John Benjamins Publishing Company This is a contribution from Journal of Pidgin and Creole Languages 21:1 © 2006. John Benjamins Publishing Company This electronic file may not be altered in any way. The author(s) of this article is/are permitted to use this PDF file to generate printed copies to be used by way of offprints, for their personal use only. Permission is granted by the publishers to post this file on a closed server which is accessible to members (students and staff) only of the author’s/s’ institute. For any other use of this material prior written permission should be obtained from the publishers or through the Copyright Clearance Center (for USA: www.copyright.com). Please contact rights@benjamins.nl or consult our website: www.benjamins.com Tables of Contents, abstracts and guidelines are available at www.benjamins.com Universal constraints and local conditions in Pidginization Case studies from New Guinea William Foley University of Sydney This paper looks at the processes involved in the genesis of three pidgins of New Guinea. I will argue that the main principle at work in the formation of the morphosyntax of pidgins is a general human capacity for language simplification, but that these processes of simplification are subject both to the effects of specific linguistic features in the local multilingual contact situation that gives rise to the pidgin and to constraints resulting from a universal linguistic endowment. By a comparative look across a range of morphosyntactic features at the processes of simplification that produced three unrelated pidgins of New Guinea, Yimas Pidgin, Hiri Motu and Tok Pisin, the article exemplifies some of the types of structures that can result in pidgins from this general human capacity for language simplification, sifted through both local conditions and universal constraints. Keywords: pidgins, language contact, language simplification, grammatical relations, tense-aspect, New Guinea, Papuan languages, Austronesian languages . The processes of Pidginization It is generally agreed that pidginization is a process that occurs in a multilingual contact situation (Siegel 2000; Thomason 2001). A pidgin language is one that results in a multilingual contact situation in order to aid communication between groups speaking distinct languages. Of course, pidginization is only one result, and perhaps an unusual one at that, of a multilingual contact situation; others include stable multilingualism, with or without language Journal of Pidgin and Creole Languages : (), ‒. ‒ ⁄ - ‒ © John Benjamins Publishing Company © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.2 (146-194) William Foley convergence, as in Kupwar village in India (Gumperz & Wilson 1971), mixed languages like Media Lengua (Muysken 1988, 1997) or Michif (Bakker 1997), a lingua franca or koineization (Siegel 1985, 1993) (this perhaps restricted to situations involving closely related languages or dialects of the same language) and language death (Gal 1979; Kulick 1992). The results of pidginization are an amalgam of the linguistic elements of two or more languages, as with a number of the processes enumerated above, but what crucially distinguishes pidginization from all of these others is a necessary simplification of the resulting amalgam language from the source languages in the original multilingual contact situation. One language, typically that of the socioeconomically dominant cultural group usually contributes the bulk of the vocabulary to the pidgin: this is the superstrate language. The other contributing language(s) are substrate languages. A fundamental question of pidginization concerns the linguistic processes that give rise to the grammatical structure of the resulting pidgin language. An overarching process here is what Thomason (2001) calls negotiation. In the classic pidginization context, none of the contributing ethnolinguistic groups knows the others’ language(s), or has the desire or means to acquire it. Still communication must take place, and as speakers of the various languages try to accomplish this, they make guesses about what will work. If these guesses hit the mark, there is a good chance they will stick; if they fail, something else will be tried. Gradually through this process of hits and misses across the language lines, in short, a negotiation, an agreed linguistic format for these exchanges emerges, and, if the contact situation remains of long standing, decades or generations, a stable pidgin will likely be crystallized and be passed on to succeeding generations. But the real question in this scenario is what guides these guesses. Speakers actually have two types of knowledge. First, they know their own language, so one option available is to simply lexify the sentences of their own language with word forms they learn from the other language(s), a type of extensive borrowing. Because of the typical differences in social status among the speakers of the languages in pidginization contexts, the lexical items will mainly be drawn from the superstrate language, the highest status language in the multilingual interactions (though apparent exceptions do occur, e.g. Chinook Jargon, whose lexicon is mainly drawn from Native languages of the northwest coast of North America). This in essence is the relexification theory of pidgin genesis, a current and influential view advanced by Lefebvre (1998), who claims that pidgin and creole languages result from relexification of the substrate language(s) by © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.3 (194-282) Universal constraints and local conditions in Pidginization lexical items from the superstrate language, an analog of the processes that gave rise to mixed languages like Media Lengua (Muysken 1988, 1997). In this view the basic structure of a pidgin may be simplified vis a vis the source languages, but mainly along the lines laid down by the substrate language(s), so that its structure will parallel in many respects that of one or more of the substrate languages. While Lefebvre (1998) makes a case for such a process at work in the development of Haitian Creole, this approach cannot be a complete account of the processes of pidginization. Indeed an independent process of simplification, over and above one of relexification (and reanalysis and leveling, other processes appealed to by Lefebvre (1998)), is necessary to account for the origins of most pidgins. So while relexification does and must play a role in the processes of linguistic negotiation in pidginization, it cannot be generally the whole or even the main factor. Pidgin languages are typically simpler than both their superstrate and substrate languages, and processes of simplification are crucial in their formation. So the claim here is that the basic grammar of a pidgin is typically arrived at by radical simplification of the contributing languages (this claim can be modulated in the case where the contributing languages, and especially the substrate language(s) are both relatively homogeneous and already rather morphologically simple, isolating languages, as is possible in the linguistic processes that led to the genesis of Haitian Creole (Lefebvre 1998)). Other processes discussed by Siegel (1997, 2000), such as leveling, e.g. the elimination of variant forms and the retention of others in the resulting pidgin, clearly play a role in the crystallization and stabilization of a pidgin, but in my view, a secondary one to simplification: the processes of simplification generate the variants in the incipient code in the first place, among which the leveling forces then select. But this notion of simplification itself needs spelling out: what specifically is entailed when we talk about processes of linguistic simplification? Siegel (2000) provides a summary of simplification processes under four rubrics: a) reduction; b) increased regularity; c) greater transparency; and d) lack of markedness. A general thrust of much of Charles Ferguson’s work is the view that it is part of the linguistic competence of adult speakers of a language to know how to simplify it, if social circumstances should warrant this. Besides multilingual contact situations, another social context which may require this competence is baby talk. Table 1, adapted from Ferguson (1982) displays the features of simplified language forms in contrast to those of the source language: © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.4 (282-307) William Foley Table 1. Linguistic simplification More complex linguistic structure Simpler or simplified linguistic structure Lexicon Larger vocabulary in a given domain or overall; specialized registers. Compounds and polymorphemic words Smaller vocabulary, generic terms rather than specific. Monomorphemic words; paraphrases of semantically complex words Syntax Subordination; hypotaxis. Free or variable word order conditioned by syntax. Presence of copula, pronouns, function words Coordination; parataxis. Invariable word order. Absence of copula, pronouns, function words Morphology Extensive systems of inflectional and derivational affixes. Allomorphy, including of stems Heavily reduced or no inflectional and derivational affixes. No allomorphy, invariant stems Phonology Consonant clusters. Polysyllabic words CV monosyllables and CVCV disyllables There is a close correlation between Siegel’s (2000) four rubrics of simplification and the processes listed in Table 1: reduction, exemplified by smaller vocabulary and heavily reduced or no affixation; increased regularity, in invariable word order and no allomorphy; greater transparency, also in no allomorphy and the presence of monomorphemic words or paraphrases of semantically complex words. Markedness is more difficult, in that what counts as marked versus unmarked is often strongly theoretically informed, but one rather uncontroversial way to get a handle on this is through the widespread distribution of typological features. On this metric, parataxis, i.e. coordinate in preference to subordinate clauses, exhibits an unmarked setting, as no language lacks coordination as a clause combining option, but some do lack subordination, such as many polysynthetic languages. CV syllable structure is another unmarked setting: all languages have this syllable type, but some, e.g. Polynesian languages, only have it. So we can take Siegel’s four rubrics, as exemplified in the more specific processes of Table 1, as a set of parameters for linguistic simplification. But is there a still more general principle we can reduce these to? I believe so, and this is it: the restriction of words to generic terms, the elimination of allomorphy, the use of parataxis rather than embedding, and the adoption of invariant word order for grammatical relations all ease the processing load of the hearer. There is no need to cope with variations and irregularities. Furthermore, the loss of in- © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.5 (307-355) Universal constraints and local conditions in Pidginization flections and function words and their replacement by full lexemes maximizes the area of the language easiest for the foreigner to acquire – the lexicon. Indeed, early writers (Silverstein 1972; Mühlhäusler 1979) noted how some early pidgins consist of no more than a reduced lexicon put through the grammatical rules of one’s native language. Ultimately, of course, this insight is the basis of the relexification hypothesis, but this will only look like pidginization if the rules of the native language are simplified during the process; otherwise the result would be a mixed language like Media Lengua, and these two results must not be conflated. Media Lengua (Muysken 1988, 1997) is a morphologically complex language, like Quechua, but with a Spanish root lexicon. It is a far cry from any pidgin (or creole) in structure. Simplification is the crucial component in pidginization; relexification is operative, but on structures produced by simplification. The idea that linguistic simplification (in all simplified registers: pidgins, foreigner talk, baby talk, etc.) rests on a intention to ease the processing load of the hearer is elaborated in recent work by Kusters (2003). He distinguishes two types of speech communities: Type 1, more closed, isolated and largely monolingual; and Type 2, open communities with a history of extensive contacts and much second language acquisition. He argues that there is a statistical (not absolute, statistical) correlation between linguistic complexity (measured, say, by the features of Table 1) and the type of speech community: languages spoken in Type 2 communities show higher features of linguistic simplification than those of Type 1 (it hardly needs saying that pidgins are prototypically associated with Type 2 communities). The reason is obvious: speakers in Type 2 communities need to communicate efficiently to non-native speaking hearers and to do so, they employ the parameters of linguistic simplification exemplified above. Kusters (2003) develops his own three principles of linguistic simplification, which re-iterate the above framework: 1) the Economy Principle, that demands as few inflectional categories as possible (this is reduction); 2) the Transparency Principle, which requires that the relation betweeen form and meaning be as clear as possible (this is greater transparency); and 3) the Isomorphy Principle, which requires that the order of elements be the same in different domains (this is increased regularity). Ultimately, then, pidgins are linguistic accommodations arrived at by simplification of one’s native language along the lines laid out by the basic principles enunciated above, combined with relexification of the output simplified structure with formal items mainly derived from the superstrate language. This definition represents a general, plausibly universal, potential, and one well © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.6 (355-411) William Foley within adult linguistic competence, but, being rather vague, it allows a very wide latitude in its actual realization in any given pidginization context. Thus, pidgins may be structurally quite different depending on the local linguistic resources, i.e. the typology of the source languages contributing to the pidginization process. For example, we would expect to find different basic word orders in pidgins whose source languages were all verb final from those which are verb medial. The difference in word order in Hiri Motu and Tok Pisin in exactly this feature and in their source languages respectively amply confirms this claim. Further, the amount of simplification necessary to reach adequate linguistic accommodation among speakers can also be variable. To the extent that the languages or most speakers of them share a feature, albeit not a maximally invariant or simple one, that feature may appear in the resulting pidgin. The suffix marking transitive verbs in Tok Pisin is one such example, an analog being present in nearly all substrate Oceanic languages. The central question in the genesis of pidginization through linguistic simplification concerns the role of local versus universal constraints on outputs. Under the general principle of accommodate to the hearer and the rubrics of reduction, increased regularity, greater transparency and lack of markedness, are there universal constraints that operate over various components of grammar leading to similar results regardless of the input languages? Or are local conditions always determinative, so that the structure of pidgins is a localized compromise from the specific structures of the input languages? We do not know enough to answer this question, because there are as yet insufficient data from the crucial key witnesses – pidgins without European superstrates and particularly those developing prior to European colonial contact. This paper is intended to offer some data to start answering this question from one small corner of the globe, but that which was last to feel the blowtorch of the European colonial expansion. In the next three sections I will investigate grammatical structures in three pidgins of Papua New Guinea, Yimas Pidgin, Hiri Motu and Tok Pisin, and compare these to parallel structures in their source substrate and superstrate languages. These pidgins run the gamut from stable, but restricted (Yimas Pidgin) to fully expanded and creolized (Tok Pisin), with Hiri Motu somewhere in between. I will demonstrate that the resulting patterns of grammatical simplification differ significantly in the three languages and reflect the different types of structural patterns contributed by the local source languages. © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.7 (411-443) Universal constraints and local conditions in Pidginization . Yimas Pidgin As mentioned in the previous section, pidgins arise in attenuated language contact situations, either due to a wide social gulf between the speakers of the superstrate and substrate languages or to social contacts of a more transient nature, such as trading networks. Yimas Pidgin is an example of the latter cause. It is a pidgin derived from Yimas and Arafundi, two languages of the Sepik area in Papua New Guinea (all data from my own fieldwork). It is actually one of a family of Yimas based pidgins used by male Yimas villagers in their trading relations with neighboring groups. There are at least three others (Williams 2000), but Yimas is the superstrate, dominant lexifier language in all of them (see Foley 1991) for a grammar of Yimas). As in keeping with the general Sepik pattern (Gewertz 1983) the primary trading network is Arafundi sago, the carbohydrate staple, exchanged for Yimas fish. These trade contacts are not sporadic, but are sustained and passed down in clans from generation to generation. Trading partners as well as the proper pidgin language used in interactions with them was part of the clan based patrimony passed from father to son. A clear status differential marks these exchanges: the Yimas fish suppliers occupy a superior position socially and economically vis à vis the sago suppliers. Other goods besides fish and sago are exchanged between the two groups, but in all cases it is the Yimas which supply the prestige goods. Given these social circumstances, it is expected that the pidgin used will have Yimas as the superstrate language and Arafundi as the substrate, hence Yimas Pidgin. Yimas Pidgin is a relatively stable pidgin; native speakers of both Yimas and Arafundi speak it much the same, the major differences being phonological. It is not creolized, having no native speakers; indeed it is on the verge of dying out, its function having been usurped by Tok Pisin (New Guinea Pidgin English). Lexically, it is largely Yimas: over 80% of the lexicon is Yimas derived, the rest coming from Arafundi or from unplaced sources (plus a small mixture of Karawari forms, probably a transfer from the Yimas-Karawari Pidgin). Grammatically, Yimas Pidgin represents a massive simplification of Yimas, a highly morphologically complex polysynthetic language, with some admixture of simplified Arafundi structures. It is demonstrably not Arafundi relexified with Yimas content morphemes. It is, in turn, quite different from the Yimas pidgin used to trading partners speaking Alamblak (Williams 2000), demonstrating that it is indeed a true conventionalized pidgin, not just a foreigner talk register of Yimas. It is particularly noteworthy that still another Yimas lexified pidgin was used in trading contacts with Karawari speakers (Williams 2000). © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.8 (443-490) William Foley Yimas and Karawari are closely related languages (Yimas is not related genetically to either Arafundi or Alamblak, nor they to each other), almost dialects of the same language: Karawari speakers within earshot could often understand a good deal of what I was saying when engaged in casual conversations with Yimas villagers. The fact that a pidgin was employed in Yimas-Karawari trade encounters, not stable bilingualism or a koine (Siegel 1997) derived from the two, is powerful testimony to the salience of these clan based conventionalized codes, the pidgins, as indexes of the intervillage trading encounter. In this section I will focus on the radical simplification of the grammars of the contributing languages to the Yimas-Arafundi Pidgin (just Yimas Pidgin in the following), with a view to comparison with related processes in the formation of Hiri Motu and Tok Pisin. Yimas is a highly complex language morphologically, with extensive allomorphy and suppletion and a great plethora of bound forms, especially in the verbal system. Yimas Pidgin is derived from this by massive reduction, by eliminating all allomorphy and suppletion and almost all bound forms, and by drastically cutting most distinctions in any category. Let me consider some examples in greater detail. . Nominal inflection Yimas is a super gender language parallel to Bantu languages, with ten major noun classes and six minor ones. All nouns are marked for gender and as well take inflection for number: singular, dual and plural (often portmanteau with gender indication). Noun modifiers like adjectives and possessives agree in gender and number with the nouns they modify (however, forming a NP constituent with the noun as head is not necessary; Yimas is a nonconfigurational ‘free scrambling’ language like Warlpiri (Hale 1983)). Consider the following examples: (1) a. patn ama-na-kn betelnut.V.SG 1SG-POSS-V.SG ‘My betelnut is big.’ ama-na-ra b. par kat betelnut.V.PL 1SG-POSS-V.PL ‘My betelnuts are big.’ ama-nac. tr tooth.VI.SG 1SG-POSS-VI.SG ‘My tooth is big.’ © 2006. John Benjamins Publishing Company All rights reserved kpa-n anak big-V.SG COP.V.SG (Y) kpa-ra arak big-V.PL COP.V.PL (Y) kpaakk big-VI.SG COP.VI.SG (Y) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.9 (490-547) Universal constraints and local conditions in Pidginization d. tr ki ama-na- ki kpa- ki akiak tooth.VI.PL 1SG-POSS-VI.PL big-VI.PL COP.VI.PL ‘My teeth are big.’ e. tanm ama-na-m kpa-m apk bone.VII.SG 1SG-POSS-VII.SG big-VII.SG COP.VII.SG ‘My bone is big.’ (Y) (Y) Yimas Pidgin loses all gender distinctions and all inflection for number. The number of a noun can be marked by kundamwin ‘two’ and manba ‘many’, both from Arafundi, placed after it (reduction combined with increased transparency and regularity). The forms for class V singular (the unmarked class in Yimas with the largest membership) are extended to cover all nouns, as in (2) (2) a. SG: patn trG] ama-nakGn kGpan anak tan>m 1SG-POSS big COP ‘My betelnut/tooth/bone is big.’ b. PL: patn trG] manba ama-nakGn kGpan anak tanGm many 1SG-POSS big COP ‘My betelnuts/teeth/bones are big.’ (YP) (YP) Thus, the pervasive allomorphy of number marking conditioned by gender class for nouns and noun modifiers is completely eliminated, leading to a more transparent linking of meaning and form in Yimas Pidgin, i.e. manba (many) unambiguously always signals plural number in Yimas, regardless of noun modified. In some ways, the resulting pidgin structure reflects Arafundi grammatical patterns; Arafundi nouns lack gender and are not inflected for number. Number can be explicitly indicated by kGndamu\ ‘two’ or manba ‘many’, but this is optional. This much parallels Yimas Pidgin. However, Arafundi nouns are inflected for their possessor with a suffix, kot-k hand/foot-1SG.POSS ‘my hand/foot’, so the Yimas Pidgin structure in (2), in which the possessor is expressed with an independent word, is sourced in the parent Yimas structure of (1). . Expression of grammatical relations Being a polysynthetic language, Yimas encodes almost all grammatical information about a clause with the verb (Boas 1911; Baker 1996; Evans & Sasse © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.10 (547-615) William Foley Table 2. Yimas agreement affixes DL 1 PL SG DL 2 PL 1 SG SG 3 PL DL ↓ PRO kapa ipa ama kapwa ipwa mi A ]krakayka]krannannnmpump- ↓ S kapaipaamakapwaipwamanapuimpa- O ]krakra]a]kulkulnannapuimpa- D ]krakra]a]kulkulnan-nakn -mpun -mpn 2002). The verb can occur with many prefixes and suffixes, with the former predominating. The suffixes largely express notions of tense, aspect and mood; there are five potentially filled suffixal slots to the verb. The prefixes are more numerous, with eight potentially filled positions. The prefixes express notions like modality; adverbial notions like place, direction, duration and manner; and valence alternations to the verb like reciprocal formations, causativization and applicatives. Prefixes to the verb also indicate core grammatical relations. Core NPs lack all case marking (there is a single oblique case marker -n ∼ -nan for all non-core NPs) and word order at the clause level is remarkably free (in any case the great bulk of clauses consist of just a verb anyway). Core grammatical relations are signaled in a very complex way by prefixes which indicate gender class, number and person. These exhibit a very complex person-based split for their case marking schema: first and second person forms follow an underlying nominative/accusative pattern and third person forms, an ergative/absolutive one (see Dixon (1994) or Tallerman (2005) for a discussion of nominative/accusative versus ergative/absolutive alignments and the use of A, S, and O as notation for core grammatical relations; D represents the recipient argument of a ditransitive verb like ‘give’). The relevant prefixes are set out below: Examples (3)–(8) illustrate the use of these affixes: (3) ama-wa-t pu-ka-tay pu-]a-tay 1SG.S-go-PERF 3PL.O-1SG.A-see 3PL.A-1SG.O-see ‘I went’ ‘I saw them’ ‘They saw me’ © 2006. John Benjamins Publishing Company All rights reserved (Y) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.11 (615-674) Universal constraints and local conditions in Pidginization (4) pu-wa-t pu-n-tay na-mpu-tay 3PL.S-go-PERF 3PL.O-3SG.A-see 3SG.O-3PL.A-see ‘They went’ ‘He saw them’ ‘They saw him’ (5) ta-ka-wa-t ta-ka-tay-c-um NEG-1SG.A-go-PERF NEG-1SG.A-see-PERF-3PL.O ‘I didn’t go’ ‘I didn’t see them’ ta-]a-tay-c-um NEG-1SG.O-see-PERF-3PL.A ‘They didn’t see me’ (6) ta-pu-wa-r-um ta-pu-n-tay-c-um NEG-3-go-PERF-3PL NEG-3-3SG.A-see-PERF-3PL.O ‘They didn’t go’ ‘He didn’t see them’ ta-mpu-tay-c-ak NEG-3PL.A-see-PERF-3SG.O ‘They didn’t see him’ (Y) (Y) (Y) (7) ura] k-mpu-]a-tkam-t coconut.VI.SG VI.SG.O-3PL.A-1SG.D-show-PERF ‘They showed me the coconut.’ (Y) (8) ura] k-mpu-tkam-r-mpun coconut.VI.SG VI.SG.O-3PL.A-show-PERF-3PL.D ‘They showed them the coconut.’ (Y) Arafundi is less complex morphologically, but still exhibits extensive crossreferencing for core grammatical relations. Arafundi is a straightforward nominative/accusative language in cross-referring. Subjects are obligatorily marked by suffixes to the verb but these are portmanteau, representing basic tense/aspect/mood distinctions. There are three basic sets of subject suffixes, illustrated in ow-k wash-1SG.PERF ‘I washed’ versus ow-] wash-1SG.IRR ‘let me wash’ versus ow-p-cGk wash-PROG-1SG.IMPERF ‘I am washing’. Objects are optionally also indicated by suffixes immediately after the verb root: pokonanci-na-cGk hit-2SG.O-FUT-1SG.A.IMPERF ‘I will be hitting you.’ Pronominal objects may be marked with an accusative case marker instead of being cross-referenced: poko-nci-wam hit-1SG.O-3PL.A.PERF versus c-i poko-wam 1SG-ACC hit-3PL.A.PERF ‘they hit me’. There is also an optional case marker for nouns. This occurs most frequently on subjects of transitive verbs, so it looks like an ergative case marker, but its basic function is to mark the subject in cases of potential ambiguity (although I will gloss it ERGative for convenience). Because word order among NPs in a clause is free, this could potentially occur © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.12 (674-737) William Foley whenever the subject or object are both animate; hence the disambiguating function of this marker: (9) a. nu]gum-Gn nam poko-e man-ERG woman hit-3SG.PERF ‘The man hit the woman.’ b. nam-Gn karGma yGp-e woman-ERG cassowary see-3SG.PERF ‘The woman saw the cassowary’. (Ar) (Ar) Interestingly, this case marker can even be used on subjects when they cooccur with pronominal objects, which, being accusatively case marked, leave no ambiguities as to the assignment of grammatical relations: (10) nu]gum-Gn c-i poko-e man-ERG 1SG-ACC hit-3SG.PERF ‘The man hit me.’ (Ar) Finally, this ergative suffix is typically used with ditransitive verbs, to mark the subject, as the dative NP is not case marked, although it is cross referenced in the verb (the verb ‘give’ has a zero stem allomorph for 3PL recipients): (Ar) nam-Gn nu]gum mGnkis Ø-makG-na-tam woman-ERG man sago give-3PL.O-FUT-3PL.A.IMPERF ‘The women will give the men sago.’ (Ar) b. nu]gum-Gn nam mGnkis Ø-makG-na-tam man-ERG woman sago give-3PL.O-FUT-3PL.A.IMPERF ‘The men will give the women sago.’ (11) a. All cross-referencing is eliminated in Yimas Pidgin. And contrary to most pidgin languages, word order among NPs is still free. NPs are generally not case marked either: (12) a. ama patn awt-nan 1SG betelnut get-NFUT ‘I got betelnut.’ b. tupwi mGn am-bi ta-nan sago 3SG eat-DEP PROG-NFUT ‘He’s eating sago.’ (YP) (YP) In cases of potential ambiguity of grammatical relations, i.e. more than one animate NP, an optional case marker is employed, but in this case in a manner opposite to that of the Arafundi system. Thus, in the ditransitive clause it is © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.13 (737-795) Universal constraints and local conditions in Pidginization the dative NP which is marked with a postposition namban, derived ultimately from a Yimas allative postposition nampan ‘toward’: (13) a. mGn tupwi ama namban asa-nan 3SG sago 1SG DAT give-NFUT ‘He gave me sago.’ b. ama mGn manba namban mariawkGn anak 1SG 3SG many DAT talk FUT ‘I will tell them’ (YP) (YP) This postposition is also optionally extended to transitive clauses to mark animate objects (à la Spanish a) to disambiguate the subject and object grammatical relations, given the free word order: (14) a. ama mGn namban 1SG 3SG DAT ‘I hit him.’ b. mGn ama namban 3SG 1SG DAT ‘He saw me.’ kratGkG-nan hit-NFUT (YP) tay-\an see-NFUT (YP) The use of namban is a move toward greater transparency and regularity in addition to reduction; instead of the myriad of allomorphs of affixes of variable position in the verb to indicate objects and datives, there is just a single form. The use of a postposition to mark dative NPs and animate object NPs has no precedent in either source language (although the pattern, a conflation of dative and object, is exemplified by the use of the same prefix forms for first and second person dative and object arguments in Yimas; see Table 2). However, this system is widespread in the more isolating languages of the Middle and Lower Sepik, like Yessan-Mayo, Watam and Iatmul, and the last of these is itself a language of high prestige, which formed the basis of an earlier trade pidgin in the Sepik. Earlier generations of Yimas speakers were fluent in this pidgin, and it is quite possible that it provided a model for this system of indicating grammatical relations in Yimas Pidgin. In fact, one cannot completely rule out the possibility that the Yimas based pidgins are themselves relexifications of the earlier, but unfortunately unattested and now extinct, Iatmul Pidgin. But the fact that there are very significant structural differences between the Yimas-Arafundi Pidgin discussed here and the Yimas-Alamblak Pidgin (Williams 2000) suggests otherwise: that they are both local developments. If they were simply relexifications of an earlier Iatmul Pidgin, we would © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.14 (795-827) William Foley expect them to be more alike. The very feature under discussion here illustrates this point. While both pidgins extend an ostensive dative indicator to mark animate objects, they do so in very different ways, with quite distinct sources in the parent Yimas superstrate language: the Yimas-Arafundi Pidgin, as we have seen, adapts the Yimas allative postposition nampan ‘toward’ to this function; the Yimas-Alamblak Pidgin has a more complex system, but essentially uses a bound suffix, -nakn, derived either from the third person singular verb agreement suffix for dative, -nakn (see Table 2) or a possessive agreement suffix for class V singular nouns (see example (1a)), also -na-kn (Williams 2000). The crucial point here is that Yimas Pidgin does not employ word order or hierarchical configurational phrase structure to signal grammatical relations (note invariable word order in Table 1). Put simply, word order or configurational phrase structure was just unavailable for this function, as neither source language used it to signal grammatical relations. At the clause level, both languages are non-configurational with free order of NPs, Yimas being much more extreme in this area than Arafundi, as Yimas generally lacks even coherent NP structure (on this, see Foley (1991: 180–191)). . Tense-aspect marking Being a highly wrought polysynthetic language, Yimas mainly lacks independent adverbial words for direction, place, time, or manner, these being mostly expressed through verbal affixes. Tense is highly complex with many distinctions (remote past, far past, near past, immediate past, present, habitual, near future, remote future), but is reduced in Yimas Pidgin to future versus nonfuture (past and present): (15) a. ama mGn namban kratGkG-nan 1SG 3SG DAT hit-NFUT ‘I hit/am hitting him.’ b. ama mGn namban kratGkGn anak 1SG 3SG DAT hit FUT ‘I will hit him.’ (YP) (YP) The non-future is derived from the Yimas near past tense suffix -nan, for events occurring yesterday, while the future comes from the unmarked form of the copula, which is used in Yimas for timeless unbounded statements. Thus, the source semantics is bounded versus unbounded events (Yimas -nan NEAR PAST is the most temporally bounded of its tense suffixes, denoting events © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.15 (827-889) Universal constraints and local conditions in Pidginization occurring yesterday; the others have a wider range), which developed into non-future (present/past) versus future. Aspect in Yimas is also complex and indicated in a number of different ways. Some aspectual inflections occupy the final position of the tense suffixes: (16) amtra ya-ka-am-t food.V.PL V.PL.O-1SG.A-eat-PERF ‘I have eaten the food.’ (Y) While other aspectual inflections are indicated by incorporated adverbial prefixes: (17) pu-yakal-taw-ant-ntut 3PL.S-CONT-sit-hear-REM.PAST ‘They were waiting and listening.’ (Y) Aspect in Arafundi is also indicated in diverse ways, either as a portmanteau suffix with the subject’s person and number: (18) poko-Ø-e hit-3SG.O-3SG.A.PERF ‘He hit him.’ (AR) Or as an independent suffix: (19) \G] yGm ow-pG-cGk 1SG water wash-PROG-1SG.IMPERF ‘I am bathing.’ (AR) Yimas Pidgin, in keeping with many pidgins around the world, including the other two New Guinea based pidgins discussed below, expresses aspect much more transparently and regularly, via serial verb like constructions. The Yimas verb mntk- ‘finish’ provides the perfective marker mGndGk, while the Yimas Pidgin marker for imperfective aspect, ta-, is multiply sourced as either the Yimas verbs taw- ‘sit’, tay- ‘see’ or ta- ‘put’ or the Arafundi verb te- ‘get’. These aspectual verbs are used in serial verb constructions following the main predicating verb, which is tenseless and suffixed with the dependent verb marker -mbi derived from Yimas -mpi of the same function. The aspectual verb carries the tense ending for the clause: (20) a. ama mi namban andG-mbi mGndGk-nan 1SG 2SG DAT hear-DEP PERF-NFUT ‘I heard you already.’ © 2006. John Benjamins Publishing Company All rights reserved (YP) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.16 (889-949) William Foley b. ama mi namban andG-mbi ta-nan 1SG 2SG DAT hear-DEP IMPERF-NFUT ‘I’m listening to you.’ (YP) . Clause linkage This use of -mbi to mark tenseless dependent verbs also figures prominently in Yimas interclausal connections. Clauses in Yimas Pidgin are typically loosely joined in a sentence using this dependent suffix; there is no switch reference marking, and clauses linked through the dependent suffix may have shared or disjoint subjects: (21) ya]i kandGk kura-mbi am-bi tandaw-nan pot LOC put-DEP eat-DEP sit-NFUT ‘(they) put (it) in a pot and ate it sitting down.’ (YP) This pattern is also found in Yimas, but is complemented by an extensive system of clausal linkage through nominalization, both finite and non-finite, as exemplified in this bit of text: (22) (Y) ma-y mnta i-mp-awl-k]A pot.VIII.SG other-VIII.SG then VII.SG.O-3PL.A-get-IRR mnta warapak-mpi-ca-k-mp-n B [kpa-y big-VIII.SG then flay-SEQ-put-IRR-VII.SG-OBL ya]i-\an wul-k-mp-n pot.VIII.SG-OBL put down-IRR-VII.SG-OBL wut-mpi na-mpu-t-krapak-mpi-]a-k]B put down-SEQ 3SG.O-3PL.A-RCP-divide-SEQ-give-IRR [kalk na-mpu-wul-k]C C V.SG.O-3PL.A-put down-IRR sago pudding.V.SG ‘They got another pot, a big one, and then having flayed and put (her) in the pot, having put (her) down there, they took their share, and they made sago pudding.’ A [ya]i Arafundi in keeping with many central highlands languages of New Guinea employs clause chaining with switch reference, a construction exemplified by the following: (23) a. \G] nam yepe-a ambuk 1SG woman see-SR go.1SG.PERF ‘I saw the woman and left.’ © 2006. John Benjamins Publishing Company All rights reserved (AR) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.17 (949-1021) Universal constraints and local conditions in Pidginization b. \G] nam yepe-tum-u] a\ 1SG woman see-DR-1SG go.1SG.PERF ‘I saw the woman and (she) left.’ (AR) The suffix -a indicates the subjects of the two clauses linked together are the same, while -tum indicates they are different. In addition -u] in (23b) expresses that the subject of its clause is first person singular. If it were second singular, the proper suffix would be -man. Nominalizations, finite and non-finite, and switch reference are totally lacking in Yimas Pidgin, which exclusively employs clause linkage through simple juxtaposition or marking dependence with -mbi. . Summary The processes of language simplification we have demonstrated in the genesis of Yimas Pidgin can be summarized according to the rubrics of Siegel (2000) and Kusters (2003): A. reduction (Siegel 2000) or the Economy Principle (Kusters 2003): loss of gender and number distinctions of nouns; loss of verbal agreement affixes for core grammatical relations; loss of a complex person-based split for the alignment of grammatical relations; collapse of a eight tense system to a binary distinction. B. increased regularity (Siegel 2000) or the Isomorphy Principle (Kusters 2003): number of all nouns indicated by either kundamwin ‘two’ and manba ‘many’; single invariant forms of adjectives and possessives; animate objects of all persons signaled by namban; aspect indicated by serial verbs, as opposed to a complex array of prefixes and suffixes. C. greater transparency (Siegel 2000) or the Transparency Principle (Kusters 2003): number of nouns indicated by separate words, rather than bound suffixes portmanteau with gender; grammatical function signaled overtly by separate word namban on animate objects and recipients, situations in which potential ambiguity could obtain; aspect signaled by serial verbs in consistent postverbal position, rather than a mix of prefixes and suffixes. D. lack of markedness (Siegel 2000): clause linkage by unelaborated clause chaining, rather than nominalizations, either finite or non-finite, or switch reference systems. © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.18 (1021-1046) William Foley . Hiri Motu Hiri Motu has a better documented, if somewhat more complex, history than Yimas Pidgin. The superstrate language is Motu, spoken around Port Moresby in Papua New Guinea, but determining the substrate language(s) is rather tricky, with probably multiple sources. The Motu are a coastal Austronesian people who migrated to the central Papua area from further east, as its closest linguistic relatives largely lie to the east. When the Motu arrived on the central Papuan coast they came into contact with the Papuan language speaking Koita, who inhabited the area. Motu shows some innovations vis a vis the related Austronesian languages further east which probably reflect Koita influence. The Motu entered into an elaborate trading relationship with the Koita, who provided game and bush products for Motu pots and sea products; on the whole the Motu and Koita were closely integrated socially and economically. As a coastal, sailing people, the Motu also traded further afield. The most spectacular of their extensive trading networks involved the hiri trading voyages to the Gulf of Papua, a dangerous journey of some 400 km. There live the Eleman speaking peoples and the Koriki, with whom the Motu traded pots and received lumber and sago in return. In the trading contacts of the Gulf of Papua, pidginized version of the Eleman languages or Koriki were used by the Motu to communicate with their trading partners (Dutton 1983). What the Motus used in their contacts with the Koita is not known, but with the arrival of Europeans in the Port Moresby area in 1874, clear reports emerged of a simplified form of Motu used in talking to foreigners. It is most implausible that this grew up with the arrival of Europeans, and it is probably the continuation of an indigenous tradition, a contact language for the Koita. As more immigrants arrived in the Port Moresby area, their highly diverse linguistic backgrounds of European languages, Chinese, Malay and other Oceanic Austronesian languages insured that simplified Motu would have been the only language in common, and it quickly assumed the role of a contact language between them, ultimately gelling into what we know today as Hiri Motu. Thus, the history of Hiri Motu differs from that of Yimas Pidgin in that, while Hiri Motu has a single superstrate language, it has multiple substrate languages, prominent among them Koita, Eleman languages, other Oceanic Austronesian languages, and English. The lexicon reflects this; while most of its basic vocabulary is derived from Motu, the rest is composed of words that come from Polynesian languages, other Austronesian languages of Papua and pidginized forms of English. There are also at least some words © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.19 (1046-1105) Universal constraints and local conditions in Pidginization from the Gulf of Papua languages, e.g. nakimi ‘brother-in-law’ from Koriki (Dutton 1985). . Nominal inflection Grammatically, Hiri Motu is a simplified version of Motu, reflecting its source in some variety of Simplified Motu, but Motu itself, while an Austronesian language, already shows some divergence from other Austronesian languages of Papua in the direction of Koita, as we shall see below. Nouns in Motu are generally not inflected for gender or number, a given noun functioning vaguely as singular or plural according to context. (This is true of Koita and the Eleman languages, as it is of Arafundi and many other Papuan languages). Adjectives, however, do mark number for their modified nouns when definite (Dutton 1985). (M) (24) a. dabua kurokuro-na cloth white-SG ‘the white cloth’ b. dabua kurokuro-dia cloth white-PL ‘the white clothes. (M) As with Yimas Pidgin agreement between modifier and head noun disappears in Hiri Motu, the only vestiges being agreement for number with namo ‘good’ and kida ‘bad’: (Dutton 1985): (HM) (25) dala maoromaoro ta path straight one ‘a straight path’ Motu, like Oceanic Austronesian languages generally, has a type of gender distinction in possessive NPs. These languages distinguish between inalienable and alienable possession and, within alienable possession, between things which are edible or otherwise. Inalienable nouns are marked for possession simply by suffixing the possessor suffix (indicating person and number) to the noun, as in the typical Austronesian Oceanic pattern (Pawley 1976): (26) a. natu-dia child-3PL.POSS ‘their child/children’ b. ima-gu hand-1SG.POSS ‘my hand’ © 2006. John Benjamins Publishing Company All rights reserved (M) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.20 (1105-1150) William Foley Alienable possession is again marked in the typical Austronesian Oceanic pattern, by suffixing the possessor suffix to a classifier preceding the possessed noun (Pawley 1976): (27) a. a-gu boroma edible-1SG.POSS pig ‘my pig/pork for eating’ b. e-gu boroma property-1SG.POSS pig ‘my pig (as property)’ (M) (M) The substratum Papuan languages are more variable in this domain. Eleman languages, represented here by Toaripi, mark the possessor with a possessive suffix/particle attached to the basic pronoun forms, which precedes the possessed noun; there is no classification of nouns for possession (Brown 1973): (28) a. ara-ve mai 1SG-POSS hand ‘my hand’ b. Mirou ve marehari Mirou POSS younger brother ‘Mirou’s younger brother’ (T) (T) Koita is more complicated. It has a number of gender-type clauses into which possessed nouns fall, leading to different morphological forms for possessive constructions according to the head noun (Dutton 1975): (29) Normal form class 1: omoto class 3: ihiko class 4: deγa class 5: uri class 8: kuhi a. possessed -te class omote -ke class ihikoke -ye class deγaye -de class uride -ve class kuhive form (K) ‘head’ ‘ear’ ‘buttocks’ ‘nose’ ‘sore’ di ihikoke 1SG.POSS ear ‘my ear’ © 2006. John Benjamins Publishing Company All rights reserved (K) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.21 (1150-1218) Universal constraints and local conditions in Pidginization b. yau vai>aue 3PL.POSS spear ‘their spear’ (K) Hiri Motu possessive constructions are radically simplified from Motu in the direction of the Eleman languages, but the actual constructions involved are quite different, so the Eleman languages could not have been a model for them. Rather they reflect the parent Motu system neutralized, so that the unmarked Motu ‘property’ possessive class is extended to cover all possessives regardless of the possessed noun (Dutton 1985): (30) a. lau-egu 1SG-1SG.POSS ‘my father’ b. lau-egu 1SG-1SG.POSS ‘my house’ c. lau-egu 1SG-1SG.POSS ‘my food’ tamana father (HM) ruma house (HM) aniani food (HM) . Expression of grammatical relations The indication of grammatical relations in Motu is a mix of general Austronesian Oceanic patterns and probable imports from Koita. Most Austronesian Oceanic languages signal subject and object via pronominal agreement affixes to the verb with somewhat variable word order for subject and object NPs across the languages: SVO (Tolai), VOS (Fijian, Kilivila) or SOV (Motu). As the great bulk of Austronesian languages are VO in grammatical structure, the switch to OV in Motu and other languages of mainland New Guinea probably reflects diffusion from adjoining Papuan languages, which typically do have this constituent order. Both Koita and the Eleman languages are OV languages. Motu fairly closely preserves the subject and object verb agreement affixes inherited from the proto-language of the Austronesian Oceanic languages. Subjects are marked by a set of prefixes and object by suffixes (third person object suffixes are restricted to animate NPs), as set out in Table 3 (Lister-Turner & Clark 1930): © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.22 (1218-1270) William Foley Table 3. Motu agreement affixes 1 2 3 1 INCL 1 EXCL 2 3 SG PL PRO SUBJ OBJ lau oi ia ita ai imui idia naoetaaoe- -gu -mu -(i)a -da -mai -mui -dia These are illustrated in the following examples: (31) a. na-gwadai-a 1SG.SUBJ-pierce-3SG.OBJ ‘I pierced him’ b. e-ubu-gu 3SG.SUBJ-feed-1SG.OBJ ‘She fed me.’ (M) (M) With a true ditransitive verb like heni- ‘give’ the object suffix actually agrees with the dative NP: (32) buka e-heni-gu book 3SG.SUBJ-give-1SG.OBJ ‘He gave me a book.’ (M) The substrate languages, Koita and Toaripi, also have verb agreement to signal grammatical relations, but along very different lines. Verb agreement works on a ergative-absolutive system, so that the verb agrees in number but not person with absolutive NPs, either the object of a transitive verb or the subject of an intransitive one: Koita (Dutton 1975): (33) a. da-ka mu 1SG-SUBJ PERF ‘I stood’ b. yau-ka mu 3PL-SUBJ PERF ‘They stood’ c. da-ka mu 1SG-SUBJ PERF ‘I saw you (SG)’ ra-ma-nu stand-SG-PAST (K) ra-ha-nu stand-PL-PAST (K) a era->a-nu 2SG see-SG-PAST (K) © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.23 (1270-1331) Universal constraints and local conditions in Pidginization d. da-ka mu 1SG-SUBJ PERF ‘I saw you (PL)’ e. *da-ka mu 1SG-SUBJ PERF ya era-geve-nu 2PL see-PL-PAST (K) ya era->a-nu 2PL see-SG-PAST (K) Toaripi (Brown 1973): (34) a. are itoi b. ere itoteai 3SG stand.SG 3PL stand.PL ‘He stands’ ‘They stand’ c. ere ere-ve etau oroti voa vuopōpe 3PL 3PL-POSS thing canoe in put.SG ‘They put their thing in the canoe’ d. ere ere-ve etauroro oroti voa ōpōpe 3PL 3PL-POSS things canoe in put.PL ‘They put their things in the canoe’ e. *are are-ve etauroro oroti voa vuopōpe 3SG 3SG-POSS things canoe in put.SG (T) (T) (T) (T) The (e) examples in both the Koita and Toaripi are ungrammatical because the verb is agreeing in number with the singular transitive subject (ergative), rather than the plural transitive object (absolutive). With ditransitive verbs like ‘give’, the dative NP controls the object verb agreement (Dutton 1975; Brown 1973): (35) a. 1. da mo-ima 2. no mo-higeue 1SG give-SG 1PL give-PL ‘Give it to me!’ ‘Give it to us!’ b. ava mea fere la a-ro miarai 3SG this betelnut OBJ 2SG-OBJ give.SG ‘He gave this betelnut to you.’ (K) (T) Hiri Motu follows Yimas Pidgin in losing almost all verbal agreement. The pronominal forms in the language are derived directly from the Motu independent pronouns (ita- ‘see’ has the allomorph itai- when it co-occurs with the transitive suffix -a, as in 36b, c): (36) a. ia noho 3SG stay ‘He’s here.’ b. lau itaia oi 1SG see 2SG ‘I saw you’ © 2006. John Benjamins Publishing Company All rights reserved (HM) (HM) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.24 (1331-1380) William Foley c. (HM) boroma lau itaia pig 1SG see ‘I saw a pig.’ The only vestiges of the Motu agreement system in Hiri Motu is a suffix -a derived from Motu -a 3SG.OBJ, which is a fossilized marker of transitive verbs (see example (36b)) and -dia from Motu -dia 3PL.OBJ which marks plural objects and replaces -a on transitive verbs (Dutton & Brown 1975): (37) boroma lau ita-dia pig 1SG see-3PL.OBJ ‘I saw pigs’ (Compare (36c)) This agreement for number of the object of a transitive verb could reflect influence from Koita or Eleman languages, which also do this by following their absolutive NP agreement pattern. The above examples demonstrate some variation in Hiri Motu word order (SVO, SOV, OSV, etc.), so clearly word order is not a reliable guide to grammatical relations in Hiri Motu (see Dutton (1997) for discussion on parameters for this variation). Nor indeed is pronominal verb agreement, given its total lack in the language. Rather the language makes use of a few mutually exclusive case marking and topic and focus marking particles to keep the grammatical relations straight. These are derived from Motu (and the Motu system in turn is probably due to diffusion from Koita, as Koita like all languages of the Koiarian family of languages makes heavy use of such particles (Olson 1981); these are largely lacking in Austronesian Oceanic languages other than Motu). To understand what is happening it is best to start with Motu, which has three basic particles: ese, na, and be. Ese is the most straight forward of these; it marks the subject of transitive verbs, especially in cases of possible ambiguity. It is, thus, an ergative case: (38) a. hahine ese natu-na e-ubu-dia woman ERG child-3SG.POSS 3SG.SUBJ-feed-3PL.OBJ ‘The woman fed her children’ b. sisia ese boroma e-kori-a dog ERG pig 3SG.SUBJ-bite-3SG.OBJ ‘The dog bit the pig.’ (M) (M) na is a type of topicalization marker. It is used when objects are topicalized, preceding the subject in word order: © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.25 (1380-1443) Universal constraints and local conditions in Pidginization (39) boroma na sisia ese e-kori-a pig TOP dog ERG 3SG.SUBJ-bit-3SG.OBJ ‘The pig, the dog bit.’ (M) Or na can mark the dative NP with a ditransitive verb: (40) tau ese mero na buka e-heni-a man ERG boy TOP book 3SG.SUBJ-give-3SG.OBJ ‘The man gave the boy a book.’ (M) And na can even mark subjects of intransitive verbs if they are topicalized or there is some doubt as to their role: (41) tau na vada e-la man TOP PERF 3SG.SUBJ-go ‘The man has gone.’ (M) na also marks subordinate clauses, in the manner of topicalizing particles in many Papuan languages (Haiman 1978): (M) (42) ia bain-e-ma (a)i na bai-na-hamaoro-a 3SG FUT-3SG.SUBJ come time TOP FUT-1SG.SUBJ-tell-3SG.OBJ ‘When he comes, I’ll tell him.’ be marks focussed, new information NPs, in contrast to the topicalized, presupposed NPs marked by na (Dutton 1985) (43) a. lada-mu be name-2SG.POSS FOC ‘What’s your name?’ b. lada-gu na name-1SG.POSS TOP ‘My name’s Asi.’ daika? who (M) Asi Asi (M) This Motu system appears to be a truncated version of the much richer related system of Koita topic and focus particles. Dutton (1975) provides a list of over a dozen of these particles, and it is much too elaborate to go into any detail here. Suffice to say there is a set including -ka, which appears to be restricted to marking subjects (see examples (33) above), and a topicalizer -na ∼ -ra, common in equational sentences and overtly topicalized NPs in verbal clauses, as in: (44) a. ata vire-ra haroro koita-vara man that-TOP preach person-SPEC ‘That man’s a pastor.’ © 2006. John Benjamins Publishing Company All rights reserved (K) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.26 (1443-1498) William Foley b. au-na oregenu? 3SG-TOP where ‘Where is it?’ c. a-na oregene a oti-Ø-ma 2SG-TOP where 2SG go-SG-PRES ‘Where are you going?’ (K) (K) and a focus marker -ki: (45) a. da-ka-ki roku i-Ø-nu 1SG-SUBJ-FOC papaya eat-SG-PAST ‘(It was) I (who) ate the papaya.’ b. da-ki era-@a-nu, a-ka veitera 1SG-FOC see-SG-PAST, 2SG-SUBJ NEG ‘I saw it, you didn’t.’ (K) (K) Strikingly, -ki also marks dative NPs with verbs other than mo- ‘give’: (46) Madi-ra-ki Kila-ki ro-Ø-nu Madi-SPEC-FOC Kila-FOC tell-SG-PAST ‘(It was) Madi (who) told Kila (something).’ (K) Toaripi lacks such a syntactized use of focus/topic particles (although, interestingly, the Gulf of Papua trade language used by the Motu and based on the Eleman languages does have them (Dutton 1983), probably as a result of Motu providing the syntactic model for them in the pidginization process). Rather like Yimas Pidgin, Toaripi uses a case marker to mark dative NPs and object NPs, especially animate or otherwise syntactically, ambiguous ones; the particle is la (a suffix -ro occurs with pronouns) (Brown 1973): (47) a. ara mea fere la a-ro miarai 1SG this betelnut OBJ 2SG-OBJ give.SG ‘I gave you these betelnuts.’ b. ara-ro ere sa elori 1SG-OBJ 3PL cut open ‘They cut me open.’ c. mea o a-ro leisa moita this word 2SG-OBJ who said ‘Who said this to you?’ d. lea marisa teve uta la sape this girls not hole OBJ mend.PROG ‘These girls were mending holes in the net.’ © 2006. John Benjamins Publishing Company All rights reserved (T) (T) (T) (T) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.27 (1498-1557) Universal constraints and local conditions in Pidginization But: e. Meavea uvi turōpe Meavea house built ‘Meave built a house.’ (T) Note that the object marker is required on dative NPs (a, c) and animate object NPs (b), but appears to be optional for inanimate objects (d, e), in this case probably reflecting discoursal factors. Note that Toaripi word order is fairly flexible: while usually SOV (a, d, e), an OSV order is possible (b, c), necessitating the marking of at least one of the animate NPs functioning as grammatical relations. Toaripi marks the object in contrast to Motu and Koita, which both mark the subject, Motu on an ergative basis and Koita on an accusative one. Hiri Motu follows Motu and Koita in using particles to mark grammatical relations, but instead of three Motu particles (ese, na and be), it only uses two, so that be expands to take over the functions of na (Dutton 1985). ese is used, like in Motu, as a disambiguating ergative case marker: (48) tau ese hahine ia botaia man ERG woman 3SG hit.TR ‘The man hits the woman.’ (HM) Note that these Hiri Motu data would lead us to query the value of McWhorter’s (2001) observation that creole languages do not exhibit ergativity. No known creole language may, but certainly at least one pidgin language does, and would lead us to expect that there is no impediment to this feature in creoles. The lack is just a contingent fact; after all, Hiri Motu could easily be acquired natively as a creole. The particle be marks both topicalized and focussed NPs (Dutton 1985): (49) a. oiemu ladana be daika your name who ‘What’s your name?’ b. lauegu ladana be Asi my name Asi ‘My name is Asi.’ c. inai sisia be lauegu boroma ia lulua this dog my pig 3SG chase.TR ‘This dog chased my pig.’ d. gabu ia daroa negana(i) be mero maraki ia tai place 3SG sweep time boy small 3SG cry ‘While she was sweeping the place, the small boy cried.’ © 2006. John Benjamins Publishing Company All rights reserved (HM) (HM) (HM) (HM) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.28 (1557-1608) William Foley be does not, however, extend in Hiri Motu to yet another use of na in Motu, the marking of dative NPs with ditransitive verbs. In this case the ubiquitous postposition dekanai, derived from Motu dekana(i) ‘beside/towards’ is used: (50) oi henia lau dekanai 2SG give.TR 1SG OBL ‘You give (it) to me.’ (HM) Note that dative use of Hiri Motu dekanai ‘toward’ parallels that of Yimas nampan ‘toward’, but in Hiri Motu it does not extend to marking animate object NPs. . Tense-aspect marking Motu has two basic tenses: future versus non-future (present or past). This is signaled by a prefix in b- preceding the subject pronominal prefix versus Ø: bai-na-gini FUT-1SG.SUBJ-stand ‘I’ll stand’ versus na-gini 1SG.SUBJ-stand ‘I stand’ or ‘I stood’. (This could perhaps be better analyzed as a realis-irrealis distinction, in line with many other Austronesian languages of New Guinea). Koita and Toaripi are both much more complex, with many tense distinctions, including multiple distinctions within past and future. Hiri Motu follows Motu with a basic binary future/non-future distinction, but the means to indicate it are rather different. In keeping with the general morphological simplification in Hiri Motu, non-future is not marked, but future is expressed with an adverbial dohore (sometimes shortened to do), derived from Motu dohore ‘by and by’ (this usage may have been influenced by the Tok Pisin future marker baimbai). So: (51) a. lau itaia 1SG see.TR ‘I see/saw it.’ b. dohore/do lau itaia FUT 1SG see.TR ‘I’ll see it.’ (HM) (HM) Aspect in Motu is expressed in various ways. Imperfective aspect is indicated by suffixes to the verb, -va for present continuous and -mu for past continuous: (52) na-itaia-mu/-va 1SG.SUBJ-see.TR-PRES.IMPERF/-PAST.IMPERF ‘I am/was seeing it.’ © 2006. John Benjamins Publishing Company All rights reserved (M) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.29 (1608-1664) Universal constraints and local conditions in Pidginization Perfective aspect is marked with a adverb vada ‘completed’ preceding the verb: (53) vada na-itai-a PERF 1SG.SUBJ-see-3SG.OBJ ‘I have seen it.’ (M) In Koita, aspect is expressed with a set of particles immediately following the subject constituent: (54) a. au-ki gure ni-vi-ma 3SG-FOC IMPERF cry-SG-PRES ‘(It is) he (who) keeps on crying.’ b. da-ka mu i-Ø-nu 1SG-SUBJ PERF eat-SG-PAST ‘I’ve already eaten it.’ (K) (K) In Toaripi, aspect is again indicated in various ways: the imperfective aspect is expressed by compounding a verb with the verb root -pea ‘be, dwell’, while perfective is marked with the preverbal adverbial evera ‘finished, already’, much like Motu. Hiri Motu is most like Toaripi in its expression of aspect and not too dissimilar from Yimas Pidgin. The Motu imperfective aspect suffixes are lost and this is indicated with a serial verb like construction involving noho ‘dwell, stay’: (55) lau itaia noho 1SG see.TR IMPERF ‘I am/was looking at it.’ (HM) Perfective is indicated by the adverbial vadaeni (derived like true Motu vada from vadaeni ‘enough’), also placed after the verb: (56) lau itaia vadaeni 1SG see.TR PERF ‘I have seen it.’ (HM) The source model for the Hiri Motu aspectual system could have been the Toaripi system, but that is not likely, given the general lack of Toaripi contributions to Hiri Motu. Another possibility is Tok Pisin, as its aspectual system, involving i stap ‘stay/IMPERF’ and pinis ‘finish/PERF’, is closely parallel, as we shall see in the next section. But the Hiri Motu aspectual system is also similar to that of Yimas Pidgin, and this suggests general forces are at work here. © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.30 (1664-1719) William Foley . Clause linkage Finally, clause linkage in Hiri Motu is largely similar to Motu, with straightforward conjunction and embedding, although it favors a more paratactic style. It is somewhat simplified structurally and uses fewer conjunctions, but is otherwise not hugely different from Motu. The following bit of Hiri Motu text illustrates (Dutton 1985): (57) vada ai be patrol loaloa edeseni Japani idia noho (HM) then 1PL.EXCL patrol roam where Japanese 3PL dwell edeseni noho where dwell ‘We used to go on patrol to where the Japanese were staying.’ lau kakana ruosi, ai ruosi lau inei dala ta, dala 1SG elder brother two 1PL.EXCL two go this path one path ta ia mai Sananda dekenai ia lau Doboduru one 3SG come Sananda OBL 3SG go Doboduru ‘My elder brother and I we both went along this path; a path it runs from Sananda, it goes to Doboduru.’ Note the simple juxtaposition of clauses in this last sentence, representing a preference against embedded structures. Motu in this case would have prenominal relative clauses (Lister-Turner & Clark 1930): (58) [biku e-ani ore] boroma-dia na vada e-heau banana 3SG.SUBJ-eat finish pig-PL TOP PERF 3SG.SUBJ-run (M) boio away ‘The pigs that ate up the bananas have run away.’ Such a convoluted structure would be strongly disfavored in Hiri Motu, the weak paratactic juxtaposition of clauses exemplified above is typical. In point of contrast, the clause linkage systems of both Koita and Toaripi are very different from Motu and Hiri Motu. They make extensive use of clause chaining (Foley 1986) with dependent verbs, and, in the case of Koita, switch reference morphology as well (Brown 1973; Dutton 1975): (59) uamori ere-ve larietau ovi pat-ei elavo voa women 3PL-POSS food up bring-DEP men’s house inside mi-ōri, ere larietau kokoruk-a av-i au lo-pe give-DEP 3PL food gather-DEP sit-DEP eat do-PAST © 2006. John Benjamins Publishing Company All rights reserved (T) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.31 (1719-1773) Universal constraints and local conditions in Pidginization ‘The woman brought the food up and gave (it) (to them) inside the men’s house; they (the men) gathered the food, sat down and ate it up.’ (60) a. da-ka mu oro-γi eva-γa-nu 1SG-SUBJ PERF come-SR see-SG-PAST ‘I came and saw him.’ b. au-ki da ihirom-ege da-ki oti-Ø-nu 3SG-FOC 1SG call-DR 1SG-FOC go-SG-PAST ‘(It was) he(who) called me and I (who) went.’ (K) (K) In the (a) example the subjects of the two clauses are the same, hence -γi SR, but in (b) they are different (-ege DR). Note how similar, particularly the Toaripi example, is to clause linkage in Yimas Pidgin, and the difference of both of these styles to the clause linkage employed in Hiri Motu, which is more Oceanic Austronesian in type. The difference in clause linkage style between these two pidgins, Yimas Pidgin and Hiri Motu, clearly reflects the language mix in the contact situations giving rise to them: only OV, clause chaining languages in the case of Yimas Pidgin, but a wider typological mix for Hiri Motu, consisting of VO, paratactically conjoining Austronesian languages and English, but also OV clause chaining Papuan languages. The fact that Hiri Motu is more like the former type reflects either its general unmarkedness in pidgin genesis (to be overriden in certain situations like that of Yimas Pidgin, where all contributing languages share a particular discourse style) or the greater influence of Oceanic Austronesian speakers and their language type in the specific genesis of Hiri Motu. . Summary The processes of language simplification we have demonstrated in the genesis of Hiri Motu can also be summarized according to the rubrics of Siegel (2000) and Kusters (2003): A. reduction (Siegel 2000) or the Economy Principle (Kusters 2003): loss of adjective agreement; loss of possessive gender class distinctions for nouns; loss of verbal agreement affixes for subject and object ; contrast between be and na neutralized, with be taking over functions of both; loss of a tense contrast in imperfective aspect. B. increased regularity (Siegel 2000) or the Isomorphy Principle (Kusters 2003): single invariant forms of adjectives and possessives; no suppletive verb forms for object number agreement; use of only one particle be for © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.32 (1773-1826) William Foley subjects of intransitive verbs, instead of the contrast between na TOP and be FOC; extension of the oblique postposition dekanai to indicate dative NPs of ditransitive verbs, making these parallel to other locative roles; future tense marking invariable, instead of a prefix with a number of persondetermined allomorphs; imperfective aspect also invariable, rather than portmanteau with tense; aspect marking always postverbal. C. greater transparency (Siegel 2000) or the Transparency Principle (Kusters 2003): grammatical relations signaled by overt particles, the ergative ese, the oblique dekanai, and be, rather than a mix of particles and bound verbal affixes; future tense indicated by a preverbal independent word, rather than a bound prefix; aspect signaled by independent forms in consistent postverbal position. D. lack of markedness (Siegel 2000): clause linkage by simple parataxis, rather than embedding, which is common in Motu, or clause chaining, as in Koita or Toaripi. . Tok Pisin Tok Pisin is by far the best known of the pidgin languages spoken in Papua New Guinea. It is an English derived pidgin, so that English is its superstrate language (with a small contribution from German), but its substrate languages are numerous Oceanic Austronesian languages of the Pacific islands, especially Polynesian languages like Samoan and the Austronesian languages of the New Ireland/Rabaul area, notably Tolai, and the Australian Pidgin English of colonial eastern Australia (see Mühlhausler (1983), Keesing (1988) and Baker (1993) for social histories of the genesis of Tok Pisin). Because Tok Pisin is better known than the other two pidgins discussed and because its superstrate language is English (over 80% of the basic vocabulary is derived from English), I will be somewhat briefer in my consideration of this language. Also, because the morphological structure of both its superstrate language and substrate languages are comparatively simple, unlike for Yimas Pidgin and Hiri Motu, I will not be focusing on the specific issue of simplification in the genesis of Tok Pisin, but rather the likely origins of its grammatical structures or their apparent lack in its source languages; the similarity or difference of these structures to comparable domains in Yimas Pidgin and Hiri Motu will be discussed in Section 5. Though the superstrate language for Tok Pisin is English, its substrate languages are highly varied. I will focus on the Oceanic substrate of Melanesian © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.33 (1826-1876) Universal constraints and local conditions in Pidginization languages, particularly those from which most of the plantation laborers were drawn: New Britain, New Ireland, Solomon Islands and Vanuatu. While there is good evidence for the pivotal role of Australian Pidgin English in the genesis of Tok Pisin (Baker 1993), it is still generally accepted that Oceanic languages are major substrate contributors (Siegel 1999). Sorting out the relative contribution of superstrate and substrate languages to the syntactic patterns of Tok Pisin is a relatively tricky task for this pidgin due to the similarity of syntactic structure between English and these Oceanic languages, although, as we shall see, when these diverge, Tok Pisin typically exhibits Oceanic-like structures, not English ones. . Nominal inflection Tok Pisin has no gender distinctions for nouns, but most current dialects do mark number for animate nouns, by procliticizing the third person plural pronoun ol ‘they’ to the noun: ol man ‘the men’, ol pato ‘the ducks’. This is clearly different from English, which pluralizes with an inflectional suffix /-6z/, but does closely parallel usage in many Oceanic languages: Raga (Vanuatu) ira vavine 3PL woman ‘the women’, Lau (Solomons) genera ngwane 3PL man ‘the men’ (Keesing 1988). Possession is indicated in Tok Pisin by bilong from English belong in the order possessed bilong possessor: dok bilong pikinini: dog of child ‘the child’s dog’. This is the opposite of the usual order in English possessive NPs, nor does it reflect the typical Oceanic order (Fijian nu-qu vale POSS-1SG house ‘my house’, Mota (Vanuatu) no-k o matip POSS-1SG.DET coconut ‘my coconut’, so this must be an independent development during the pidginization process. Baker (1993) plausibly demonstrates its source in Australian Pidgin English). Note, too, that the common Oceanic gender-type distinctions among possessed nouns (see Section 3.1) is neutralized. . Expression of grammatical relations Tok Pisin is a canonical SVO language, as is both its superstrate English, and very many of its substrate Oceanic languages: (61) a. Tok Pisin pikinini i paitim dok asde child AGR hit dog yesterday ‘The child hit the dog yesterday.’ © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.34 (1876-1928) William Foley b. Tolai (New Britain) (Capell 1969) a tutuna i gire ra pap DET man 3SG.SUBJ see DET dog ‘The man is seeing the dog.’ c. Kwaio (Solomon Islands) (Keesing 1988) ma’a a-gu ka aga-si-a boo naaboni father POSS-1SG 3SG.SUBJ see-TR-3SG.OBJ pig yesterday ‘My father saw a pig yesterday.’ d. Sie (Vanuatu) (Lynch & Capell 1983) etm-en yi-tai lau father-3SG.POSS 3SG.SUBJ-make canoe ‘His father made a canoe.’ Other Oceanic languages may have a verb initial order, but derive a SVO order via topicalization of the subject, e.g. Fijian. But this similarity of Tok Pisin syntax to that of English is more apparent than real, for when the language is more deeply scrutinized, it is clear that the actual encoding of grammatical relations in the language is radically different from that of English and these differences, point for point, resemble patterns in the Oceanic languages. First, Tok Pisin, unlike English, exhibits overt morphological encoding for a distinction between transitive and intransitive verbs. Almost all transitive verbs are marked by an overt transitivizing suffix: (62) VINTR boil bruk pas pulap slek VTR boil-im bruk-im pas-im pulap-im slek-im ‘boil’ ‘break’ ‘close’ ‘full, fill’ ‘loose, loosen’ (TP) This pattern is widespread among Oceanic languages (see also discussion of Motu in Section 3.2): (63) a. Tolai (Mosel 1980) aliv aliv-e ‘float’ momo mom-e ‘drink’ b. Kwaio (Keesing 1988) aga aga-si ‘see’ gumu gumu-ri ‘pound’ fana fana-si ‘shoot’ © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.35 (1928-1982) Universal constraints and local conditions in Pidginization Intransitive verbs which denote activities are also morphologically marked in Tok Pisin, by reduplication: (64) VINTR lukluk waswas pekpek pispis puspus VTR luk-im ‘see’ was-im ‘wash’ ‘defecate’ ‘urinate’ ‘copulate’ (TP) This pattern of reduplicating intransitive activity verbs is also widespread in Oceanic languages, but totally lacking in English, here illustrated by Tolai (Mosel 1980): (65) kukul kul ‘buy’ tutumu tumu ‘write’ pekapeke peke ‘excrete’ (To) Secondly, Tok Pisin has a preverbal clitic particle i used with person singular and non-singular subjects, which is essentially a subject agreement form. This may be derived from English he (although there are also clear Oceanic precursors with the form i (see Keesing 1988)), but the overall syntactic structure is completely alien to English: (66) a. mipela i go nau 1PL AGR go now ‘We’re going now.’ b. ol man i pait-im ol dok PL man AGR hit-TR PL dog ‘The men are hitting the dogs.’ (TP) (TP) Essentially such a system is found in nearly every Oceanic language. Regardless of their basic constituent ordering of NPs, Oceanic languages exhibit SVO order in their agreement morphemes for the grammatical relations of subject and object. Subjects are cross-referenced by an agreement morpheme proclitic to the verb and objects by an enclitic. Kwaio (Keesing 1988) is absolutely prototypical: (67) ma’a a-gu ka aga-si-a boo naaboni (Kw) father POSS-1SG 3SG.SUBJ see-TR-3SG.OBJ pig yesterday ‘My father saw a pig yesterday.’ © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.36 (1982-2044) William Foley As these clitics function as anaphoric agreement markers (Bresnan & Mchombo 1987), full NPs realizing the subject and object grammatical relations are freely dispensable (Oceanic languages are ‘pro-drop’ languages): (68) a. ka aga-si-a naaboni 3SG.SUBJ see-TR-3SG.OBJ yesterday ‘He saw it yesterday.’ b. ka aga-si-’adauru 3SG.SUBJ see-TR-1PL.INCL.OBJ ‘He saw us.’ (Kw) (Kw) It seems plausible to claim that while both the subject agreement predicate i and the transitive suffix -im may have parent English morphemes, he and him respectively, the system within which they function is wholly derived from Oceanic patterns, not English ones. That Tok Pisin has often been viewed as clearly reflecting English source structures is simply due to the superficial similarity of its SVO word order to that of English. A deeper look reveals that this SVO structure, while parallel to English, is probably not due to English, but to substrate Oceanic languages which are also SVO, for the remainder of the system for encoding grammatical relations in Tok Pisin is wholly non-English, but fairly clearly is an analog of Oceanic patterns. . Tense-aspect marking Like Hiri Motu, but unlike Yimas Pidgin, Tok Pisin does not have obligatory inflection for tense. However, tense notions can be expressed by a set of preverbal particles (this is also how Oceanic languages do it): (69) Compare Tolai (Mosel 1984): © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.37 (2044-2099) Universal constraints and local conditions in Pidginization (70) ‘The man saw/will see the dog.’ Interestingly, there are traces of the future/non-future system found earlier in Hiri Motu and Yimas Pidgin, for a sentence unmarked for any time specification will typically be freely interpretable as present or past: (71) man i pait-im dok man AGR hit-TR dog ‘The man is hitting/hit the dog.’ (TP) While a future reading requires an overt tense marker or other future time specifier like tumora ‘tomorrow’: (72) man i laik pait-im dok man AGR FUT hit-TR dog ‘The man is gonna hit the dog.’ (TP) Aspect in Tok Pisin is expressed in ways exactly parallel to Yimas Pidgin, by using serial verb constructions involving stap ‘stay, live’ for imperfective and pinis ‘finish’ for perfective. (73) a. ol man i wok-im haus i stap PL man AGR make-TR house AGR IMPERF ‘The men are building the house.’ b. ol man i wok-im haus pinis PL man AGR make-TR house PERF ‘The men have built the house.’ (TP) (TP) Interestingly, while this structure is clearly not derived from English, it also is not all that common among Oceanic languages. What is most common in Oceanic languages is that the formations of the two aspects are not parallel. In Tolai (Mosel 1984), for instance, imperfective aspect is commonly signaled by reduplication of the verb (this is the most widespread way to signal this semantic domain among Oceanic languages): (74) u ia-ian kau-gu vudu 2SG RED-eat POSS-1SG.POSS banana ‘You are eating my bananas.’ © 2006. John Benjamins Publishing Company All rights reserved (To) JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.38 (2099-2179) William Foley While perfective aspect is indicated by a preverbal perfective particle parallel to the tense markers in (70) above (Mosel 1984): (75) iau na tar ian kau-m vudu 1SG FUT PERF eat POSS-2SG.POSS banana ‘I will have eaten your bananas.’ (To) There are, however, some Oceanic languages, although an apparent minority, which do form this aspectual contrast in parallel in a Tok Pisin-like way. Northeast Ambae of Vanuatu is one such language (Hyslop 2001), employing eno ‘lie’ for the imperfective and rovo ‘finish’ for the perfective: (76) a. gu wehi-eu mo eno 2SG.TEL hit-1SG.O REAL IMPERF ‘You were beating me.’ b. ra-ru mo laqa mo rovo 3NSG-DL REAL speak REAL PERF ‘When they had spoken/finished speaking.’ (NEA) . Conclusion: Local and universal constraints on pidginization In the title of this paper I posed the question of the relative role of universal constraints versus local conditions on the outputs of the pidginization process. I stated my intention to investigate this through a comparative study of three pidgins of New Guinea, particularly non-European derived ones. Having looked at what has happened in the formation of Yimas Pidgin, Hiri Motu and Tok Pisin, it appears that both may be relevant, and it is now certainly clear in these New Guinea case studies that the latter is a very important factor. It is true that all three languages show similar universal processes of language simplification: reduction and increased regularity in the loss of inflection and the elimination of allomorphy, i.e. loss of noun class in Yimas Pidgin, loss of possessed noun gender in Hiri Motu and Tok Pisin, loss of number marking in Yimas Pidgin and Hiri Motu, loss of agreement or cross-referring to signal core grammatical relations in Yimas Pidgin and Hiri Motu. But the grammatical systems resulting from such losses are quite different in each of the three languages, and these differences reflect the local conditions, i.e., the structures of the source languages, for the genesis of each pidgin. Thus, the source languages for Yimas Pidgin and Hiri Motu are non-configurational languages with free constituent © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.39 (2179-2196) Universal constraints and local conditions in Pidginization order for NPs functioning as core grammatical relations. The source languages for Tok Pisin, on the other hand, are not and typically have a more fixed VO order (usually SVO or more rarely VSO or VOS). Consequently, Yimas Pidgin and Hiri Motu have relatively free order of core NPs, but Tok Pisin is a fixed SVO language. Motu and Koita both use a complex system of case marking and topicalization particles to signal core grammatical functions; Hiri Motu does the same, unusually for a pidgin employing an ergative case marking system like its superstrate Motu, but this is also exhibited in the verbal agreement patterns of its substrates. Yimas Pidgin, on the other hand, chooses to mark animate direct objects and dative NPs and to leave other core NPs unmarked. This reflects a widespread case marking schema of other languages of the Middle Sepik; but interestingly, it is not a feature of Yimas, which completely lacks case marking and signals core grammatical functions via verb agreement, nor of the substrate language Arafundi, which does have case marking, but of an ergative type, marking the subject of a transitive verb, rather than the object, as Yimas Pidgin does. But the mark-the-object rule is around in the local linguistic area; it is in the air (or water!), so to speak. Many of the morphologically simpler, more isolating languages in the region indicate core grammatical relations in this way, particularly the languages spoken by the high status cultures of the Middle Sepik, such as Iatmul; consequently it was a readily available option to solve the problem of signaling core grammatical functions in Yimas Pidgin. The fact that the Yimas-Alamblak Pidgin does not signal grammatical relations in this way does argue against a hypothesis that this solution was sourced in a relexification of an Iatmul-based pidgin. The high status of this cultural group would lead us to expect a Iatmul Pidgin to serve as a model for all Yimas based pidgins if this were true. This divergence between the two varieties indicates that both were local solutions to the problem of negotiating a language of exchange through linguistic simplification with speakers of different kinds of linguistic backgrounds and knowledge. Finally, Tok Pisin, like its source Oceanic languages and English, leaves the core grammatical relations of subject and object unmarked, letting the fixed SVO order perform the signaling function. However, unlike Yimas Pidgin and Hiri Motu, and rather like its local source Oceanic languages, it also uses verbal morphology to encode grammatical relations, namely the proclitic i for subject agreement and the transitive suffix -im for object. All of these developments across the three pidgins indicate the importance of local conditions for the genesis of pidgins. Another example concerns clause linkage patterns across the three pidgins, and this probably reflects constraints of greater transparency and lack of © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.40 (2196-2244) William Foley markedness. Hiri Motu and Tok Pisin have a preference for weakly integrated, paratactic structures, relying on coordination or simple juxtaposition. In this, they reflect the favored patterns of their substrate Oceanic Austronesian languages, but also perhaps the typical method of pidgins around the world, and, in any case, this is the unmarked setting for interclausal connections. But Yimas Pidgin employs the more typologically unusual clause chaining structure in clause linkage, a choice no doubt due to the fact that all languages in the multilingual mix that gave rise to it were Papuan languages, for which this schema is normative (although Yimas itself does still have other devices, like nominalization, which may be at least as frequently used in discourse). This difference among the three pidgins again demonstrates the important role of local conditions in pidgin genesis. But this is not the whole story. For when we look at the systems of tense and aspect in the three pidgins, they are remarkably similar in spite of enormous differences in this area in the various sets of source languages. For example, all three pidgins have a neutral future/non-future tense system, with non-future the unmarked option, in spite of the significant formal differences between the three pidgins: Yimas Pidgin Hiri Motu Tok Pisin Non-FUT -nan Ø Ø FUT -n anak dohore bai The fact that the end result of the process of pidginization in the three languages is so similar, regardless of the very different local conditions (e.g. Yimas has eight tense distinctions; Motu has basically two; English essentially has two, but past versus non-past; and other Oceanic languages are highly variable in their degree of elaboration in this domain) suggests that universal constraints are operative here. Furthermore, in the specific case of Yimas Pidgin, none of the contributing languages to it or its sisters, the Yimas-Alamblak or Yimas-Karawari pidgins, have a basic binary tense system like this; it has no pre-existing models and could only have arisen during the pidginization process itself. A similar point can be made for aspect. Again, all three languages show remarkable similarities: serial verb-like constructions involving a verb like ‘sit, stay’ for imperfective and ‘finish’ for perfective This is especially striking for Tok Pisin because such a feature is not pronounced in most of its Oceanic source languages. The same point can also be made for Yimas Pidgin, as models for this aspectual system are lacking in all its source languages, although © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.41 (2244-2332) Universal constraints and local conditions in Pidginization again they are found in some other languages of the Sepik region. The similar way of marking aspect in all three languages in spite of their very different local conditions strongly suggests universal constraints are at work here. What is the source of these constraints in the domain of tense-aspect-mood is not clear. I believe it has something to do with the high level of abstractness of these semantic categories; there is not much in the perceivable world that corresponds to perfect aspect or irrealis mood, unlike actors and objects. In second language acquisition, the difficulties in acquiring control over categories like the imperfect or the subjunctive as opposed to nominative or accusative are well known. But pinning this intuition down more precisely and indeed ascertaining whether there is any truth to it will require a great deal more work. I end my speculations here. The process of pidginization is an interesting and important one in linguistic change. It reveals in a fascinating way the creative skills humans use to arrive at linguistic accommodations with linguistic others. Comparative studies of pidginization results may ultimately reveal the delicate interplay of local linguistic conditions with more robust universal linguistic constraints, and in doing so, provide an arena in which linguists can usefully study the role of both. Abbreviations A ACC AGR Ar CONT COP D DAT DEP DET DL DR ERG EXCL FOC FUT HAB subject of a transitive verb accusative case agreement marker Arafundi continuative copula dative/recipient of a ditransitive verb dative case dependent determiner dual number different referent ergative case exclusive focus future habitual HM IMPERF INCL IRR K Kw LOC M NEA NEG NFUT NSG O OBJ OBL PAST PERF PL © 2006. John Benjamins Publishing Company All rights reserved Hiri Motu imperfective inclusive irrealis Koita Kwaio locative case Motu Northeast Ambae negative non-future non-singular object of a transitive verb object oblique past perfective plural JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.42 (2332-2413) William Foley POSS PRES PRO PROG RCP REAL RED REM.PAST S SEQ SG SPEC SR possessive present pronoun progressive reciprocal realis reduplication remote past subject of an intransitive verb sequential singular specific same referent SUBJ T To TOP TP TR V–VIII Y YP 1 2 3 subject Toaripi Tolai topic Tok Pisin transitive Yimas gender classes Yimas Yimas Pidgin first person second person third person References Baker, Mark (1996). The polysynthesis parameter. New York: Oxford University Press. Baker, Phillip (1993). Australian influence on Melanesian Pidgin English. Te Reo, 36, 3–67. Bakker, Peter (1997). “A language of our own”. The genesis of Michif – The mixed Cree-French language of the Canadian Métis. New York: Oxford University Press. Boas, Franz (Ed.) (1911). Handbook of American Indian languages. Washington: Bureau of American Ethnology Bulletin 40.1. Bresnan, Joan, & Sam Mchombo (1987). Topic, pronoun and grammatical agreement in Chichewa. Language, 63, 741–82. Brown, Harold (1973). The Eleman languages. Pacific Linguistics, C26, 279–376. Capell, Arthur (1969). A survey of New Guinea languages. Sydney: Sydney University Press. Dixon, Robert M. W. (1994). Ergativity. Cambridge: Cambridge University Press. Dutton, Tom (1975). A Koita grammar sketch and vocabulary. Pacific Linguistics, C29, 281– 412. Dutton, Tom (1983). Birds of a feather: A pair of rare pidgins from the Gulf of Papua. In Ellen Woolford & William Washabaugh (Eds.), The social context of creolization (pp. 77– 105). Ann Arbor: Karoma Press. Dutton, Tom (1985). Police Motu: iena sivarai (its story). Port Moresby: University of Papua New Guinea Press. Dutton, Tom (1997). Hiri Motu. In Sarah G. Thomason (Ed.), Contact languages: A wider perspective (pp. 9–41). Amsterdam: Benjamins. Dutton, Tom, & Harold Brown (1975). Hiri Motu: The language itself. Pacific Linguistics, C40, 759–793. Evans, Nick, & Hans-Jürgen Sasse (Eds.) (2002). Problems of polysynthesis. Berlin: Akademie Verlag. Ferguson, Charles (1982). Simplified registers and linguistic theory. In Loraine Obler & Lise Wise (Eds.), Exceptional language and linguistics (pp. 49–66). New York: Academic Press. © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.43 (2413-2520) Universal constraints and local conditions in Pidginization Foley, William A. (1986). The Papuan languages of New Guinea. Cambridge: Cambridge University Press. Foley, William A. (1991). The Yimas language of New Guinea. Palo Alto: Stanford University Press. Gal, Susan (1979). Language shift: Social determinants of linguistic change in bilingual Austria. New York: Academic Press. Gewertz, Deborah (1983). Sepik river societies. New Haven: Yale University Press. Gumperz, John, & Robert Wilson (1971). Convergence and creolization: A case from the Indo-Aryan-Dravidian border. In Dell Hymes (Ed.), Pidginization and creolization of languages (pp. 151–169). Cambridge: Cambridge University Press. Haiman, John (1978). Conditionals are topics. Language, 54, 564–589. Hale, Kenneth (1983). Warlpiri and the grammar of non-configurational languages. Natural Language and Linguistic Theory, 1, 5–47. Hyslop, Catriona (2001). The Lolovoli dialect of the North-east Ambae language, Vanuatu. Pacific Linguistics, 515. Keesing, Roger (1988). Melanesian Pidgin and the Oceanic substrate. Palo Alto: Stanford University Press. Kulick, Donald (1992). Language shift and cultural reproduction: socialization, self and syncretism in a Papua New Guinea village. Cambridge: Cambridge University Press. Kusters, Wouter (2003). Linguistic complexity: The influence of social change on verbal inflection. Leiden: University of Leiden Center for Linguistics. Lefebvre, Claire (1998). Creole genesis and the acquisition of grammar; The case of Haitian Creole. Cambridge: Cambridge University Press. Lister-Turner, R., & J. B. Clark (1930). Revised Motu grammar and vocabulary. Port Moresby: Government Printer. Lynch, John, & Arthur Capell (1983). Sie grammar outline. Pacific Linguistics, C79, 75–147. McWhorter, John (2001). The world’s simplest grammars are creole grammars. Linguistic Typology, 5, 125–166. Mosel, Ulrike (1980). Tolai and Tok Pisin: The influence of the substratum on the development of New Guinea Pidgin. Pacific Linguistics, B73. Mosel, Ulrike (1984). Tolai syntax and its historical development. Pacific Linguistics, B92. Mühlhäusler, Peter (1979). Growth and structure of the lexicon in New Guinea Pidgin. Pacific Linguistics, C52. Mühlhäusler, Peter (1983). Samoan Plantation Pidgin English and the origin of New Guinea Pidgin. In Ellen Woolford & William Washabaugh (Eds.), The social context of creolization (pp. 28–76). Ann Arbor: Karoma Press. Muysken, Peter (1988). Media Lengua and linguistic theory. Canadian Journal of Linguistics, 33, 409–422. Muysken, Peter (1997). Media Lengua. In Sarah G. Thomason (Ed.), Contact Languages: A wider perspective (pp. 365–426). Amsterdam: Benjamins. Olson, Michael (1981). Barai clause junctures. Canberra: Australian National University dissertation. Pawley, Andrew (1976). Austronesian languages: Western part of southeastern mainland Papua. Pacific Linguistics, C39, 301–319. © 2006. John Benjamins Publishing Company All rights reserved JB[v.20020404] Prn:10/04/2006; 15:45 F: JPC21101.tex / p.44 (2520-2577) William Foley Siegel, Jeff (1985). Koines and koineization. Language in Society, 14, 357–378. Siegel, Jeff (1993). Controversies in the study of koines and koineization. International Journal of the Sociology of Language, 99, 1–8. Siegel, Jeff (1997). Mixing, leveling and pidgin/creole development. In Arthur K. Spears & Donald Winford (Eds.), The structure and status of pidgins and creoles (pp. 112–149). Amsterdam: Benjamins. Siegel, Jeff (1999). Transfer constraints and substrate influence in Melanesian Pidgin. Journal of Pidgin and Creole Languages, 14, 1–44. Siegel, Jeff (2000). Introduction: The processes and language contact. In Jeff Siegel (Ed.), Processes in language contact: Studies from Australia and the South Pacific (pp. 1–11). Sant-Laure Quebec: Fides. Silverstein, Michael (1972). Chinook Jargon: Language contact and the problem of multilevel generative systems. Language, 48, 378–406, 596–625. Spears, Arthur K., & Donald Winford (1997). The structure and status of pidgins and creoles. Amsterdam: Benjamins. Tallerman, Maggie (2005). Understanding syntax. London: Arnold. Thomason, Sarah G. (Ed.) (1997). Contact languages: A wider perspective. Amsterdam: John Benjamins. Thomason, Sarah G. (2001). Language contact: An introduction. Edinburgh: Edinburgh University Press. Williams, Jeffrey (2000). Yimas-Alamblak tanim tok: An indigenous trade pidgin of New Guinea. Journal of Pidgin and Creole Languages, 15, 37–62. Woolford, Ellen, & William Washabaugh (1983). The social context of creolization. Ann Arbor: Karoma Press. © 2006. John Benjamins Publishing Company All rights reserved