|Home||Feature Inventory||Related Issues||Search||Contact|
The semantic category corresponding the most closely to the central function of grammatical 'definiteness' is identifiability - that is, the expression of whether or not a referent is familiar or already established in the discourse. C. Lyons (1999:278) observes that "[i]n languages where identifiability is represented grammatically, this representation is definiteness; and definiteness is likely to express identifiability prototypically" (note, however, that there may be instances of identifiability, such as generics, which are not treated in a given language as definite; Lyons 1999:278). As with other grammatical categories, it is also to be expected that there are other uses of definiteness which do not relate to identifiability - one of such uses is inclusiveness (a term due to Hawkins 1978), which is particularly appropriate for non-referential uses of definiteness with plural and mass noun phrases. Inclusiveness expresses the fact that the reference is made to the totality of the objects or mass in the context which satisfy the description (C. Lyons 1999:11).
Definiteness as a semantic and/or pragmatic concept has been a subject of much debate among both linguists and philosophers, within various theoretical frameworks. Apart from identifiability and inclusivenes, other concepts which have been considered significant for the understanding of definiteness include familiarity, uniqueness, and reference (see C. Lyons 1999 for a thorough overview of the relevant research traditions). Furthermore, in the discussion of the nature of definiteness, various other distinctions are drawn in addition to that between definite and indefinite, among them specific versus non-specific, and deictic versus non-deictic; specificity and deixis may interact with definiteness affecting its grammatical realisation.
In sum, it is to be expected that there will be considerable variation between languages in the use of the grammatical category of definiteness. C. Lyons (1999:278) gives the following examples of variation that is found: some languages require generics to be definite while others do not; in some languages definiteness is optional even in noun phrases clearly interpreted as identifiable (e.g. in Hausa); in languages like Maori which show an article combining obviously 'definite' (i.e. identifiable) uses with something akin to specificity, this article too can be treated as encoding definiteness - in this case certain types of noun phrase occurrence which in other languages are treated as indefinite are grammatically definite (see also §5 below). However, despite all this variation, there is always a central core of uses of grammatical definiteness which relates directly to identifiability.
Definiteness is a category of the noun phrase. Following C. Lyons (1999:278), I assume that the semantic/pragmatic concept of identifiability underlying grammatical definiteness is probably universal (see also Lambrecht 1994 on identifiability as an information structure element). It is demonstrable that a 'definite' interpretation plays an important part even in languages which show no formal marking of definiteness. For example, in Mandarin, a noun phrase in subject position must be a topic and therefore 'definite', while a noun phrase in the existential construction must be understood as 'indefinite'. Therefore, even though definiteness is not a formal category in Mandarin, it is nevertheless an element of discourse organisation which corresponds to the identifiability of the referent.
Thus, not all languages have a grammaticalised concept of definiteness. Definiteness as a grammatical category is only present in languages which show some overt marking of definiteness, for example some kind of a definite article. Since definiteness may be thought of as only one of a number of categories which serve to guide the hearer in working out how the discourse is structured and how entities referred to fit into it, definiteness marking is not essential to communication. Some languages which do not grammaticalise definiteness may be argued to compensate by marking other distinctions with a similar function (e.g. topic and focus - as exemplified above) (C. Lyons 1999:48). Furthermore, many languages have grammaticalised definiteness only in pronominal, but not full, noun phrases. C. Lyons (1999:280) offers the following typology of languages with regard to the grammaticalisation of definiteness:
Definiteness can be encoded using a wide range of lexical, syntactic and morphological devices. C. Lyons classifies definiteness encoding broadly into two categories: 'simple' and 'complex'. 'Simple' definiteness encoding occurs when the definite and indefinite noun phrases are marked with some type of article which are either affixes or free-form determiners (see C. Lyons 1999:47-106 for comprehensive discussion, examples and references; see Dryer 2005a-b for classification and information on geographical distribution of languages with various types of definite and indefinite articles). 'Complex' definiteness encoding occurs when the definiteness of the noun phrase is due to something other than presence or absence of an article; the range of encoding methods includes proper nouns, personal pronouns, and noun phrases containing a demonstrative or possessive modifier (C. Lyons 1999:107-156).
