long’) and the roof (Nuku Hiva, lit. ‘assemble roof/side’). For constructing the backside roof they needed nine plaited coconut fronds (Fatu Iva, lit. ‘plaited coconut frond nine’). The work was done at night and had to be finished before daybreak, so towards the end of the construction Atanua shouted that dawn had broken (Tahuata, lit. ‘light up image’). They heard the morning bird (Mohotani,12 lit. ‘bird crow/ sing’) and they cried ‘‘attention, it will be light!’’ (Eiao,13 lit. ‘hey/attention daylight’)14 So Oatea and Atanua rushed to finish the work by digging a hole to clear up the left-overs of the house-building (0Ua Huka, lit. ‘hole left-overs/rubbish’). This kind of island naming is unique in French Polynesia and it does not only demonstrate the strong cultural union between the islands – despite their linguistic and cultural diversity – but also the importance of the concept of HOUSE in Pacific cultures (cf. Hill, 1996).

3. Formal classification and ontological status of landscape and body part terms This section discusses how the linguistic marking of landscape and body part terms is revealing with respect to how Marquesans categorise ‘‘what’’ and ‘‘where’’ in their language. The linguistic marking of landscape and body part terms is contrasted with the prepositional marking of location-denoting nominals such as place names and local nouns. Furthermore, this section will draw a clearer picture of how Marquesans classify landscape terms. In particular, it will address the question whether geographical entities such as river, mountain, sea, etc. are more perceived and conceptualised like places or more like first-order entities. Landscape terms are compared with body part terms which develop into locatives (e.g. ma te tua ‘behind’, lit. ‘along the back/spine’).

3.1. The ontological status of landscape entities and features Lyons (1977, p. 693) suggests that geographical entities or landscape features are an interesting ontological category because they hold an intermediate position between ‘being a first-order entity’ (e.g. an object) and ‘being a place’. According to Lyons landscape features such as cliffs, mountains, rivers etc. can be perceived and conceptualised as first-order entities, but need not be. Lyons (1977) comments There are some first-order entities that are either permanently or normally static, rather than self-moving or moveable: but they will not count as first-order entities unless the language so classifies them and they stand out from their environment with respect to their colour, shape or texture. Such aggregates, collections or conglomerations of matter as cliffs, clouds, lakes, and so on, may or may not be perceived and conceptualized as first-order entities: their status is indeterminate; and they may be treated differently by different languages. 12 13 14

Mohotani is an uninhabited island south-east of Hiva 0Oa (cf. Fig. 1). Eiao is an uninhabited island north-west of Nuku Hiva (cf. Fig. 1). Ei is an interjection expressing astonishment and warning.

214

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Smith and Mark (1998, 2001) claim that geographical entities are ontologically different from non-geographical entities (i.e. Lyons’ first-order entities) and that this distinction is not only a matter of scale. Unlike manipulable objects, geographical entities are not merely located in space but they are part of the landscape, i.e. they constitute space itself. On the other hand, empirical experimental data show that geographical objects (i.e. geographical entities) ‘‘are organized into categories in much the same way as are detached, manipulable objects’’ (Smith and Mark, 2001, p. 596). This research has provided valuable insights for the kinds of questions which are relevant for linguistic research. One of the questions which are of relevance for our purposes is if places and geographic entities are two superordinate kinds of things in the landscape and, if so, if they are treated differently in language and thought. In what follows I will show that the ontological ambiguity concerning geographical entities is quite obviously expressed in Marquesan. The Marquesan language is in fact a good testing ground because speakers make a linguistic distinction when talking about the location of things (or first-order entities) or about locations as such. Words denoting mere locations such as local nouns, place names and body part terms used as locatives15 always take the preposition 0 i (see Example 2), whereas words denoting first-order entities always receive the preposition 0 io (see Example 3):16

