Descriptive and discourse-referential modifiers in a layered model of the noun phrase.
This article argues that adnominal modifiers in a layered model of the noun phrase can be divided into two major subcategories: descriptive modifiers and discourse-referential modifiers. Whereas descriptive modifiers can be subdivided into classifying, qualifying, quantifying and localizing modifiers (Section 2), discourse-referential modifiers in the noun phrase are concerned with the status of entities as referents in the world of discourse (Section 3). I will pay particular attention to three issues: (i) formal reflections of the layered, semantic structure of the noun phrase (Section 4), (ii) the special relationship between localizing and discourse-referential modifiers (Section 5), and (iii) semantic and morphosyntactic parallels between modifier categories in the noun phrase and the clause (Section 6).
In addition this sample-based typological study shows (contra Hawkins's Universal 20') that there are languages with the adjective before and the demonstrative or numeral after the head noun. These word order patterns provide additional support for the layered model of the noun phrase defended here in that it can now be shown for the first time that all patterns that iconically reflect the layered structure of the simple noun phrase are actually attested in the languages of the world.
1. Introduction (1)
Noun phrases (henceforth NPs) serve a dual purpose: on the one hand they provide a physical description of the referent and on the other hand they are constructions that speakers use to refer to an entity. (2) This double function of NPs is reflected, among other things, in the two major kinds of modifiers we find inside the noun phrase: descriptive modifiers and discourse-referential modifiers (called "discourse modifiers" in Rijkhoff [2002a: 229]; cf. also Hengeveld 2004b: 372-373). (3)
Descriptive noun modifiers, which are discussed in Section 2, specify properties of the referent of the NP in terms of the categories KIND (or CLASS), QUALITY, QUANTITY and LOCATION. By contrast, discourse-referential modifiers (discussed in Section 3) pertain to the referential status of entities (objects, events) in conversational space. The linear organization of descriptive and discourse-referential modifiers is investigated in more detail in Section 4, which focuses on the degree to which NP-internal ordering patterns reflect the organization of modifiers in the layered model of the noun phrase (ICONICITY). Section 5 is concerned with the special relationship between discourse-referential modifiers and certain descriptive modifiers (LOCALIZING MODIFIERS), whereas Section 6 explores parallels between the layered structure of NPs and the layered structure of clauses.
Notice, finally, that certain parameters (appositional vs. non-appositional modifiers, embedded vs. nonembedded modifiers, free vs. bound modifiers, etc.) were fixed so as to be able to make a valid cross-linguistic comparison. Hence I only considered noncomplex NPs in which the demonstrative, numeral and adjective are free constituents of the integral (whole) NP. Note furthermore that I will restrict myself mainly to simple NPs which are headed by an underived noun and which are used to refer to a concrete object (cf. Rijkhoff 2002a: 19-27).
2. Descriptive modifiers in the noun phrase
Layered representations of clausal structures, which are designed to reflect scopal differences amongst modifier categories, have been proposed in various theories of grammar (for an overview, see Butler 2003). In Dik's Functional Grammar (FG), layering was introduced by Hengeveld (1989, 1990), whose proposals were immediately incorporated into the general FG framework (Dik 1989, 1997). Subsequently Rijkhoff (1988, 1990, 1992, 2002a) put forward a layered model of the NP, arguing that, to some extent, NPs and clauses can be analyzed in a similar fashion (see Figure 4 below). Recently Dik's Functional Grammar has been succeeded by Functional Discourse Grammar (Hengeveld and Mackenzie 2008), which is also the framework within which the current proposal is situated (see Note 3 and the introduction of the current special issue).
In the layered NP model DESCRIPTIVE MODIFIERS are distributed over four nested layers (notice that each language uses only a subset of the modifier categories mentioned below):
--the "kind" layer has scope over the head noun and accommodates modifiers (CLASSIFYING MODIFIERS) that only relate to the property that is designated by the noun, further specifying what kind of entity is being referred to by the speaker;
--the quality layer has scope over the layer with classifying modifiers and accommodates modifiers (QUALIFYING MODIFIERS) specifying properties concerning size, color, weight, value, age etc.;
--the quantity layer has scope over the quality layer and accommodates modifiers (QUANTIFYING MODIFIERS) having to do with number distinctions (singular, plural) and cardinality (one, two, etc.);
--the location layer has scope over the quantity layer and accommodates modifiers (LOCALIZING MODIFIERS) specifying properties concerning the location of the referent, such as demonstratives.
There is no one-to-one relationship between form and function of modifiers. For example, demonstratives can serve as localizing or discourse-referential operators (Section 5) and possessive modifiers can be employed as classifying, qualifying, or localizing satellites (Rijkhoff forthcoming); vice versa, a qualifying satellite can take the form of an adjective, a possessive NP or a relative clause (Rijkhoff 2008a).
The layered organization of descriptive noun modifiers is shown in Figure 1, where grammatical modifier categories or OPERATORS (such as Definiteness and Number) are represented to the left of the head noun and phrasal modifiers or SATELLITES, which typically involve members of lexical word classes (Verb, Noun, Adjective or Adverb), appear on the right side. The distinction between operators and satellites is relevant for several reasons (Dik 1997: 63-67, 159-161; Rijkhoff 2002a: 332-334); for example, certain diachronic developments only involve operators (see Rijkhoff 2008b). The relevance of the operator-satellite distinction is also discussed in Section 6, where it is shown that in some languages discourse-referential operators in the NP and in the clause are expressed by the same element.
Thus, in the NP those three black sniffer dogs in the corner the classifying satellite sniffer specifies the special kind of dog that is being referred to here. The qualifying satellite black relates to the combination sniffer dog, not the quantity or the location of the entity. The quantifying operator three specifies the number of black sniffer dog entities, not the number of locations; and both those (localizing operator) and in the corner (localizing satellite) specify the location of the sniffer dog entities with all their qualitative and quantitative properties.
[FIGURE 1 OMITTED]
2.1. Classifying modifiers in the noun phrase
Classifying modifiers specify what KIND of entity is being denoted by the head noun. Some common examples of classifying modifiers in English are the adjectives urban in "urban hero," presidential in "presidential election," or electric in "electric train." These modifiers differ from qualifying modifiers (Section 2.2) in that they do not specify some more or less objective property of an entity (OLD car) or the speaker's subjective attitude towards the entity (UNCOMFORTABLE chair), but rather a particular subclass of the category denoted by the head noun.
2.1.1. Classifying operators. Classifying operators are members of grammatical rather than phrasal or lexical modifier categories that (further) specify what KIND Of entity is being referred to. Oromo, fox example, has certain nominal affixes (dubbed "nominal aspect markers" in Rijkhoff [1992, 2002a]) that indicate whether the speaker refers to a collective set of entities or a singleton set (a set with only one member). They are typically used with so-called set nouns (Rijkhoff 2002a: 38-41, 100-121), which occur in many languages and which are transnumeral in that the unmarked form can be used to refer to any number of individuals (one or more than one). Hence the unmarked form also appears when the noun is modified by a numeral (on the difference between number markers and nominal aspect markers, see Rijkhoff 2002b: 219-220, 2002a: 101-103).
Oromo: set noun (1) gaala lamaani camel(s) two 'two camels' (Stroomer 1987: 59)
An affix indicates what KIND of set is being referred to: a collective set (containing more than one member) or a singleton set (with only one member). (4)
Oromo Unmarked transnumeral form Singulative form (2) nama nam-ica 'man/men' (N-sing.sx) 'a/the man' unmarked transnumeral form collective form (3) farda fard-oollee 'horse/horses' (N-coll.sx) 'horses' (Stroomer 1987: 76-77, 84-85)
Sometimes (erstwhile) numeral classifiers also serve as classifying operators, as in Hmong Njua. Example (4) illustrates the transnumeral character of Hmong Njua nouns, (5a) shows that the classifier indicates that the speaker refers to a singleton set (notice that the classifier also marks definiteness). Example (5b) demonstrates that the group classifier coy now marks collectivity (see Ratliff 1991, but notice that I follow Harriehausen's original glossing).
Hmong Njua (4) Kuv yuav tsev ISG buy house 'I buy a house / (some) houses' (5) a. Kuv yuav lub tsev 1SG buy CLF house 'I buy the house' b. Kuv yuav cov tsev 1SG buy PL house 'I buy (the) houses' (Harriehausen 1990: 117)
2.1.2. Classifying satellites. Classifying satellites are lexical or phrasal modifiers that further specify what KIND of entity (thing, event) is being referred to by the speaker. In English classifying satellites may take the form of an adjective (6a), but also of a possessive construction (6b)-(6c):
(6) a. a CORPORATE lawyer
b. a WOMAN'S hat (5)
c. a house OF WORSHIP
Classifying modifiers seem to be restricted in various ways as compared to qualifying modifiers (Section 2.2). For example, a classifying adjective does not admit an intensifier (7), comparison (8), or predicative position (9): (6)
(7) a. an ELECTRIC train
b. * a very ELECTRIC train
(8) a. a MEDICAL examination
b. * a more MEDICAL examination
(9) a. the PRESIDENTIAL election
b. * the election is PRESIDENTIAL
It can be difficult, however, to draw the line between a noun + classifying satellite and compounds or quasi-compounds (on the "hybrid status" of these modifiers, see e.g., Giegerich 2005).
2.2. Qualifying modifiers in the noun phrase
Qualifying satellites are phrasal or lexical noun modifiers that specify more or less inherent properties (qualities or attributes) of the referent of the noun phrase, such as size (a BIG house, telescopes OF ENORMOUS SIZE), value or quality (a CHEAP suit, wine OF AN INCREDIBLE RICHNESS), age (a YOUNG child, youths UNDER AGE 16), or color (BLUE curtains, a Jovian moon OF INCREDIBLE REDNESS). These examples illustrate, once again, that the same function (here: qualifying satellite) can be performed by members of different form classes, such as adjectives or possessives. Languages without adjectives, such as Galela, often employ relative clauses (headed by a stative verb) as qualifying satellites (see, e.g., Bhat 1994 on parts of speech systems: see also Hengeveld et al. 2004; Rijkhoff 2003). In Galela the first syllable of the verbal modifier is reduplicated, yielding the participial form.
Galela (10) awi dohu i lalamo his foot it be_big:PRT 'his big foot' (van Baarda 1908: 35)
Qualifying operators are not deemed to exist (Rijkhoff 2008a, forthcoming). (7) By definition, members of a grammatical word class constitute a smallish, closed set of items, capturing only a limited number of crucial, absolute distinctions (Dik 1997: 160). For example, if Definiteness is a grammatical category in some language, there are only two choices: +Definite or -Definite (indefinite). This makes operators rather unsuitable to express gradable notions (e.g., big/bigger/biggest; fairly/rather/ very etc. big); a house can be rather big but not rather SINGULAR.
2.3. Quantifying modifiers in the noun phrase
Quantifying modifiers specify quantitative properties of the referent and relate to the material in the quality layer (including the head noun). Grammatical expressions of the notion "quantity" are referred to as QUANTIFYING OPERATORS; phrasal or lexical expressions as QUANTIFYING SATELLITES (due to a lack of sufficient information, I will ignore universal and existential quantifiers as well as ordinal numerals).
2.3.1. Quantifying operators. In many languages "plural" and other number distinctions are either optional or altogether absent (Rijkhoff 2002a: 106-119, 146 155), but if nominal number is a relevant category in some language it is commonly expressed by some grammatical element like a nominal affix (e.g., a suffix, as in Dutch), or a clitic (Mupun):
Dutch (11) fiets-en bicycle-PL 'bicycles'
Notice that the plural marker in Mupun (which is identical with the third person plural pronoun) is a clitic, which means that, strictly speaking, it is not an integral part of the NP:
Mupun (12) jirap d'e wuran mo girl REL tall PL 'tall girls' (Frajzyngier 1997: 200)
In many languages cardinal numerals are grammatical modifiers (QUANTIFYING OPERATORS). For example, Dutch numerals are categorized as grammatical modifiers because the simple atoms in the numerical system (i.e., the cardinal numerals from 'one' to 'twelve'; cf. Greenberg 1978b) constitute a closed class and do not display properties associated with lexical word classes like verb, noun or adjective. It will be shown in the next section, however, that there are also languages in which expressions of cardinality are explicitly categorized as lexical items (QUANTIFYING SATELLITES).
2.3.2. Quantifying satellites. In quite a few languages cardinal numerals may or must appear as predicates or they are classified as lexical elements (Rijkhoff 2002a: 168-172). For example, Krongo numerals are verbs that appear in the imperfective when they modify a noun:
Krongo (13) noo-coori nk-ootoono PL-house CN.PL-IMPF:be_three 'three houses' (Reh 1985: 252)
In Samoan the numeral appears as the head of a special kind of relative clause ("attributive numeral clause") introduced by the general tense-aspect- mood marker e if the NP has specific reference.
Samoan (14) Sa fau=sia e Tagaloaalagi fale e PAST build=ES ERG Tagaloaalagi house GENR tolu ... three 'Tagaloaalagi built three houses ...' (Mosel and Hovdhaugen 1992: 318)
2.4. Localizing modifiers in the noun phrase
Localizing modifiers specify locative properties of the referent and relate to the material in the quantity layer (which has scope over the quality layer and head noun).
2.4.1. Localizing operators. Adnominal demonstratives are grammatical manifestations of the notion LOCATION. As is the case with other modifier categories, they do not occur in all languages (see e.g., Derbyshire [1979: 131] on Hixkaryana), and in some languages they require the presence of a classifier or a definite article (this is discussed in more detail in Section 4.2.2: see also Rijkhoff 2002a: 182-185).
Mandarin Chinese (15) nei-tiao niu DEM-CLF cow 'that cow' (Li and Thompson 1989: 105) Hungarian (16) ez a piros alma this the red apple 'this red apple' (Moravcsik 1997: 315)
2.4.2. Localizing satellites. An example of a lexical modifier that specifies the location of the referent of the matrix NP is the adnominal prepositional phrase on this carpet in (17).
(17) The stain [on this carpet] was difficult to remove.
Restrictive relative clauses and possessive modifiers are also typically used as localizing satellites, but recall that there is no one-to-one relationship between the form of a modifier and its position in the layered representation of a linguistic structure.
(18) Could you pass me the book [[that's lying on that table].sub.Rel.cl]?
(19) I'd like to talk to the teacher [[of my daughter].sub.Poss.NP].
It is essential for the localizing satellite to provide a referential anchor for the addressee, i.e., the localizing satellite must contain a reference to another entity, one that is easily identifiable for the addressee and which makes it possible to locate the referent of the matrix NP (Rijkhoff forthcoming). In the examples above the referential anchors are: this carpet, that table and my daughter. They enable the hearer to locate (and identify) the referent of the embedding matrix NP, here the stain, the book, and the teacher respectively. Referential anchors also play an important role in presupposition; this is discussed in Section 5.
3. Discourse-referential modifiers in the noun phrase
It has been mentioned in the introduction that lexical NPs may have a descriptive as well as a referential value: they can be used to describe an entity and to refer to that entity (or introduce it into conversational space). In the previous section we have dealt with noun modifiers that give (what one might call) a "physical" description of a referent. This section is concerned with the modifiers that relate to the interpersonal function of modifiers in the NPs, i.e., noun modifiers that are concerned with the pragmatic status of the referent of the NP in the shared world of discourse, here referred to as DISCOURSE-REFERENTIAL MODIFIERS. As in the case of the descriptive noun modifiers at the representational level, we distinguish between grammatical and lexical manifestations of discourse-referential modifiers.
3.1. Discourse-referential operators
Definite and indefinite articles, which indicate whether or not the referent of the NP is identifiable in the world of discourse, are good examples of grammatical instances of DISCOURSE-REFERENTIAL MODIFIERS. (8) There is basically one reason why the referent of an indefinite NP is not identifiable: because its referent has not been properly introduced in the shared world of discourse. There are, however, various reasons why the referent of a definite NP is deemed identifiable. For example, because the referent was introduced earlier in the discourse (previous mentioning), because the real world correlate of the referent is available in the physical context, or because in the given context the referent is believed to be a unique entity.
--Referent has been mentioned earlier:
(20) I just bought a book and a calendar. Surprisingly, the book was much cheaper than the calendar.
--Referent is available in physical context:
(21) Now tell me--what do you see on the monitor?
--Referent is unique in given context
(22) Let's meet in front of the town hall.
Definite NPs can also be used to introduce new discourse referents that have no unique reference or that are not available in the linguistic or extralinguistic context: this is discussed in Section 5.
3.2. Discourse-referential satellites
DISCOURSE-REFERENTIAL SATELLITES are lexical modifiers that provide the addressee with information about the referent as a discourse entity, such as same or other. (9)
(23) I gave her the same / the other book.
In the case of (the) same the addressee is explicitly instructed to find a particular referent that was mentioned a little earlier in the conversation. In the case of (the) other the addressee should identify the second member of a referent set that was introduced before. By contrast, an indefinite modifier such as another (as in: I gave her another book) is an instruction for the addressee to construe a new referent of the same type (book) as a discourse referent that already exists as a distinct referent in the discourse world and was mentioned not long before.
Other examples of DISCOURSE-REFERENTIAL SATELLITES are modifiers that are used for discourse deixis, such as Dutch laatstgenoemde 'last mentioned' or zojuist genoemde 'just mentioned'. They typically specify when or where a referent was introduced in the previous (spoken or written) discourse:
Dutch (24) De zojuist genoemde personen krijgen allemaal een the just mentioned persons get all a prijs prize 'The persons whose names have just been mentioned will all get a prize'
A modifier such as zojuist genoemde in the example above does not tell us anything about the descriptive or physical properties (concerning KIND, QUALITY, QUANTITY, LOCATION) of the referent in the world of discourse. Instead it indicates that the entities involved already exist in the world of discourse and are now collectively being referred to again. In some sense zojuist genoemde does tell us something about the "where", in that it points to a certain place (or moment) in the discourse. However, whereas localizing satellites specify the location of the referent in the world of discourse, discourse-referential satellites such as zojuist genoemde 'just mentioned' (also modifiers such as the former, the latter), relate to the place or time the referent was mentioned before in the actual conversation (DISCOURSE DEIXIS).
3.3. Discourse-referential modifiers in a layered model of the NP
Since discourse-referential modifiers relate to the referent of the NP with all its descriptive properties (kind, quality, quantity, location), they appear in the outermost layer of the underlying (semantic) representation of an NP, embracing all four descriptive layers of modification (Figure 2). (10)
[FIGURE 2 OMITTED]
4. The ordering of descriptive and discourse-referential modifiers
This section is concerned with iconicity in that certain formal properties of noun modifiers (their linear organization in the NP) are shown to correspond to certain semantic (scopal) properties (the position of a modifier in the layered model of the NP). More specifically, it will be shown that the linear organization of modifiers in the noun phrase mirrors the layered organization of the underlying semantic representation. Section 4.1 discusses the place of descriptive modifiers in the NP; Section 4.2 is concerned with the position of one particular discourse-referential operator: the definite article.
4.1. Descriptive modifiers. the order of demonstrative, numeral, adjective, and noun
Since information about NP-internal ordering patterns in the grammars and in the typological literature is in most cases restricted to the relative order of demonstrative, numeral and adjective, I will confine myself here to these three descriptive noun modifiers in their function of localizing, quantifying, and qualifying modifiers, respectively. Furthermore, due to the restrictions mentioned in the Introduction (Section 1) I will ignore, for example, (a) bound modifiers, (b) appositional forms of modification, (c) lexical expressions of cardinality, as well as (d) verbal and nominal forms that are used to express adjectival notions such as relative clauses and adnominal NPs (see examples  and  above), since the employment of such embedded modifiers would result in complex NPs.
4.1.1. Iconic basic patterns. If demonstrative, numeral and adjective are ordered according to the scopal relations as captured in the layered organization of the underlying structure (Figure 1), only eight combinations (of the 24 that are logically possible) are predicted to occur (Rijkhoff 2004).
(25) dem num A N dem A N num num A N dem A N num dem dem num N A dem N A num num N A dem N A num dem
Notice that these NP internal orders are all instances of a more abstract symmetrical structure, given in (26), in which the adjective always appears immediately before or after the noun and the demonstrative always appears in the periphery.
(26) DEM NUM A N A NUM DEM
When this hypothesis is tested against crosslinguistic facts, we find, first of all, that some languages do not have true classes of morphologically free, syntactically integrated adnominal demonstratives, numerals or adjectives (Section 1). Furthermore, there are languages in which some of the modifier categories we are dealing with here are mutually exclusive. For example, speakers of Wambon avoid having a demonstrative, a numeral and an adjective in the same NP. Instead these modifiers appear in juxtaposed NPs, as in:
Wambon (27) ev-o kap ambalopkup ev-o kap kaimombalin [[that-CN man five] [that-CN man good]] 'those five good men' (Lourens de Vries, p.c.)
In the case of certain other languages (e.g., Hittite, Ika) the relevant information was simply not available. For these reasons not all languages in the basic sample (see Appendix) could be used to investigate the relative order of demonstrative, numeral, adjective, and noun (this problem is discussed in great detail in Rijkhoff (2002a), in particular Section 10.2.4). Notice furthermore that none of the remaining languages in my basic sample had [dem A N num] or [A N num dem] as the basic order (Rijkhoff 2002a: 331). (11)
(28) dem num A N Alamblak, Dutch, Georgian, Hungarian, Kayardild, Ket, Nama Hottentot, Imbabura Quechua, Pipil, Tamil, Turkish dem A N num -- num A N dem Berbice Dutch Creole A N num dem -- dem num N A Burushaski, Guarani dem N A num Bambara (12) num N A dem Basque, Hmong Njua N A num dem Oromo
In other words, languages with the numeral following and the adjective preceding the noun are absent in my sample. The same is true for languages in the sample used in a study by Hawkins, who stated in his Universal 20' (Hawkins 1983:119-129): "'In no case does the adjective precede the head when the demonstrative or numeral follow". Notice, however, that Berbice Dutch Creole constitutes a counterexample to Hawkins's Universal 20', since is has [num A N dem]. Furthermore, it has recently been claimed in a study by Haddican (2002) that the same order [num A N dem] is also attested in other creole languages, such as Bislama and Sranan. As a matter of fact, the same source mentions a creole language with the order [A N num dem]: Sango. Finally, Zande is a language that appears to have the order [dem A N num]: (13)
Zande (29) gi rarai a-mangu biata-re DEM heavy PL-box three-DEF/here 'these three heavy boxes' (Christopher Leone Daffalla, p.c.)
When we add these languages to the original sample, there are no gaps in the "iconic" patterns listed in (30):
(30) dem num A N Alamblak, Dutch, Georgian, Hungarian, Kayardild, Ket, Nama Hottentot, Imbabura Quechua, Pipil, Tamil, Turkish dem A N num Zande num A N dem Berbice Dutch Creole, Bislama, Sranan A N num dem Sang dem num N A Burushaski, Guarani (also e.g., French and other Romance languages) dem N A num Bambara num N A dem Basque, Hmong Njua N A num dem Oromo, Fa d'Ambu, Nubi
4.1.2. Non-iconic basic patterns. In the previous section it was shown that all iconic ordering patterns involving a demonstrative, a numeral, an adjective and a noun are attested in the world's languages.
Iconic patterns (31) dem num A N dem A N num num A N dem A N num dem dem num N A dem N A num num N A dem N A num dem
The other sixteen logically possible orders are listed in (32).
Non-iconic patterns (32) num A dem N A num N dem dem N num A N dem num A A num dem N num dem N A num N dem A N num dem A A dem num N A dem N num A N dem num N A dem num dem A num N num dem A N N num A dem N dem A num
Greenberg's (1966) Universal 20 already indicates, however, that there is a clear preference for the iconic patterns.
Universal 20. When any or all of the items (demonstrative, numeral, and descriptive adjective) precede the noun, they are always found in that order. If they follow, the order is either the same or its exact opposite. (Greenberg 1966: 86-87)
The reason why Greenberg had to allow for one non-iconic pattern was due to Kikuyu, the only one language in his sample with the order [N dem num A] (according to Seiler (1978: 322), however, this is a "less popular variant" in Kikuyu, the normal order being [N A num dem]).
Almost two decades later Hawkins (1983), using a large sample of over 300 languages, mentioned two other Bantu languages with a non-iconic basic pattern in the NP: Aghem [N A dem num] and Noni (which has [N dem num A] as well as [N dem A num]), both spoken in Cameroon. Consequently he reformulated Greenberg's Universal 20 as follows: (14)
Universal 20'. When any or all of the items (demonstrative, numeral, and descriptive adjective) precede the noun, they (i.e., those that do precede) are always found in that order. For those that follow, no predictions are made, though the most frequent order is the mirror-image of the order for preceding modifiers. In no case does the adjective precede the head when the demonstrative or numeral follow. (Hawkins 1983:119-120)
Thus, whereas all iconic patterns are attested, only a small fraction of the non-iconic orders has been found so far. Apart from the three non-iconic patterns listed above (IN dem num A], [N A dem num] and IN dem A num]), Heine (1980) mentions a number of languages spoken in Kenya with non-iconic patterns. Notice that the word order sequences given below are not the only possible patterns; all the languages mentioned are said to display an enormous amount of variation.
Rendille (Afro-Asiatic; Cushitic): N-dem num A Gabra (Afro-Asiatic; Cushitic): N num A dem Sampur (Nilo-Saharan; Eastern Nilotic): dem N num A Camus (Nilo-Saharan: Eastern Nilotic): dem N num A Turkana (Nilo-Saharan: Eastern Nilotic): N dem num A Luo (Nilo-Saharan: Western Nilotic): N num A dem Logoli (Niger-Congo; Central Bantu): N num A dem
Detailed analyses of NPs in the languages listed above are not available (Heine's article is only six pages long), but some of Heine's comments strongly suggest that we may not be dealing with simple, integral NPs in which modifiers are expressed as free forms (Rijkhoff 2002a: 19-23, 273-276, 329-332). For example, adjectives in Sampur and probably also in Camus (both are dialects of Nilo-Saharan Maa) are probably better analyzed as verbs heading a relative construction (Heine 1980: 182). Generally speaking, it seems to be the case that at least one of the following statements is true for non-iconic patterns:
--the items treated as adjectives are actually verbs or nouns, i.e., adnominal relative clauses or NPs, turning the NP into a complex structure (see Example ); (15)
--numerals are expressed as phrasal modifiers, also turning the NP into a syntactically complex construction (see Examples  and );
--modifiers are expressed as bound rather than free elements, which means their expression is a matter of morphology rather than syntax, as in (see, for example, the demonstrative suffix in Rendille above);
--modifiers are in apposition (rather than fully integrated constituents). For example, the Ngalakan NP "exhibits a fairly loose sort of structure. It is possible for constituents of what could be considered the "same" NP to be separated from each other by other clausal constituents, or for many NP constituents having the same referent to be strung together in a fairly loose sort of appositional structure: [...]" (Merlan 1983: 71, 83; but cf. McGregor 1997);
--modifiers are assigned a special pragmatic function like Focus, indicating we are dealing with a marked pattern (Rijkhoff 2002a: 272-273).
Thus, whereas all iconic patterns are attested, only some of the non-iconic patterns are believed to occur as the basic word order. Furthermore, I have suggested that the small number of languages that are deemed to use a non-iconic pattern as the basic word order might in fact only constitute apparent counterexamples to the strong tendency to iconically align demonstrative, numeral and adjective relative to the head noun in that they involve forms or constructions with embedded, apposed, or bound modifiers (see also Rijkhoff 1990).
4.2. Discourse-referential modifiers." the position of definite article"
In the previous section we discussed the relative order of three descriptive modifiers (demonstrative, numeral and adjective) in a simple, integral NP and found that there is a strong preference to adhere to patterns that reflect scopal relations in the underlying structure of the NP (see Figure 4). Thus, if the layered model of the NP presented above is correct, we may also expect discourse-referential modifiers to appear in the first or in the last position in the NP (i.e., not between a descriptive modifier and the noun), as in the following example.
Koyra Chiini (33) har hinka woo di man two DEM DEF 'these (those) two men' (Heath 1999: 61)
Here we will only concentrate on the order demonstrative and (free) definite article. This restriction is partly due to the fact that an adequate account of the internal syntax of NPs is often lacking in the grammatical description of a language. It is also true, however, that articles (like other modifier categories) are not universally attested (Rijkhoff 2002a: 185-88). For example, in my sample (see Appendix) Bambara, Georgian, Hixkaryana, Lango, Nivkh, Imbabura Quechua, Tsou, Turkish and West Greenlandic are all examples of languages without a distinct class of definite articles. Notice furthermore that in many languages demonstratives are mutually exclusive with articles (Rijkhoff 2002a: 329-332; see Haspelmath (1999) on article-possessor complementarity). One such language is Dutch:
Dutch (34) a. het boek DEF book 'the book' b. dat boek that book 'that book' c. * het dat boek the that boek d. * dat het boek that the book
Nevertheless my sample contains two counterexamples to the hypothesis formulated above, each of which involves the order definite article-- demonstrative: Abkhaz and Hungarian. (16)
4.2.1. Abkhaz. Abkhaz has a so-called "definite-generic article," which is actually a prefix that occurs in between the demonstrative and the noun (Hewitt 1979: 57, 225).
Abkhaz (35) w[??]y a-j[??]ab that_one ART-girl 'that girl' (Hewitt 1979: 57)
Since the hypothesis formulated above only applies to free modifiers (Section 1), Abkhaz does not constitute a true counterexample. Notice furthermore that this definite-generic article prefix differs from the typical definite article in that it is used in NPs with definite and indefinite reference and also appears on nouns in their citation form: (17)
Abkhaz (36) a-j[??]ab rac[??]a ART-girl many 'many girls' (Hewitt 1979: 57)
4.2.2. Hungarian. Hungarian seems to constitute a true counterexample: a free article appears between the demonstrative and the noun:
Hungarian (37) ez a barna kalap DEM ART brown hat 'this brown hat' (Moravcsik 1997: 319)
Moravcsik (1997: 319) has argued that this pattern arose "from a pronominal demonstrative to which a noun phrase was appositionally adjoined (see also McCool 1984, 1993; Lehmann 1995: 37-39). Thus, the origin of an expression like ez a barna kalap 'this brown hat' was ez, a barna kalap 'this, the brown hat'". Whereas the syntax of adnominal grammatical modifiers that have developed from other adnominal grammatical modifiers (e.g., definite articles from distal demonstratives, plural markers from collective markers) seems to adhere to iconic patterning, erstwhile appositional elements that became fully integrated NP-internal modifiers may appear in positions that reflect their original status as an appositive, resulting in non-iconic patterns. Thus, in the case of Hungarian we would be dealing an instance of syntactic entrapment (on entrapment in morphology, see Moravcsik 2000: 549): an NP external element is reanalyzed as part of an integral NP. Since this does not involve a shift of position, the erstwhile outer modifier (the adnominal definite article) is now trapped between the erstwhile NP-external element (the demonstrative) and the other constituents of the NP.
5. The relationship between the localizing and discourse-referential modifiers
The distinction between descriptive and discourse-referential modifiers in the NP may suggest that we can always draw a hard and last line between the two modifier categories, but in reality there tends to be quite a bit of overlap between discourse-referential modifiers and one descriptive modifier category: the localizing modifiers. This is perhaps best exemplified by the attributive distal demonstrative, which in many languages functions both as a spatial deictic (localizing operator) and as a marker of definiteness (discourse-referential operator).
Pipil (38) ne ta:ka-t DEF man-ABS 'the/that man' (Campbell 1985: 56)
In some of these languages this categorial vagueness is disambiguated in syntax, as in Old Georgian where the modifiers igi, ese, ege preceded the noun when they served as demonstratives and followed the noun as definite articles (Testelec 1997: 247).
Other examples of the close connection between localizing and discourse-referential modifiers may be found in the Uralic and Turkic languages (Menges 1968: 113; Siewierska et al. 1998: 811-812, Notes 14 and 32). For example, Comrie (1988: 465), referring to Tauli (1966: 148), writes that some Uralic languages "use the third person singular possessive suffix as a general marker of definiteness."
Komi, Southern Permyak dialect (39) et-piris secce woktis ruc. Ruc-is cig. Once then came fox. fox-POS.3SG hungry 'Once the fox came that way. The fox was hungry.' (Fraurud 2001: 252; taken from Redei 1978: 474)
As the change in function has not (yet) clearly resulted in a category switch (from suffixed possessive pronoun to suffixed definite article), it is, however, not entirely clear whether or not the possessive affix in these languages has actually grammaticalized into a bound marker of definiteness (cf. Fraurud 2001). (18)
In Section 4.1.1 we saw that the speaker uses (noncomplex) definite NPs to refer to entities that (s)he assumes to be immediately identifiable for the hearer, i.e., the identifiability or existence of these referents in the shared world of discourse is presupposed. This is typically the case for referents that (i) were mentioned before, (ii) are available in the physical context, and (iii) are uniquely identifiable in a given context (e.g., 'the sun'). It is, however, also possible to use definite NPs for entities that cannot be identified in this way. In such cases the speaker must provide a referential anchor to "ground" the new entity in the shared world of discourse. Such referential anchors are typically provided by localizing satellites: they are the constructions that mention the entity through which the referent of the matrix NP can be "anchored" in conversational space (see below). The name for this discourse strategy is due to Prince (1981: 236): "a discourse entity is Anchored if the NP representing it is LINKED, by means of another NP, or "Anchor", properly contained in it, to some other discourse entity."
To put it differently, localizing satellites make it possible to introduce a new referent through a definite NP, allowing the speaker to present a new referent as an identifiable entity. Consider these examples (* = pragmatically marked):
(40) * Tomorrow I'll buy THE BOOK.
(41) * I went to see THE APARTMENT last night.
(42) * They went to THE RESTAURANT to have lunch.
Each of these sentences contains a definite NP (in small caps), suggesting that their referents are assumed to be identifiable for the hearer. However, in each case the hearer has the inclination to ask "What book?", "What apartment?", or "What restaurant?" Now compare these sentences:
(43) Tomorrow I'll buy THE BOOK YOU WANT SO MUCH.
(44) We went to see JOHN'S APARTMENT last night.
(45) They went to THE RESTAURANT IN THE VAN GOGH MUSEUM for lunch.
In these examples each definite NP is provided with an embedded localizing satellite, which enables the hearer to properly identify the referent of the matrix NP, even though by itself this referent is not immediately recoverable from the linguistic or extra-linguistic context. Definiteness marking does not depend on whether or not the hearer knows the book, the apartment or the restaurant, but rather on the hearer's ability to identify an entity mentioned in the localizing satellite. Now the book can be identified as 'the book THE ADDRESSEE wanted so much'. In a similar vein the hearer can now identify the apartment as 'the one owned or occupied by JOHN' (with whom the hearer is presumed familiar), mentioned in the possessive modifier; and the restaurant can now be identified as 'the restaurant in THE VAN GOGH MUSEUM' (also assumed known). The examples show that localizing satellites cannot only be used to specify the location of an entity (a descriptive property), but also to license or explain the existence of an entity in the shared world of discourse of the interlocutors (a referential property). (19)
The notions location, existence, and identifiability are of course closely related. For an entity to be identifiable it has to exist somewhere, i.e., it must have a location. And vice versa, if an entity has a location it must exist (cf. Lyons 1977: 718-724; Bugenhagen 1986). The relationship between possessive and locative constructions has been discussed in various studies. For example, Clark (1970) observed that from a cognitive perspective possessors can be regarded as locations:
[...] psychologically it would appear quite plausible to argue that if an object is in some place, and the "place" is actually an animate being, then the object is possessed by the "place". In other words, it is the [+Animate] feature added to the locative phrase that transforms it into a Possessor-nominal. (Clark 1970: 3)
Various studies have also established that across languages there are some remarkable grammatical similarities between existential, locative and possessive constructions. (20) Lyons (1967) was the first to suggest that these constructions are related, both synchronically and diachronically. The nature of this relation was investigated in more detail in a sample of 65 languages by Clark (1970, 1978), who argued that these constructions are
[...] systematically related within each language in word order, in the verb used, and in their locative nature. The relationship between them can generally be accounted for by two discourse rules. These determine the word order of each construction depending (a) on the definiteness of the nonlocative nominal, and (b) on the animacy of the other, locative, nominal. (Clark 1978: 85)
In other words, the constructions under consideration all contain what Clark calls LOCATIONALS. Consider the following examples from English and French. In (46) and (47) the locational is the locative on the table; in (48) and (49) it is the possessor John('s) (for a full overview of the similarities between languages in the sample, I refer to Clark [1978: 93-121]).
(46) a. English: There is a book on the table existential b. French: Il y a un livre sur la table existential (47) a. English: The book is on the table locative b. French: Le livre est sur la table locative (48) a. English: John has a book possessive I b. French: Jean a un livre possessive I (49) a. English: The book is John's possessive II b. French: Le livre est a Jean possessive II
The localist interpretation of possession also holds for adnominal possessive modifiers. This is illustrated by the following example from Ewe (Niger-Kordofanian), in which a locative element has developed into a marker of possession.
Ewe (50) fofo nye [??]e x[??] father my place house "nay lather's house" ('the house at my lather's place') (Claudi and Heine 1986: 316)
In this section we saw that there are close links between localizing and discourse-referential modifiers (e.g., demonstratives may serve as discourse-referential or localizing operators), which are represented in two adjacent layers in the underlying structure the NP. The next section is concerned with relationships between the layered model of the NP and the layered model of the clause.
6. Parallels between noun phrases and clauses
In this section I will argue that in Functional [Discourse] Grammar, NPs and clauses can be analyzed in a similar fashion in that they have the same kind of layers, accommodating the same kind of modifier categories (Figure 3).
Section 6.1 gives an outline of parallels between NPs and clauses at the descriptive level: Section 6.2 discusses similarities at the interpersonal level.
6.1. Descriptive modifiers in the NP and in the clause
For the NP we distinguished between four kinds of descriptive modifiers, in order of increasing scope: classifying modifiers, qualifying modifiers, quantifying modifiers, and localizing modifiers. It appears that the same functional modifier categories are relevant for the analysis of the clause. This is shown in Figure 4.
[FIGURE 3 OMITTED]
Tense, as a deictic category, is of course the localizing operator at the level of the clause. Thus, the past tense of buy in (51) indicates that the event took place before the moment of speaking. (21) Time and place adverbials such as in London and yesterday are examples of localizing satellites, which provide more precise descriptions of the location of the buying event (in space or time).
(51) I [BOUGHT.sub.Past] this book IN [LONDON.sub.Place] [YESTERDAY.sub.Time]
Languages also have quantifying satellites to specify how often an event occurs, i.e., lexical expressions of the notion Quantity, as in:
(52) EVERY ONCE IN A WHILE our cat catches a mouse.
Some languages employ grammatical means (quantifying operators) to indicate how often an event takes place. When reference is made to a single occurrence this is called semelfactive aspect and in the case of multiple occurrences this is called iterative, repetitive, or frequentative aspect (sometimes subsumed under the label "Pluractional"; Newman 1990). The following example from Hidatsa (Amerind) contains such a quantifying element (glossed as ITERative MOOD in the grammar):
Hidatsa (53) Wi i hirawe ksa C woman she sleep INGR ITER.MOOD 'The woman kept falling asleep' (Matthews 1965: 158)
[FIGURE 4 OMITTED]
Lexical forms and constructions expressing qualitative properties of an event are, for example, adverbs of speed (qualifying satellites).
(54) She read the paper QUICKLY.
Finally, verbal aspect markers (classifying operators) specify what kind of event is being referred to, indicating for example, that an event is bounded (perfective) or open-ended (imperfective).
Hungarian (55) EL-olvastam az ujsag-ot PERF-read-I the paper-ACC 'I read the paper" (from beginning to end) (Judit Horvath, p.c.)
So-called stripped nouns are instances of classifying satellites at the level of the clause. As opposed to an incorporated noun, a stripped noun is a separate word (according to phonological criteria such as stress placement), which must appear next to the verb (Miner 1986, 1989; Gerdts 1998). In the following example, mitmit sac 'the knife' is a distinct argument NP, as can be seen by the fact that is separated from the verb.
Kusaiean (56) a. Sah el twem upac mitmit sac Sah he sharpen diligently knife the 'Sah is sharpening the knife diligently' (Gerdts 1998: 94; original example in K. Lee 1975)
In the next example, mitmit 'knife' is a stripped noun that serves as a classifying satellite. Instead of a general "sharpening action" we are now dealing with a special kind of sharpening: knife sharpening.
(56) b. Sah el twetwe MITMIT upac Sah he sharpen knife diligently 'Sah is diligently knife sharpening'
Even though we are concerned with free modifiers here (Section 1), it may be interesting to mention that Bybee (1985) found that the order of morphological markers of aspect, tense and mood relative to the verb usually reflects their position in the layered organization of the clause (Figure 4). At least for English similar things can be said about the relative order of qualifying, quantifying and localizing satellites. According to Quirk et al. (1985: 551), different kinds of temporal satellites tend to occur in the order time duration (for a short while = qualifying satellite), time frequency (every day or so = quantifying satellite) and time position (in January = localizing satellite), as in:
(57) I was there FOR A SHORT WHILE EVERY DAY OR SO IN JANUARY
Diachronic developments also seem to confirm the hypothesis that there are both semantic and morphosyntactic parallels between NPs and clauses (el. Rijkhoff 2008b). For example, Gildea (1993) has shown how over time demonstratives ("location in space") may turn into tense markers ("location in time") and Von Garnier already showed in 1909 that elements with a collective meaning (specifying a "mode of being"; Section 2.1.1) can turn into markers of perfectivity (specifying a "mode of action").
A question that has not been addressed so far is why we should find parallels between NPs and clauses in the first place. Apart from the least interesting zero hypothesis (i.e., similarities are due to chance), there are basically three possible ways to account for parallels between NPs and clauses (Rijkhoff 2008b):
(a) clause structure is derived from NP structure:
(b) NP structure is derived from clause structure;
(c) NP structure and clause structure are due to a single cognitive procedure that deals with spatio-temporal entities.
Currently the first possibility has the best empirical foundation. Although relatively little is known about the way the human cognitive system deals with spatial and temporal entities, linguistic evidence indicates that our conceptual system is largely metaphorical in nature (Lakoff 1987). Since there are many examples which show that spatial metaphors are used to express temporal and other nonspatial notions, it is assumed that spatial conception plays a fundamental role m human cognition (Lyons 1977: 718: Levinson 1992). Thus perhaps it is because temporal entities are understood in terms of (cognitively less complex) spatial entities that NPs and clauses can be analyzed in a similar fashion. (22)
6.2. Discourse-referential modifiers in the NP and in the clause
Parallels between NPs and clauses can also be drawn at the interpersonal level (Figure 3; Rijkhoff 2002a: 229 238). It was observed in the introduction that lexical NPs serve a dual purpose in that they are used both to provide it description of an entity and to serve as an instrument to refer to that entity. In other words, we can distinguish between the description of an entity (the linguistic construction: the NP) and the referent of that description (the object). This way we can account for sentences such as the following, in which different descriptions are used to refer to the same entity (i.e., Venus; Frege 18921:
(58) The Morning Star is the Evening Star.
The practical need for this distinction is also demonstrated by the following example, in which he and him are used to the refer to referent of my neighbor and the superstitious old fool, whereas that only refers to the negative description (for a more formal account, see Rijkhoff 2002a: 225-238):
(59) A: My neighbor just saw a black cat and now the superstitious old fool believes he is in for some bad luck today.
B: Why do you call him that?
Similarly, a clause gives both a description of an event and refers to that event in the world of discourse. Thus at the level of the clause we can also distinguish between the speaker's description of an entity (the clause) and the entity to which the speaker refers by means of that linguistic construction (the event). The following dialogue illustrates the usefulness of this distinction for anaphoric purposes.
(60) A: Fred saw that Harry was talking to the police this morning.
B: Yes, I saw that too.
(61) A: Fred saw that Harry was talking to the police this morning.
B: That is not a good description of what really happened; the police were actually interrogating him.
In (60) person B uses that to refer to the referent of the object clause (what one sees is an event), whereas in (61) person B uses that to refer to Fred's description of the event, i.e., the linguistic expression (here: the object clause as such).
At the level of the clause, discourse-referential operators indicate that the event is either actual (or realis) or nonactual (or irrealis).
Takelma (62) Mena yap'a t'om[??]-k'wa bear man kill(REALIS)-3HUMAN_OBJECT 'The bear killed the man' (Chung and Timberlake 1985: 244; from Sapir 1922: 158)
Actual events are grounded (or "anchored") in the world of discourse: they can be located in time and the clause will be tensed (provided tense is a grammatical category in the language in question). Conversely, if the speaker refers to a nonactual event, it will not be possible to locate the event in time. (23) There are several reasons why an event may be nonactual; for instance, because the speaker is using a command (referring to an event to be brought about by the addressee), because the possible occurrence of the event is hoped, feared, or doubted, or because the speaker wants to indicate that (s)he has no direct evidence for the actual occurrence of the event.
Takelma (63) Mena yap'a dom-k' wa-k' bear man kill(IRREALIS)-3HUMAN_OBJECT-INFERENTIAL 'It seems that the bear killed the man / The bear must have, evidently has, killed the man' (Chung and Timberlake 1985: 245; from Sapir 1922: 158) Palauan (64) Ku-rael el mo er a blik ISG.IRR-travel COMP go LOC ART my_house 'Perhaps I should go home' (Chung and Timberlake 1985: 248-249: from Josephs 1975) (65) Mo-lim a kerum! 2SG.IRR-drink ART your_medicine 'Drink your medicine!'
This also explains why nonactual predications are commonly tinder the scope of certain modal or illocutionary operators. In Kobon, for example, "the tense distinctions between simple past, remote past, present, and future are made only in the indicative mood [...]" (Davies 1981: 168) and in West Greenlandic tense affixes do not occur in the imperative/optative mood, at least not "in their strictly temporal sense"' (Fortescue 1984: 275). (24)
As was already indicated in Figure 3, some typical examples of discourse-referential satellites at the level of the clause are adverbs such as actually and really, which emphasize the actual existence of the event in conversational space (for other examples, see Rijkhoff 2002a: 236). As to the parallels between operator categories of the NP and the clause at the Interpersonal Level, there are two symmetries, one linking the grammatical categories REALIS and DEFINITE, the other linking IRREALIS and INDEFINITE (Table 1).
The discourse-referential operators DEFINITE and ACTUAL are similar because they indicate that the thing or event or referred to by the speaker is (already) grounded in the world of discourse. By contrast, their negative counterparts IRREALIS and INDEFINITE, or rather NONSPECIFIC-INDEFINITE, signal that the entity referred to by the speaker does not have a proper location (is not "grounded") in the world of discourse--at least not yet. (25)
The idea that at some abstract level of representation discourse-referential operators in the NP and in the clause essentially express the same notion (something like: determined or undetermined existence of an entity in the shared world of discourse of Speaker and Addressee) is supported by the fact that in certain languages these operators are expressed by the same element in the NP and in the clause. For example, Fongbe uses the same morpheme to mark DEFINITENESS in the NP and ACTUALITY in the clause and in Jacaltec both the exhortative mood and nonspecificity are expressed by the suffix/-OJ/, "[...] the general suffix of irrealis [...]" (Craig 1977: 93; notice that variation between -oj and -uj is due to vowel harmony):
Fongbe (66) N du as[??]n [??] I eat crab DET 'I ate the crab (in question/that we know of).' (67) Jan wa [??] John arrive DET 'Actually, John arrived.' (Lefebvre 1998: 94, 99) Jacaltec (68) Way-oj ab naj sleep-OJ EXH CLF/he 'Would that he slept!' (69) X-[??]-'oc heb ix say- a' hun-uj ASP-ABS.3-start PL woman look_for-FUT a-OJ munlabel pot 'The women started looking for a pot' (Craig 1977: 93)
In (68) -OJ is suffixed to the intransitive verb in the exhortative mood; in (69) it marks the referent of the term phrase as nonspecific. (26)
Interestingly, there is also an antisymmetrical relationship between discourse-referential modifiers in the NP and in the clause, which has to do with the number of ways that an entity can be definite/indefinite (thing) or actual/non-actual (event).
Here Definite aligns with Irrealis in that both definite things (referents of definite noun phrases) and non-actual events (referents of irrealis clauses) can occur in the world of discourse for many different reasons. As we saw above ill examples (20)-(22) there are several reasons why an NP is definite. Similarly, there are many reasons why an event is nonactual, as is shown in these examples from the Papuan language Amele (see also examples - above).
Amele (70) Ho bu-busal-eb age qo-qag-an pig SIM-run.out-3SGDS.IRR 3PL hit-3PL-FUT "They will kill the pig as it runs out.' (71) Ho bu-busal-eb age qu-ig-a pig SIM-run.out-3SGDS.IRR 3PL hit-3PL-IMP 'Kill the pig as it runs out!' (72) Ho bu-busal-eb cain qu-wain pig SIM-run.out-3SG.DS.IRR PROH hit-NEG.F.3PL 'Don't kill the pig as it runs out!' (Roberts 1990:371 372)
There is, however, basically only one reason why an NP is (specific) indefinite: because the entity (thing) designated by the indefinite NP has not been properly introduced in the world of discourse--hence the speaker assumes that the addressee does not know (yet) what particular thing is being referred to. There is also only one reason why a sentence is in the realis mood: because the entity (event) designated by the sentence is real, i.e., it has happened (or is happening). Thus antisymmetry between discourse-referential operators in the NP and in the clause is due to the fact that
(a) referents of DEFINITE NPs and referents of IRREALIS clauses are part of the world of discourse for many different reasons;
(b) referents of SPECIFIC-INDEFINITE NPs and referents of REALIS clauses exist in the world of discourse for one and the same reason: they "ground themselves" in the world of discourse when being referred to for the first time. (27)
To sum up, both noun phrases and clauses have a descriptive and a referential value and this is reflected, among other things, in the various kinds of modifier categories. There are both semantic and morphosyntactic arguments to distinguish between modifiers that describe certain physical properties of a referent of a noun phrase or a clause on the one hand and modifiers that relate to discourse-referential properties of an entity in the discourse world on the other.
In this article I have addressed various issues concerning a layered analysis of the noun phrase in Functional [Discourse] Grammar. Firstly, I have argued that there are two major kinds of modifier categories in the NP: descriptive modifiers at the representational level and discourse- referential modifiers at the interpersonal level (Sections 2 and 3). Descriptive noun modifiers specify properties of the referent of the NP in terms of Kind, Quality, Quantity and Location, whereas discourse-referential modifiers are concerned with the referential status of entities in the world of discourse.
Secondly, I have argued that crosslinguistically the position of adnominal modifiers relative to the head noun strongly tends to reflect the differences in scope as captured in the layered model of the noun phrase (Section 4). It was shown that all eight "iconic" patterns are attested as the basic order in at least one of the world's languages. Only few of the other 16 logically possible orders are deemed to occur, but there is evidence to suggest that these orders are either nonbasic (e.g., due to focus assignment to a constituent inside the NP; Rijkhoff 2002a: 334-335), include morphologically bound modifiers, or do not involve simple, integral NPs (but rather complex NPs or apposed modifiers). Thirdly, I have argued that there is a special relationship between localizing and discourse-referential modifiers, due to the fact that there is an intricate cognitive relationship between the notions Location and Existence (Section 5). Finally I demonstrated that NPs and clauses can be analyzed in a similar fashion in that they have the same kind of layered structure, accommodating the same kind of modifier categories (Section 6).
Appendix. The sample
The basic sample contains 52 languages, which are distributed across (sub)families in such a way that the genetic (historic) distance between them is always maximal (I used Ruhlen's  classification; for details about the sampling method I refer to Rijkhoff and Bakker ):
FAMILY LANGUAGE(S) (subfamily) Afro-Asiatic 2: Gude (Chadic), Oromo (Cushitic) Altaic 1: Turkish Amerind 7: Pipil (Central Amerind), Hixkaryana (Ge-Pano-Carib), Cayuga (Northern Amerind, Almosan-Keresiouan), Koasati (Northern Amerind, Penutian), Guarani (Equatorial- Tucanoan), Ika (Chibchan-Paezan), Imbabura Quechua (Andean) Australian 3: Ngalakan (Gunwinygun), Kayardild (Tangkic), Nunggubuyu (Nunggubuyu) Austric 5: Tsou (Austro-Tai, Austronesian, Tsouic), Samoan (Austro-Tai, Austronesian, Malayo-Polynesian), Nung (Austro-Tai, Daic), Vietnamese (Austroasiatic), Hmong (Miao-Yao) Caucasian 1: Abkhaz Chukchi-Kamchatkan 1: Chukchi Elamo-Dravidian 1: Tamil Eskimo-Aleut l: West Greenlandic Indo-Hittite 2: Dutch (Indo)-European), Hittite (Anatolian) Indo-Pacific 5: Wambon (Trans-New Guinea), Alamblak (Sepik- Ramu), Galela (West Papuan), Bukiyip (Torricelli), Nasioi (East Papuan) Kartvelian 1: Georgian Khoisan 1: Nama Hottentot Korean-Japanese-Ainu 1: Korean Na-Dene 1: Sarcee Niger-Kordofanian 4: Babungo (Niger-Congo, Niger-Congo Proper, Central Niger-Congo), Kisi (Niger-Congo, Niger-Congo Proper, West Atlantic), Bambara (Niger-Congo, Mande), Krongo (Kordofanian) Nilo-Saharan 2: Lango (East Sudanic), Ngiti (Central Sudanic) Pidgins and creolen 1: Berbice Dutch Creole Sino-Tibetan 2: Mandarin Chinese (Sinitic), Burmese (Tibeto-Karen) Uralic-Yukaghir 1: Hungarian Language isolates 9: Basque, Burushaski, [dagger] Etruscan, Gilyak (= Nivkh), [dagger] Hurrian, Ket, [dagger] Meroitic, Nahali, [dagger] Sumerian.
Received 22 February 2006
Revised version received 25 January 2007
University of Aarhus
Bhat, Darbhe Narayana Shankara (1994). The Adjectival Category: Criteria for differentiation and identification. Studies in Language Companion Series 24. Amsterdam: John Benjamins.
Boas, Franz (1911). Introduction. In Handbook of American Indian Languages: Part 1, Bulletin 40 of the Smithsonian Institution, Franz Boas (ed.), 5-83. Washington: Bureau of American Ethnology. (Reprinted 1969 in Oosterhout: Anthropological Publications.)
Bot Ba Njock, Henri Marcel (1977). L'adjectif qualicatif dans trois langues bantu du nordouest: duala, basaa, bulu. In Language and Linguistic Problems in Africa, Proceedings of the VII Conference on African Linguistics, Paul F. A. Kotey and Haig Der- Houssikian (eds.), 207-225. Columbia, SC: Hornbeam.
Boyer, Carl B. (1968). A History of Mathematics. New York: John Wiley.
Breban, Tine and Davidse, Kristin (2003). Adjectives of comparison: the grammaticalization of their attribute uses into postdeterminer and classifier uses. Folia Linguistica XXXVII (3 4), 269-317.
Bugenhagen, Robert D. (1986). Possession in Mangap-Mbula: its syntax and semantics. Oceanic Linguistics 25(1/2), 124-166.
Butler, Christopher S. (2003). Structure anti Function: A Guide to Three Major Structural-Functional Theories. Part 1: Approaches to the Simplex Clause. Part 2: From Clause to Discourse and Beyond. Studies in Language Companion Series 63/64. Amsterdam: John Benjamins.
Bybee, Joan L. (1985). Morphology: A Study of the Relation between Meaning and Form. Typological Studies in Language 9. Amsterdam: John Benjamins.
Campbell, Lyle (1985). The Pipil Language of El Salvador. Berlin and New York: Mouton de Gruyter.
Chappell, Hillary and McGregor, William (1989). Alienability, inalienability and nominal classification. In Proceedings of the Fifteenth Annual Meeting of the Berkeley, Linguistics Society, February 18-20. 1989: General Session and Parasession on Theoretical Issues in Language Reconstruction, Kiria Hall, Michael Meacham, and Richard Shapiro (eds.), 24-36. Berkeley: Berkeley Linguistics Society.
Christie, J. J. (1970). Locative, possessive and existential in Swahili. Foundations o[" Language 6, 166-177.
Chung, Sandra and Timberlake; Alan (1985). Tense, aspect, and mood. In Language Typology and Syntactie Description: Volume III: Grammatical Categories and the Lexicon, Timothy Shopen (ed.), 202-258. Cambridge: Cambridge University Press.
Clamons, Robbin (2001). Oromo. In Facts about the World's Languages: An Ecyclopedia of the World's Major Languages, Past and Present, Jane Garry and Carl Rubino (eds.), 531-533. New York: H. W. Wilson.
Clark, Eve V. (1970). Locationals: a study of the relation between "existential', 'locative', and 'possessive' constructions. Working Papers on Language Universals 3, LI-L37.
Clark, Eve V. (1978). Locationals: existential, locative, and possessive constructions. In Universals of Human Language: Vol. 4: Syntax, Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds.), 85-126. Stanford: Stanford University Press.
Claudi, Ulrike and Heine, Bernd (1986). On the metaphorical basis of grammar. Studies in Language 10(2), 297 335.
Comrie, Bernard (1988). General features of the Uralic languages. In Handbuch der Orientalistik, Achte Abteilung; V-1: The Uralic Languages: Description. History and Foreign Influences, Denis Sinor (ed.), 451-477. Leiden: Brill.
Craig, Colette Grinevald (1977). The Structure of Jacaltec. Austin: University of Texas Press.
Creissels, Denis (2000). Typology. In African Languages: An Introduction, Bernd Heine and Derek Nurse (eds.), 231-258. Cambridge: Cambridge University Press.
Davies, John (1981). Kobon. Lingua Descriptive Studies 3. Amsterdam: North- Holland.
Derbyshire, Desmond C. (1979). Hixkaryana. Lingua Descriptive Studies 1. Amsterdam: North-Holland.
Dik, Simon C. (1989). The Theory of Functional Grammar. Functional Grammar Series 9. Dordrecht: Foils.
Dik, Simon C. (1997). The Theory of Functional Grammar: Part 2: Complex and Derived Constructions, Kees Hengeveld (ed.). Berlin and New York: Mouton de Gruyter.
Du Feu, Veronica M. (1987). The determinants of the noun in Rapanui. Journal of the Polynesian Society 96, 473-495.
Du Feu, Veronica M. (1989). Verbal parameters expressed in the NP in Rapanui. Paper presented at the Colloquium on NP Structure, University of Manchester, September 1989.
Du Feu, Veronica M. (1996). Rapanui. London: Routledge.
Elliot, Jennifer R. (2000). Realis and irrealis: forms and concepts of the grammaticalization of reality, Linguistic Typology 4(1), 55-90.
Farsi, A. A. (1968). Classification of adjectives. Language Learning 18, 45-60.
Fortescue, Michael (1984). West Greenlandic. London: Croom Helm.
Frajzyngier, Zygmunt (1997). Grammaticalization of number: from demonstratives to nominal and verbal plural. Linguistic Typology 1(2), 193-242.
Fraurud, Karl (2001). Possessives with extensive use: a source of definite articles? In Dimensions of Possession, Irene Barton, Michael Herslund, and Finn Sorensen (eds.), 243- 267. Amsterdam: John Benjamins.
Frege, Gottlob (1970 ). On sense and reference. In Translations from the Philosophical Writings of Gottlob Frege, P. Geach, M. Black (eds.), 56 78. Oxford: Oxford University Press.
von Gamier, K. (1909). COM--als perfektierendes Praefix bei Plautus, SAM- im Rigveda, CYN- bei Homer. Indogermanisehe Forschungen 25, 86-109.
Gerdts, Donna B. (1998). Incorporation. In The Handbook of Morphology, Andrew Spencer and Arnold M. Zwicky (eds.), 84-100. Oxford: Blackwell.
Giegerich, Heinz J. (2005). Associative adjectives in English and the lexicon- syntax interface. Journal of Linguistics 41(3), 571-591.
Gildea, Spike (1993). The development of tense markers from demonstrative pronouns in Panare (Cariban). Studies in Language 17(1). 53-73.
Grant, Anthony P. (1995). Article agglutination in Creole French: a wider perspective. In From Contact to Creole, and Beyond, Philip Baker (ed.), 149-176. Westminster Creolistic Series 1. London: University of Westminster Press.
Greenberg, Joseph H. (1966). Some universals of grammar with particular reference to the order of meaningful elements. In Universals of Language, 2nd ed., Joseph H. Greenberg (ed.), 73-113. Cambridge, MA: MIT Press.
Greenberg, Joseph H. (1978a). How does a language acquire gender markers? In Universals of Human Language: Vol. 3: Word Structure, Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds.), 48-82. Stanford: Stanford University Press.
Greenberg, Joseph H. (1978b). Generalizations about numeral systems. In Universals of Human Language: Vol. 3: Word Structure, Joseph H. Greenberg, Charles A. Ferguson, and Edith A. Moravcsik (eds.), 250 295. Stanford: Stanford University Press.
Greenberg, Joseph H. (1981). Nilo-Saharan moveable -k as a Stage III article (with a Penutian typological parallel). Journal of African Languages and Linguistics 3, 105-112.
Greenberg, Joseph H. (1991). The last stages of grammatical elements: contrastive and expansive desemanticization. In Approaches to Grammaticalization, Typological Studies in Language 19, Elizabeth Closs Traugott and Bernd Heine (eds.), 301-314. Amsterdam: John Benjamins.
Haddican, Bill (2002). Aspects of DP word order across creoles. Paper presented at the CUNY/SUNY/NYU linguistics Mini-Conference, April 20, 2002 [abstract available at: http: //www.nyu.edu/gsas/dept/lingu/events/minico02/abstract/haddican.pdf].
Hagege, Claude (1974). The 'adjective' in some African languages. Studies in African Linguistics (Supplement 5), 125-133.
Halliday, Michael Alexander Kirkwood (2004). An Introduction to Functional Grammar, 3rd ed. Christian M. I. M. Matthiessen (ed.). London: Arnold.
Harder, Peter (1996). Functional Semantics: A Theory of Meaning, Structure and Tense in English. Berlin and New York: Mouton de Gruyter.
Harriehausen, Bettina (1990). Hmong Njua: syntaktische Analyse einer gesprochenen Sprache mithilfe datenverarbeitungstechnischer Mittel und sprachvergleichende Beschreihung des sudostasiatischen Sprachraumes. Linguistische Arbeiten 245. Tubingen: Niemeyer.
Haspelmath, Martin (1999). Explaining article-possessor complementarity: economic motivation in noun phrase syntax. Language 75(2), 227-243.
Hawkins, John A. (1983). Word Order Universals: Quantitative Analyses of Linguistic Structure. New York: Academic Press.
Heath, Jeffrey (1999). A Grammar of Koyra Chiini. Mouton Grammar Library 19. Berlin and New York: Mouton de Gruyter.
Heine, Bernd (1980). Determination in some East African languages. In Wege zur Universalienforschung: Sprachwissenschaftliche Beitrage zum 60. Geburtstag yon Hansjakob Seiler, Gunter Brettschneider and Christian Lehmann (eds.), 180-186. Tubingen: Narr.
Heine, Bernd (1997). Possession: Cognitive Sources, Forces, and Grammaticalization. Cambridge: Cambridge University Press.
Hengeveld, Kees (1989). Layers and operators in Functional Grammar. Journal of Linguistics 25(1), 127-157.
Hengeveld, Kees (1990). The hierarchical structure of utterances. In Layers and Levels of Representation in Language Theory: A Functional View, Jan Nuyts, A. Machtelt Bolkestein, and Co Vet (eds.), 1-23. Amsterdam: John Benjamins.
Hengeveld, Kees (2004a). The architecture of a Functional Discourse Grammar. In A New Architecture for Functional Grammar, J. Lachlan Mackenzie and Maria A. Gomez Gonzalez (eds.), 1-22. Berlin and New York: Mouton de Gruyter.
Hengeveld, Kees (2004b). Epilogue. In A New Architecture for Functional Grammar, J. Lachlan Mackenzie and Maria A. Gomez Gonzalez (eds.), 365-378. Berlin and New York: Mouton de Gruyter.
Hengeveld, Kees (2008). The noun phrase in Functional Discourse Grammar. In The Noun Phrase in Functional Discourse Grammar, Daniel Garcia Velasco and Jan Rijkhoff (eds.), 43-62. Berlin and New York: Mouton de Gruyter.
Hengeveld, Kees and Mackenzie, J. Lachlan (2008). Functional Discourse Grammar: A Typologically Based Theory of Language Structure. Oxford: Oxford University Press.
Hengeveld, Kees; Rijkhoff, Jan; and Siewierska, Anna (2004). Parts-of-speech systems and word order. Journal of Linguistics 40(3), 527-570.
Hewitt, B. George (1979). Abkhaz. Lingua Descriptive Studies 2. Amsterdam: North-Holland.
Josephs, Lewis S. (1975). Palauan Reference Grammar. Honolulu: University of Hawaii Press.
Karttunen, Lauri (1976). Discourse referents. In Notes from the Linguistic Underground, Syntax and Semantics 7, James D. McCawley (ed.), 363-387. New York: Academic Press.
Kimball, Geoffrey D. (1991). Koasati Grammar. Lincoln: University of Nebraska Press.
Laka, Itziar (1993). Unergatives that assign ergative, unaccusatives that assign accusative. In Papers on Case and Agreement L MIT Working Papers in Linguistics 18, Jonathan D. Bobalijk and Colin Davies (eds.), 149-172. Cambridge, MA: MIT Dept. of Linguistics.
Lakoff, George (1987). Women, Fire and Dangerous Things: What Categories Reveal about the Mind. Chicago: University of Chicago Press.
Lee, Kee-Dong (1975). Kusaiean Reference Grammar. Honolulu: University Press of Hawaii.
Lefebvre, Claire (1998). Multifunctionality and variation among grammars: the case of the determiner in Haitian and in Fongbe. Journal of Pidgin and Creole Languages 13(1), 93-150.
Lehmann, Christian (1995). Thoughts on Grammaticalization. Munich: Lincom Europa.
Levi, Judith N. (1973). Where do all those other adjectives come from? In You Take the High Node and I'll Take the Low Node: Papers from the 9th Regional Meeting of the Chicago Linguistic Society, April 13-15, 1973, Claudia W. Corum, Thomas Cedric Smith- Stark, and Ann Weiser (eds.), 332-345. Chicago: Chicago Linguistic Society.
Levinson, Stephen C. (1992). Primer for the field investigation of spatial description and conception. Pragmatics 2(1), 5-47.
Li, Charles N. and Thompson, Sandra A. (1981). Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press.
Lyons, John (1967). A note on possessive, existential and locative sentences. Foundations of Language 3, 390-396.
Lyons, John (1977). Semantics, 2 vols. Cambridge: Cambridge University Press.
Mackenzie, J. Lachlan (1987). The representation of nominal predicates in the fund. Working Papers in Functional Grammar 25, 1-18.
Matthews, George Hubert (1965). Hidatsa Syntax. The Hague: Mouton.
McCool, George J. (1984). The Old French demonstrative system: a semantic analysis. Word 35, 205-217.
McCool, George J. (1993). The French demonstrative system: from Old to Modern French. Word 44, 31-40.
McGregor, William B. (1997). Functions of noun phrase discontinuity in Gooniyandi. Functions of Language 4(1), 83-114.
Menges, Karl Heinrich (1968). The Turkic Languages and Peoples: An Introduction to Turkic Studies. Wiesbaden: Harrassowitz.
Merlan, Francesca (1983). Ngalakan Grammar, Texts, and Vocabulary. Pacific Linguistics B-89. Canberra: Australian National University.
Miner. Kenneth L. (1986). Noun stripping and loose incorporation in Zuni. International Journal of American Linguistics 52(3), 242-254.
Miner, Kenneth L. (1989). A note on noun stripping, International Journal of American Linguistics 55, 476-477.
Moravcsik, Edith (1997). Parts and wholes in the Hungarian noun phrase: a typological study. In Item Order in Natural Languages. Proceedings of LP '94, Bohumil Palek (ed.), 307-324. Prague: Charles University Press.
Moravcsik, Edith (2000). Infixation. In Morphologie: ein internationales Handbuch zur Flexion und Wortbildung: Morphology: An International Handbook on Inflection and Word- Formation: Volume 1, Geert E. Booij, Christian Lehmann, and Joachim Mugdan (eds.), 545-552. Berlin and New York: Walter de Gruyter.
Mosel, Ulrike and Hovdhaugen, Even (1992). Samoan Reference Grammar. Oslo: Universitetsforlaget AS.
Newman, Paul (1990). Nominal and Verbal Plurality in Chadic. Publications in African Languages and Linguistics 12. Dordrecht: Foils.
Nordlinger, Rachel and Sadler, Louisa (2004). Nominal tense in crosslinguistic perspective. Language 80(4), 776-806.
Prince, Ellen F. (1981). Toward a taxonomy of Given-New information. In Radical Pragmatics, Peter Cole (ed.), 223 255. New York: Academic Press.
Quirk, Randolph; Greenbaum, Sidney, Leech, Geoffrey; and Svartvik, Jan (1985). A Comprehensive Grammar of the English Language. London: Longman.
Ratliff, Martha (1991). Coy, the underspecified noun, and syntactic flexibility in Hmong. Journal of the American Oriental Society 111(4), 694-703.
Redei, Karoly (1978). Zyrian Folklore Texts. I. Gombos (trans.). Budapest: Akademia Kiodo.
Reh, Mechthild (1985). Die Krongo-Sprache (Niino Mo-di): Beschreibung, Texte, Worterverzeichniss. Berlin: Reimer.
Rijkhoff, Jan (1988). A typology of operators. Working Papers in Functional Grammar 29, 1-34.
Rijkhoff, Jan (1990). Explaining word order in the noun phrase. Linguistics 28, 5-,2.
Rijkhoff, Jan (1992). The Noun Phrase: A typological study of its form and structure. Unpublished doctoral dissertation, University of Amsterdam.
Rijkhoff, Jan (2002a). The Noun Phrase. Oxford: Oxford University Press.
Rijkhoff, Jan (2002b). On the interaction of linguistic typology and Functional Grammar. Functions of Language 9(2), 209 237.
Rijkhoff, Jan (2003). When can a language have nouns and verbs? Aeta Linguistica Hafniensia 35, 7-38.
Rijkhoff, Jan (2004). Iconic and non-iconic word order patterns: on symmetry in the NP and counterexamples to Universal 20'. In Words in their Places': A Festschrift for J. Lachlan Mackenzie, Henk Aertsen, Mike Hannay, and Rod Lyall (eds.), 169-180. Amsterdam: Free University Faculty of Arts.
Rijkhoff, Jan (2008a). Layers, levels and contexts in Functional Discourse Grammar. In The Noun Phrase in Functional Discourse Grammar, Daniel Garcia Velasco and Jan Rijkhoff (eds.), 63-116. Berlin and New York: Mouton de Gruyter.
Rijkhoff, Jan (2008b). Synchronic and diachronic evidence for parallels between noun phrases and sentences. In Interdependence of Diachronic and Synchronic Analyses, Studies in Language Companion Series 103, Folke Josephson and Ingmar Sohrman (eds.), 13- 42. Amsterdam and Philadelphia: John Benjamins.
Rijkhoff, Jan (forthcoming). On the co-variation between form and function of adnominal possessive modifiers in Dutch and English. In The Expression of Possession, The Expression of Cognitive Categories 2, William B. McGregor (ed.). Berlin and New York: Mouton de Gruyter.
Rijkhoff, Jan and Bakker, Dik (1998). Language sampling. Linguistic Typology 2(3), 263-314.
Rijkhoff, Jan and Seibt, Johanna (2005). Mood, definiteness and specificity: a linguistic and a philosophical account of their similarities and difl'erences. Tidsskrift jor Sprogforskning 3(2), 85-132. [http://ojs.statsbiblioteket.dk/index.php/tfs/index].
Roberts, John R. (1990). Modality in Amele and other Papuan languages. Journal of Linguistics 26, 363-401.
Ruhlen, Merritt (1991). A Guide to the World's Languages: Vol. 1: Classification. London: Arnold.
Sapir, Edward (1922). The Takelma language of southwestern Oregon. In Handbook of American Indian languages: Part 2, Bulletin 40 of the Smithsonian Institutiom Franz Boas (ed.), 1-296. Washington, DC: Bureau of American Ethnology.
Searle, John (1969). Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press.
Seiler, Hansjakob (1978). Determination: A functional dimension for interlanguage comparison. In Language Universals: Papers from the Conference at Gummersbach/Cologne, Germany Oct. 3-8, 1976, Hansjakob Seiler (ed.), 301-328. Tubingen: Narr.
Siewierska, Anna; Rijkhoff, Jan; and Bakker, Dik (1998). Appendix: 12 word order variables in the languages of Europe. In Constituent Order in the Languages of Europe, Empirical Approaches to Language Typology: Eurotyp 20-1, Arma Siewierska (ed.), 783-812. Berlin and New York: Mouton de Gruyter.
Stokes, Bronwyn (1982). A description of Nyigina: A language of the West Kimberley, Western Australia. Unpublished doctoral dissertation, Australian National University.
Stroomer, Harry (1987). A Comparative Study of Three Southern Oromo Dialects in Kenya: Phonology, Morphology and Vocabulary. Cushitic Language Studies 6. Hamburg: Buske.
Tauli, Valter (1966). Structural Tendencies in the Uralic Languages. Uralic and Altaic Series 17. Bloomington: Indiana University Press.
Taylor, John R. (1996). Possessives in English. Oxford: Oxford University Press.
Testelec, Yakov G. (1997). Word order in Kartvelian languages. In Constituent Order in the Languages of Europe, Empirical Approaches to Language Typology: Eurotyp 20-1, Anna Siewierska (ed.), 235-256. Berlin and New York: Mouton de Gruyter.
van Baarda, M. J. (1908). Leiddraad bij het Bestuderen van 't Galela'sch Dialekt, op het eiland Halmaheira [Manual for the study of the Galela dialect, on the island of Halmahera]. The Hague: Nijhoff.
von Heusinger, Klaus (2003). Cross-linguistic implementations of specificity. In Meaning through Language Contrast. Vol. 2, Katarzyna M. Jaszczolt and Ken Turner (eds.), 405-421. Amsterdam: John Benjamins.
Welmers, William E. (1973). African Language Structures. Berkeley: University of California Press.
Willemse, Peter (2005). Nominal reference-point constructions: Possessives and esphoric NPs in English. Unpublished doctoral dissertation, Catholic University of Leuven.
(1.) I am grateful to Christopher S. Butler and Miriam Taverniers (the editors of this issue), Bill McGregor and two anoymous reviewers for comments. Abbreviations: 1 = first person, 2 = second person, 3 = third person, ABS absolutive. ACC = accusative. ALLAT = allative, ART = article. ASP aspect, CLF = classifier, CN = connector, coll.sx = collective suffix. COMP = complementizer. DEF = definite, DEM = demonstrative, ERG = ergative, ES = ergative suffix, EXH = exhortative, FUT = future, GENR general tense-aspect-mood marker, IMPF = imperfective, INGR ingressive. IRR = irrealis. N = noun. LOC = locative, NEG = negative, NP = noun phrase, NONFUT = nonfuture, PAST = past, PERF = perfective, PL = plural, POS = possessive, PRT = participle, REL = relative clause marker, Rel.cl. = relative clause, sing.sx = singulative suffix, SG = singular. Correspondence address: Department of Linguistics, University of Aarhus, Ndr. Ringgade 1, Building 1410, DK-8000 Aarhus C, Denmark. E-mail: firstname.lastname@example.org.
(2.) On referring and predicating as part of an illocutionary act, see also Searle (1969), Mackenzie (1987), Dik (1997: 127f.).
(3.) The division between the descriptive and discourse-referential function corresponds more or less to information specified at the representational and the interpersonal level in Functional Discourse Grammar (Hengeveld 2004a, 2004b, 2008; Hengeveld and Mackenzie 2008). Note, however, that discourse-referential modifiers differ from modifiers specified at Hengeveld's interpersonal level in that they are only concerned with the pragmatic status of entities in the world of discourse. In Systemic Functional Grammar (Halliday 2004) the interpersonal component mostly serves to express social relations. See Butler (2003) for a detailed comparison of Dik and Hengeveld's Functional [Discourse] Grammar, Van Valin's Role and Reference Grammar, and Halliday's Systemic Functional Grammar.
(4.) There are, of course, also lexical ways to change or specify the kind of entity denoted by the noun, such as derivational morphology (cf. child vs. childhood); here I am only interested in grammatical expressions of such distinctions (cf. Rijkhoff 2002a on Seinsart). On the inflectional status of the Oromo affixes, see Clamons (2001: 532).
(5.) The phrase a woman's hat is potentially ambiguous, meaning either 'the hat of an unidentified woman' (nonclassifying) or 'a particular kind of hat' (classifying), see in this difference e.g., Taylor (1996: 665) and Willemse (2005); see also Section 3.4.
(6.) Cf. Quirk et al. (1985: 1339); on nonpredicative adjectives, see also e.g., Farsi (1968), Levi 1973).
(7.) The grammatical elements I erroneously regarded as qualifying operators in the "old" four-layered model presented in Rijkhoff (2002a: 101) have to do with "kind" rather than "quality".
(8.) Other good examples of interpersonal operators are markers of specificness, which can be found in many languages spoken in Africa and Polynesia (cf. Rijkhoff 2002a: 235; von Heusinger 2003). It is not clear whether genericity should also be regarded as operator category at the interpersonal level of the NP (Figure 3), one reason being that there does not seem to be a single language that has developed a special article (i.e., a "dedicated" marker) for generic reference.
(9.) Breban and Davidse (2003) present a detailed investigation of the semantics and pragmatics of "other".
(10.) A formal representation of the underlying structure of the noun phrase is given in Rijkhoff (2002a); see Rijkhoff (2008a) for a revised version.
(11.) I used a variety sample (Rijkhoff and Bakker 1998), which does not necessarily give a good indication of the relative frequencies of the patterns. As to the less frequently attested patterns in my sample: the pattern [dem num N A] is also found in the Romance languages. According to Hawkins (1983: 119) the pattern [dem N A num] is also found in Kabardian and Warao and the pattern [N A num dem] in Selepet and Yoruba, although it remains to be seen if we are really dealing with simple, integral NPs here.
(12.) In Rijkhoff (2002a) certain details about Bambara NPs were not available, but according to information supplied by Bernd Heine the basic order is [dem N A num], as in (the diacritics are: a = low tone, a = high tone, on all syllables of a word):
Bambara: (i) Nin jiri belebele saba these tree big three 'these three big trees' (Bernd Heine, p.c.)
There is also a marked alternative order with the demonstrative following the numeral:
Bambara: (ii) Jiri belebele saba ninnu tree big three these.PL 'THESE three big trees" (Bernd Heine, p.c.)
(13.) The deictic demonstrative gi and the definiteness marker re always appear together. Furthermore, the numeral may be preceded by du, whose meaning or function is unclear [Christopher Leone Daffalla, p.c.).
Zande: (i) Gi rarai a-lnangu (du) biata-re DEM heavy PL-box DU three-DEF/here 'these three heavy boxes' (Christopher Leone Daffalla, p.c.)
(14.) Recall, however, that Zande [dem A N num], Berbice Dutch Creole, Bislama, Sranan [num A N dem] and Sango [A N hum dem] all constitute counterexamples to the claim that the adjective never precedes the head when the demonstrative or numeral follow (Universal 20').
(15.) On the absence of a major class of true adjectives in many African languages, see e.g., Welmers (1973), Hagege (1974), and Bot Ba Njock (1977), Creissels (2000: 249).
(16.) See also Rijkhoff (2002a: 182 185). I will ignore cases in which nouns are borrowed with their definite article, since in such cases the article can be regarded as part of the noun (see e.g., Grant  on this phenomenon in creole languages). For example, English lariat (lasso) was originally la reata in Spanish and many Arabic nouns were borrowed into various European languages in tandem with the article al (alcohol, algorithm). Thus algebra derives from Arabic al-jabr 'restoration', 'completion' (cf. also in Don Quixote: algebrista 'bone-setter', 'restorer': Boyer [968: 251-253).
(17.) These properties are characteristic of erstwhile definite articles that have grammaticalized further (Greenberg 1978a, 1981, 1991 ): see also e.g., Laka (1993) and Frajzyngier (1997: 238, note 5) on the Basque "determiner" -a.
(18.) Furthermore, in Turkish the possessive suffix only seems to mark definiteness on terms that are not headed by a noun
(i) Bun-u iste-mi-yor-um, baska-sin-i ver this-ACC want-NEG-PRES-1S other-POS.3SG-ACC give 'I don't want this [one), give me the other one' (Gerjan van Schaaik, p.o.)
(19.) Using the Internet (Google), I checked all definite/indefinite combinations of the matrix NP headed by the Dutch noun fiets (de fiets 'the bike' and een fiets 'a bike') and definite and indefinite adnominal possessives headed by the nouns father, mother, husband wife, child, son, daughter, man, woman, boy or girl. As it turned out, localizing (i.e., definite) adnominal possessives are almost exclusively attested in definite matrix NPs (23,957 out of 23,988 cases, i.e., 99.87%,; see Rijkhoff [forthcoming] for details). See also Haspelmath (1999), who did a text count in several languages (English, Italian, Modern Greek) and found that adnominal possessives serve as (what 1 would call) localizing satellites in approximately 95% of the definite NPs.
(20.) On the relationship between location, possession and existence, see also Lyons (1967: 390), Clark (1978: 86). Christie (1970), Claudi and Heine (1986:316 317), Heine 11997: 47).
(21.) Halliday also draws a parallel between the clause and the NP (or rather between his "verbal group" and his "nominal group"), when he argues that the Finite element (i.e., tense) "is the verbal equivalent of the Deictic, relating the process to the 'speaker-now'." See Halliday (2004: 336) for the most recent statement of this parallelism.
(22.) This assumption may, however, reflect a Eurocentric perspective.
(23.) Notice that a tensed clause that is used to refer to a non-actual event specifies when an event was non-actual.
Nyigina: (i) marlu wa-la-ma-na burrula-nana NEG 3SG-NONFUT.IRR-go-PAST Derby-ALLAT 'He didn't go to Derby' (Stokes 1982: 248)
(24.) Elliott (2000:69 70) writes that there are four broad contexts that are "frequent targets for irrealis marking. These are (i) potential events, (ii) events whose occurrence is dependent on certain conditions being fulfilled (conditionals), including counterfactuals, (iii) events which are qualified by modality, and (iv) commands. There are a number of additional semantic contexts where, in at least some languages, irrealis marking has saliency. These are (v) negation, (vi) habituals, and (vii) interrogatives. The distinction between these various semantic contexts is not always clear cut, and there is frequently overlap in their description."
(25.) In order to explain symmetrical and antisymmetrical relations between (ir)realis and (re)definiteness we need to distinguish between specific and nonspecific indefinite reference. Consider the following examples, which show that there are two ways to continue Max wants a dog (Karttunen 1976). The difference is due to the fact that in (a) reference is made to a specific dog ('Max wants a certain dog, which is known to be black'), whereas in (b) the speaker does not refer to any particular dog ( English has no special article for specific or nonspecific reference unlike, for example, many Polynesian or sub-Saharan African languages).
(a) Specific-indefinite reference: Max wants a dog. It is black
(b) Nonspecific-indefinite reference: Max wants a dog. It must he black.
See, e.g., yon Heusinger (2003) on the connection between specificity and the "anchoring" or "grounding" of referents.
(26.) A relationship between discourse-referential modifiers in the NP and the clause is also attested in, e.g., Rapanui and Koasati. Rapanui he is used to mark both "indefinite action' (tense) and nonspecificity (Du Feu 1987, 1989, 1996). Koasati has live so-called article suffixes. These suffixes serve to "locate a noun in time and indicate that it was previously mentioned": -.:saya 'the aforesaid', -:yolli 'the very', -:ka "the long ago', -:kitla 'the former', and -o:to 'the deceased: the long ago' (Kimball 1991: 405).
Koasati: (i) am-okha-:kitta 1SG.POS-friend-ART 'my former friend' (Kimball 1991: 409)
The precise status of the "article suffix" is not quite clear. Kimball (1991 ) observes that four of the five article suffixes are identical to the participial suffixes (-.'shya 'present participle', -:yolli "habitual participle', -:ka 'preterite participle', and - :kitta 'imperfect participle'). There is no participial suffix corresponding to -o:to, nor is there an article corresponding to the future participle (cf. also Boas [1911: 39] on "'tense classes on nouns" in American Indian languages and Nordlinger and Sadler  on nominal tense).
(27.) This antisymmetry may point to a fundamental difference between first and second order (spatial and temporal) entities (Rijkhoff 1988: 25; see also Harder 1996: 280; Rijkhoff and Seibt 2005).
Table 1. Symmetry between definite/realis and nonspecific-indefinite/ irrealis Noun phrase Clause (thing) Occurrence in the world of discourse (event) definite thing or event (already) has a location in realis the discourse world, i.e., the entity is 'grounded' nonspecific- thing or event does not have a location in irrealis indefinite the discourse world (yet). i.e., the entity is not 'grounded' Table 2. Anti-symmetry between definite/realis and speck-indefinite/irrealis Number of reasons for an entity to be Noun Phrase marked as (specific-in) definite Clause (thing) or (ir)real (event) definite many one one (grounds itself) realis specific- (grounds itself) many irrealis indefinite
|Printer friendly Cite/link Email Feedback|
|Publication:||Linguistics: an interdisciplinary journal of the language sciences|
|Date:||Jul 1, 2008|
|Previous Article:||The status of purpose, reason, and intended endpoint in the typology of complex sentences: implications for layered models of clause structure.|
|Next Article:||Functional discourse grammar--multifunctional problems and constructional solutions.|