Most commonly, definiteness is not a feature by our definition, but an additional piece of information selected for the noun phrase that may be expressed through a free-form determiner, an affixal marker, or a clitic. This information may be expressed more than once within the noun phrase. See C. Lyons (1999:77-85) for examples and discussion of the so-called 'double determination' in Hausa, Ewe, Danish, written Icelandic, Albanian, and Romanian, and definite adjectives in Arabic, Albanian, Romanian and a number of Slavonic, Baltic, and Germanic languages. Also, see Corbett (2006:135-137) for discussion of multiple marking of definiteness in Modern Hebrew, Maltese, and Norwegian.
It has been suggested that in some languages the definiteness of a noun phrase is expressed by an agreement marker elsewhere in the sentence. Several of the Uralic languages, for example, have been cited as having object-verb agreement in definiteness (e.g. C. Lyons 1999:86-87, 207-208). However, Corbett (2006:91-93) argues that definiteness in Hungarian is a condition on agreement forms, not an agreement feature. The verbal suffix in Hungarian has a distinct form when it occurs in the environment of a definite object, and three contrasting analyses of this phenomenon have been suggested: (1) that the suffix is a fused subject (person, number) agreement and object (definiteness) agreement marker; (2) that the verb agrees with its object under the condition of definiteness, but otherwise it does not; (3) that Hungarian verbs do not agree with their objects, but rather they take a special kind of subject agreement in the presence of 'definite' objects and then only with third person objects (except for the 1.SUBJ-2(familiar) OBJ marker -lek/-lak); that is, there is subject agreement which has a realisation conditioned by the presence of a 'definite' object. Corbett favours the analysis of defineteness as a condition on agreement and points out that a similar analysis may also be appropriate for another, rather complex, instance of subject agreement having different realisations according to the presence of a 'definite' object in Muna (Austronesian), as described by van den Berg (1989:59-60).
Despite the fact that in most instances definiteness is not a value of a feature, and perhaps at best it might occur as a morphosemantic feature, we have come across one instance where we need to posit definiteness as a morphosyntactic feature. In German, in order to describe nominal inflection, we need gender, number, and case. However, in order to describe adjectival inflection, after separating out gender, number and case, we are still left with three different adjectival paradigms, referred to as 'strong', 'mixed' and 'weak'. An adjective inflecting according to the strong paradigm shows full agreement features. The following is an example listing the strong paradigm for gut 'good' (all examples from Corbett 2006:95-96, and the discussion follows in part Zwicky 1986):
The mixed paradigm, exemplified below, shows partially reduced agreement. It shares some forms with the strong paradigm (these are marked '(S)'), and some with the weak paradigm (these are marked '(W)'). The remaining forms (unmarked) are shared across all three paradigms:
Finally, the following is the weak paradigm for the same adjective. The weak paradigm shows reduced agreement:
Corbett notes that, as we progress from the strong paradigm to the weak, in each there are fewer distinct inflections (five in the strong, four in the mixed, and two in the weak). 'However, the sets of cells which are distinguished in the strong paradigm are not simply collapsed: the weak paradigm has different forms for the feminine singular and the plural, which are identical in the strong paradigm' (2006:96).
Therefore, we have to treat the choice of one of the paradigms as a choice of one of three distinct options, perhaps values of a feature. What dictates the choice of the paradigm for the adjective is the type of element in the determiner position in the phrase. The choice of the adjectival paradigm correlates with the choice of the determiner in the following way:
The correlation can be understood in terms of definiteness, even though there is no unique marker of definiteness in German on adjectives - instead, definiteness is expressed through the choice of the determiner and the selection of inflectional endings on the adjective. One way of analysing definiteness in German nominal phrases would be to see it as assigned to the whole phrase together with the (optional) determiner. However, we still have to account for the selection of the adjectival paradigm, and the observed correlations suggest strongly that we should recognise a morphosyntactic feature. It is not completely clear, however, whether we are dealing with agreement or government.