The prepositions 0 i and 0 io both mark nominals for location, goal or direction towards a goal. Both prepositions seem to be distributed in different lexical classes: 0 i typically marks location-denoting nominals such as local nouns, body part terms and place names, whereas 0 io marks proper names of persons, personal pronouns and common full words.18 However, the difference between 0 i- and 0 io-marking is not a simple matter of distribution across different lexical classes, but one based on a semantic distinction, namely that speakers mark the difference between locations as such ( 0 i-marking) and the locations of things, 15 Note that the same body part terms often denote body parts as well as object parts. In Marquesan, the different semantic functions of body part terms are distinguished by the different morphosyntactic construction types in which they occur (cf. Cablitz, 2006, for details). 16 Note that the 0 i- and 0 io-distinction – as described in this article – is true for the North Marquesan vernaculars on 0Ua Pou and Nuku Hiva islands. In the South Marquesan dialects 0 io is not used in the same way. The usage of 0 io in South Marquesan still has to be examined in more detail. 17 The sample sentences are taken from a corpus of naturally spoken language. The data of the corpus consists of narratives, interactive tasks and elicitations. These data were collected on field trips to the Marquesas between 1996 and 1999 on the islands of ’Ua Pou and Nuku Hiva in the North Marquesas. 18 I prefer the term ‘common full word’ to ‘common noun’ because there is no real verb/noun-distinction in Marquesan (cf. Cablitz, 2006, for detailed analysis).

G.H. Cablitz / Language Sciences 30 (2008) 200–226

215

animals or persons ( 0 io-marking). This is in particular evident with respect to a number of nominals which allow 0 i- as well as 0 io-marking such as landscape terms, body part terms and the local noun tai ‘sea’. In Marquesan, landscape terms do not form a coherent lexical class. In my analysis (Cablitz, 2006) lexical classes are characterised by formal or morphosyntactic criteria alone. In some cases formal and semantic criteria go hand in hand in that lexemes with the same semantic criteria also manifest the same formal properties. This is e.g. the case with place names. The term ‘landscape term’ is based on semantic criteria alone and includes all expressions which denote geographical entities as described in Section 2.1. Some landscape terms show more formal similarities with the class of common full words, whereas other landscape terms – such as tai ‘sea’ and uta ‘ashore, inland’ – belong to the class of local nouns. 3.2. Marquesan landscape terms: when ‘‘what’’ is ‘‘where’’ This subsection discusses whether landscape terms are linguistically treated as first-order entities or as places. As just mentioned the landscape terms tai ‘sea’ and uta ‘ashore, inland’ belong to the class of local nouns which are location-denoting nominals exclusively used to express spatial relations. In fact tai ‘sea’ and uta ‘inland’ form part of a so-called geocentric system which is used for the description of spatial configurations in small-scale as well as in large-scale reference, be it to distinguish two mangos on a tree or to localise an object in tabletop space. The two landscape terms tai ‘sea’ and uta ‘inland’ form a system of directional oppositions in that tai-constructions are always semantically opposed to uta-constructions. In the Marquesan geocentric system,19 the main SEA/INLAND-axis is complemented by an orthogonal axis which is lexically undifferentiated and labelled in both directions as k o ‘left or right side of valley’. In the literature this axis is often labelled as ACROSS (cf. Levinson, 1996). Although k o denotes a region within a Marquesan valley, the etymology of k o is not really clear (cf. Cablitz, 2006, pp. 555–558). In any case, the local noun ko cannot be classified as a landscape term. Being local nouns used as geocentric descriptors tai ‘sea’, uta ‘inland’ and ko ‘across’ are marked by the preposition 0 i:

19

In geocentric – or absolute – systems some salient landscape features are used as a secondary reference point for spatial description in macro- as well as micro-level space (see in particular Levinson, 1996).

216

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Landscape terms which are not used in the geocentric system such as mata 0 ae ‘cape’, tuaivi ‘mountain’, mouka ‘inland mountain’, etc. mostly take the same prepositional marking as common full words denoting first-order entities, namely 0 io. Note the contrast in prepositional marking with the local noun uta ‘inland’ and the landscape terms mouka ‘inland mountain’:

A number of landscape terms such as ka 0 avai ‘river, valley’, moana ‘far out at sea’ and henua ‘land, terrain’ can take 0 i- as well as 0 io-marking:

