1. PREAMBLE
Many Danish linguists doing language-particular analysis use the terms ‘marked’ and ‘unmarked’ in the manner originally suggested by Jakobson and Hjelmslev in the 1930s (e.g. Bjerrum Reference Bjerrum1966; P. A. Jensen Reference Jensen1985; Basbøll Reference Basbøll2003, Reference Basbøll2008; Hansen Reference Hansen2006:18–19).
Acknowledging the fact that the term ‘markedness’ in the course of history has come to be used in several competing ways, and that a common understanding of the theory cannot be taken for granted, in recent years there has been a renewed effort among Danish linguists to make explicit just how the markedness theory of the early works by Jakobson and Hjelmslev is conceived. In a number of papers, Henning Andersen gives an outline of markedness theory in a synchronic as well as a diachronic perspective, and he points to contributions by Hjelmslev, in particular the notion of participation, as particularly fruitful in linguistic analysis (e.g. Andersen Reference Andersen2001a, b). Christensen (Reference Christensen2002, Reference Christensen2007) integrates the theory in her work on ‘hyperparadigms’. Jensen (Reference Jensen, Jørgensen and Widell2007, Reference Jensen, Jakobsen, Jansen, Jensen and Strudsholm2009, Reference Jensen2011) makes explicit reference to the Jakobson–Hjelmslev markedness theory in her works on grammatical analysis, and, taking these last-mentioned studies as the point of departure, Petersen (Reference Petersen2010) renders additional evidence for the Jakobson–Hjelmslev theory of markedness being useful in the description of grammar. The present paper seeks to make a further contribution to these efforts and to demonstrate how the theory of markedness as presented here is useful in the analysis of grammatical paradigms in Modern Danish. Thus, the work in this paper may be seen as an example of ‘categorial particularism’ (Haspelmath Reference Haspelmath2010), i.e. as language-particular analysis whereby the understanding of the relationship between categories within an individual language is pursued.
On a theoretical level, there are two important points in this paper. The first is to show how the notion of participation is a necessary prerequisite to the Jakobson–Hjelmslev markedness theory. The second is to discuss points of critique in order to refine the theory. This last point pertains to the notion of Gesamtbedeutung (Jakobson Reference Jakobson1936) and Begrebszone (Hjelmslev Reference Hjelmslev1934 [1972], Reference Hjelmslev1939).Footnote 1
The main body of the paper is divided into three major parts. Section 2 contains a historically-based discussion of the theories of markedness, aiming at reinforcing the notion of participation as a necessary part of markedness theory. Section 3 presents a number of cases where the Jakobson–Hjelmslev markedness theory has proved useful as a means of analyzing grammatical paradigms in Danish. In Sections 4 and 5, points of criticism are discussed, leading to a suggestion concerning how to further refine the theory.
2. STATE OF THE ART. . .?
‘Marked’ and ‘unmarked’ are terms often found in works on linguistics (as well as other disciplines such as anthropology and cultural studies). The terms have an immediate appeal to many people, often resulting in uses that are short of definition and criteria, and leaving the linguist at a loss as to how to use them as linguistic terms proper. There is, unfortunately, a lack of consensus among the linguists who do use the words as real linguistic terms as to how to define and use the notion of markedness and the terms marked and unmarked. Croft identifies (at least) three quite different uses, the early European approach, the approach found in generative linguistics, and finally what he calls ‘typological markedness’ (Croft Reference Croft2003:87); Haspelmath (Reference Haspelmath2006) lists no fewer than 12 senses of markedness, some of which are mutually incompatible.
Some scholars study linguistic categories in order to uncover presumed universal properties of categories such as definiteness, number, tense, case, etc. As a part of this endeavor, the members of the categories under scrutiny are placed in implicational hierarchies of markedness, like Oj/NSpec > Oj/Spec > Oj/Def > Oj/PN > Oj/Pro, thus allowing for formulations like ‘personal pronouns are the most marked type of object, and non-specific indefinites, the least marked’ (Aissen Reference Aissen2003:445). In this approach, markedness relations are considered relative values and the categories of the paradigm belong to a continuum (see also Croft Reference Croft2003:127ff.).
Others, while basing their understanding of markedness on binary oppositions, by no means do so on uniform grounds. Givón, to mention one, determines which member of a grammatical paradigm is marked on the basis of hypotheses of information processing as one criterion.Footnote 2 According to this author negative clauses, for example, are marked in comparison with affirmative clauses, as the former hold a higher degree of cognitive complexity and are thus comparatively more difficult to process than the latter (Givón Reference Givón1995:33, 42–43). Lyons, to mention another, bases his work on sense relations on lexical specificity. Taking pairs of words like ‘lion’: ‘lioness’, ‘dog’: ‘bitch’, etc. as his point of departure, he writes: ‘A semantically marked lexeme is one that is more specific in sense than the corresponding semantically unmarked lexeme’ (Lyons Reference Lyons1977:307). Thus, the lexemes ‘lioness’ and ‘bitch’ are the marked members, being the ones of the word pairs with the more specific sense.
However, even though many linguists take pains to explain their understanding of the terms ‘marked’ and ‘unmarked’, great diversity – to the point of general confusion – as to how to define and use the notion of markedness may be observed, and Haspelmath (Reference Haspelmath2006) can hardly be blamed for airing his frustrations about the many ways the terms are used. Instead of abandoning the notion of markedness altogether, as Haspelmath suggests, I shall make an attempt at showing the usefulness of early structuralist markedness theory in the analysis of grammatical paradigms, and as in the original structuralist works on markedness theory, the analyses offered here are language-particular. In order to distinguish this interpretation of markedness theory from other interpretations, I shall try to extract certain patterns in the ways the terms are used, and with a point of departure in early works of Trubetzkoy, Jakobson and Hjelmslev, I shall show how the patterns stem from these works. I shall then focus on Jakobson's and Hjelmslev's early criteria and how these are useful in analyzing grammatical paradigms of binary oppositions.
2.1 The legacy from early European structuralism
In works that acknowledge a legacy of markedness theory, there is a noticeable recurrence of certain words and observations in the attempts to define the notion of markedness. Definitions and characterizations frequently include such observations and criteria as the following: marked forms are less frequent than unmarked; marked forms have more properties than unmarked; marked forms have a narrower range of use than unmarked; marked forms are more complex than unmarked; marked forms are more prominent than unmarked.
Even though one particular source of some of the most influential approaches to markedness in modern linguistics is easily identified in Joseph Greenberg and his work on universals (Greenberg Reference Greenberg and Greenberg1963, Reference Greenberg1966), the influence of all of these observations and criteria may be traced back to early work on structural linguistics as developed by members of the Linguistic Circles of Prague and Copenhagen in the 1930s. Many linguists make explicit reference to the Prague School of structural linguistics (see Bußmann 1983:310; Givón Reference Givón1995:26ff.; Andersen Reference Andersen2001a, b; Christensen Reference Christensen2002:18f.; Croft Reference Croft2003:87ff.). In particular Trubetzkoy and his work on phonology are often cited as a source of inspiration.Footnote 3 In the American literature, the reference to Trubetzkoy is often to the English translation (Trubetzkoy Reference Trubetzkoy1969) of his monograph on phonology (Trubetzkoy Reference Trubetzkoy1939). But part of the confusion about markedness stems, on the one hand, from Trubetzkoy's (Reference Trubetzkoy1931) seminal paper and the equally influential paper on Russian verbs published the following year by Jakobson (Reference Jakobson1932), and, on the other hand, from the ways the insights of these works were passed on in later works, including some by Jakobson himself (see also Battistella Reference Battistella1996:56ff.).
Trubetzkoy and Jakobson were both members of the Linguistic Circle of Prague, founded in the 1920s, a society for debate and discussion concerning contemporary linguistic ideas. In his 1931 paper, Trubetzkoy aimed at constructing an outline for the description of phonological systems based on consistent criteria. Instead of looking at the different phonemes as entities unrelated to other phonemes, he adopted a modular approach whereby each individual phoneme could be seen as holding a number of properties, these properties to be further identified as voiced: unvoiced, rounded: unrounded, etc. Any phoneme could then be analyzed as being part of a larger system where each specific phoneme would differ from another phoneme with respect to one or more properties, and at the same time would be similar with respect to some other properties. In the course of describing the relevant properties, Trubetzkoy wrote the following (since then much cited) passage on what he called privative oppositions:
Die zwei Glieder eines korrelativen Gegensatzes sind nicht gleichberechtigt: das eine Glied besitzt das betreffende Merkmal (oder besitzt es in seiner positiven Form), das andere besitzt es nicht (oder besitzt es in seiner negativen Form). Wir bezeichnen das erste als merkmalhaltig, das zweite – als merkmallos. (Trubetzkoy Reference Trubetzkoy1931:97)Footnote 4
Almost immediately after Trubetzkoy's seminal paper was printed, Jakobson published his seminal paper on Russian verbs (Jakobson Reference Jakobson1932), actually using the above passage from Trubetzkoy (Reference Trubetzkoy1931) as the opening lines of his own paper. Although Jakobson also worked on phonology, his paper was not on this subject. Rather, it was a certain way of thinking in Trubetzkoy's work corresponding to some of his own observations on semantics and grammar, in particular the properties of the Russian verb forms, that inspired Jakobson to publish his own paper.Footnote 5 With due reference to earlier research, Jakobson mentioned how in the previous century Russian linguists had noticed that the relationship between two members of a morphological paradigm was not symmetrical. In many cases the one form would possess a property (thus being ‘merkmalhaltig’, see Trubetzkoy Reference Trubetzkoy1931:97) which the other form would not (being ‘merkmallos’, see Trubetzkoy Reference Trubetzkoy1931:97). Thus, the perfective in Russian would be merkmalhaltig, and the imperfective merkmallos as regards the property of what might be called ‘demarcation’. The perfective would signal a situation with a beginning or an ending, whereas the imperfective would not convey such a signal. Jakobson did not (at this point in time) use the terms ‘marked’ and ‘unmarked’.Footnote 6 They only appeared when he had embarked on his American career. But the groundwork of what might be called the early (or the first version of) Jakobson's markedness theory was laid in this 1932 paper on Russian verbs and in the further developments in his 1936 paper on Russian case morphology.
At the time, some of the scholars of the Linguistic Circle of Prague and of the Linguistic Circle of Copenhagen were interested in one another's work, and Hjelmslev contributed greatly to the early theory of markedness.Footnote 7 In particular his lectures from 1934 on linguistic theory in a synchronic as well as a diachronic perspective (unpublished until 1972), his works on case (Hjelmslev Reference Hjelmslev1935, Reference Hjelmslev1937), and his paper on categories (Hjelmslev Reference Hjelmslev1939) contain important insights and refinements, adding to the precision of the theory of markedness.
2.2 Participation
One of the most important parts of the theory of markedness in the Jakobson–Hjelmslev tradition is the notion of participation.Footnote 8 In the 1932 paper, Jakobson makes the link from Trubetzkoy's phonological work to Russian verbal morphology through a lexical example. Two words, for example duck and drake, may not have the same distributional pattern in English.Footnote 9 The word duck may be used to denote both the female of the species and also the species itself; it may even be used to denote a male member of the species! Drake, on the other hand, has limited use – it is only applicable to male members of the species. Now, this distributional pattern results in a paradigm where the two members are indeed not symmetrical, but furthermore – and this is the Jakobson–Hjelmslev insight that enables the whole theory of markedness to be useful in linguistic analysis – the asymmetrical relationship between the two members is of a nature where one member covers a certain part of a domain, whereas the other covers the entire domain including the domain covered by the first member. This has been illustrated graphically in a number of ways. Andersen (Reference Andersen2001a:43) and Jensen (Reference Jensen, Jørgensen and Widell2007, Reference Jensen2011) draw boxes within boxes and they use inclusion brackets. In the rest of this paper, I use inclusion brackets to convey this participative nature of the relationship between the two members of a paradigm, as in (1):
(1)
It could be said that the Jakobson move was to introduce the notion of ‘usability’ or ‘domain of use’ into the theory. Whereas for Trubetzkoy the main interest lay in creating a set of terms and criteria in order to make adequate and consistent descriptions of phonological systems, Jakobson was also interested in the distributional patterns of grammatical forms, and he made a point of the fact that, for instance, the verbal form of the present in Russian had a wider distributional range than the preterite due to the different semantics of the forms (Jakobson Reference Jakobson1932:8). Hjelmslev, too, saw the ‘usability’ or ‘domain of use’ of the forms as central to the theory of markedness, for instance exemplified by adjective forms in French, where the positive covered a wider domain than the comparative-superlative (Hjelmslev Reference Hjelmslev1934 [1972]:70–71, 77–80).Footnote 10
Although Jakobson made explicit reference to Trubetzkoy and his observation that one member of a paradigm has something which the other has not, Jakobson's original contribution to the theory might, in fact, be phrased differently, namely that one member can be used in contexts that the other cannot.
What took place as a consequence of Jakobson's move was, in fact, that two different – in some ways interrelated – interpretations of marked and unmarked (i.e. ‘merkmalhaft’ and ‘merkmallos’) were now current. They might be called the ‘has-something-more’ and the ‘can-something-more’ interpretations of markedness.
These two interpretations are not incompatible. In fact, with a point of departure in (1) above, it might be said that in having something more (a semantic specification of ‘male’) the marked lexeme drake is the member of the paradigm with the greater intension, and conversely that the unmarked lexeme duck, with the wider distributional range, was the member of the paradigm with the greater extension.Footnote 11 Thus, the paradigmatic structure and the syntagmatic use are interrelated.
Later on, the ‘has-something-more’ interpretation became a central part of the theory of linguistic expressions as composed of distinctive features. This is seen in later works by Jakobson himself (Fischer-Jørgensen Reference Fischer-Jørgensen1995:144f.) and even later in the Chomsky–Halle tradition of phonology (Gregersen Reference Gregersen1991; Fischer-Jørgensen Reference Fischer-Jørgensen1995:218f.).Footnote 12 The original modular approach by Trubetzkoy was the point of departure in this tradition, where the ‘has-something-more’ interpretation became dominant. For decades this approach to phonology was something to be reckoned with and could not be ignored.Footnote 13 The ‘has-something-more’ interpretation is also found in the theories of markedness based on some linguistic expressions being more inherently complex than others. The inspiration from the modular approach is found in the componential analysis in semantics (Lyons Reference Lyons1977:245ff.; Cramer et al. Reference Cramer, Henriksen, Kunøe, Larsen, Togeby and Widell1996:138–139), and in grammatical analysis (Givón Reference Givón1995).
3. JAKOBSON–HJELMSLEV MARKEDNESS AS A MEANS OF ANALYZING GRAMMATICAL PARADIGMS IN MODERN DANISH
Leaving the ‘has-something-more’ understanding of marked and unmarked forms and following the ‘can-something-more’ understanding instead – and in particular the part played by the notion of participation – I shall now turn to analyzing grammatical paradigms in binary oppositions.
As pointed out by Haspelmath (Reference Haspelmath2006:41), the early works on markedness by Jakobson and Trubetzkoy (and, we might add, Hjelmslev), did not aspire to offer universally applicable analyses. Instead, Jakobson's articles on Russian verb forms (Reference Jakobson1932) and on Russian case morphology (Reference Jakobson1936) were restricted to – Russian. In accordance with this practice, the analyses in the following sections are language-particular in that they are based on a single language, i.e. Modern Danish. This may be seen as heeding recent warnings on universals as formulated by Evans & Levinson (Reference Evans and Levinson2009) and Haspelmath (Reference Haspelmath2010).
3.1 Marked and unmarked tenses
The first example concerns the temporal meanings of the present and the preterite in Modern Danish. Non-temporal meanings used in children's make-believe worlds, in polite requests, etc. do not form part of the analysis.Footnote 14
Regarding temporal meaning, the preterite can only be used for situations that took place in the past, as in (2). This is not the case for the present tense. It may be used for situations taking place in the future, here-and-now, in the past, and as part of generic descriptions covering what might be called ‘all-time’, illustrated in (3)–(6), respectively.Footnote 15
(2)
(3)
(4)
(5)
(6)
As illustrated by examples (2)–(6), the preterite is the marked member of the paradigm and the present is unmarked. This follows from the observation that the preterite has limited use, whereas the present tense may be used for a number of situations including the kind of situation where the preterite may be found (i.e. situations that took place in the past). This is illustrated in (7):
(7)
This kind of markedness analysis is not only relevant in the study of morphological paradigms (and issues in lexicon). It may also be used in the analysis of word order and syntax. Examples of this are given in Sections 3.2 and 3.3.
3.2 Marked and unmarked word orderFootnote 16
Danish being a V2 language, the first position of the sentence is open to almost all constituent types, including cohesive adverbials. Cohesive adverbials may also appear in the central part of the sentence in close proximity to the finite verb. Certain words lead a double life as either modal particles (MPs for short; see (8a)–(10a)) or cohesive adverbials (see (8b)–(10b)). Now, if they are placed in the central part of the sentence, this ambiguity is not resolved.Footnote 17 If, however, the word in question appears in the first position of the sentence, it loses its potential as a modal particle and may only be read as a cohesive adverbial:
(8)
(9)
(10)
(11)
(12)
(13)
This pattern may be described as follows: the occurrence of words like ellers, altså, and dog in the central part of the sentence is unmarked (see (8)–(10) above), whereas the fronting of those words is marked (see (11)–(13) above). As regards words like these, the functional potential of fronting them is not as great as placing them in the central part of the sentence. A similar pattern is seen in words leading a double life as either manner adverbials or sentence adverbials, such as sikkert ‘safely/certainly’, rigtigt ‘correctly/really’, and the like.Footnote 18
This can be illustrated as in (14):
(14)
In Section 4.3, another example of markedness in word order will be shown.
3.3 Marked and unmarked noun phrases
Jensen (Reference Jensen, Jørgensen and Widell2007) offers an analysis along the same lines in a study on nouns and articles in Danish. In this case the notion of referentiality is taken into account in order to explain certain patterns in the nominal morphology.
The point of departure is that in Danish a noun may or may not co-occur with an article. When a noun appears without an article, it only carries the descriptive content of the noun category:Footnote 19
(15)
(16)
Functioning as a subject complement in (15), the noun does not refer to any entity but is used solely to assign a quality to the subject (represented by the pronoun han ‘he’). In (16), the noun does not refer to any entity either. Instead, it is incorporated in the phrase kan-lide-fisk ‘liking-fish’, and, as in (15), it only carries the descriptive content of the category fish.Footnote 20
In a sentence like (17) below, however, the noun cum article may carry the descriptive content of the category (the species of lion), resulting in the kind of generic utterance one might read in a textbook on zoology, or it may refer to a specific entity in the world:
(17)
As a consequence of this, the bare noun is analyzed as the marked member of the paradigm, and the noun phrase consisting of noun cum article is analyzed as the unmarked member, as the latter can be used to cover more functions than the former, including the function also covered by the former (i.e. to carry the descriptive content of the category). This results in the paradigm in (18):
(18)
However, Jensen (Reference Jensen, Jørgensen and Widell2007) makes no attempt to explain in markedness terms that subject complements sometimes occur in the form of noun cum article in Danish. Indeed, had she taken a closer look at this empirical fact, it might have revealed certain shortcomings in the Jakobson–Hjelmslev theory as described so far. I shall address this issue in the next section.
4. REFINING THE JAKOBSON–HJELMSLEV THEORY
In two studies, Jensen (Reference Jensen, Jakobsen, Jansen, Jensen and Strudsholm2009) and Petersen (Reference Petersen2010), the authors point to the flaws in the Jakobson–Hjelmslev theory. However, when the way in which the theory is formulated is examined more closely, one finds that it is easily mended, and not mended merely in order to save a flawed theory, but refined to make it reach its full potential as a means of analyzing grammatical paradigms in actual language use.
4.1 Critique of the analysis of the noun phrase
First, let us take a look at some points of the critique. With reference to Jensen (Reference Jensen, Jørgensen and Widell2007), Petersen (Reference Petersen2010) states that it cannot be the case that bare nouns are marked in all contexts. In phrases where a noun is incorporated in prepositional phrases, a bare noun, rather than noun cum article, must be the unmarked member. She gives the example of spille på (en) fløjte ‘play on (a) flute’.
(19)
(20)
In these examples, the sentence containing a bare noun may be used to answer the question in (19) as well as that in (20), whereas the sentence containing a noun phrase cum article may only serve as an answer to the here-and-now question in (19). Following the line of thought as presented in Jensen (Reference Jensen, Jørgensen and Widell2007), Petersen then comes to the conclusion that in phrases where a noun is incorporated in prepositional phrases, bare nouns must be the unmarked case, as the construction covers both the meaning pertaining to the situation here-and-now and the meaning where a certain quality is assigned to the subject.
Inspired by Hansen (Reference Hansen1927:52–56; Reference Hansen1967:170), Jensen herself (Reference Jensen, Jakobsen, Jansen, Jensen and Strudsholm2009) points to the inadequacies of the analysis of nouns with and without articles. In the case of nouns functioning as subject complements, it is not obvious that bare nouns should be the marked member of the paradigm. In fact, actual usage points to the bare noun being the unmarked member of the paradigm. Consider (21)–(24):
(21)
(22)
(23)
(24)
In (21) and (22) above, we are presented with two statements, both open to two different interpretations. Either the sentences in question convey information of a descriptive nature (the subject of (21) may very well be a writer by profession, and the subject of (22) may just as well be American by nationality), or the sentences are used as the speaker's personal opinion of the subject. Or, to put it differently, the sentences in (21) and (22) may be interpreted as providing either objective information or subjective evaluation. These options are illustrated by (21a, b) and (22a, b). The sentences in (23) and (24), on the other hand, can only be interpreted as providing subjective evaluation. When functioning as subject complement, then, a bare noun is the unmarked member of the paradigm, precisely because it covers more meanings than the noun cum article – including the meaning covered by the marked member (noun cum article).
The analysis of (19)–(20) and of (21)–(24) clearly raises some serious doubts concerning the analysis presented in (18) above. However, it does not necessarily follow that the notion of markedness should be abandoned altogether (as in Haspelmath Reference Haspelmath2006); rather, it should prompt the linguist to refine the theory.
4.2 Paradigms and subparadigms
In the 1930s, Jakobson and Hjelmslev were looking for so-called Gesamtbedeutungen. As case studies, they made in-depth analyses of the verbal systems of Russian (Jakobson Reference Jakobson1932), of the comparison of adjectives in French (Hjelmslev Reference Hjelmslev1934 [1972]), of the case system of Russian (Jakobson Reference Jakobson1936), Eskimo and Lak (Hjelmslev Reference Hjelmslev1935, Reference Hjelmslev1937), etc. The aim was to find the common denominators that would encompass the full range of, say, the nominative, the genitive, or the ergative case in a language. Every occurrence of, for example, the genitive could then be seen as an instantiation pointing back to the overall meaning of the genitive in that language – its Gesamtbedeutung.
However, in Jakobson's (Reference Jakobson1936) paper on case in Russian, a move towards a differentiation of the cases is also found, or rather, the relationship between the different cases. He sets up a multidimensional diagram of eight cases: the nominative (N), the accusative (A), the dative (D), the instrumental (I), the genitive 1 (G1), and the genitive 2 (G2), the locative 1 (L1), and the locative 2 (L2). I shall not discuss Jakobson's number of cases, nor how he arrived at these specific ones.Footnote 21 What is relevant to the matter at hand is the very way of conceptualizing the paradigm, shown in Figure 1.
Figure 1. The paradigm and subparadigms of Russian case as set up by Jakobson (Reference Jakobson1936).
Each pair of cases constitutes a small paradigm with two members. N: A is such a subparadigm and the same applies to I: D, to G1: G2, and so on. At the same time, these small paradigms themselves constitute the members of other paradigms. Thus, NA, for example, constitutes a member in the paradigm consisting of the two members NA: ID. And furthermore, the four cases NAG1G2 constitute a member in the subparadigm NAG1G2: IDL1L2. In this manner, Jakobson manages to make a description of the case system in Russian, whereby each case is at one and the same time a part of the case system as such, and correlative to the other cases in the subparadigms. Note also that all the (sub)paradigms are broken down into binary oppositions even though the morphological case system contains more than two cases.
The different paradigms revolve around semantic-functional distinctions such as ‘directionality’ and ‘quantification’, and ‘marginal’ and ‘central’. This, however, is of less importance for the matter at hand. The important point in the present paper on markedness is that there is a marked and an unmarked member in every subparadigm. A case like the accusative, for instance, is the marked member in the subparadigm N: A, while at the same time it is the unmarked member in the subparadigm A: D. Along with the other cases of a larger paradigm, it is also a part of the unmarked member of the paradigm NAG1G2: IDL1L2. A single linguistic phenomenon such as a case form (e.g. the accusative) may thus be part of more than one paradigm at a time, and it may differ from paradigm to paradigm whether it is the marked or the unmarked member.
The important lesson to be learned from this analysis is that a linguistic form (e.g. a case form) is not inherently marked or unmarked. Its markedness status depends on the (sub)paradigm it belongs to.
Following this insight, the theoretical point of a linguistic form not being in itself marked or unmarked can be moved one step further, as will be shown in what follows. We now return to the analysis of nouns in Danish – with or without articles: in some environments, the bare noun will be the marked member (see (25)), while in others it will be the unmarked one (see (26)).
(25)
(26)
The further step taken in (25)–(26) concerns the internal relationship among the members of the paradigm in different contexts. Whereas, according to Jakobson, a case like the accusative in Russian could be marked in a paradigm constituted by the two members N: A, and unmarked in a paradigm constituted by the two members A: D, and thus display different markedness relations, the analysis in (25)–(26) concerns different markedness relations between the same two members of the paradigm. In order not to confuse this phenomenon with so-called ‘reverse’ markedness relations (see next section), I shall refer to this specific kind of different markedness relations as opposite markedness.
This mechanism is, in fact, very common. In the next section, I shall present another example of a grammatical paradigm with opposite markedness relations.
4.3 Word order and opposite markedness relations
For a long time it was conventional wisdom that in Danish a specific word order would be the signal of subordinate clauses, whereas another word order was taken to be the signal of main clauses (see Wiwel Reference Wiwel1901:297f.; Diderichsen Reference Diderichsen1946 [1987]:186ff.). In the 1980s, however, Lars Heltoft conducted an analysis revealing that word order and clause type were two separate, codependent phenomena, resulting in a 2 × 2-way distinction of illocutionary potential. In main clauses both word order patterns were possible, and the same was true for subordinate clauses. The result as regarding illocutionary potential, however, was different (Heltoft Reference Heltoft and Jacobsen1990).
The examples in (27)–(30) show the four possibilities exemplified by clauses all containing a subject (han ‘he’), negation (ikke ‘not’), a finite verb (sover ‘sleeps’) and an adverb (endnu ‘yet’). The points regarding the interplay between word order and clausal type only concern the first three: subject, negation and verb. I shall call the word order subj–verb–neg in (27) and (30) word order 1 (WO1), and the word order subj–neg–verb in (28) and (29) word order 2 (WO2).Footnote 22
(27)
(28)
(29)
(30)
The first two examples, (27) and (28), show what has been ‘canonized’ as the word order for main clauses and subordinate clauses, respectively. The examples in (29) and (30) show the other two possibilities. The former, (29), constitutes a bona fide main clause all by itself, and according to Heltoft, when the WO2 is used in main clauses, the result is never an assertive utterance, but, for example, an utterance of the emotive kind.Footnote 23 Conversely, in (30), the clause han sover ikke endnu (WO2) constitutes not a main clause but an embedded one, i.e. a subordinate clause. According to Heltoft (Reference Heltoft and Jacobsen1990), whenever WO1 is used in embedded clauses, this results in sentences where some of the assertive potential of (27) is retained. Whereas (28) shows a clear instance of indirect speech, (30) may be seen as a pseudoquotation in the sense of Dubois (Reference Dubois1989). See also Ågerup (Reference Ågerup1996), Hansen & Heltoft (Reference Hansen and Heltoft2011:1677f.).
This analysis by Heltoft was followed up by Christensen (Reference Christensen2007), who showed that this 2 × 2-way distinction resulting in four combinations may be explained in terms of markedness. WO1 in main clauses and WO2 in subordinate clauses are the unmarked (UM) combinations, whereas WO2 in main clauses and WO1 in subordinate clauses are the marked (M) combinations, as illustrated in Figure 2.
Figure 2. The opposite markedness relations of Danish word order.
Christensen (Reference Christensen2007) arrives at this conclusion in the following way. First, regarding illocutionary potential, in main clauses, WO2 is only compatible with not-assertive speech acts, e.g. emotional outbursts, as in (31):
(31)
In contrast, in subordinate clauses, WO2 is not restricted with respect to the kind of speech act it is compatible with. In (32)–(35) the WO2 in the subordinate clause is compatible with assertive as well as not-assertive speech acts, including the emotive kind in (33):
(32)
(33)
(34)
(35)
WO1 demonstrates the opposite markedness relations. Whereas (28) above showed an instance of indirect speech, (36) below is one step closer to direct speech. Indeed, with complementizer-deletion it would not be possible to know whether (36) was to count as an instance of direct speech conveying someone's assertion or whether (36) was to count as an instance of direct or indirect speech. In (36) one can almost hear the quotation marks around the clause.
(36)
In main clauses, however, WO1 is not restricted to speech acts of the assertive kind. In (37)–(40) the word order is part of all kinds of speech acts, including the assertion in (37).
(37)
(38)
(39)
(40)
This may be expressed as in (41) and (42):
(41)
(42)
It should be noted that the analysis in (41) does not take precedence over the one in (42) or vice versa; they do not belong to a continuum or a scale of ‘more and less’ marked forms. Both analyses are parts of the description of the relationship between word order and clause type in Danish.
The sentences in (25)–(26) and (41)–(42) conveyed not only different markedness relations but opposite markedness relations. The notions of ‘reverse markedness’ and ‘general vs. local markedness’ are often found in discussions on language universals and typology, see inter alia Tiersma (Reference Tiersma1982), Battistella (Reference Battistella1996:56–60ff.), Aissen (Reference Aissen2003). However, the kind of analysis offered here for (25)–(26) and (41)–(42) differs fundamentally from those found in works on typology in the following way.
In classical typological studies on universals, the aim is to find evidence of linguistic categories common to all languages (Croft Reference Croft2003:88), and as a part of this endeavor, taking their inspiration from Greenberg, many linguists have been working on establishing sets of implicational hierarchies, see inter alia Comrie (Reference Comrie1981), Croft (Reference Croft2003). One example of this was already mentioned in Section 2 above; in (43) a second example, involving number in nouns (and pronouns) is given:
(43) singular < plural < dual < trial/paucal
According to Croft (Reference Croft2003:126–127ff.), the scale in (43) conveys that ‘the plural is more marked than the singular but less marked than the dual’ and that it follows from this that markedness is a relative property, not an absolute one.
No such claim is made in the present paper with reference to (25) and (26). Instead, the two analyses are seen as belonging to the description of the relation between noun ± article, and none of them is claimed to take precedence over the other. The same is true for (41) and (42). Thus, statements along the lines of ‘X is more marked than Y’ simply have no place in this line of thought.
5. THE TASK AHEAD
The reader may have noticed that all of the analyses in the previous sections demonstrate some kind of delimitation. The analysis of the present tense and the preterite limits itself to the temporal uses; the analysis of placing a word or a phrase in the front or in the central part of the sentence only concerns a handful of words with ambiguous meaning; the analysis of noun with or without article either limits itself to subject complements or – conversely – limits itself to everything but the subject complement and so forth. This way of establishing some kind of delimitation is common to all linguistic analyses. Scores of studies contain an early paragraph describing the extent to which the analysis applies. This may be regarded as a symptom of the complex structure of language. The characteristics of a linguistic phenomenon such as a certain use of the noun or a certain kind of word order may not be taken to hold in all contexts. Rather, the demarcation is apparently not only necessary to any given analysis but may be a symptom of a wider, more universal characteristic of language as such. The rules on how to use articles are not universal, not for all languages of the world, and not for the individual language either, be it Danish, English, Swedish or Greek; the same applies to personal pronouns, to tense, to number, etc.Footnote 24 This challenges the classical typologist approach of the Greenberg tradition, whereby certain linguistic phenomena are looked for in many languages, sometimes at the cost of capturing all the distributional details, but it also challenges the notion of Gesamtbedeutung (and to some extent Begrebszone), whereby a single semantic-functional description – often of a very abstract kind – is sought for all instances of a grammatical category in a particular language.
Instead, the present paper advocates a view whereby one might study the empirical data and find out how the paradigms and subparadigms are organized, and as a part of this endeavor investigate the functional semantic properties around which the two members of a (sub)paradigm revolves. For the lack of a better term, this final point might be called ‘with respect to . . .’. In the case of the two tenses in Danish, the analysis was ‘with respect to the ability to denote situations in time’. In the case of placing certain words in the front or in the central part of the sentence, the analysis was ‘with respect to the potential to function as a modal particle or a cohesive adverbial’. In the case of nouns with or without article outside subject complement, the analysis was ‘with respect to the potential of referring to an entity in the world or not’. In the case of nouns with or without article in subject complement, the analysis was ‘with respect to the potential of conveying objective description vs. subjective evaluation’. In the case of WO1 and WO2, the analysis was ‘with respect to the compatibility of certain speech acts’. This approach is functional-semantic with the emphasis on a broad definition of functional: How may a certain grammatical phenomenon be used? What is its functional potential and what are its limitations? How are the functional limitations related to the syntax-semantics of the phenomenon? Finding out how these things function, then, is a matter of extensive work on analyzing grammatical phenomena in context.
6. CODA
In this paper, I have presented a number of cases where the early theories of markedness as formulated by Jakobson and Hjelmslev are used in grammatical analysis. I have, moreover, suggested that the implications of these analyses may give due cause to reflect on what may be considered universal properties of language.
It is the task of all linguists to look for common denominators, and most of us harbor (perhaps secret) wishes to find universal properties of – say – the grammatical subject, the indicative mood, the word class of nouns, etc. But time and time again real language stands in the way of a beautiful analysis.
ACKNOWLEDGEMENTS
I wish to thank the members of ‘Grammatiknetværket’ for discussing these issues with me on several occasions. I also wish to thank the anonymous reviewers for suggestions and corrections; in particular one very critical and observant reviewer has prompted me to clarify many of the assumptions and claims of this paper. Finally, I wish to thank Ewa Jaworska for many good suggestions pertaining to the exposition.