Zwicky (1986:984-987) analyses it as government: the determiners govern the feature of definiteness on the adjectives by requiring the selection of a particular type of adjectival paradigm. The questions which arise are: if it is definiteness that is the governed feature, we should not expect to find its values on the governors. Furthermore, apart from saying that the particular determiners require the selection of the particular adjectival paradigms, it is difficult to characterise this feature in terms of its values. The best characterisation that can be given is: the definite articles govern the 'weak', or 'reduced' value of the feature of definiteness, and the indefinite articles govern the 'mixed', or 'partially reduced' value of the feature of definiteness. It appears that the appropriate assessment of this analysis has to be postponed until we have a theory of syntactic government comparable to the theory of canonical agreement proposed by Corbett (2006).
The alternative view, which is adopted here tentatively (and is open to further argumentation), is to analyse the correlations as agreement: there is systematic covariance between the controllers (the two types of determiners: definite and indefinite) and the targets (the adjectives). The exponence of definiteness on the adjectives is non-autonomous, but it is expressed through the selection of the required adjectival paradigm. In each case, the result is a particular pattern of distribution of information relevant to the concept of definiteness throughout the phrase. It seems easy to accept that definite and indefinite articles themselves express a value of definiteness (and act as controllers of agreement on adjectives), even though, on this view, we also have to accept the fact that German adjectives agree in number and gender with one controller (the noun), but in definiteness with another (the determiner). Comments on this issue will be very welcome.
The fact that the values of the proposed feature may not always correspond to (in)definiteness semantically does not pose a problem to give the feature the label 'definiteness'. In a parallel way, what is labelled as 'gender' often does not correspond to a semantically assigned gender or class. The feature of 'gender' does have a semantic core, or basis, but there are few languages with purely semantically assigned gender values. Similarly, definiteness in Germanic has some semantic basis, but we do not necessarily expect it to be semantically assigned throughout.
It would seem that definiteness feature should have two values: definite and indefinite. However, C. Lyons (1999:49-51) points out that, in languages which distinguish simple definites and indefinites, the correct analysis of definiteness marking may be that only definiteness is directly encoded. In such cases, we regard definiteness as an additional piece of information selected for the noun phrase, not a feature value by our definitions (see the 'Feature Inventory' page for clarification). What complicates the analysis is the phenomenon of quasi-indefinite cardinal articles, such as the English a and the reduced some (often conventionally referred to as sm, e.g. C. Lyons 1999:34). Although it can be argued that a and sm are cardinality words, not indefinite articles, they "do indirectly signal indefiniteness while not encoding it: a is obligatory in singular indefinite noun phrases in the absence of any other determiner, and neither a nor sm ever appears in definite noun phrases. This 'indirect signalling' of indefiniteness by a cardinality determiner, leading to a strong intuition that it contrasts with definite determiners, is widespread" (C. Lyons 199:48-49).
Hence, although potentially there are three ways in which definiteness distinction can be expressed in languages which distinguish simple definites and indefinites (C. Lyons 1999:49):
For a discussion of value options available for the only possible instance of morphosyntactic definiteness identified so far, necessary for the description of the adjectival paradigms in German, see §3 above. In German, the choice of the definiteness value for a noun phrase affects the distribution of information relating to definiteness over the noun phrase.
The inflectional markers which appear on attributive elements in German, traditionally described as case, gender and number agreement ('CGN') markers, but arguably also encoding definiteness in a non-autonomous way (see §3 above for argumentation), are typically found on attributive adjectives or participles and relative pronouns. However, they are also capable of attaching to at least some nouns and prepositions. If the inflectional marker is treated as a portmanteau for case, gender, and number, it is perhaps surprising that nouns (as in examples 1a and 1b) inflect for gender, and prepositions (as in examples 2a and 2b, of German found in the Rhein area) inflect at all (Struckmeier 2007a:2, with his annotation):
Modern Hebrew has a definiteness marker -ha which can occur on most of the elements of the noun phrase. The most interesting examples of a complex mechanism determining the marking of definiteness within the noun phrase in Modern Hebrew are found in the so-called 'construct states', which are noun-noun constructs. See Corbett (2006:135-136) for a brief overview of this phenomenon analysed as multiple marking of definiteness within the noun phrase, and Danon (2001; 2010) for comprehensive discussion of the distribution of definiteness marking within the Hebrew construct state, including an overview and evaluation of theoretical syntactic analyses which have been proposed to model this phenomenon. Danon also offers argumentation in support of treating definiteness in Modern Hebrew as a 'monovalent feature whose presence alternates with a lack of specifi¯cation' - which corresponds to our analysis of definiteness as an additional piece of information selected for the noun phrase.