G.H. Cablitz / Language Sciences 30 (2008) 200–226

217

Comparing the 0 i- and 0 io-examples used with the landscape term ka 0 avai ‘river, valley’, we see that the 0 io-marked ka 0 avai refers to the river, i.e. a smaller space, whereas the 0 imarked ka 0 avai refers to a vaster space, namely a valley or village. However, with respect to the landscape terms moana ‘far out at sea’ and henua ‘land, terrain, country’ we can no longer argue that the difference of 0 i- and 0 io-marking – as in the case of ka 0 avai – can be explained by the size of location occupied by the geographic entity. With respect to henua ‘land, terrain, country’ we can observe different morphosyntactic properties in 0 i- and 0 iomarked NPs (as far as the data can tell). 0 I-marked henua ‘land, terrain’ can take a range of articles (e.g. te, titahi, hua20) and collective nominals (e.g. tau, pito21) and it is often modified by possessive attributes with proper names of persons, common full words or personal pronouns ( 0 i to ıa henua ‘on his land’). 0 Io-marked henua only occurs with the article he or with pre- and post-positioned possessive attributes (e.g. 0 io to ıa henua ‘on his land’, 0 io he henua o Toto ‘on the land of Toto’). In general, the article he occurs in most 0 io-marked NPs (see Examples 9–11, 14 and 16, see also below). However, there are no

20

The articles te, titahi and hua can be glossed in the following way: te is a kind of general default article marking definite, indefinite and non-specific NPs alike; titahi marks indefinite NPs and hua is an anaphoric article best glossed as ‘the same’. 21 Tau and pito are prenuclear collective nominals which never occur on their own and mark the lexical head of a NP for plurality. Tau is the most neutral collective nominal which can precede any common full word. Pito, on the other hand, means ‘piece’ which is often used in connection with a coherent entity.

218

G.H. Cablitz / Language Sciences 30 (2008) 200–226

clear patterns as to whether specific morphosyntactic properties trigger 0 i- or 0 io-marking. Furthermore, there seems to be no semantic difference between 0 i or 0 io-marked henua. However, the size of the geographical entity can explain the difference between 0 i- or 0 iomarking of landscape terms. In general we can say that landscape terms which denote smaller confined spaces such as mata 0 ae ‘cape’, ava ‘passage’, papua ‘garden’, opata ‘cliffs’, tuaivi ‘mountain’ or tahuna ‘gravel beach’ seem to be marked exclusively by 0 io (except henua ‘land, terrain’), whereas vaster spaces such as motu ‘island’, vao ‘bush, interior of island’ or moana ‘far out at sea’ are preferably marked by 0 i:

That size matters in the choice between 0 i and 0 io is also demonstrated by the fact that small islets or lava rocks in the sea, which are also called motu, are preferably marked by 0 io as shown in the following example:

Not surprisingly small confined spaces such as man-made landmarks (e.g. ha 0 e ‘house’, tohua ‘public place’, paepae ‘ancient stone platform of houses’) are always marked by 0 io. It seems that the boundaries of man-made landmarks and geographical entities such as ka 0 avai ‘river’, mouka ‘inland mountain’ or mata 0 ae ‘cape’ are more salient for the purpose of categorisation, i.e. they can be visually perceived in the same way as first-order entities and therefore receive preferably 0 io-marking like the class of common full words which typically denote objects or persons. Or in other words: 0 io-marked geographical entities are simply perceived and conceptualised more as things than as places. Geographical entities such as vao ‘interior of island’, motu ‘island’ and moana ‘far out at sea’ are vast spaces whose boundaries cannot – in most cases – be visually perceived and therefore are less likely to be conceptualised as objects or things. This does not mean that locations are per se conceptualised as such vast spaces. Locations can also be regions of objects on a small-scale level such as the ‘front’ region of a chair or the ‘inside’ region of a bowl. In other words: we can distinguish between different ontological categories, i.e. ‘‘things’’ or ‘‘places’’ regardless of their size, but when the ontological status is less clear – as in the case of geographical entities – size can become an important factor in the categorisation of these entities. We can conclude so far that there is a gradient based on size with respect to the 0 i- and 0 io-marking of geographical entities which can be illustrated as follows (see Fig. 3).

G.H. Cablitz / Language Sciences 30 (2008) 200–226

219

Fig. 3. Gradient of size with respect to 0 i- and 0 io-marking.

Although landscape terms which refer to vaster spaces like moana ‘far out at sea’ are likely to receive 0 i-marking, there are, however, also occurrences with 0 io-marking (see Example 14). Whereas the 0 i- and 0 io-marking of motu ‘island, islet’ can be undoubtedly explained by the difference in size (‘island’ [ 0 i-marking] vs. ‘islet’ [ 0 io-marking]), the difference between 0 i- and 0 io-marking of moana cannot. In Examples (13) and (14) the landscape term moana refers to exactly the same geographical entity without an implied difference in size. According to my consultants 0 io-marking is more felicitous when participants have actual physical contact with the geographical entity. Otherwise – as a kind of default-marking – 0 i is used. The local noun tai ‘sea’ is the only local noun which can take 0 i- as well as 0 io-marking, and the reasoning behind it bears some similarity to the difference between the 0 i- and 0 iomarking of moana ‘far out at sea’:

220

G.H. Cablitz / Language Sciences 30 (2008) 200–226

There is a clear difference between the 0 i- and 0 io-marking of tai. Whereas 0 i-marked tai refers to the sea-region around the geographical entity SEA which includes the beach and also the land close to the beach, but not the geographical entity SEA as such; 0 io-marked tai, on the other hand, can only refer to the geographical entity SEA, thus when the participants have actual physical contact with the geographical entity. The local noun tai basically has two denotational domains: (1) the actual geographical entity SEA, and (2) the location around the geographical entity SEA (i.e. the beach and its surroundings), but not the geographical entity SEA as such. 0 I- and 0 io-marking with tai, in fact, distinguishes two distinct places. Tai not only takes 0 io-marking, it is also distinguished from uta ‘inland’ and ko ‘across’ (with which it forms the geocentric system) by a number of different morphosyntactic and syntactic properties. Apart from being marked by 0 io, tai can function as subject and object NP in verbal clauses, it is found in noun-incorporating constructions ( 0 i te tai kaukau ‘at the sea-swimming’) and it occurs in a number of compounds (tahatai ‘shoreline’, pa 0 atai ‘salt’, taiheke ‘tide out’, taipı ‘tide in’, etc.). Some of the morphosyntactic and syntactic properties which distinguishes tai ‘sea’ from other local nouns, are also shared by landscape terms. Apart from the 0 io-marking, landscape terms can also function as core arguments. Thus, landscape terms which are more likely to receive 0 i-marking are rather conceived as ‘‘places’’ than as ‘‘things’’, and the syntactic function as core argument is restricted to subject NP position in verbal clauses with state verbals. This also holds for tai ‘sea’ (see Example 26):

However, the majority of NPs containing landscape terms mostly function as locative complement phrases or adjuncts. Tai ‘sea’ shows all signs of a mixed word classification as it shares many morphosyntactic properties with the class of common full words as well as local nouns (cf. Cablitz, 2006, pp. 497ff.). However, tai only shares morphosyntactic and syntactic properties with common full words when it denotes the actual geographical entity SEA, but not the region around the geographical entity SEA. In the latter meaning – thus when marked by 0 i – tai only functions as locative complement phrase or adjunct.

G.H. Cablitz / Language Sciences 30 (2008) 200–226

221

3.3. Marquesan body part terms: when ‘‘what’’ is ‘‘where’’ This subsection describes the use of body part terms as grammaticalised markers of location. The grammaticalisation process of these terms is also mirrored in the use of 0 iand 0 io-marking. The border between a ‘‘what’’- and ‘‘where’’-category is blurred in a similar way as demonstrated for landscape terms. In many Oceanic languages the largest lexical source of spatial expressions derive from body part terms (Bowden, 1992, p. 35), a phenomenon which is also observed in a number of other (unrelated) languages (cf. Brugman and Macauley, 1986; de Leo´n and Levinson, 1992, for Mesoamerican languages; Heine, 1989, for African languages). In Marquesan, there is only a small class of four body part terms which are used as location-denoting nominals (e.g. ‘the ball is at the face of the car’):

One could argue that there are six body part terms used as locatives if one includes 0 ima a e ‘left hand’ and 0 ima oko ‘right hand’. However, the usage of these two body part terms is very restricted and the constructions types differ to a large extent from those of the body part terms listed in Example (27). These four body part terms have not changed their form as such in the process of grammaticalisation – as often observed in studies on grammaticalisation of body part terms – but they are part of complex noun phrase constructions. It is the use of certain construction types which indicates whether they are location-denoting nominals or not. The important aspect in the development of Marquesan body part terms into locatives is the semantic change or extension to different denotational domains. An intermediate stage in the development is the semantic extension of the body part term to denote an object part. This is a phenomenon which can be observed in many languages. The semantic extension to the denotational domain of OBJECT PART is, in fact, a crucial bridge from the domain OBJECT to the domain SPACE. This intermediate stage is also characterised by a state of adjacency or contact of the theme – the object to be localised – with the relatum – the object against which the theme is localised (cf. Bowden, 1992; Heine, 1989).22 In Marquesan, body part terms denoting body parts and object parts can take the preposition 0 io, whereas the same body part terms used as location-denoting nominals (e.g. ‘at the back-region (=behind) of a house’) cannot be marked by 0 io, but only by 0 i: 0

22

The terms ‘relatum’ and ‘theme’ – used e.g. by Klein (1991) – correspond to Talmy’s (1983) notions of ‘ground’ and ‘figure’.

222

G.H. Cablitz / Language Sciences 30 (2008) 200–226

There is a subtle difference between the two denotational domains ‘body part’ and ‘object part’ with respect to 0 io-marking. When body part terms denoting real body parts are marked by 0 io, contact between theme and relatum is implied. As for 0 io-marked body part terms denoting object parts the theme needs to be close to the relatum, but contact between theme and relatum is not a necessary condition for the usage of 0 io. It seems that the 0 io-marking of body part terms denoting object parts clearly represent an intermediate stage between the domain OBJECT and SPACE, but a stage in which contact between theme and relatum does not play a role anymore (see Heine, 1989; Levinson, 1992; Bowden, 1992; Svorou, 1994). The border between a ‘‘what’’- and ‘‘where’’-category is simply blurred. This is manifested in the 0 io-marking as well as its condition of usage (i.e. the non-adjacency between theme and relatum). Furthermore, body part terms denoting real body parts and object parts can function as core arguments in any type of clause; body part terms used as location-denoting nominals, on the other hand, can only occur as locative complement phrases or adjuncts. We will come back to this point in the following subsection. 3.4. 0 Io- and 0 i-marking and the implication for ontological categorisation of landscape and body part terms In the previous subsections I have discussed the various uses of the locative prepositions i and 0 io which are quite revealing with respect to how speakers are sensitive to the different ontological categories of first-order entities and places. When talking about spatial relations, speakers quite clearly make a distinction between the two ontological categories by marking NPs either by 0 i (places/locations) or 0 io (first-order entities). First of all, it is interesting to see that this distinction is linguistically expressed in Marquesan because other languages such as English, German or Hindi do not make this distinction at all (Narasimhan and Cablitz, 2002). So far, such a neat distinction between a ‘‘what’’- and ‘‘where’’-category when talking about spatial relations has not been reported for other languages. Note though that there are some similarities to Hawaiian (Cook, 2002). In focus of the present analysis were those lexical items which represent so-called borderline cases between ‘being a first-order entity’ and ‘being a place’, namely landscape and body part terms. Although these lexical items are often candidates for developing into locatives, we have two very different kinds of borderline cases at hand. The shift from denoting body parts to denoting spatial regions (PLACE/LOCATION) via an intermediate stage of denoting object parts is a shift based on a ‘‘categorial metaphor’’, namely that from OBJECT to SPACE (Heine, 1989, p. 101). The body and the relationship between its different parts play an important role with respect to how we understand spatial relationships and how we apply this knowledge when talking about locations (see Bowden, 1992, p. 30). The development of landscape terms into locatives (e.g. to be used in a geocentric system) is not based on the same kind of metaphorical understanding of SPACE, 0

G.H. Cablitz / Language Sciences 30 (2008) 200–226

223

although we have seen above that some landscape terms are based on the metaphor of the body (e.g. tuaivi ‘mountain’, mata 0 ae ‘cape’). As noted, Lyons (1977, p. 673) states that geographical entities are indeterminate whether they are perceived and conceptualised as first-order entities or as places, and he proposed to count them as first-order entities unless the language classifies them differently. Whether a geographical entity receives 0 i- or 0 io-marking seems to be dependent on the size of the geographical entity and the ability to perceive its boundaries or not. Natural geographical entities whose boundaries are less likely to be perceived (e.g. moana ‘far out at sea’ or motu ‘island’) are more likely to receive 0 i-marking – i.e. the same as location-denoting nominals; on the other hand, smaller geographical entities such as ka 0 avai ‘river’, ana ‘cave’ or kapua ‘mountain crater’ have clearer boundaries and therefore are more salient for the purpose of categorisation. With regard to the linguistic marking in Marquesan we cannot readily state that geographical and non-geographical entities are ontologically distinct – as proposed by Smith and Mark (1998) – and that this difference is not only a matter of scale or size. It seems that perceptual boundedness is an important criterion of ontological distinction as well: from the point of view of speakers of Marquesan the categorisation and ontological status of landscape feature seems to be a matter of being able to perceive an entity in its totality or not. However, Smith and Mark’s (1998) notion of scale or size does matter as well: the larger a geographical entity is, the less likely it is perceived as a first-order entity. But is there further evidence that geographical entities are more likely to be categorised as places than as first-order entities? And what is the connection to body part terms with regard to syntactic and morphosyntactic properties? As for the body part terms which have become locatives size is not a determining factor because they denote small locations or object regions in small-scale space. The shift from the domain OBJECT to the domain SPACE is also reflected in the syntactic transition from being used as core arguments to being used just as complement phrases or adjuncts. This kind of transition is often found in the grammaticalisation processes of body part terms. As for body part terms the domains OBJECT and SPACE are differentiated with respect to syntactic functions and case-marking. As for landscape terms, the division between OBJECT and SPACE is less clear. On the one hand, the majority of landscape terms is marked by 0 io. This speaks for a categorisation of them as first-order entities. On the other hand, landscape terms mostly function as locative complement phrases or adjuncts only – except for those occurrences in subject NP position with state verbals (see Examples 23–25). This might be evidence that they are more likely to be perceived and conceptualised as places. All in all, the data indicate that Marquesan landscape terms form an ontological in-between category. 4. Some concluding remarks In this paper, I have described how Marquesans label their landscape and name places in their environment. We have also looked at the relationship between landscape terms and place names and seen that the two domains can be semantically associated: landscape terms occur in place names (e.g. Motukio 0 e ‘rat island’) and these place names refer to the same geographical entities as the generic label included in the place name. For the most part, place names are derived from descriptive labels whose etymology is semantically

224

G.H. Cablitz / Language Sciences 30 (2008) 200–226

transparent. We can conclude that the naming of places goes back to past events, expresses a relationship of possession (Teanaotetua ‘the cave of Tetua’), is connected with mythology or simply depicts a very characteristic feature of that place (e.g. Papatoetoe ‘black crab lava rock’, i.e. a rock where one finds a lot of black crabs, etc.). Furthermore, the fact that the labelling of some landscape terms derive from body part terms is interesting because it shows that the categorisation of SPACE – i.e. the space we live in – can be based on metaphor (cf. Brown, 2008; Burenhult, 2008). A different kind of metaphorical labelling of geographical entities and phenomena was observed with local nouns. The usage of local nouns such as 0 oto ‘inside’ and vaho ‘outside’ for ‘bay’ and ‘ocean’, and 0 uka ‘up’ and 0 a 0 o ‘down’ to denote the flowing direction of sea currents shows that specific spatial concepts can serve as a model for structuring the environment. The second part of the paper addressed the question of ontological status of geographical entities and their corresponding labels. As previously observed in the literature (Lyons, 1977; Smith and Mark, 1998, 2001) there seems to be some kind of ambiguity with regard to the ontological status of these entities. Landscape features have an ambiguous ontological nature because, on the one hand, they constitute the landscape and therefore space itself, but, on the other hand, having distinctive features and boundaries, they tend to be categorised as first-order entities. The linguistic marking of landscape terms in Marquesan expresses this ambiguity, but factors such as size and the ability to perceive boundaries of these entities do play a role when speakers mark landscape terms either as location-denoting nominals or nominals denoting first-order entities. There should be more investigations to find out if there are other languages like Marquesan, which distinguish structurally between first-order entities and place? If so, do they categorise landscape features according principles similar to those of Marquesan? Further research can hopefully contribute to a better understanding of cross-linguistic variation, on the one hand, and basic cognitive principles, on the other. Acknowledgements This research has been generously supported by the Max Planck Institute for Psycholinguistics (Nijmegen) and a DoBeS grant of the Volkswagen foundation. I have greatly benefited from discussions with Tehoatahiiani Bruneau, Niclas Burenhult, Lucien and Tahia Mataiki, Ulrike Mosel, Bhuvana Narasimhan, Eric Pederson, Gunter Senft, Pierre Tahiatohiupoko, Benjamin Teikitutoua, Edgar Tetahiotupa, the participants of the Language and Space workshop in Bielefeld (July 2002) and the 2nd Conference on Austronesian Languages and Linguistics in Oxford (June 2006). I would like to thank in particular Ian Thomson for proof-reading my paper and Niclas Burenhult and two anonymous reviewers for their helpful comments on earlier versions of this paper. References Bauer, W., 1997. The Reed Reference Grammar of Maori. Reed Publishing, Auckland. Blanchard, P., 1997. Libres propos techniques et simples sur quelques aspects actuels de la de´mographie de la Polyne´sie franc¸aise. Bulletin de la Socie´te´ des Etudes Oce´aniennes 273–274, 71–87. Bowden, J., 1992. Behind the Preposition – Grammaticalisation of Locatives in Oceanic Languages. Pacific Linguistics, Canberra.

G.H. Cablitz / Language Sciences 30 (2008) 200–226

225

Broschart, J., 1991. Noun, Verb and Participation: a Typology of the Noun/Verb-Distinction. In: Seiler, H., Premper, W. (Eds.), Partizipation: das sprachliche Erfassen von Sachverhalten. Narr, Tu¨bingen, pp. 65–137. Brown, P., 2008. Up, down, and across the land: landscape terms, place names, and spatial language in Tzeltal. Language Sciences 30, 151–181. Brugman, C., Macauley, M., 1986. Interacting semantic systems: mixtec expressions of location. Proceedings of the Berkeley Linguistics Society 12, 315–327. Burenhult, N., 2008. Streams of words: hydrological lexicon in Jahai. Language Sciences 30, 182–199. Cablitz, G.H., 2005. S’orienter aux Marquises: une analyse linguistique des re´fe´rences spatiales du Marquisien. ´ ditions Au vent des ˆıles, Singapore, pp. 163–206. In: Rigo, B. (Ed.), L’Espace-Temps. E Cablitz, G.H., 2006. Marquesan – a Grammar of Space. Mouton de Gruyter, Berlin. Cook, K.W., 2002. The case markings of Hawaiian locative nouns and place names. In: Bennardo, G. (Ed.), Representing Space in Oceania: Culture in Language and Mind. Pacific Linguistics, Canberra, pp. 91–104. de Leo´n, L., Levinson, S.C., 1992. Introduction – spatial description in mesoamerican languages. Zeitschrift fu¨r Phonetik, Sprachwissenschaft und Kommunikation (special edition) 45 (6), 527–529. Dumond-Fillon, R., 1996. Le´gende de la Terre des Hommes – regards d’un linguiste. In: Dubus, D. Duponchel, JP., Heitaa, A., Pugibet, V., Tarroux, P., (Eds.), Marquises. E´ditions Polye`dre Culture, Tahiti. Elbert, S.H., Pukui, M.K., 1979. Hawaiian Grammar. University of Hawaii Press, Honululu. Green, R., 1966. Linguistic subgrouping within Polynesia: the implications for prehistoric settlement. Journal of the Polynesian Society 75, 6–38. Handy, E.S.C., 1923. The native culture in the marquesas. Bishop Museum Bulletin 9, Honolulu. Heine, B., 1989. Adpositions in african languages. Linguistique Africaine 2, 77–127. Hill, D., 1996. Distinguishing the notion of ‘place’ in an oceanic language. In: Pu¨tz, M., Dirven, R. (Eds.), The Construal of Space in Language and Thought. Mouton de Gruyter, Berlin, pp. 307–328. Hughes, H.G.A., Fischer, S.R., 1998. Introduction. In: Crook, W.P., Greatheed, S., Te’ite’i, T. (Eds.), An Essay Toward a Dictionary and Grammar of the Lesser-Australian Language, According to the Dialect Used at the Marquesas (1799). Linguistic Society of New Zealand, Auckland. ´ volution des structures de l’habitat a` Kellum-Ottino, M., 1971. Arche´ologie d’une valle´e des ˆıles Marquises: E Hane, Ua Huka. Publication de la Socie´te´ des Oce´anistes, Paris. Klein, W., 1991. Raumausdru¨cke. Linguistische Berichte 132, 77–114. Landau, B., Jackendoff, R., 1993. ‘‘What’’ and ‘‘where’’ in spatial language and spatial cognition. Behavioral and Brain Sciences 16, 217–265. Laporte, J., 1995. Toponymie des ˆıles Marquises (Fenua Enata). Bulletin de la Socie´te´ des Etudes Oce´aniennes 268, 2–31. Lavonde`s, H., 1983. Le vocabulaire marquisien de l’orientation dans l’espace. L’Ethnographie 79 (1), 35–42. Lerner, J-Y., Zimmermann, E., 1991. Eigennamen. In: von Stechow, A., Wunderlich, D. (Eds.), Semantik: Ein internationales Handbuch der zeitgeno¨ssischen Forschung. Walter de Gruyter, Berlin, pp. 349–369. Levinson, S.C., 1992. Primer for the field investigation of spatial description and conception. Pragmatics 2 (1), 5– 47. Levinson, S.C., 1996. Frames of reference and Molyneux’s question: crosslinguistic evidence. In: Bloom, P., Peterson, M., Nadel, L., Garrett, M. (Eds.), Language and Space. MIT Press, Cambridge, MA, pp. 109–170. Lyons, J., 1977. Semantics. Cambridge University Press, Cambridge. Marck, J.C., 1996. Eastern Polynesian subgrouping today. In: Davidson, J.M., Irwin, G., Leach, F., Pawley, A.K., Brown, D. (Eds.), Oceanic Cultural History: Essays in Honor of Roger Green. New Zealand Journal of Archaeology Special Publication, Wellington, pp. 491–511. Miller, G.A., Johnson-Laird, P.N., 1976. Language and Perception. Cambridge University Press, Cambridge. Mosel, U., Hovdhaugen, E., 1992. Samoan Reference Grammar. Oxford University Press, Oslo. Narasimhan, B., Cablitz, G.H., 2002. Granularity in the cross-linguistic encoding of motion and location. In: Paper presented at the Third Annual Workshop on language and Space, University of Bielefeld, July 8–9, 2002. Ottino, P., 1997. Naissance d’une culture oce´anienne. Bulletin de la Socie´te´ des Etudes Oce´aniennes 273–274, 3– 22. Pawley, A.K., 1966. Internal relationships of Polynesian languages and dialects. Journal of the Polynesian Society 75, 39–64. Senft, G., 2008. Landscape terms and place names in the Trobriand Islands – the Kaile’una subset. Language Sciences 30, 340–361.

226

G.H. Cablitz / Language Sciences 30 (2008) 200–226

Smith, B., Mark, D.M., 1998. Ontology and geographic kinds. In: Poiker, T.K., Chrisman, N. (Eds.), Proceedings of the 8th International Symposium on Spatial Data Handling (SDH’98), pp. 308–320. Smith, B., Mark, D.M., 2001. Geographical categories: an ontological investigation. International Journal of Geographical Information Science 15 (7), 591–612. Svorou, S., 1994. The Grammar of Space. John Benjamins, Amsterdam. Talmy, L., 1983. How language structures space. In: Pick, H.L., Acredolo, L.P. (Eds.), Spatial Orientation: Theory, Research and Application. Plenum Press, New York, pp. 225–320. Thomas, N., 1990. Marquesan Societies. Inequality and Political Transformation in Eastern Polynesia. Clarendon Press, Oxford. Vonen, A.M., 2000. Polynesian multifunctionality and the ambitions of linguistic description. In: Vogel, P., Comrie, B. (Eds.), Approaches to Typology of Word Classes. Mouton de Gruyter, Berlin, pp. 479–488.