Articles marking specificity, or something close to specificity, rather than definiteness are fairly widespread (C. Lyons 1999:59). However, sometimes the meaning and use of articles in a language may be based on still other distinctions. For example, Maori nominals introduced by the article he are usually interpreted as indefinite or nonspecific, but it is doubtful whether either definiteness or specificity offers a consistent enough basis for the distinctions observed in the use of this article (Bauer 1993; C. Lyons 1999:58-59). Syntactically, the he nominal cannot occur as subject of a transitive clause or as direct object; it is only allowed as subject of an active intransitive clause or subject of a passive clause. The interpretation of the article includes generic, indefinite, semidefinite (as in the English a friend of mine), and even definite. Polinsky (1992) suggests that the description of he should be maintained in terms of the opposition referential/nonreferential, rather than definite/indefinite. Thus, she suggests that the major distinction is between indication of an individuated entity (referential) and indication of a class of entities or any entity in this class or property of the class (nonreferential). The nouns introduced by he are not specified for referentiality; this is consistent with their typical use in the predicative function. Being unspecified for referentiality, nouns marked by he are primarily nonreferential; however, if the noun is conceptually related to some referential nominal in the clause, it can acquire the referential interpretation, due to and contingent upon the referential interpretation of the other noun. To explain the syntactic distribution of he nouns, Polinsky introduces the category of localisation. This category implies that individuated events involve individuated participants; the primary individuation is associated with the active participant in a situation. Thus, if the agent in a transitive clause is individuated, the patient or undergoer in the same clause also has to be individuated. This explains the inability of a he nominal to occur as direct object, in the presence of a referential transitive subject.
Definiteness in German: a feature of agreement or government? The descriptive generalisation regarding German noun phrases appears to be that in the sequence determiner-adjective-noun, the inflectional material appears on the leftmost modifying element, whether it is a determiner or an adjective, except that the cardinal article ein, kein 'no/ not a', or a possessive cannot accept certain inflections. In such cases, the inflections appear further to the right. C. Lyons concludes (1999:219) that "[t]his means that articles have no special status in the realization of inflectional features". However, the identification of the 'mixed' paradigm of adjectival inflection (discussed in §3 above) suggests that the picture is more complex: the 'indefinite' articles and possessive pronouns do not just reject inflections, but they also induce adjectives to appear with or without inflection depending on the case the phrase is in. The question remains, however, about exactly how the inflectional paradigm for the adjective is selected: through government by the determiner (as suggested by Zwicky 1986 and Corbett 2006), or through agreement with the determiner (as tentatively suggested in this entry). An additional complication that has to be considered is that the distribution of strong versus mixed paradigm adjectives is not completely clear, especially when there are two adjectives in the phrase. For a recent discussion of attributive agreement in German, see Struckmeier (2007b).
Definiteness in tense-aspect distinctions? While discussing parallels between tense and (definite) pronouns, Partee (1984) observes that just as pronouns can relate to a referent introduced in the previous discourse or to a referent understood on the basis of the context, so tense can relate to an antecedent time or to an understood time. C. Lyons picks up on this observation and suggests that it may be possible to attribute some tense-aspect distinctions to definiteness (1999:45-46). Traditional grammars sometimes describe the distinction between the past (historic/preterite) tense, as in I read that book, and the corresponding perfect tense, as in I have read that book, in terms of a distinction between 'definite' and 'indefinite' past. C. Lyons argues that the preterite indeed makes a definite time reference: "[i]n the absence of a time adverbial which identifies the time of the event (I read that book yesterday), the hearer is assumed to be able to locate the event temporally on the basis of contextual knowledge. The perfect also presents the event as past, but, while the speaker may know when she read the book, there is no implication that the hearer knows or can work out (or needs to) when the event occurred" (1999:45-46). Hence, this disctinction is parallel to that between the car and a car, except that referent identification is substituted with event identification, or, more precisely, with the identification of the temporal location of the event. Note that this observation is consistent with the analysis of the semantics of tense offered in this Inventory (see the entry on 'Tense'), captured with Reichenbachian primitives. On the view offered in this Inventory, all simple tense meanings have a semantic component that can be expressed as the simultaneity of the event time and reference time (E=R), while the perfect meanings have a semantic component that can be expressed as the dissociation of the event time and reference time (E ≠ R) (Kibort 1997). See the entry on 'Tense' (§4) for more details.
'Degree modifiers' as 'determiners'? Since the structural position of determiners in the noun phrase is paralleled in the adjective phrase and in adverb and quantifier phrases by 'degree modifiers' (as/so/that/too [big]), C. Lyons (1999:46) hypothesises that these words can also be treated as being of category 'determiner' (analysed as either a specifier of the relevant phrase, or a head of a functional phrase containing the adjective/quantifier phrase). He points out that both that and this operate both as definite determiners in noun phrases and as degree modifiers in adjective phrases, e.g. Tom is stupid but not that stupid, The fish I almost caught was this big!. As degree modifiers, they are colloquial, though they have a more formal counterpart: so. C. Lyons argues that "there is little reason to doubt that this and that have demonstrative meaning in this use; the degree they convey of the property expressed by the adjective is accessed anaphorically [in Tom is stupid but not that stupid] or communicated by means of an ostensive gesture [in The fish I almost caught was this big!], exactly parallel to what happens with noun phrase demonstratives" (1999:46). So must also be a demonstrative degree modifier, though it lacks the deictic distinction. As a candidate for a simple definite degree modifier C. Lyons offers as, as in: (a) Joe is as bright, (b) Joe is as bright as Ann, and (c) Joe is not as bright as you think. In such contexts, as is phonologically weak, with a normally reduced vowel, and its use is close to that of the: in (a) the degree of brightness referred to is accessed by the hearer from the context or the preceding discourse, and in (b) and (c) it is provided in a relative-like modifier (C. Lyons 1999:46).
Are demonstratives universally definite in meaning? This question is taken up by C. Lyons in several places in his (1999) monograph on definiteness. Demonstratives are generally considered to be definite, but it is clear that their definiteness is not a matter of inclusiveness. Identifiability is only part of their semantic content, another part being deixis. In (1999:17-21), C. Lyons argues that "a demonstrative signals that the identity of the referent is immediately accessible to the hearer, without the inferencing often involved in interpreting simple definites", because "the work of referent identification is being done for the hearer by the speaker". This suggests that demonstratives are necessarily definite. C. Lyons (1999:151-152) then considers a possible counterexample in the form of such, arguably an indefinite variant of the demonstrative, but rejects this analysis in favour of treating such as a non-demonstrative containing a demonstrative element as part of its meaning ('of this/that kind', or 'like this/that'). This means that the assumption that demonstratives are inherently definite is maintained, which implies that definiteness exists in some form in all languages (1999:107).
Definiteness is not inherent in possessives. Although, as argued briefly above, demonstrativeness seems to be semantically incompatible with indefiniteness, C. Lyons argues that definiteness is not inherent in possessives (1999:22-26; 124-133). This runs counter to the traditional assumption that possessives are definite determiners, found in many descriptive grammars and in much theoretical work. However, it is demonstrable that, while in some languages a possessive imposes a definiteness interpretation in the matrix noun phrase, in other languages it does not. This difference is now discussed in terms of a typological distinction between 'DG languages' (e.g. English, Irish), and 'AG languages' (e.g. Italian, Greek) (where 'DG' and 'AG' stand for 'determiner-genitive' and 'adjectival-genitive', even though it is not claimed that possessives are necessarily determiners in the first type and adjectives in the second type of language; C. Lyons 1999:24).
Definiteness and animacy. See C. Lyons (1999:213-215) for an interesting discussion of definiteness and animacy with respect to the widely used 'animacy hierarchy' which helps express cross-linguistic generalisations.
How to cite this entry: