Printer Friendly

Degrees of clause integration: from endotactic to exotactic subordination in Dutch (1).

Abstract

This article presents a fine-grained syntactic categorization of Dutch subordinating conjunctions, both simple and phrasal. By means of a hierarchical set of formal criteria, this article provides a detailed description of the syntactic behavior of 87 conjunctions with a total of 154 different readings. These formal criteria test the degree of dependence on the main clause predicator. The presence or absence of proportionality with paradigms of pro-forms is introduced as a fundamental criterion for syntactic analysis (Section 2). This concept is further refined by distinguishing between external and internal proportionality (Section 3). The possibility of clefting the subordinate clause is applied as a syntagmatic criterion (Section 4) and correlated with the proportionality criteria (Section 5). The next step examines the possibility of fronting the subordinate clause (Section 6) and its relation to the various clefting types (Section 7). Finally, the different correlation results are integrated into a hierarchically structured general overview (Section 8). In the epilogue, a first attempt is made at mapping the syntactic categorization elaborated in the body of the article onto the more general (semantic and functional) typologies of clause integration presented in the literature by different authors. This comparison wants to stress the general methodological applicability of the approach presented (Section 9).

1. Introduction

1.1. Approaches to clause combining

Over the past decades, clause combining mechanisms have been the subject of intensive study. Within this area of research, the classification of subordinate clauses and (subordinating) conjunctions has played a prominent role. Roughly speaking, three major approaches can be distinguished: a primarily semantic, a semantico-syntactic, and a predominantly syntactic one. These will briefly be introduced in the following paragraphs.

In their landmark typological paper, Thompson and Longacre (1985: 172) define three types of subordinate clauses: "those which function as noun phrases (called complements), those which function as modifiers of nouns (called relative clauses), and those which function as modifiers of verb phrases or entire propositions (called adverbial clauses)." They furthermore divide the adverbial clauses into a number of basic semantic types belonging to two distinct groups. Clauses in the first group can be substituted for by "monomorphemic non-anaphoric adverbs" and express relations of time, location or manner (Thompson and Longacre 1985: 177). The second group of clauses, which express purpose, concession, or condition, among others, do not have such a substituting adverb. A notion similar to this type of substitutability, that is, that of proportionality, will play a crucial role in the present analysis. More recently, the semantic space of interclausal relations has received more fine-grained descriptions. In his historical-typological work on subordinators in European languages, for instance, Kortmann (1997: 138, 210, 1998: 464ff.) distinguishes 32 of these relations, which are organized into a number of networks, such as time, modality, or cause/condition/concession.

A second line of approach investigates the correlations between a hierarchy of interclausal semantic relations and a hierarchy of syntactic bondedness. Concrete proposals, such as Foley and Van Valin (1984) or Hengeveld (1998), however, differ as to which of the two hierarchies is spelled out in more detail.

In their seminal analysis, Foley and Van Valin (1984: 264) primarily elaborate the concept of syntactic bondedness. Central is the idea that the functional-semantic notion of dependence is NOT equivalent to the purely syntactic notion of embeddedness, that is, "whether a clause is dependent in some way upon another clause is independent of whether it is embedded as an argument of another clause" (Foley and Van Valin 1984: 243). The interaction between embeddedness and dependency yields the tripartition of coordination, subordination, and cosubordination. Coordination is neither embedded nor dependent, subordination is both embedded and dependent, whereas the intermediate type of co-subordination is dependent but not embedded. The present article also emphasises the distinction between embeddedness and dependence, but restricts its scope to the transition from subordination to co-subordination. In addition to the tripartition of nexus types, that is, the syntactic relationship between the clauses, Foley and Van Valin distinguish three functional levels: nucleus (verb level), core (within the clause), and periphery (above the simple clause). The interaction of nexus types and functional levels results in a hierarchy of syntactic bondedness.

In Hengeveld (1998: 335ff.) the semantic typology of adverbial clauses is worked out more extensively than in Foley and Van Valin (1984). His typology is defined in terms of semantic primitives which function as interacting parameters. The first dimension--resembling the functional levels of Foley and Van Valin--concerns the (semantic) layers of clause structure or entity types: properties or relations (predicates), states of affairs (predications), propositions, and speech acts. The second dimension combines the three binary parameters of (non)factuality, (non)presupposedness, and (in)dependent time reference. Hengeveld then argues for a systematic correlation between semantic types of adverbial clauses -defined in terms of the interaction between the two dimensions above-on the one hand, and the way they are expressed on the other hand.

A third, more formal approach to clause combining focuses on the syntactic properties of clause integration. A key exponent of this type of analysis is Lehmann (1988)

while this [the semantic nature of the relation between the two clauses] has always played a prominent role in the classification of subordinate clauses, it does not appear to be constitutive of cross-linguistically valid types of clause linkage. That is, there is no cross-linguistic notion of, say, the concessive clause, which would possess any constant structural correlates. It rather appears that the grammatical types that will emerge [... ] cut across semantically different clause linkage relations. (Lehmann 1988: 183)

As the present article subscribes to this view, it is essential to spell out which of Lehmann's (1988: 183) six "semantosyntactic parameters," listed below, are relevant to our analysis.

1. The explicitness of the linking

2. The main clause syntactic level of the subordinate clause

3. The hierarchical downgrading of the subordinate clause

4. The desententialization of the subordinate clause

5. The grammaticalization of the main verb

6. The interlacing of the two clauses

First of all, the scope of this research is restricted to complex sentences containing a subordinator. Hence, as far as the criterion of EXPLICITNESS OF LINKING is concerned, attention is confined to "specific conjunctions," ignoring other linking devices such as prepositional phrases or connective adverbs (Lehmann 1988: 213). As a corollary, the main clause SYntACTIC LEVEL which the subordinate clause belongs to is purely sentential, that is, inside/outside VP or main clause, thus disregarding complex predicate formations such as verb serialization or derivation (Lehmann 1988: 192). The parameter of HIERARCHICAL DOWNGRADING, that is, the continuum from (paratactic) independent clause to (embedded) governed clause (Lehmann 1988: 189), is crucial to the present proposal. Due to the central role assigned to the main clause predicator, however, this criterion will in a sense be reversed into "hierarchical upgrading," a general tendency towards increasing detachment from the main clause.

Along the lines of Lehmann (1988), the cline of clause combining in Hopper and Traugott (1993: 170-171) integrates the criteria of (overt vs. covert) formal linking and (semantico-pragmatic) integration: in parataxis, maximal overt linking correlates with minimal integration whereas in subordination, minimal overt linking correlates with maximal integration. The intermediate stage of hypotaxis or interdependency concerns combinations in which one clause cannot stand by itself, and is therefore relatively dependent. Adverbial clauses, which include temporals, causals, conditionals, and concessives (and which constitute the bulk of the subordinating structures to be analyzed in this article) are considered to be hypotactic elements. As argued by Hopper and Traugott (1993: 176), "adverbial clauses themselves show a continuum of looser-to-tighter integration, a continuum that correlates with their function." The central aim of this article is to provide a detailed description of this continuum in terms of a number of formal criteria. In other words, rather than analyzing integration primarily as a semantic-pragmatic relation between events or states, the present proposal describes the subordinate clause with regard to its syntactic relation to the main clause predicator. This shift of focus with respect to the notion of integration also has implications for the remaining three criteria of Lehmann's listed above.

First of all, maximal integration--referred to as "embedding" by Lehmann or "subordination" by Hopper and Traugott--is standardly associated with the shift from finite to nonfinite verb forms. The distinction between independent and dependent verb forms, the latter being those which are used in subordinate constructions only, also plays a central role in Hengeveld (1998: 339). For Hopper and Traugott (1993: 170) "clausal remnants such as infinitives and participles" are the endpoint of an evolution towards maximal integration into the main clause without overt linking. For Lehmann (1988: 200), on the other hand, this process of DESENTENTIALIZATION of the subordinate clause extends beyond the nonfinite verb forms to include verbal nouns. Although both finite and nonfinite subclauses feature in the Dutch data discussed below (but verbal nouns do not), the dimension of finiteness as such is not a decisive criterion in our syntactic analysis. As pointed out by Lehmann (1988: 204), desententializing the subordinate clause is only one way of reducing a complex sentence to a simple one, the other being the GRAMMATICALIZATION OF THE MAIN VERB. In view of the central role we assign to the verb functioning as the (real syntactic) predicator of the main clause, the grammaticalization process of reducing a main verb to an evidential or modal verb, or even further to an auxiliary verb or affix, plays no role whatsoever in our proposals.

The same holds for the various types of INTERLACING TWO CLAUSES which arise with control verbs or raising verbs (Lehmann 1988: 207208). As these verbs strongly tend to occur without overt subordinating conjunction and to combine with nonfinite verb forms, the resulting structures fall outside the scope of the present article, which crucially concerns the degrees of syntactic connectedness between the main clause predicator and the subclause it combines with.

1.2. The database

The starting point for drawing up the list of expressions to be analyzed and classified was the exhaustive list of conjunctions in the electronic edition of the Dutch standard dictionary van Dale Groot Woordenboek der Nederlandse Taal (2000). Elimination of the coordinating conjunctions and of the subordinating ones marked as "marginal" (in terms of labels such as "archaic," "regional," "formal," "literary," or "rarely used") resulted in a list of 55 simple subordinating conjunctions. The standard Dutch grammar Algemene Nederlandse Spraakkunst (Haeseryn et al. 1997: 543) mentions the existence of phrasal conjunctions (combinations of words which together function as a conjunction) but does not furnish any list. A (preliminary) list of 32 phrasal subordinators was drawn up from various sources. The detailed syntactic analysis of these 87 elements resulted in 154 distinct readings, which are listed alphabetically in the Appendix together with their syntactic feature analysis, the (different) English translations, and a glossed example. Most examples were taken or adapted from the van Dale dictionary. Although examples construed by the authors were checked on the Internet, some individual variation in acceptability judgements may remain.

2. The concept of proportionality

2.1. From pronominal paradigm to lexicalization

In the study of verbal valency within the framework of the pronominal approach (2) a crucial role is assigned to PARADIGMS OF PRONOUNS that are proportional to constituents containing lexical (i.e. non-pronominal) elements. The concept of PROPORTIONALITY is used in order to reduce the huge number of possible combinations between lexical elements, as in (1a), to a much smaller number of combinations with pronouns in (1b).
(1) a. Marie betwijfelf zjn oprechtheid
 de buurvrouw onze eerlijkheid
 ... ... ...
 'Mary/the has doubts his/our ... sincerity/
 neighbor/... about honesty'

 b. Zjh/hij betwijfelt dat
 wie? wat?
 iemand iets
 ... ...
 'He/she/who?/ has doubts that/what?/
 somebody/... about something/...'


The various subject noun phrases in (1a) are specific instantiations of the more general pronominal elements in the same subject paradigm hij/zij/ wie/iemand/ ... in (1b). In the pronominal approach, the former are referred to as the LEXICALIZATIONS of particular pronouns.

Notice that pronominal paradigms may contain various types of MINIMAL REFERENTIALS (3) belonging to the traditional categories of personal, demonstrative, possessive, interrogative, and indefinite pronouns. The group of pronouns is further subdivided according to the type of referential specification. Pronouns whose referential specification is suspended are labelled SUSPENSIVE. These correspond roughly to the traditional category of interrogative pronoun and adverb. On the other hand, ASSERTIVE (4) pronouns such as he/him, she/her, they/them, or these are characterized by the presence of a number of syntactico-semantic features labelled as case, number, gender, and sex.

The finite nature of the resulting list(s) of pronouns makes it possible to examine their combinations with VERBAL PREDICATORS (i.e. verbal construction kernels) in a systematic and exhaustive way without having to appeal to the particular semantics of the lexicalizations. The possibility or impossibility for a particular pronoun to appear in a particular paradigm is indeed meaningful: the pronouns reveal the primary or basic selectional restrictions a verbal predicator imposes on its dependents. (5)

2.2. Valency versus rection

The pronominal paradigms in (1b) are predicator-specific and as such reflect the valency structure of the verbal predicator. However, exactly the same relation of proportionality with a pronominal paradigm exists outside the domain of valency for syntactic units which are not predicator-specific. The syntagms of temporal and spatial reference, reason and manner in (2a), for instance, have their own proportional paradigms in (2b).
(2) a. in januari in Leuven wegens de regen met veel
 'in January' 'in Leuven' 'because of the moeite
 rain' 'with great
 difficulty'

 b. nu/dan/toen hier/daar daarom zo
 'now/then' 'here/there' 'therefore' 'this/that way'
 wanneer? waar? waarom? hoe?
 'when?' 'where?' 'why?' 'how?'
 ooit ergens om een of op een of
 'ever' 'somewhere' andere reden andere manier
 'for some reason' 'somehow'


The elements in (2b), which are traditionally labelled adverbial, constitute an integral part of the system of minimal referentials of the language. (6) Although with some predicators, such as wonen in ('live in'), they are valency-bound, in general they are not. The members of such non-valency-bound paradigms will be referred to as "RECTION" elements.

2.3 Sentential lexicalization

The proportionality relation with a pronominal paradigm not only holds for the nominal phrases in (1a) and the prepositional phrases in (2a), but also for the subordinate clauses in (1') and (2') below:
(1') Hendrik betwijfelt of zijn buurman oprecht is.
 'Henry doubts whether his neighbor is sincere.'

(2') a. Het was aan het regenen toen hij vertrok.
 'It was raining when he left.'

 b. Hij vertrok omdat het regende.
 'He left because it was raining.'


The of/whether clause in (1') is proportional to the object paradigm dat/ wat/iets/ ... in (1b) whereas the toen/when clause in (2'a) and the omdat/ because clause in (2'b) are proportional to the temporal and causal paradigms in (2b) respectively. It is precisely this type of SENTENTIAL LEXICALIZATION, that is, lexicalization by means of subclauses, which will be focused on in this article.

2.4. Endotaxis versus epitaxis

Valency and rection paradigms fall within the scope of the general mechanism of proportionality and, hence, share a number of syntactic properties. Both are considered ENDOTACTIC, that is, falling "within" the pronominally controllable realm of the verbal predicator. All other constituents that do not meet this proportionality criterion, and are therefore more "peripheral" to the endotactic structure, are called EPITACTIC. The distinction between endotactic and epitactic elements can most clearly be illustrated by means of prepositional phrases and subordinate clauses with similar lexical and word order patterns but with fundamental differences in proportionality.
(3) a. Zij gelooft in haar slaagkansen/erin dat ze zal slagen.
 'She believes in her chances of success/that she will succeed.'
 b. Zij gelooft erin/daarin. Waarin gelooft ze?
 'She believes in it/that. What does she believe in?'
(4) a. Zij is vertrokken in bet besef van haar verantwoordelijkheid/in
 het besef dat zij verantwoordelijk was.
 'She left well aware of (lit. in the awareness of) her
 responsibility/of the fact that she was responsible.'
 b. *Ze is erin/daarin vertrokken. *Waarin is ze vertrokken?
 '*She left in it/that.' '*What did she leave in?'


The prepositional phrase and the subordinate clause in (3a), which are valency terms with the predicator geloven ('believe'), are fully proportional with the pronominal paradigm erin/daarin/waarin ('therein/wherein') in (3b), and, hence, are examples of endotactic subordination. Their counterparts in (4a) do not exhibit this type of proportionality in (4b), and are therefore considered epitactic. However, the prepositional phrase and the subordinate clause in (4a) do relate to the prepositional phrase in dat besef ('in that awareness'). This correspondence reveals the need for a further refinement of the proportionality criterion, as will be elaborated in the next section.

3. External versus internal proportionality

3.1. Internal proportionality without external proportionality

The mechanism of proportionality for valency in (1b and 3b) and rection in (2b) clearly does not suffice to account for all types of correspondence relations between subordinate clauses and specific prepositional phrases. In the case of complex conjunctions, that is, PHRASAL CONJUNCTIONS, (7) such as in het besef dat ('in the awareness that') in (4a), the subordinate clause dat zij verantwoordelijk was ('that she was responsible') is a further lexicalization of the demonstrative determiner dat ('that') in the prepositional phrase in dat besef ('in that awareness'). However, the suspensive counterpart in welk besef ('in which awareness?') is not available in Dutch. This type of correspondence relation between subclauses on the one hand, and assertive and suspensive determiner elements on the other, will henceforth be called "INTERNAL" proportionality and is to be distinguished from the EXTERNAL proportionality with full-fledged pronominal paradigms discussed in the previous section. This more restricted type of proportionality basically concerns the internal structure of the phrasal conjunction and thus retains the latter's core lexical components, that is, the preposition in and the noun besef. As a consequence, internal proportionality does not apply to nonphrasal or simple conjunctions, such as terwijl ('while'), which do not have any internal structure.

Notice that the in bet besef dat construction in (6) occupies an intermediate position in between combinations such as onder voorwaarde dat ('on condition that') in (5), and expressions such as met bet gevolg dat ('with the consequence that') in (7):
(5) a. Zij is vertrokken onder voorwaarde dat hij haar zou vervangen.
 'She left on condition that he would replace her.'
 b. *Zij is daaronder vertrokken. *Waaronder is zij vertrokken?
 '*She left thereunder. *Whereunder did she leave?'
 c. Zij is vertrokken onder die voorwaarde. Onder welke
 voorwaarde ...?
 'She left on that condition. On which condition ...?'
(6) a. Zij is vertrokken in het besef dat zij verantwoordelijk was.
 'She left well aware of the fact that she was responsible.'
 b. *Zij is daarin vertrokken. *Waarin is zij vertrokken?
 '*She left therein. *Wherein did she leave?'
 c. Zij is vertrokken in dat besef. *In welk besef ...?
 'She left well aware of that. *Well aware of what ...?'
(7) a. Zij is vertrokken met het gevolg dat her werk niet af is.
 'She left with the consequence that the work is not finished.'
 b. *Zij is daarmee vertrokken. *Waarmee is zij vertrokken?
 '*She left therewith. *Wherewith did she leave?'
 c. *Zij is vertrokken met dat gevolg. *Met welk gevolg ...?
 '*She left with that consequence. *With which consequence ...?'


The lack of external proportionality in the (b) examples reflects the epitactic status of all three phrasal conjunctions in (5)-(7). In (5c) both assertive and suspensive internal proportionality hold--onder die/welke voorwaarde ('on that/which condition')--whereas in (7c), internal proportionality is lacking altogether--met *dat/*welk gevolg ('with that/ which consequence'). (8) The intermediate phrasal conjunction in (6c) shows assertive but no suspensive internal proportionality: in dat/*welk besef ('in that/*which awareness').

As for external proportionality the presence of the assertive form in (3) entails that of the suspensive form and vice versa. The absence of assertive external proportionality in (4)-(7) entails the absence of its suspensive counterpart and vice versa. However, assertive and suspensive proportionality do not necessarily coincide in the internal paradigms: assertive determiners may appear with suspensive counterparts (5c) or without (6c), but suspensive determiners always have assertive counterparts.

3.2. Internal proportionality in combination with external proportionality

The threefold distinction revealed in (5)-(7) is not restricted to the epitactic class of subordinating elements that are not externally proportional, but also exists within the endotactic class which does observe external proportionality.
 (8) a. Door van het balkon te springen heeft hij zijn voet gebroken.
 'By jumping off the balcony he broke his foot.'
 b. Zo heeft hij zijn voet gebroken. Hoe ...?
 'This way he broke his foot. How ...?'
 c. Daardoor heeft hij zijn voet gebroken. Waardoor ...?
 'Thereby he broke his foot. Whereby ...?'
 (9) a. Nadat ik alle lichten heb gedoofd ga ik naar huis.
 'After having switched off all the lights I go home.'
 b. Dan ga ik naar huis. Wanneer ...?
 'Then I go home. When ...?'
 c. Daarna ga ik naar huis. *Waarna ga ik naar huis?
 'After that [lit. thereafter] I go home. *After what [lit.
 whereafter] ...?'
(10) a. Zonder het papier te scheuren heeft zij het pakje opengemaakt.
 'Without tearing the wrapper she opened the parcel.'
 b. Zo heeft zij het pakje opengemaakt. Hoe ...?
 'This way she opened the parcel. How ...?'
 c. *Daarzonder/*zonder dat heeft zij het pakje opengemaakt.
 *Waarzonder/*zonder wat ...?
 '*Without that she opened the parcel. *Without what ...?'


The subclause introduced by door ('by') in (8a) is externally proportional to zo/hoe ('how/this way') in (8b), that is, it lexicalizes the rectional paradigm of manner introduced in (2b). In addition, this clause is internally proportional to daardoor/waardoor ('thereby/whereby') in (8c), where the original prepositional element door ('by') is still visible. (9) The example in (9a) with nadat ('after') has external proportionality with dan/wanneer ('then/when') in (9b), but only partial internal proportionality in (9c): assertive daarna ('thereafter') but not suspensive *waarna ('*whereafter'). Although the clause with zonder ('without') in (10a) is still externally proportional to hoe/zo ('how/this way') in (10b), it lacks internal proportionality with *daarzonder/*waarzonder ('*without that/what') in (10c). Notice, incidentally, that external proportionality in (8b) and (9b) reflects more general, abstract semantic features, whereas internal proportionality in (8c) and (9c) instantiates more specific ones. (10) This contrast obviously does not apply to the examples in (10).

The above discussion thus reveals that internal proportionality can arise from three different sources: (i) from the syntactic structure, as in the phrasal conjunctions onder voorwaarde dat ('on condition that') and in her besef dat ('in the awareness that') in (5)-(6), (ii) from the morphological structure, as with the compound conjunction nadat ('after') in (9), and (iii) from the homonymy between prepositions and conjunctions, as with door ('by') in (8). The mere presence of one of these formal properties, however, does not automatically imply the availability of internal proportionality, as can be inferred, for instance, from the phrasal conjunction met her gevolg dat ('with the result that') in (7) and the simple conjunction zonder ('without') in (10). Although the former does have the relevant determiner position and the latter does have a homonymous preposition, neither of them is internally proportional.

3.3. Internal and external proportionality: summary

In the summarizing table in (11) below, the independent dimensions of internal and external proportionality are integrated. Since assertive and suspensive external proportionality always coincide, only one column is needed. For internal proportionality, however, the features "assertive" and "suspensive" need to be distinguished in the second and third columns in order to account for the threefold distinction between paradigms with complete (assertive = 1, suspensive = 1), partial (assertive = 1, suspensive = 0), and no internal proportionality (assertive = 0, suspensive = 0). This tripartition can be observed for both values of external proportionality in the table in (11). In the shorthand notation of column four (which is also used in the Appendix), the value to the left of the underscore corresponds to external proportionality, whereas the two values to the fight indicate assertive and suspensive internal proportionality respectively. (11) The symbol "N" refers to the total number of conjunctive readings distinguished in the Appendix.
(11)

Extern Intern Intern Proportionality N = 154 Conjunction
 ass susp shorthand (example in
 text)

1 1 1 [1_11] 14 door to (8)

1 1 0 [1_10] 5 nadat (9)

1 0 0 [1_00] 27 zonder
 tel (10)

0 1 1 [0_11] 8 onder
 voorwaarde
 dat (5)

0 0 0 [0_10] 26 in het besef
 dat (6)

0 0 0 [0_00] 74 met het
 gevolg
 dat (7)

[1_11] full external proportionality
and full internal proportionality

[1_10] full external proportionality
and assertive internal proportionality

[1_00] full external proportionality
and no internal proportionality

[0_11] no external proportionality
and full internal proportionality

[0_10] no external proportionality
and assertive internal proportionality

[0_00] no external proportionality
and no internal proportionality


Notice that zonder te on row three is marked with a numerical index. These indices, which feature prominently in the list given in the Appendix, reflect the different types of syntactic behavior a given expression may exhibit. Consider the cases of terwijl ('while') in (12):

(12) a. [Terwijl.sub.1] de baby sliep, heb ik de krant gelezen.

'While the baby was sleeping, I read the newspaper.'

b. [Terwijl.sub.2] ik het hem uitdrukkelijk verboden had, is hij toch komen opdagen.

'Whereas/although I had explicitly forbidden him to do so, he showed up anyway.'

c. [Terwijl.sub.3] je hier nu toch bent, ik heb de zaak al met hem besproken.

'Since you are here anyway, I have already discussed the case/ matter with him.'

The main distinction in (12) is that the (a) example is proportional to dan/ wanneer ('then/when'), whereas the (b) and (c) subclauses are not. The latter two crucially differ from one another, however, with respect to the word order pattern in the main clause: [terwijl.sub.2] ('whereas/although') in (12b) has VSO, as opposed to SVO with [terwijl.sub.3] ('since') in (12c). In the case of naarmate ('as') and naar ('according as') in (13) neither proportionality nor word order yields a syntactic distinction. Instead, it will be argued that the possibility of clefting the subordinate clause distinguishes the two conjunctions:

(13) a. Naarmate iemands geestelijke horizon ruimer wordt, neemt zijn bescheidenheid toe.

'As a person's mental horizon broadens, his modesty increases.'

b. Naar we gisteren vernamen, is het daar slecht weer.

'As we learned yesterday, the weather is bad over there.'

In the next sections, the criteria of clefting and word order are dealt with in full detail.

4. Clefting of subordinate clauses

In Dutch, as well as in English and many other languages, a clefting mechanism is available, that is, a particular syntactic construction which allows the language user to change the informative weight of one or the other constituent by isolating it from the rest of the main clause structure. Consider the following examples of cleft sentences ("gekloofde zinnen" in Dutch) taken from the ANS (Haeseryn et al. 1997: 1243-1244):

(14) a. Dirk heeft gisteren aan Francine een ties jenever gegeven.

'Dirk gave a bottle of gin to Francine yesterday.'

b. Het was Dirk die gisteren aan Francine een ties jenever gegeven heeft.

'It was Dirk who gave a bottle of gin to Francine yesterday.'

c. Het was gisteren dat Dirk aan Francine een ties jenever gegeven heeft.

'It was yesterday that Dirk gave a bottle of gin to Francine.'

On the basis of the informationally neutral (14a), two types of cleft sentences can be built. Both consist of the same four components, namely, (i) the impersonal pronoun het ('it') functioning as a preliminary subject, (ii) a form of the verb zijn ('be'), (iii) the constituent with informative prominence, and (iv) a subordinate clause containing what is considered to be given information. (12) The difference between the two types of clefting then concerns the nature of the element introducing the subclause: the relative pronoun die ('who') in (14b) versus the subordinating conjunction dat ('that') in (14c). (13)

However, this clefting mechanism cannot be applied to all constituents. The sentence adverbial vermoedelijk ('presumably') in (15a) and (15b) is a case in point:

(15) a. Vermoedelijk gaf Dirk gisteren een ties jenever aan Francine. 'Dirk presumably gave Francine a bottle of gin yesterday.'

b. *Het was vermoedelijk dat Dirk gisteren een ties jenever gaf aan Francine.

'*It was presumably that Dirk gave Francine a bottle of gin yesterday.'

The general hypothesis is that the availability of clefting is an indication of the proximity relation between a given constituent and its verbal predicator. This correlation generalizes to those cases where the constituent is lexicalized as a subclause. Consider again the difference between [terwijl.sub.1] in (16) and [terwijl.sub.2] in (17):

(16) a. [Terwijl.sub.1] de baby sliep, heb ik de krant gelezen.

'While the baby was sleeping, I read the newspaper.'

b. Het was [terwijl.sub.1] de baby sliep dat ik de krant heb gelezen.

'It was while the baby was sleeping that I read the newspaper.'

(17) a. [Terwijl.sub.2] ik het hem uitdrukkelijk verboden had, kwam hij toch opdagen.

'Whereas I had explicitly forbidden him to do so, he showed up anyway.'

b. *Het was [terwijl.sub.2] ik het hem uitdrukkelijk verboden had dat hij toch kwam opdagen.

'*It was whereas I had explicitly forbidden him to do so that he showed up anyway.'

The subclause with temporal [terwijl.sub.1] in (16a), which does allow clefting in (16b), is more closely related to the predicator than the one with adversative [terwijl.sub.2] in (17a), which does not allow clefting in (17b) (see Verstraete [2002: 244-248] for a similar treatment of English while). The correlations between the various proportionality constellations--introduced in the previous sections--and the various clefting types, both of which encode degrees of proximity to the predicator, are discussed in full detail in the next section.

5. Correlations between proportionality and clefting

5.1. Pronominal and clausal clefting with endotactic subordinators

In Section 3, the distinction between endotactic and epitactic subordination is defined in terms of the presence or absence of external proportionality. Endotactic subordination comprises the subcategories of valency and rection. As for the latter, the three temporal conjunctions in (18)-(20) all admit clefting in spite of the differences in internal proportionality: (14)

(18) a. Toen hij aan de macht kwam, verbeterde de toestand.

'When he came into power, the situation improved.'

b. Het was toen hij aan de macht kwam dat de toestand verbeterde.

'It was when he came into power that the situation improved.'

c. Het was dan/toen dat de toestand verbeterde. Wanneer was het dat ...?

'It was then that the situation improved. When was it that ...?'

(19) a. Nadat hij verkozen werd, verbeterde de toestand.

'After he was elected, the situation improved.'

b. Het was (pas) nadat hij verkozen werd, dat de toestand verbeterde.

'It was (only) after he was elected that the situation improved.'

c. Het was (pas) dan dat de toestand verbeterde/Wanneer was het dat ...?

'It was (only) then that the situation improved/When was it that ...?'

d. Het was (pas) daarna dat de toestand verbeterde/*Waarna was het dat ...?

'It was (only) after that that the situation improved/*After what was it that ...?'

(20) a. Tot(dat) hij de verkiezingen verloor, verbeterde de toestand.

'Until he lost the elections, the situation was improving.'

b. Het was tot(dat) hij de verkiezingen verloor dat de toestand verbeterde.

'It was until he lost the elections that the situation was improving.'

c. Het was zo lang dat de toestand verbeterde. Hoe lang was het dat ...?

'It was that long that the situation was improving. How long was it ...?'

d. Het was tot dan dat de toestand verbeterde. Tot wanneer was het dat ... ?

'It was until then that the situation was improving. Until when was it ... ?'

The toen-clause in (18) has the [1_00] pattern: it is only externally proportional to wanneer/dan/toen ('when/then') in (18c). The nadat-clause in (19) is [1_10], that is, externally proportional to wanneer/dan ('when/ then') in (19c) and internally to the assertive daarna ('after that') but not to the suspensive *waarna ('*after what') in (19d). The tot(dat)-clause in (20) is then of type [1_11] as it is fully proportional, both externally to hoelang/zolang ('how long/so long') in (20c) and internally to tot wanneer/tot dan ('until when/then') in (20d).

The type of cleft sentences whose focus position is occupied by pronominal elements, as in the (c) and (d) examples in (18)-(20), is referred to as "PRONOMINAL CLEFTING." The (b) examples, which have the sentential lexicalization in focus position, are instances of "CLAUSAL CLEFTING." Although both clefting types strongly tend to correlate, neither type necessarily entails the other. In other words, four classes of subordinators need to be distinguished: those which can be clefted both pronominally and clausally [+p,+c], as in (18)-(20), those which can be clefted neither way [-p,-c], and the two intermediate types allowing only pronominal clefting [+p,-c] or only clausal clefting [-p,+c].

5.2. Pronominal and clausal clefting with epitactic subordinators

The examples in (18)-(20) show that with endotactic subordination the presence of external proportionality outweighs the differences in internal proportionality in that both pronominal and clausal clefting [+p,+c] are available throughout. In the group of epitactic conjunctions, on the other hand, the correlation between different degrees of internal proportionality and the availability of clefting is less predictable. In general, though, clefting possibilities decrease as internal proportionality disappears.

Epitactic subordinate clauses which have both assertive and suspensive internal proportionality--that is, type [0_11]--can be clefted without exception. Consider for instance onder voorwaarde dat ('on condition that') in (21), which is internally proportional to onder die/welke voorwaarde ('on that/which condition'):

(21) a. Onder voorwaarde dat hij zou slagen werd hem een fiets beloofd.

'On condition that he succeed, he was promised a bike.'

b. Het is (enkel) onder voorwaarde dat hij zou slagen dat hem een fiets werd beloofd.

'It is (only) on condition that he succeed that he was promised a bike.'

c. Het is (enkel) onder die voorwaarde dat hem een fiets werd beloofd./Onder welke voorwaarde was het dat ...?

'It is (only) on that condition that he was promised a bike/On what condition was it that ...?'

Although strictly speaking elements of this type cannot be said to belong to the class of rectional elements in (18)-(20)--because external proportionality is absent--they share with the endotactic elements the guaranteed availability of clausal clefting in (21b) and pronominal clefting in (21c). The fact that assertive and suspensive internal proportionality coincide causes these elements to closely resemble the externally proportional ones with their full-fledged paradigm. On the basis of this similarity, the [0_11] elements will be incorporated into the endotactic category in the overview of Section 8.

Out of the 23 epitactic subordinators which have assertive but no suspensive internal proportionality--type [0_10]--eleven still allow both pronominal and clausal clefting [+p,+c], whereas five of them allow neither type of clefting [-p,-c]. The status of the remaining seven expressions will be discussed in Section 5.4 below. The phrasal conjunctions in de veronderstelling dat ('on the assumption that') in (22) and ondanks dat ('despite the fact that') in (23), for instance, are both of type [0_10]: they are internally proportional only to the assertive in die veronderstelling ('on that assumption') and desondanks/ondanks dat ('in spite of that'), respectively. Nevertheless, the former has both clausal and pronominal clefting, as shown in (22b)-(22c), whereas the latter does not, as in (23b) and (23c):

(22) a. In de veronderstelling dat ze op bezoek kwam bleef hij thuis.

'On the assumption that she would come and visit, he stayed home.'

b. Het was in de veronderstelling dat ze op bezoek kwam dat hij thuisbleef.

'It was on the assumption that she would come and visit that he stayed home.'

c. Het was in die veronderstelling dat hij thuisbleef.

'It was on that assumption that he stayed home.'

(23) a. Ondanks dat ze op bezoek kwam, ging hij weg.

'Despite the fact that she would come and visit, he left.'

b. *Het was ondanks dat ze op bezoek kwam dat hij wegging.

'*It was despite the fact that she would come and visit that he left.'

c. *Het was ondanks dat/desondanks dat hij wegging.

'*It was in spite of that that he left.'

With the epitactic conjunctions of type [0_00], which lack any kind of proportionality, the vast majority of cases (64 out of 70) exclude clefting [-p,-c], as is illustrated for alhoewel ('although') in (24b) and (24c):

(24) a. Alhoewel ze op bezoek kwam ging hij weg.

'Although she would come and visit, he left.'

b. *Het was alhoewel ze op bezoek kwam dat hij wegging.

'*It was although she would come and visit that he left.'

c. *Het was alhoewel dat, dat hij wegging.

'*It was although that, that he left.'

5.3. Clausal clefting without pronominal clefting [-p, +c]

In (18)-(22) both clausal and pronominal clefting are acceptable [+p,+c], whereas in (23) and (24) both types are excluded [-p,-c]. This parallelism does not hold for naarmate (dat) ('according as'), however, which belongs to the first intermediate type, allowing clausal clefting in (25b) but rejecting pronominal clefting in (25c).

(25) a. Naarmate iemands geestelijke horizon ruimer wordt, neemt zijn bescheidenheid toe.

'(According) as a person's mental horizon broadens, his modesty increases.'

b. Het is naarmate iemands geestelijke horizon mimer wordt dat zijn bescheidenheid toeneemt.

'It is (according) as a person's mental horizon broadens that his modesty increases.'

c. *Het is naarmate dat, dat zijn bescheidenheid toeneemt.

'*It is (according) as that, that his modesty increases.'

The same exceptional clefting configuration [-p,+c] not only shows up with the near-synonyms (al) naargelang (dat) ('according(ly) as') and met dat ('with'), but also with vooraleer ('before') in (26):

(26) a. Vooraleer ik begreep wat er gaande was, was hij alweer vertrokken.

'Before I realized what was going on, he had already left again.'

b. Het was (nog) vooraleer ik begreep water gaande was dat hij alweer vertrokken was.

'It was (even) before I realized what was going on that he had already left again.'

c. *Het was (nog) vooraleer dat, dat hij alweer vertrokken was.

'*It was (even) before that, that he had already left again.'

Although the majority of the nonproportional epitactic elements of type [0_00] exclude clefting, as in (24), the conjunctions naarmate and vooraleer in (25) and (26), which are of the same [0_00] type, demonstrate that clefting behavior has to be considered as an autonomous syntacticosemantic criterion, independent of that of proportionality. (15) In other words, absence of pronominal clefting [-p] does not necessarily entail absence of clausal clefting [-c].

5.4. Pronominal clefting without clausal clefting [+p,-c]

The second intermediate clefting type exhibits the reverse discrepancy: allowing pronominal clefting [+p] but rejecting clausal clefting [-c]. Unlike the first type, it is not restricted to one particular proportionality constellation. With rectional elements of type [1_00], such as alsof ('as if') in (27), which have only external but no internal proportionality, clausal clefting is excluded in (27b) in spite of the acceptability of its pronominal counterpart in (27c):

(27) a. Hij keek me aan alsof hij me niet begreep.

'He looked at me as if he didn't understand me.'

b. *Het was alsof hij me niet begreep dat hij me aankeek.

'*It was as if he didn't understand me that he looked at me.'

c. Het was zo dat hij me aankeek.

'It was that way that he looked at me.'

d. Alsof hij me niet begreep, zo was het dat hij me aankeek.

'As if he didn't understand me, that way it was that he looked at me.'

e. Het was zo, (namelijk) alsof hij me niet begreep, dat hij me aankeek.

'It was that way, (namely) as if he didn't understand me, that he looked at me.'

The ungrammaticality of the standard clausal cleft in (27b) can be overcome in two different ways. In (27d) pronominal clefting is combined with the left-dislocation of the fully lexicalized subclause. (16) In addition to this "lexicalization through left-dislocation," Dutch also allows "lexicalization through apposition": in (27e) pronominal clefting of zo ('that way') combines with the apposition of the clause introduced by alsof ('as if'). Both cases can be considered as "double marking" constructions, that is, constructions in which both the pronoun and its clausal lexicalization co-occur. (17) These two alternative clefting patterns, which will henceforth be referred to as "double-marked" clefts, differ from one another in that the subclause in (27d) occupies a position outside the scope of the clefting mechanism, whereas in (27e) the subclause is within its scope.

The discrepancy between allowing pronominal clefting but rejecting clausal clefting [+p,-c] also occurs with a number of subclauses which only have assertive but no suspensive internal proportionality, that is, type [0_10]. With phrasal conjunctions such as met de beperking dat ('with the restriction that') clausal clefting is problematic in (28b) in spite of the acceptability of its pronominal counterpart in (28c):

(28) a. Ik heb de job aangenomen met de beperking dat ik alleen's avonds wil werken.

'I have accepted the job with the restriction that I only want to work evenings.'

b. ??Het is met de beperking dat ik alleen's avonds wil werken dat ik de job heb aangenomen.

'??It is with the restriction that I only want to work evenings that I have accepted the job.'

c. Het is met die beperking dat ik de job heb aangenomen.

'It is with that restriction that I have accepted the job.'

d. (Met de beperking) dat ik alleen's avonds wil werken, met die beperking is het dat ik de job heb aangenomen.

'(With the restriction) that I only want to work evenings, with that restriction is it that I have accepted the job [lit.].'

e. Het is met die beperking, (namelijk) dat ik alleen's avonds wil werken, dat ik de job heb aangenomen.

'It is with that restriction, (namely) that I only want to work evenings, that I have accepted the job.'

The reduced acceptabilityis of the clausal clefting in (28b) can be overcome by means of the same two double-marked clefting strategies as in (27d) and (27e). In (28d) pronominal clefting is combined with the left-dislocation of the fully lexicalized subclause, that is, outside the scope of the clefting mechanism. In (28e) pronominal clefting combines with the apposition of the dat-clause (which thus falls within the scope of the clefting mechanism and) which lexicalizes the demonstrative pronoun die ('that') in the phrase met die beperking ('with that restriction'). (19) The examples with type [0_11] tot(dat) ('until') in (29) run parallel to those in (28): the discrepancy between clausal and pronominal clefting in (29b) and (29c) is resolved by the two double-marked clefting constructions in (29d) and (29e):

(29) a. We stellen de vergadering uit tot(dat) ze terug is.

'We postpone the meeting until she returns.'

b.??Het is tot(dat) ze terug is dat we de vergadering uitstellen.

'??It is until she returns that we postpone the meeting.'

c. Het is tot dan dat we de vergadering uitstellen.

'It is until then that we postpone the meeting.'

d. Tot(dat) ze terug is, tot dan is het dat we de vergadering uitstellen.

'Until she returns, until then is it that we postpone the meeting [lit].'

e. Het is tot dan, (namelijk) tot(dat) ze terug is, dat we de vergadering uitstellen.

'It is until then, (namely) until she returns, that we postpone the meeting.'

So far, the discrepancy between a problematic clausal cleft and an acceptable pronominal cleft showed up with alsof ('as if') in (27)--which was only externally proportional [1_00]--and with met de beperking dat ('with the restriction that') and tot(dat) ('until') in (28) and (29)--which were only internally proportional, namely [0_10] and [0_11], respectively. The remaining subclass, illustrated by in de zin dat ('in the sense that') in (30), combines the characteristics of the two previous ones: being of the endotactic type [1_11], it is both internally proportional to in die/welke zin ('in that/which sense') and externally to hoe/zo ('how/that way'): (20)

(30) a. Ik heb zijn antwoord geinterpreteerd in de zin dat hij toch zou komen.

'I have interpreted his reply in the sense that he would be coming anyway.'

b. ??Het is in de zin dat hij toch zou komen dat ik zijn antwoord heb geinterpreteerd.

'??It is in the sense that he'd be coming anyway that I have interpreted his reply.'

c. Het is in die zin dat ik zijn antwoord heb geinterpreteerd. In welke zin is het dat ...?

'It is in that sense that I have interpreted his reply. In what sense is it that ...?'

c'. Het is zo dat ik zijn antwoord heb geinterpreteerd. Hoe is het dat ...?

'It is that way that I have interpreted his reply. How is it that ...?'

This double proportionality is reflected in the two pronominal clefting possibilities (30c) and (30c'). Consequently, in order to overcome the reduced acceptability of (30b), the two strategies of double-marked clefting illustrated in (27d)-(27e) and (28d)-(28e), are available both with internal proportionality in (30d) and (30e) and with external proportionality in (30d') and (30e'):

(30) d. (In de zin) dat hij toch zou komen, in die zin is het dat ik zijn antwoord heb geinterpreteerd.

'(In the sense) that he would be coming anyway, in that sense is it that I have interpreted his reply.'

d'. (In de zin) dat hij toch zou komen, zo is het dat ik zijn antwoord heb geinterpreteerd.

'(In the sense) that he'd come anyway, that way is it that I've interpreted his reply.'

e. Het is in die zin, (namelijk) dat hij toch zou komen, dat ik zijn antwoord heb geinterpreteerd.

'It is in that sense, (namely) that he would be coming anyway, that I have interpreted his reply.'

e'. Het is zo, (namelijk) in de zin dat hij toch zou komen, dat ik zijn antwoord heb geinterpreteerd.

'It is that way, (namely) in the sense that he would be coming anyway, that I have interpreted his reply.'

Both the internal pronominal constituent in die zin ('in that sense') in (30d) and the external zo ('that way') in (30d') combine pronominal clefting with the left-dislocation of the fully lexicalized subclause which is outside the scope of the cleft. The subclauses in (30e) and (30e'), by contrast, occur inside the cleft and in apposition to the pronominal constituent. In (30e) the dat-clause in apposition is a lexicalization of the demonstrative pronoun die ('that') in the internally proportional in die zin ('in that sense') phrase, whereas the one introduced by in de zin dat ('in the sense that') in (30e') is a further specification of the externally proportional zo ('that way').

5.5. Correlations between proportionality and clefting." summary

The table in (31) below summarizes the correlations between the different levels of proportionality and the different clefting constellations. The various proportionality configurations on the vertical axis are represented by means of the shorthand notation introduced in (11). The clefting possibilities on the horizontal axis are captured in terms of the three binary parameters [[+ or -] pronominal], [[+ or -] double-marked], and [[+ or -] clausal]. The numbers in parentheses refer to the example sentences in the preceding paragraphs, whereas the numbers between square brackets indicate how many of the 154 conjunctions listed in the Appendix belong to which particular type.
(31)

 [+pronom] [+pronom] [-pronom] [-pronom]
 [+doub-m] [+doub-m] [-doub-m] [-doub-m]
 [+clausal] [-clausal] [+clausal] [-clausal]

[1_11] tot[(dat).sub.1] in die/de zin
 (20) [dat.sub.l] (30)
 [N = 13] [N = 1]

[1_10] nadat (19)
 [N = 5]

[1_00] toen (18)
 [N = 13]

[0_11] onder tot[(dat).sub.2]
 voorwaarde dat (29)
 (21) [N = 7] [N = 1]

[0_10] in de met de ondanks
 veronderstelling beperking (het feit)
 [dat.sub.l] (22) dat (28) dat (23)
 [N = 11] [N = 7] [N = 8]

[0_00] alhoewell
 (24)
 [N = 67]


In the first four rows of (31) all elements allow both pronominal and double-marked clefting so that the fight-hand side of the table is empty. In all these cases a complete paradigm (i.e. both assertive and suspensive proportionality) is available: the first three rows have a complete external paradigm, whereas the fourth has a complete internal one. The presence of suspensive proportionality, be it external or internal, entails pronominal clefting and hence double-marked clefting as well. In other words, pronominal syntagms which can constitute the topic of a question can occupy the focus position of a cleft construction.

The sixth row in (31), which has no proportionality whatsoever, exhibits exactly the mirror-image configuration: pronominal and double-marked clefting are excluded by definition (i.e. the bottom cells in the first two columns are empty). Moreover, nearly all elements in this row oppose clefting of any type. As was discussed in connection with (25) and (26), the expressions that do exceptionally allow clausal clefting apparently exhibit a less subordinate relationship with the main predication (see the medium grey cell in row six).

The fact that in the remaining fifth row the expressions are distributed across three different columns shows that no predictions as to clefting possibilities can be made on the basis of only minimal, that is, internal assertive proportionality. This non-predictability entails that for this particular subclass of subordinating elements the pronominal clefting criterion, namely, the opposition [+p] versus [-p], constitutes a MINIMAL DISTINCTIVE FEATURE. In the other five rows, by contrast, it is merely a corollary of the proportionality configurations: presence of (suspensive) proportionality entails [+p] in the top four rows, whereas (complete) absence of proportionality entails [-p] in row six. In other words, suspensive proportionality is a sufficient but not a necessary condition for pronominal clefting: the absence of suspensive proportionality (in row five) does not entail the impossibility of pronominal clefting.

On the horizontal axis, four different clefting constellations are distinguished. Since these were at first defined in terms of three binary parameters, namely pronominal, clausal, and double-marked clefting, eight logical combinations should be available in principle. As was demonstrated above, however, double-marked clefting has to be considered as a corollary of pronominal clefting: any pronominal referent, be it clefted or not, allows double-marking with a further (lexical or pronominal) specification. Hence, the apparent eight logical possibilities are automatically reduced to the actual four of the table in (31).

With the two most important constellations, viz. columns one and four, the values for pronominal and clausal clefting coincide, whereas with the two remaining central columns, pronominal and clausal clefting do not converge. The presence of elements such as naarmate dat and vooraleer in the medium-grey cell at the bottom of the third column, which allow clausal clefting without the possibility of pronominal clefting, demonstrates that for the non-proportional expressions in row six, clausal clefting is a minimal distinctive feature. Similarly, the impossibility of (pure) clausal clefling with [1_00] rectional conjunctions such as alsof in (27b) confirms that distinctive status (see the medium-grey cell in row three, column two).

As shown in (28b), (29b), and (30b), by contrast, clausal clefting may not be impossible in the same strict sense but seems highly dispreferred with the remaining groups of expressions in this second column, that is, the light-gray cells on rows one [1_11], four [0_11], and five [0_10]. In other words, these elements might be incorporated into the neighboring cells of the first column. In sum, at least for rows three [1_00] and six [0_00], clausal clefting has to be retained as a distinctive feature: in the former case, it distinguishes between columns one and two within the [+pronominal] left-hand side of the table, in the latter case it does so between columns three and four within the [-pronominal] fight-hand side.

6. Fronting and left-dislocation of subordinate clauses

The paradigmatic test procedures of external and internal proportionality introduced in Sections 2 and 3 provided the foundation for the proposed classification of subordinating conjunctions. This categorization was further refined in Sections 4 and 5 by means of the linear (i.e. syntagmatic) criterion of clefting. In this section, the various possibilities of fronting subordinate clauses will be dealt with in detail, and in the next section the correlations between the criteria of clefting and fronting will be discussed.

Although subclauses in the unmarked word-order occupy the sentence-final position (Haeseryn et al. 1997: 1367), as in (32a) and (33a), the majority of subordinating conjunctions also allow their subclause to be fronted to sentence-initial position, as in (32b) and (33b) below:

(32) a. We hadden alle ramen dicht gedaan vooraleer we vertrokken.

'We had closed all the windows before we left.'

b. Vooraleer we vertrokken hadden we alle ramen dicht gedaan.

'Before we left, had we closed all the windows.'

c. Vooraleer we vertrokken, toen hadden we alle ramen dicht gedaan.

'Before we left, then had we closed all the windows.'

d. *Vooraleer we vertrokken, we hadden alle ramen dicht gedaan.

'Before we left, we had closed all the windows.'

(33) a. We worden milder in ons oordeel naargelang (dat) we ouder worden.

'We grow milder in our judgement as we grow older.'

b. Naargelang (dat) we ouder worden worden we milder in ons oordeel.

'As we grow older do we grow milder in our judgement.'

c. *Naargelang (dat) we ouder worden, zo/dan worden we milder in ons oordeel.

'*As we grow older, so/then do we grow milder in our judgement.'

d. *Naargelang (dat) we ouder worden, we worden milder in ons oordeel.

'As we grow older we grow milder in our judgement.'

As is well known (Haeseryn et al. 1997: 1261ff.; Zwart 1998: 369; among others), fronting in Dutch standardly triggers inversion of the finite verb and the subject if a nonsubject constituent is fronted: the subject we follows the finite auxiliary forms hadden ('had') in (32b) and worden ('be') in (33b). The vooraleer ('before') clause in (32c) can furthermore be left-dislocated, with its proportional proform toen ('then') in sentence-initial position. (21) It is precisely the occurrence of the latter element in that position which again provokes inversion in the main clause. For subclauses with non-proportional conjunctions, no such proform is available for a double-marking constellation, and hence a main clause with inversion cannot combine with a left-dislocated subclause, as with naargelang (dat) ('as') in (33c). The combination of left-dislocation and lack of inversion, that is, with the subject we ('we') in sentence-initial position, turns out to be excluded both with the proportional vooraleer ('before') in (32d) and the non-proportional naargelang (dat) ('as') in (33d). (22)

Although vooraleer ('before') in (32) and naargelang (dat) ('as') in (33) only differ with respect to the acceptability of the c-variant, this contrast reflects fundamentally distinct degrees of dependency on the verbal predicator as introduced above. Subclauses like those in (32) fall "within" the pronominally controllable realm of the verbal predicator and were labelled endotactic; those in (33) that do not meet this proportionality criterion were called epitactic. In this respect, vooraleer in (34), the reading of which crucially differs from that of vooraleer in (32), resembles the epitactic naargelang (dat) ('as') in (33). In contrast to the toen ('then') in (32c), the proform dan ('then') which is inserted in sentence-initial position in (34c) and which again correlates with inversion, cannot be interpreted as belonging to a double-marking constellation with the left-dislocated vooraleer-clause. This is due to the absence of proportionality between the dan and the subclause introduced by vooraleer in this particular reading:

(34) a. *Jan komt ook naar het feest, vooraleer ik het vergeet.

'*John is coming to the party too, before I forget.'

b. *Vooraleer ik bet vergeet, komt Jan ook naar het feest.

'*Before I forget, is John coming to the party too.'

c. *Vooraleer ik het vergeet, dan komt Jan ook naar het feest.

'*Before I forget, then John is coming to the party too.'

d. Vooraleer ik het vergeet, Jan komt ook naar het feest.

'Before I forget, John is coming to the party too.'

In spite of the similarity between (33c) and (34c) in terms of absence of proportionality, as opposed to (32c), the overall constellation in (34) is fundamentally different from those in (32) and (33). Unlike vooraleer ('before') in (32a) and naargelang (dat) ('as') in (33a), the subclause introduced by vooraleer ('before') in (34a) does not readily fit the sentence-final position. (23) Moreover, fronting of this particular type of vooraleer clause in (34b) does not combine with the main clause inversion which is obligatory in (32b) and (33b). Instead, the only acceptable pattern in (34) combines left-dislocation of the subclause in (34d) with the standard main clause word order in which the subject Jan precedes the finite verb. (24) Precisely this combination of left-dislocation and lack of inversion is excluded both with vooraleer ('before') in (32d) and naargelang (dat) ('as') in (33d).

On the basis of these fundamental differences between (33) and (34), the basic distinction between endotactic and epitactic constructions needs to be refined. The proportional constructions in (32) will still be called endotactic. Although both (33) and (34) are not endotactic, the label epitactic will be reserved for the constellation in (33), whereas that in (34) will henceforth be called "EXOTACTIC." In the epitactic realm, the link with the verbal predicator is weaker because of the reduced proportionality but is not completely absent since the fronting of the subclause still triggers inversion in the main clause. In exotactic configurations, by contrast, even this weaker form of connection is lacking. In other words, subclauses introduced by exotactic conjunctions as in (34) no longer attribute properties to the state of affairs reported on in the main clause. This lack of an explicit link with the main clause predicator leads to different interpretations of the subclause. (25) The subclause with vooraleer ('before') in (34d), for instance, functions as an independent proposition which motivates and comments on the main clause utterance. The "forgetting" in the subclause does not relate to 'John's coming to the party' but instead to the (speech) act of uttering the main clause, that is, 'my informing you of the fact that John is coming to the party.' (26)

The comparison of (32) and (34) reveals that the two overall constellations are one another's perfect mirror-image: the acceptability of (32a)-(32c) is opposed to the non-acceptability of (34a)-(34c), whereas (32d) is excluded but (34d) is not. In other words, the endotactic and exotactic configurations constitute opposite ends of the syntactic continuum, with various epitactic configurations lying in between.

It should be added that the unmarked word order of the declarative main clause in (34d) is accompanied by an obligatory (intonation) gap, namely, either a considerable pause between the fronted subordinate clause and the main clause or else a "tone rupture," both indicating the end of a syntactic section (henceforth marked as %, see Dooren and Eynde 1982: 211-212; Haeseryn et al. 1997: 1100, 1338). In contrast to the standard pattern in (32b) and (33b), where the whole of subclause and main clause counts as a single intonational domain, the obligatory (intonation) gap in (34d) reflects the fact that both clauses constitute an independent intonational domain. This can be seen as a formal indication of a weaker degree of connectedness between subordinate and main clause.

Moreover, in the unmarked case of fronting with inversion but without intonation gap, fronting may--but need not--be combined with extra stress on the fronted constituent (Zwart 1998: 373-374; Sleeman et al. 2000:297-298). In other words, presence or absence of particular tonal morphemes on the fronted constituent is not predictable. Whether or not the subclause is fronted, sentence final intonation can fall either on the subclause, as in (35), or on any main clause constituent, as in (36). (27) Henceforth, an exclamation mark before the stressed syllable will be used to generalize over both types. For the notions of tonal morpheme, stressability, and section, see Dooren and Eynde (1982) and Eynde and Dooren (1983).
(35) a. Ik heb het gezien toen hij b. Ik heb het !toen gezien
 al !weg was.

 'I noticed when he had 'I noticed then.'
 already left.'

 a'. Toen hij al !weg was, heb b'. !Toen heb ik het gezien
 ik het gezien

 'When he had already 'Then I noticed.'
 left, I noticed.'

(36) a. (!)Ik heb het ge(!)zien b. (!)Ik heb het toen
 toen hij al weg was. ge(!)zien
 'I noticed when he had 'I noticed then.'
 already left.'

 a'. Toen hij al weg was, heb b'. Toen heb (!)ik het
 (!)ik het ge(!)zien ge(!)zien
 'When he had already 'Then I noticed.'
 left, I noticed.'


If sentence-final intonation does fall on the fronted constituent the resulting configuration is another instance of the general binarization strategy introduced in connection with (14) for the clefting mechanism. (28) In other words, the intonation pattern of (35a') and (35b') sets apart one term, that is, the temporal rection constituent, from the rest of the construction. (29) Complex examples such as these, however, fall outside the scope of this article.

A small number of subordinating conjunctions, however, do not exhibit the full variation in intonational patterns illustrated in (35) and (36). More in particular, when the subclause is fronted and triggers inversion in the main clause--as in (32b) and (33b)--there is a strong tendency for the sentence stress to fall on a subclause constituent, as in (37b), and not on a main clause constituent, as in (38b): (30)

(37) a. Ik heb Jan mijn fiets geleend, tot hij een !nieuwe heeft gekocht.

'I lent John my bike until he has bought a new one.'

b. Tot hij een !nieuwe heeft gekocht, heb ik Jan mijn fiets geleend.

'Until he has bought a new one did I lend John my bike.'

(38) a. (!)Ik heb Jan mijn fiets ge(!)leend, tot hij een nieuwe heeft gekocht.

'I lent John my bike until he has bought a new one.'

b.??Tot hij een nieuwe heeft gekocht, heb (!)ik Jan mijn fiets ge(!)leend.

'Until he has bought a new one did I lend John my bike.'

A limited number of subordinating conjunctions does not allow the subclause they introduce to appear in sentence initial position. With the proportional ones, such as alsof ('as if'), the fronted version of (39b) can be improved by means of double marking with the pronominal zo (Haeseryn et al. 1997: 1392), as in (39c):

(39) a. Hij keek alsof hij me niet begreep.

'He looked as if he did not understand me.'

b. *Alsof hij me niet begreep keek hij.

'*As if he did not understand me did he look.'

c. Alsof hij me niet begreep, zo keek hij

'As if he did not understand me, that way did he look.'

Finally, with some non-proportional subordinators, such as zodat ('so that'), fronting is excluded either way, witness (40b) and (40c). Most of these convey the notions of comparison or consequence (Haeseryn et al. 1997: 1263): (31)

(40) a. We zijn te laat vertrokken zodat we de trein hebben gemist.

'We left too late so that we have missed the train.'

b. *Zodat we de trein hebben gemist zijn we te laat vertrokken.

'*So that we have missed the train did we leave too late.'

c. *Zodat we de trein hebben gemist, zo zijn we te laat vertrokken.

'*So that we have missed the train, that way did we leave too late.'

In sum, as far as fronting possibilities for subclauses are concerned, five basic subtypes have been distinguished. The majority of conjunctions display the standard fronting pattern, as in (32) and (33), triggering inversion (the [f_VS] type). A considerable number of conjunctions, such as those in (34), constitute a second subclass, characterized by an intonation gap and lack of inversion (the [f_%SV] type). Thirdly, the examples in (37) and (38) represent the smallest class, for which fronting imposes intonational constraints on the main clause (the [f_!VS] type). A fourth type illustrated in (39) excludes clausal fronting as such but does allow the double marked pattern (the [f_DM] type). Finally, those subordinators excluding any type of clausal fronting, as in (40), will henceforth be referred to as the [f_no] group. The different types of fronting are summarized in the list below:

(41) a. [f_no] clausal fronting excluded

b. [f_DM] clausal fronting only with double marking

c. [f_!VS] clausal fronting with inversion and intonational restrictions

d. [f VS] clausal fronting with inversion

e. [f %SV] clausal fronting without inversion and with obligatory (intonation) gap

The next section investigates the way in which the various types of fronting behavior in (41) interact with the four clefting constellations proposed in Section 5.

7. Correlations between clefting and fronting

In the syntactic classification of subordinating conjunctions elaborated so far, the two syntagmatic (i.e. linear) criteria of clefting and fronting have been described independently of one another. Still, both mechanisms have in common that they bring the subordinate clause into prominence. Therefore, the relationship between these two criteria should be dealt with before they can finally be integrated with those of proportionality in the following section.

The five fronting types of (41) are mapped onto the vertical axis of the summarizing table in (42). On the horizontal axis, the four clefting constellations are represented. As discussed in Section 5, these can be characterized in terms of two binary parameters, namely, [[+ or -]p] for pronominal--and double-marked--clefting and [[+ or -]c] for clausal clefting.

The fourth row of the table in (42) contains about half of the subordinating conjunctions under investigation. These elements belong to the unmarked fronting type [f_VS] without intonation gap and with subject-verb inversion in the main clause. Although this type of fronting readily combines with all four clefting constellations, more than half of the conjunctions--that is, the ones in the leftmost column--exhibit both pronominal and clausal clefting [+p, +c].
(42)

Fronting Clefting

 [+p,+c] [+p,-c] [-p,+c] [-p,-c]

[f_no] zodat (40)
 [N=9]

[f_DM] alsofl (39)
 [N = 12]

[f_!VS] tot(dat)l in de zin dat2 opdat [N = 1]
 (37) [N = 4]
 [N=3]

[f_VS] vooraleerl; in de mate datl (al) naargelang aangezien1
 toen [N = 7] (dat)2 (33) [N = 31]
 (32/35) [N = 61
 [N = 46]

[f_%SV] vooraleer4
 (34)
 [N = 35]


Both the three top rows and the bottom row, by contrast, show significant incompatibilities between the various fronting and clefting subtypes, namely, empty cells, which are not all distributed evenly over the table. The general assumption that elements which cannot be fronted do not allow clefting is borne out by the expressions in the first row of (42). In other words, non-frontable elements [f_no] do not allow clefting at all [-p, -c]. For the elements in row two, problems with both clausal clefting and clausal fronting are overcome by the same mechanism of double marking: that is, [f_DM] correlates with [+p, -c]. All twelve elements of [+p, -c] in the second row furthermore turn out to be valency elements. The restricted fronting type [f_!VS] in row three is characterized by the sentence-final intonation pattern within the fronted subclause. The empty cell in the rightmost column in row three indicates that this particular fronting type always occurs with elements for which (at least some form of) clefting is available. This distribution may relate to the fact that both pronominal and clausal clefting standardly show the sentence-final intonation pattern as well.

The three empty cells in the fifth row reflect the fact that the [f_%SV] type with intonation gap but without inversion is incompatible both with pronominal and clausal clefting. As a matter of fact, the presence of an intonation gap without inversion entails the absence of clefting (i.e. it is restricted to type [-p, -c]). Hence, by contraposition, the presence of clefting, that is, either [+p] or [+c] or both, renders an intonation gap without inversion impossible. These two correlations corroborate the general hypothesis that the proximity relation between the subclause and its verbal predicator which clefting requires, is obviously in conflict with the absence of such proximity relation, as triggered by the intonation gap.

In the general overview of the next section, the observed correlations between the syntagmatic criteria of clefting and fronting will be integrated with the preliminary classification of the subordinating conjunctions based on the paradigmatic criteria of external and internal proportionality.

8. Overview

The aim of this section is to present a fine-grained syntactic typology of Dutch subordinating conjunctions based on a hierarchical set of formal criteria. In order to do so, the correlations between proportionality and clefting proposed in (31) of Section 5 will be integrated with the correspondences between clefting and fronting summarized in (42) of Section 7.
The resulting table, containing all 154 conjunction
readings, is presented in (43) below. (32)

(43)
Endotactic

 PROP CLEFT FRONT SUBORDINATORS N = 154

Valency 1_11 +p+c f_VS van te cA = 1

 +p-c f_DM in die/de zin dat1 cB = 1

 1_00 +p-c f_DM (al)eer3; als3; als5; cC = 11
 alsofl; alvorens2; of
 (alsof); om te3; tot
 (dat)3; voor(dat)5;
 vooraleer5; wanneer3

 f_VS datl; of ... of (dat); cD = 3
 of

Rection 1_11 +p+c f_!VS tot(dat)1 cE = 1

 f_VS door te; doordat; in de cF = 11
 mate dat2; in het geval
 dat; in zover(re) (dat)1;
 om tel; omdatl; sedert
 (dat)1; sinds (dat)1;
 tegen datl; vanaf dat1

 1_10 +p+c f_VS met te; na te; nadat; cG = 5
 voor(dat)1; vooraleerl

 1_00 +p+c f_!VS zoals1 cH = 1

 f_VS alsl; als2; eenmaal cI = 12
 (dat); ingeval1; telkens
 (als/wanneer); terwijl1;
 toen; wanneer1; wanneer2;
 zodra; zolang (dat/als)2;
 zonder tel

 0_11 +p+c f_VS met de bedoeling om/dat; CJ = 8
 met het doel (om); onder
 voorwaarde dat; op
 voorwaarde dat; sedert
 (dat)2; sinds (dat)2;
 tot(dat)2; vanaf dat2

Epitactic 0_10 +p+c f_!VS met dien verstande dat cK = 1

 f_VS (al) naargelang (dat)1; cL = 10
 in afwachting dat1; in de
 overtuiging datl; in de
 veronderstelling datl; in
 de wetenschap datl; in
 het besef dat; met dat1;
 met het idee om/dat; nul;
 voor het geval (dat)1

 +p-c f_!VS in die/de zin dat2; met cM = 3
 die/de beperking dat;
 met die/derestrictie dat;

 f_VS in de mate dat1; in cN = 4
 plaats dat; in plaats
 van te; teneinde te;

 -p-c f_no met als gevolg dat; met cO = 2
 als resultaat dat;

 f_VS alvorens te; alvorens1; cP = 6
 daargelaten dat1;
 niettegenstaande (het
 feit)(dat)l; ondanks
 (het feit) dat; tegen
 dat2

Exotactic 0_00 -p+c f_!VS opdat cQ = 1

 f_VS (al) naargelang (dat)2; cR = 6
 indien1; met dat2;
 naarmate (dat);
 voor(dat)2; vooraleer2

 -p-c f_no als6; alsof2; dat2; cS = 7
 dat3; met het gevolg
 dat; met het resultaat
 dat; zodat;

 f_VS (al)eer1; aangezien1; cT = 25
 alhoewel1; als ... al1;
 alsof3; daar1; gezien
 (dat); hoewel1; in de
 hoop dat/te1; indien
 ... al1; mits; naar;
 ofschoon1; tenzij (dat)1;
 terwijl2; vermits1; voor
 zover(re) (dat)1; voor
 (dat)3; vooraleer3; zo1;
 zo2; zoals2; zolang (als/
 dat)1; zonder dat; zonder
 te2

 0_00 -p-c f_%SV (al)eer2; aangezien2; al; cU = 35
 alhoewel2; als ... a12;
 als4; daar2; daargelaten
 dat2; hoewel2; in
 afwachting dat2; in de
 hoop dat/te2; in de
 overtuiging dat2; in de
 veronderstelling dat2;
 in de wetenschap dat2;
 in zover(re) (dat)2;
 indien ... a12; indien2;
 ingeval2; met dat3;
 niettegenstaande (het
 feit)(dat)2; nu2; of
 schoon2; om te2; omdat2;
 ook al; tenzij (dat)2;
 terwijl3; vermits2; voor
 het geval (dat)2; voor
 zover(re) (dat)2; voor
 (dat)4; vooraleer4; zo
 ... al; zoals3; zonder
 te3


A first fundamental distinction captured in (43) is that between endotactic conjunctions in the upper half and non-endotactic--that is, epitactic and exotactic--elements in the lower half. It is based on the respective presence or absence of suspensive/interrogative proportionality. Within the endotactic group, conjunctions introducing (predicator-specific) valency constituents are distinguished from those introducing (predicator-general) elements of rection. Each of the three criteria used in (43), namely, proportionality, clefting, and fronting, provides an independent argument for this distinction: (i) only the rectional group is characterized by an intermediate [1_10] type in which assertive and suspensive internal proportionality do not coincide, (ii) only the valency group allows [+p, -c] clefting in which pronominal and clausal clefting diverge, and (iii) in contrast to the elements of rection, only a few valency constituents allow straightforward clausal fronting, most of them requiring double marking instead.

Both with valency, rectional and epitactic conjunctions the complex criterion of proportionality (with its external and internal dimension) yields a first major subclassification. Obviously, the central role of this criterion is prompted by the methodological assumptions of the pronominal approach the present article subscribes to. Furthermore, this predominance of proportionality is motivated by the fact that going from proportionality towards clefting has greater predictive power than taking clefting as the classificational starting point, as expressed in the entailment relations of (44).

(44) a. IF suspensive proportionality, i.e. [1_11], [1_10], [1_00], and [0_11],

THEN pronominal clefting, i.e. [+p,+c] or [+p,-c] only.

b. IF no proportionality, i.e. [0_00],

THEN no pronominal clefting, i.e. [-p,+c] or [-p,-c] only.

Due to the fact that the [0_10] proportionality subtype is compatible with three different clefting constellations, this type is attributed an intermediate position between the suspensively proportional elements and the truly epitactic ones. Moreover, this entails that both if-then correlations in (44) have to be defined in terms of sufficient but not necessary conditions. The greater predictive power of proportionality as compared to clefting thus accounts for the hierarchical ordering between the two in the table of (43). (33)

The position of clefting and fronting with respect to one another is motivated in a similar, though more indirect way. It can be demonstrated that the correlations between clefting and proportionality are stronger than those between fronting and proportionality. Within the endotactic and epitactic realm any fronting possibility combines with any proportionality configuration. This lesser degree of predictability is reflected in the subordinate status of fronting as the third parameter of the typology in (43).

With the exotactic expressions at the bottom of the table, however, the particular fronting type of [f_%SV] does allow one to predict the proportionality and clefting types, as captured in the implication rule of (45a):

(45) a. IF (intonation gap [f_%SV]), THEN (absence of proportionality [0_00] AND absence of clefting [-p,-c]).

b. IF NOT (absence of proportionality [0_00] AND absence of clefting [-p,-c]), THEN NOT (intonation gap [f_%SV]).

b'. IF NOT (absence of proportionality [0_00]) OR NOT (absence of clefting [-p,-c]), THEN NOT (intonation gap [f_%SV]).

It is important to stress that the implication in (45a) cannot be reversed: the two subclasses of conjunctions in the second and third row from the bottom are both [0_00] and clefting type [-p,-c] but do not combine with the obligatory intonation gap, because they either do not allow fronting or else still trigger main clause inversion. In other words, for those expressions which are both [0_00] and clefting type [-p,-c] (in the bottom three rows of the table), the fronting type is a distinctive feature. The formulae in (45b) and (45b'), derived from (45a) by contraposition, express the fact that the obligatory intonation gap is excluded as long as either minimal proportionality or minimal clefting remains.

The close examination of the implicational links between the various syntactic criteria in (43) yields the implication rule in (46), where "selected" refers to predicator-specific (i.e. valency) subordination, "questioning" refers to the availability of external and/or internal suspensive proportionality, and "front_VS" comprises the [f_VS], [f_DM], and [f_!VS] types of (43):

(46) selected [??] questioning [??] proform clefting [??] clause/DM clefting [??] front)_VS

Contrary to our initial stance, the presence of an assertive proform without suspensive counterpart does not exhibit any implicational connection with any of the criteria in (46). As a consequence, the feature "minimal asserting" appears as the second rightmost column in the table (47), only to be followed by the exotactic criterion of fronting with intonation gap %. The categories cA, cB, and so on, in column four correspond to those in the rightmost column of (43):

For the five characteristics in the rule of (46), a positive value in a given column of (47) entails a positive value in any of the columns to its right. The proposed hierarchy of characteristics leads to a gradual subclassification ranging from the availability of all five characteristics (at the top of the table) all the way down to the complete absence of any of the five characteristics (at the bottom of the table). In the latter case, the availability of internal assertive proportionality and the fronting with obligatory intonation gap in the two fight-most columns allow a further distinction between the bottom three rows. How the formal-syntactic continuum in (47) underpins a functional-semantic classification will briefly be shown in the next section.

9. Epilogue: syntactico-semantics versus lexico-semantics

One of the basic assumptions of the valency studies in the framework of the pronominal approach is that the observed formal distributional complexity uncovers a considerable part of the syntactico-semantics of (verbal and nonverbal) predicators. Analogously, the detailed formal description of the syntactic behavior of the subordinating conjunctions presented in the body of this article provides the foundation for a fine-grained classification of these expressions. Importantly, this syntactico-semantic classification should not be seen as replacing the standard lexico-semantic categories of temporality, conditionality and so on, but as providing an independent dimension of semantic organization cross-cutting--and hence enriching--the former. In this respect the descriptive results of this article can be seen as an independent corroboration of Lehmann's central claim that "the grammatical types that will emerge [...] cut across semantically different clause linkage relations" (Lehmann 1988: 183). First of all, it will be illustrated how the syntactico-semantic classification results in a layered structuring of the lexico-semantic category of temporality. Secondly, the present analysis will briefly be compared with other approaches exhibiting a similar layered architecture.

The hierarchical classification of subordinating conjunctions proposed in (47) can be interpreted as an overview of the different combinatorial relations between verbal predicators (i.e. verbal construction kernels) and subordinate clauses introduced by those conjunctions. It yields a continuum ranging from close proximity to and dependency on the predicator at the top end of the table (in the case of the valency-bound elements), all the way to complete absence of such connectedness at the bottom end (in the case of the non-proportional exotactic elements). Because of the central role assigned to the main clause predicator in the present analysis, Lehmann's parameter of hierarchical downgrading--which takes the possible independence of the adverbial clause as its starting point (Lehmann 1988: 189)--is as it were reversed into HIERARCHICAL UPGRADING: a general tendency towards increasing detachment or independence from the main clause (predicator).

The table in (48) first of all illustrates the complex syntactico-semantic structure or hierarchy inside a single lexico-semantic category, namely, that of temporal succession. Secondly, the polysemy of individual conjunctions such as vooraleer ('before') is reflected in different networks of characteristics. Furthermore, the table gives an example for each of the syntactico-semantic networks not covered by the readings of vooraleer, and furnishes tentative labels for the different degrees of integration with the predicator. (34) Within the category of epitactic conjunctions, the secondary distinction between "intrinsic" and "extrinsic" refers to the presence or absence of internal (i.e. minimal) assertive proportionality respectively: intrinsic refers to type [0_10], whereas extrinsic refers to type [0_00]:

(48)
(48)
 Temporal Sel Quest [p]
 succession cleft

Endo Valency [vooraleer + + +
 .sub.5]

 Rection [vooraleer - + +
 .sub.1]

 Correlative intrinsic [in - - +
 afwachting
 dat.sub.1]

 extrinsic [vooraleer - - -
 .sub.2]

Epitaxis Interpretative intrinsic [tegen dat - - -
 .sub.2]

 extrinsic [vooraleer - - -
 .sub.3]

 Commentative intrinsic (met als - - -
 gevolg dat)

 extrinsic (zodat) - - -

Exo Illocutive [vooraleer - - -
 .sub.4]

 [p] [c]
 cleft cleft

Endo Valency + +

 Rection + +

 Correlative intrinsic + +

 extrinsic - +
 -
Epitaxis Interpretative intrinsic -

 extrinsic - -

 Commentative intrinsic - -

 extrinsic - -

Exo Illocutive - -

 Front Min. Front
 + VSO assert + %

Endo Valency + + -

 Rection + + -

 Correlative intrinsic + + -

 extrinsic + - -

Epitaxis Interpretative intrinsic + + -

 extrinsic + - -

 Commentative intrinsic - + -

 extrinsic - - -

Exo Illocutive - - +


On the basis of the three formal syntactic criteria of proportionality, clefting, and fronting, five readings have been distinguished for the conjunction vooraleer ('before'):

(49) a. [Vooraleer.sub.1] ik vertrok, heb ik alle ramen dicht gedaan. 'Before I left, I had closed all the windows.'

b. [Vooraleer.sub.2] ik begreep wat er gaande was, was hij alweer vertrokken. 'Before I understood what was going on, he had already left again.'

c. [Vooraleer.sub.3] je een oplossing vindt, moet je soms flink zoeken. 'Before you find a solution, you sometimes need to search hard.'

d. [Vooraleer.sub.4] ik het vergeet, je komt toch ook naar het feest? 'Before I forget, you are coming to the party too, aren't you?'

e. Het duurde een tijdje [vooraleer.sub.5] ik begreep water gaande was. 'It took a while before I understood what was going on.'

[Vooraleer.sub.1] in (49a) can be set apart from the other four since it is the only rectional type, that is, proportional to the pronominal paradigm of dan/ wanneer ('then/when'). It constitutes the core reading of temporal succession between the two events in the main and subclause that are thus conceived of as only being connected to one another through the dimension of time. [Vooraleer.sub.2] in (49b), which belongs to the class of (extrinsic) CORRELATIVES in (48), then differs from its counterparts in (49c) and (49d) in that it allows clefting. Although the relation of temporal succession is still present, it does not fully cover the semantic load of this entry. More in particular, the event in the vooraleer-clause is evaluated negatively as taking place too late (i.e. at the moment of leaving there is no understanding) and is no longer seen as purely factual (i.e. the understanding need not have taken place at all). This extra dimension of unfulfilled expectation may relate to the fact that the main clause in (49b) cannot be negated, whereas that of (49a) can.

The difference between [vooraleer.sub.3] in (49c) and [Vooraleer.sub.4] in (49d) is that the former causes inversion in the main clause but does not allow an intonation gap between main and subclause, whereas the latter does not trigger inversion but does require an intonation gap. As in (49b), the concept of succession in time is only part of the complex reading in (49c). In this case, the main clause event is presented as a (necessary) condition or cause for the subclause event to become a fact (i.e. sometimes, if one does not search hard, one does not find a solution). Because of this extra semantic dimension, [vooraleer.sub.3] is assigned to the class of (extrinsic) INTERPRETATIVES.

With [vooraleer.sub.4] in (49d) the notions of temporal succession and causality are intertwined as well, but this time the subclause event of forgetting is not related to the event of coming to the party, but to the speech act event of asking a question which underlies the interrogative main clause (i.e. if I ask now, then I will not forget). The latter type of relationship between main and subclause, which is to be situated on the level of discourse or pragmatics rather than on that of semantics stricto sensu, exclusively shows up with those exotactic conjunctions which are of the non-proportional [0_00] class and which require an intonation gap. They are labelled ILLOCUTIVES in the table in (48).

With [vooraleer.sub.5] the temporal reference as such is taken as a participant (i.e. subject VALENCY) of the main clause predicator. There is no proportionality with the temporal pronominal paradigm of dan/wanneer ('then/ when'), but with the participant paradigm of dat/wat ('that/what'), as is illustrated by the double-marked fronting pattern in (49e'):

(49) e'. [Vooraleer.sub.5] ik begreep water gaande was, dat duurde een tijdje. 'Before I understood what was going on, that took a while.'

As far as the correlative elements are concerned, in afwachting [dat.sub.1] ('in the expectation of') counts as the intrinsic counterpart of extrinsic [vooraleer.sub.2] since it has internal assertive in afwachting daarvan ('in the expectation thereof'). Similarly, the intrinsic interpretative tegen [dat.sub.2] ('by the time that') is distinguished from the extrinsic [vooraleer.sub.3] because of its assertive proform tegen dan ('by that time'). Although with the COMMENTATIVES met als gevolg dat ('with the consequence') and zodat ('so that') temporal succession and causality necessarily coincide, the consequence reading predominates. In these cases fronting of the subclauses is excluded.

The fact that different expressions within one and the same category are situated in various positions along the continuum from endotactic to exotactic subordination, as illustrated in (48), is obviously not restricted to the realm of temporality. Analogous gradual transitions can, for instance, be observed within the lexico-semantic categories of conditionality, causality, degree, and purpose.

The main conclusion of this article is that the concept of a continuum or a hierarchy, which shows up in many recent approaches to clause combining, can convincingly be accounted for in terms of the complex interaction between the independent syntactic criteria of proportionality, clefting, and fronting. Foley and Van Valin's (1984) hierarchy of syntactic bondedness, for instance, is based both on the fundamental distinction between embeddedness and dependency and on that between the three (functional) nexus types of nucleus, core, and periphery. Whereas embeddedness corresponds to proportionality and dependency to both clefting and fronting, the nexus types as such are not used as a classifying criterion in the present analysis. Nevertheless, they show up as the result of the interplay between the formal criteria in that nucleus and core basically correspond to the endotactic level of valency and rection, whereas periphery comprises both (the different subtypes of) the epitactic and the exotactic level. Similar correlations can be established with the layered approach proposed by Hengeveld (1998). His four (entity type) levels of predicate-predication-proposition-speech act to a great extent correspond to the levels of valency, rection, epitaxis, and exotaxis.

On the continuum from subordination to parataxis, Hopper and Traugott (1993) locate adverbial clauses in the intermediate realm of hypotaxis, which itself displays a "continuum of looser-to-tighter integration" (Hopper and Traugott 1993: 177). Within their framework of grammaticalization, however, the level of hypotaxis received relatively little attention as compared to that of subordination, which stands for maximal integration into the main clause. The exhaustive analysis of Dutch adverbial subordinators in the present article, by contrast, describes the syntactico-semantic richness of these hypotactic elements in more detail. As demonstrated in (48), the complex interaction between various formal-syntactic characteristics leads to the further subdivision of the epitactic realm into the more tightly integrated CORRELATIVE elements, the intermediate INTERPRETATIVE elements, and the least integrated COMMENTATIVE elements.

It is important to stress that the complex interaction between the independent syntactic criteria of proportionality, clefting, and fronting is language-specific. In other words, the particular implicational links for Dutch in (46) and (47) do not as such carry over to the analysis of subordinating elements in other languages, not even within the group of closely related Germanic languages. However, a fundamental distinction needs to be made between the (paradigmatic) criterion of proportionality on the one hand, and the two (syntagmatic) criteria of clefting and fronting on the other hand. The mechanism of proportionality is much more general since all languages have a system of minimal referentials--either pronominal or lexical (see Hiz 1970)--serving as its basis. Although the notion of substitutability plays a role in Thompson and Longacre's (1985: 177) semantic classification of adverbial clauses, the present proposal crucially differs from theirs in two respects. First of all, minimal referentials, that is, simple or complex proforms, are used instead of monomorphemic, non-anaphoric adverbs. Secondly, the notion of proportionality is refined by distinguishing between the external and the internal type. Unlike proportionality, the clefting and fronting mechanisms are language-specific. As will have become clear from the examples discussed, even two closely related languages such as Dutch and English, in which both mechanisms do occur, differ with respect to the actual realization. The difference in word order between Dutch main clauses and subclauses has no counterpart in English, and the same holds for inversion triggered by the fronting of a nonsubject constituent. Nevertheless, languages in general do possess morphological, syntactic, or prosodic strategies allowing the language user to change the informative weight of one or the other constituent.

Therefore, the language-specific mechanisms for bringing the subclause into prominence need to be identified and related to the proportionality mechanism in order to arrive at comparable classifications of subordinating elements.

Appendix

The syntactic characteristics of (the distinct readings of) every subordinator are given between square brackets. First of all, the major category is indicated: endotactic valency (ENV), endotactic rection (ENR), epitaxis (EPI), or exotaxis (EXO). After that, the particular constellations for proportionality, clefting, and fronting are listed, using the shorthand systems introduced in the body of the article. For apparently identical constellations see note 32.

aangezien1: "since, as, seeing (that), given (that)" [EPI, 0_00, -p-c, f_VS] 'Aangezien hij ziek was is hij moeten thuisblijven.' ('Since he was ill, he had to stay at home.')

aangezien2: "since" [EXO, 0_00, -p-c, f_%SV] 'Aangezien je het me vraagt, ik heb het gedaan.' ('Since you ask me about it, I've done it.')

al: "even though" [EXO, 0_00, -p-c, f_%SV] 'Al doe je nog zo je best, je haalt het toch niet.' ('Even though you do so your best, you will not/cannot succeed.')

alhoewel1: "although, even though" [EPI, 0_00, -p-c, f_VS] 'Alhoewel iedereen er tegen is, doe ik het toch.' ('Although everyone is against it, I will do it anyway.')

alhoewel2: "although, even though" [EXO, 0_00, -p-c, f_%SV] 'Alhoewel iedereen er tegen is, ik doe het toch.' ('Although everyone is against it, I will do it anyway.')

als ... al1: "if ... at all" [EPI, 0_00, -p-c, f_VS] 'Als hij al komt, zal hij toch weer dwarsliggen.' ('If he comes at all, he will be a troublemaker.')

als ... al2: "if ... at all" [EXO, 0_00, -p-c, f_%SV] 'Als hij al komt, hij zal het toch niet gedaan hebben.' ('If he comes at all, he won't have done it.')

als1: "if" [ENR, 1_00, +p+c, f_VS] 'Als je je best doet mag je mee op reis.' ('If you do your best, you can come along travelling.')

als2: "when" [ENR, 1_00, +p+c, f_VS] 'Als ik haar zie moet ik lachen.' ('When I see her, I have to laugh.')

als3: "if" [ENV, 1_00, +p-c, f_DM] 'Ze zouden het betreuren als Pieter opstapte.' ('They would regret it if Peter were to go away.')

als4: "if" [EXO, 0_00, -p-c, f_%SV] 'Als je mijn mening dan toch wil weten, Jan wil (weer) gelijk hebben.' ('If you really want to know my opinion, John wants to be fight once again.')

als5: "as" [ENV, 1_00, +p-c, f_DM] 'Ze doet als hoort ze me niet.' ('She acts as if she doesn't hear me.')

als6: "as if" [EPI, 0_00, -p-c, f_no] 'Hij bleef zitten, als had hij me niet begrepen.' ('He remained seated, as if he hadn't understood me.')

alsof1: "as if" [ENV, 1_00, +p-c, f_DM] 'Hij keek alsof hij me niet begreep.' ('He looked as if he didn't understand me.')

alsof2: "as if" [EPI, 0_00, -p-c, f_no] 'Hij bleef zitten, alsof hij me niet begrepen had.' ('He remained seated, as if he hadn't understood me.')

alsof3: "as if" [EPI, 0_00, -p-c, f_VS] 'Alsof dat nog niet voldoende was heeft hij ook nog een nieuwe fiets gekocht.' ('As if that was not enough, he also bought a new bike.')

alvorens te: "before" [EPI, 0_10, -p-c, f_VS] 'Alvorens de opdracht te aanvaarden, overlegde ze met haar medewerkers.' ('Before accepting the task, she consulted her assistants.')

alvorens1: "before" [EPI, 0_10, -p-c, f_VS] 'Alvorens ze de opdracht aanvaardde, overlegde ze met haar medewerkers.' ('Before she accepted the task, she consulted her assistants.')

alvorens2: "before" [ENV, 1_00, +p-c, f_DM] 'Het duurde een tijd alvorens ik het begreep.' ('It took a while before I understood.')

daar1: "since" [EPI, 0_00, -p-c, f_VS] 'Daar ervaring een factor van intelligentie is, komt het verstand pas met de jaren.' ('Since experience is a factor of intelligence, brains only come with the years.')

daar2: "since" [EXO, 0_00, -p-c, f_%SV] 'Daar het nu eenmaal zo is, het verstand komt pas met de jaren.' ('Since things are going the way they do, brains only come with the years.')

daargelaten dat1: "notwithstanding the fact that" [EPI, 0_10, -p-c, f_VS] 'Daargelaten dat hij gespiekt heeft was zijn examen niet schitterend.' ('Notwithstanding the fact that he cheated, his exam was not excellent.')

daargelaten dat2: "notwithstanding the fact that" [EXO, 0_00, -p-c, f_%SV] '(Nog) daargelaten dat hij gespiekt heeft, zijn examen was toch niet schitterend.' ('Notwithstanding the fact that he cheated, his exam was not excellent anyway.')

dat1: "that" [ENV, 1_00, +p-c, f_VS] 'Ik betreur het dat hij zich vergist.' ('I regret that he made a mistake.')

dat2: "so that/in order to" [EPI, 0_00, -p-c, f_no] 'Ik zal hollen, dat ik de trein nog haal.' ('I will run, so that I catch the train.')

dat3: "because, since" [EPI, 0_00, -p-c, f_no] 'Hij is zeker kwaad dat hij niet opbelt.' ('He must be angry, since he doesn't phone.')

door te: "by" [ENR, 1_11, +p+c, f_VS] 'Door op deze knop te drukken steek je alle lichten tegelijk aan.' ('By pressing this button, you switch on all the lights at once.')

doordat: "since, because" [ENR, 1_11, +p+c, f_VS] 'Doordat er gebrek aan geld was, moest hij thuisblijven.' ('Because there was not any money, he had to stay at home.')

eenmaal (dat): "once, as soon as" [ENR, 1_00, +p+c, f_VS] 'Eenmaal (dat) de regen voorbij is kunnen we terug naar buiten.' ('Once it stops raining, we can go outside again.')

(al)eer1: "before" [EPI, 0_00, -p-c, f_VS] 'Eer ze dat gespaard hebben zullen ze nog lang moeten werken.' ('They will still have to work for a long time before they have saved that much.')

(al)eer2: "before" [EXO, 0_00, -p-c, f_%SV] 'Eer ik het vergeet, je komt toch ook naar bet feest?' ('Before I forget, you are coming to the party too, aren't you?')

(al)eer3: "before" [ENV, 1_00, +p-c, f_DM] 'Het duurde een tijd eer ik het begreep.' ('It took a while before I understood.')

gezien (dat): "given, since" [EPI, 0_00, -p-c, f_VS] 'Gezien hij een slechte gezondheid had, kon hij niet rondreizen.' ('Given that he had a poor health, he couldn't travel around.')

hoewel1: "(al)though" [EPI, 0_00, -p-c, f_VS] 'Hoewel jij me verveelt blijf ik hier toch zitten.' ('Although you annoy me, I stay here anyway.')

hoewel2: "(al)though" [EXO, 0_00, -p-c, f_%SV] 'Hoewel jij me verveelt, ik blijf hier toch zitten.' ('Although you annoy me, I stay here anyway.')

in afwachting dat1: "awaiting, in the expectation (that)" [EPI, 0_10, +p+c, f_VS] 'In afwachting dat de laatsten aankomen, beginnen we alvast.' ('In the expectation that the last ones arrive, we start already.')

in afwachting dat2: "awaiting, in the expectation (that)" [EXO, 0_00, -p-c, f_%SV] 'In afwachting dat hij je een antwoord geeft, ik vind dat het niet kan.' ('In the expectation that he gives you an answer, I think that it is impossible.')

in de hoop dat/te1: "in the hope that" [EPI, 0_00, -p-c, f_VS] 'In de hoop dat zij meer zou verdienen is zij van werk veranderd.' ('In the hope that she would earn more, she changed jobs.')

in de hoop dat/te2: "in the hope that" [EXO, 0_00, -p-c, f_%SV] 'In de hoop dat het je vooruit kan helpen, hij heeft het gedaan.' ('In the hope that this may help you out, he has done it.')

in de mate dat1: "to the extent that, as far as" [EPI, 0_10, +p-c, f_VS] 'Ik zal je helpen in de mate dat ik tijd heb.' ('I will help you to the extent that I have time.')

in de mate dat2: "according as" [ENR, 1_11, +p+c, f_VS] 'Je wordt beloond in de mate dat je je hebt ingezet.' ('You are rewarded according as you have dedicated yourself.')

in de overtuiging dat1: "in the conviction that" [EPI, 0_10, +p+c, f_VS] 'Ik heb de deur gesloten, in de overtuiging dat iedereen al naar huis vertrokken was.' ('I have closed the door, in the conviction that everybody had already gone home.')

in de overtuiging dat2: "in the conviction that" [EXO, 0_00, -p-c, f_%SV] 'In de overtuiging dat je het begrijpt, de zaken staan er niet goed voor.' ('In the conviction that you understand, business does not look very promising.')

in de veronderstelling dat1: "on the assumption that" [EPI, 0_10, +p+c, f_VS] 'Ik heb de deur gesloten, in de veronderstelling dat iedereen al naar huis vertrokken was.' ('I have closed the door on the assumption that everybody had already gone home.')

in de veronderstelling dat2: "on the assumption that" [EXO, 0_00, -p-c, f_%SV] 'In de veronderstelling dat je het dit keer zal begrijpen, de zaken staan er niet goed voor.' ('On the assumption that you understand this time, business does not look very promising.')

in de wetenschap dat1: "in the knowledge that" [EPI, 0_10, +p+c, f_VS] 'Hij heeft niets gezegd in de wetenschap dat het voor paniek zou zorgen.' ('He hasn't said anything, in the knowledge that it would provoke panic.')

in de wetenschap dat2: "in the knowledge that" [EXO, 0_00, -p-c, f_%SV] 'In de wetenschap dat je het niet leuk zal vinden, de zaken staan er niet goed voor.' ('In the knowledge that you won't like it, business does not look very promising.')

in die/de zin dat1: "in the sense that" [ENV, 1_11, +p-c, f_DM] 'Ik heb het geinterpreteerd in de zin dat hij toch zou komen.' ('I have interpreted it in the sense that he would be coming anyway.')

in die/de zin dat2: "in the sense that" [EPI, 0_10, +p-c, f_!VS] 'Ik heb hem geholpen in die/de zin dat ik de helft van zijn taak heb opgelost.' ('I have helped him, in the sense that I solved half of his assignment.')

in bet besef dat: "aware of the fact that" [EPI, 0_10, +p+c, f_VS] 'Hij heeft zijn aandelen verkocht in het besef dat de zaak dreigt failliet te gaan.' ('He sold his shares aware of the fact that the company is going bankrupt.')

in het geval dat: "in case" [ENR, 1_11, +p+c, f_VS] 'Ik blijf thuis in het geval dat hij meegaat.' ('I stay at home in case he joins us.')

in plaats dat: "instead of" [EPI, 0_10, +p-c, f_VS] 'In plaats dat hij zelf kwam, stuurde hij een briefje.' ('Instead of coming himself, he sent a note.')

in plaats van te: "instead of" [EPI, 0_10, +p-c, f_VS] 'In plaats van zelf te komen, stuurde hij een briefje.' ('Instead of coming himself, he sent a note.')

in zover(re) (dat)1: "in as/so far as" [ENR, 1_11, +p+c, f_VS] 'Hij heeft het recht om in te grijpen in zoverre zijn belang ermee gemoeid is.' ('He has the right to intervene in as far as his own interest is involved.')

in zover(re) (dat)2: "in as/so far as" [EXO, 0_00, -p-c, f_%SV] 'In zoverre je er lets mee te maken hebt, ik doe toch mee.' ('In so far as you have anything to do with that, I am participating anyway.')

indien ... al1: "if ... at all" [EPI, 0_00, -p-c, f_VS] 'Indien hij al zou komen, zal hij toch niet werken.' ('If he were to show up at all, he won't work anyway.')

indien ... al2: "if ... at all" [EXO, 0_00, -p-c, f_%SV] 'Indien hij al komt, hij zal het toch niet gedaan hebben.' ('If he were to show up at all, he will not have done it.')

indien1: "in case" [EPI, 0_00, -p+c, f_VS] 'De cursus wordt nog eens ingericht indien er voldoende belangstelling is.' ('The course will be organized again, in case there is sufficient interest.')

indien2: "if" [EXO, 0_00, -p-c, f_%SV] 'Indien je het toch wil weten, ik heb gisteren het contract getekend.' ('If you want to know anyway, I signed the contract yesterday.')

ingeval1: "in case" [ENR, 1_00, +p+c, f_VS] 'Ingeval men u niet toelaat, mag u zich op mij beroepen.' ('In case you are not admitted, you can appeal to me.')

ingeval2: "in case" [EXO, 0_00, -p-c, f_%SV] 'Ingeval het u interesseert, ik heb hier nog een interessant boek liggen.' ('In case you are interested, I have an interesting book here.')

met als gevolg dat: "with the consequence that" [EPI, 0_10, -p-c, f_no] 'We hebben de reis niet grondig voorbereid met als gevolg dat we nu nergens terecht kunnen.' ('We didn't prepare the journey properly with the consequence that we cannot go anywhere now.')

met als resultaat dat: "with the result that" [EPI, 0_10, -p-c, f_no] 'We hebben de reis niet grondig voorbereid met als resultaat dat we nu nergens terecht kunnen.' ('We didn't prepare the journey properly with the result that we cannot go anywhere now.')

met dat1: "as" [EPI, 0_10, +p+c, f_VS] 'Met dat hij naar huis ging konden we geen vragen meer stellen.' ('As he went home, we couldn't ask any further questions.')

met dat2: "as" [EPI, 0_00, -p+c, f_VS] 'Met dat iemands geestelijke horizon ruimer wordt, neemt zijn bescheidenheid toe.' ('As a person's mental horizon broadens, his modesty increases.')

met dat3: "as, since" [EXO, 0_00, -p-c, f_%SV] 'Met dat we bier onder vier ogen kunnen spreken, ik heb het gedaan.' ('As/since we are alone here, I've done it.')

met de bedoeling om/dat: "with the intention of, in order that" [ENR, 0_11, +p+c, f_VS] 'We hebben de reis goed voorbereid met de bedoeling dat de groep zoveel mogelijk zou kunnen zien.' ('We have prepared the trip very well, in order that the group could see as much as possible.')

met die/de beperking dat: "with the restriction that" [EPI, 0_10, +p-c, f_!VS] 'Ik heb de job aangenomen met de beperking dat ik alleen's avonds kan werken.' ('I have accepted the job with the restriction that I can only work evenings.')

met die/de restrictie dat: "with the restriction that" [EPI, 0_10, +p-c, f_!VS] 'Ik heb de job aangenomen met de restrictie dat ik alleen's avonds kan werken.' ('I have accepted the job with the restriction that I can only work evenings.')

met dien verstande dat: "on the understanding that" [EPI, 0_10, +p+c, f_!VS] 'Ik heb hem geld geleend, met dien verstande dat hij het terugbetaalt zodra hij kan.' ('I have lent him money on the understanding that he pays it back as soon as he can.')

met het doel (om): "with a view to" [ENV, 0_11, +p+c, f_VS] 'Hij kwam met het doel om lets te verkopen.' ('He came with a view to selling something.')

met het gevolg dat: "with the result that" [EPI, 000, -p-c, f_no] 'We hebben de reis niet grondig voorbereid met bet gevolg dat we nu nergens terecht kunnen.' ('We didn't prepare the journey properly with the result that we cannot go nowhere now.')

met het idee om/dat: "with the idea that/of" [EPI, 0_10, +p+c, f_VS] 'Hij kwam met het idee om iets te verkopen.' ('He came with the idea of selling something.')

met het resultaat dat: "with the result that" [EPI, 0_00, -p-c, f_no] 'We hebben de reis niet grondig voorbereid met het resultaat dat we nu nergens terecht kunnen.' ('We didn't prepare the journey properly with the result that we cannot go nowhere now.')

met te: "by" [ENR, 1_10, +p+c, f_VS] 'Met onbeleefd op te treden heb je alle krediet verspeeld.' ('By being impolite, you have lost all credit.')

mits: "if, provided that" [EPI, 0_00, -p-c, f_VS] 'Je kunt mijn fiets even lenen, mits jeer goed zorg voor draagt.' ('You can borrow my bike, provided that you take good care of it.')

na te: "after" [ENR, 1_10, +p+c, f_VS] 'Ik ging naar huis na alle lichten te hebben gedoofd.' ('I went home after having turned out all the lights.')

naar: "according to, as" [EPI, 0_00, -p-c, f_VS] 'Naar we vernemen is het daar slecht weer.' ('As we were told, the weather is bad there.')

(al) naargelang (dat)1: "as, depending on" [EPI, 0_10, +p+c, f_VS] 'Naargelang (dat) we jong of oud zijn, zijn we strenger of milder in ons oordeel.' ('Depending on whether we are young or old, we are stricter or milder in our judgement.')

(al) naargelang (dat)2: "as" [EPI, 0_00, -p+c, f_VS] 'Naargelang (dat) we ouder worden, worden we milder in ons oordeel.' ('As we grow older, we become milder in our judgement.')

naarmate (dat): "(according) as" [EPI, 0_00, -p+c, f_VS] 'Naarmate iemands geestelijke horizon mimer wordt, neemt zijn bescheidenheid toe.' ('As a person's mental horizon broadens, his modesty increases.')

nadat: "after" [ENR, 1_10, +p+c, f_VS] 'Ik ging naar huis nadat ik alle lichten had gedoofd.' ('I went home after having turned out all the lights.')

niettegenstaande (her feit)(dat)1: "notwithstanding/despite/in spite of the fact that" [EPI, 0_10, -p-c, f_VS] 'De partijen bereikten een compromis niettegenstaande hun meningen sterk uiteenlopen.' ('The parties achieved an agreement in spite of the fact that their opinions diverge a lot.')

niettegenstaande (het feit)(dat)2: "notwithstanding/despite/in spite of (the fact)" [EXO, 0_00, -p-c, f_%SV] 'Niettegenstaande je het niet leuk zal vinden, de zaken staan er niet goed voor.' ('Notwithstanding the fact that you won't like it, business does not look very promising.')

nu1: "now that" [EPI, 0_10, +p+c, f_VS] 'Nu hij Parijs veroverd had kon hij zich voorbereiden voor Londen.' ('Now that he had conquered Paris, he could prepare for London.')

nu2: "now that" [EXO, 0_00, -p-c, f_%SV] 'Nu ik je zie, Jan heeft zich weer vergist.' ('Now that I see you, John has made a mistake again.')

of (alsof): "as if" [ENV, 1_00, +p-c, f_DM] 'Hij deed of hij me niet hoorde.' ('He acted as if he didn't hear me.')

of ... of (dat): "whether or not" [ENV, 1_00, +p-c, f_VS] 'Ik weet niet of ze erin geslaagd is of niet.' ('I don't know whether she has succeeded or not.')

of: "whether" [ENV, 1_00, +p-c, f_VS] 'Ik weet niet of ik het u al gezegd heb.' ('I don't know whether I have already told you.')

ofschoon1: "(al)though" [EPI, 0_00, -p-c, f_VS] 'Ik merk niet dat ik verder kom, ofschoon ik mijn best doe.' ('I don't see that I make any progress, although I do my best.')

ofschoon2: "(al)though" [EXO, 0_00, -p-c, f_%SV] 'Ofschoon je het niet leuk zal vinden, de zaken staan er niet goed voor.' ('Although you won't like it, business does not look very promising.')

om te1: "in order to" [ENR, 1_11, +p+c, f_VS] 'Ik heb bet gedaan om er zo vlug mogelijk vanaf te zijn.' ('I have done it in order to get rid of it as soon as possible.')

om te2: "to" [EXO, 0_00, -p-c, f_% SV] 'Om een lang verhaal kort te maken, hij heeft bet wel degelijk gedaan.' ('To make a long story short, he has done it indeed.')

om te3: "to" [ENV, 1_00, +p-c, f_DM] 'Het is moeilijk om zoiets te begrijpen.' ('It is difficult to understand such a thing.')

omdat1: "because" [ENR, 1_11, +p+c, f_VS] 'Ik werd beloond omdat ik mijn werk goed had gedaan.' ('I was rewarded because I had done a good job.')

omdat2: "because" [EXO, 0_00, -p-c, f_%SV] 'Omdat je het toch wil weten, ik beb het gedaan.' ('Because you really want to know it, I have done it.')

ondanks (hetfeit) dat: "despite the fact that" [EPI, 0_10, -p-c, f_VS] 'Ondanks dat ze op bezoek kwam ging hij weg.' ('Despite the fact that she would come and visit, he left.')

onder voorwaarde dat: "under the condition that" [ENR, 0_11, +p+c, f_VS] 'Ik ga met je mee, onder voorwaarde dat je vriendelijk bent.' ('I'll join you under the condition that you are kind.')

ook al: "even though" [EXO, 000, -p-c, f_%SV] 'Hij geeft veel aan de armen, ook al is hij zelf niet rijk.' ('He gives a lot to the poor, even though he's not rich himself.')

op voorwaarde dat: "provided that/on condition that" [ENR, 0_11, +p+c, f_VS] 'Ik wil je helpen op voorwaarde dat je eerst je kamer opruimt.' ('I want to help you on condition that you clean up your room first.')

opdat: "in order that" [EPI, 0_00, -p+c, f_!VS] 'We hebben de reis grondig voor-bereid opdat niemand zou klagen.' ('We prepared the journey properly in order that nobody would complain.')

sedert (dat)1: "since" [ENR, 1_11, +p+c, f_VS] 'Ik werk daar al sedert (dat) ik terug in Belgie ben komen wonen.' ('I've been working there ever since I moved back to Belgium.')

sedert (dat)2: "since" [ENR, 0_11, +p+c, f_VS] 'Zij is ongelukkig sedert (dat) ze naar de stad verhuisd is.' ('She has been unhappy [ever] since she moved into town.')

sinds (dat)1: "since" [ENR, 1_11, +p+c, f_VS] 'Ik werk daar al sinds (dat) ik terug in Belgie ben komen wonen.' ('I've been working there ever since I moved back to Belgium.')

sinds (dat)2: "since" [ENR, 0_11, +p+c, f_VS] 'Zij is ongelukkig sinds (dat) ze naar de stad verhuisd is.' ('She has been unhappy [ever] since she moved into town.')

tegen dat1: "by the time that" [ENR, 1_11, +p+c, f_VS] 'Tegen dat je thuiskomt, zal het huis opgeruimd zijn.' ('By the time you come home, the house will be cleaned up.')

tegen dat2: "by the time that" [EPI, 0_10, -p-c, f_VS] 'Tegen dat de zaak voorkomt, heeft bet schip de haven al lang verlaten.' ('By the time the case gets to court, the ship will have left the port for a long time.')

telkens (als/wanneer): "whenever" [ENR, 1_00, +p+c, f_VS] 'Telkens (als/ wanneer) ik haar aankijk, lacht zij.' ('Whenever I look at her, she smiles.')

teneinde te: "in order to" [EPI, 0_10, +p-c, f_VS] 'Hare Majesteit gaf toe, teneinde te voorkomen dat er een opstand zou uitbreken.' ('Her Majesty surrendered in order to avoid that a rebellion would arise.')

tenzij" (dat)1: "unless" [EPI, 0_00, -p-c, f_VS] 'De winkel is altijd open tenzij de straat opengebroken wordt.' ('The shop is always open unless the street is broken up.')

tenzij (dat)2: "unless" [EXO, 0_00, -p-c, f_%SV] 'Tenzij je bezwaren hebt, ik zal de brief wel schrijven.' ('Unless you have any objections, I will write the letter.')

terwijl1: "while" [ENR, 1_00, +p+c, f_VS] 'Ik heb de krant gelezen terwijl de baby sliep.' ('I read the newspaper while the baby was sleeping.')

terwijl2: "whereas/although" [EPI, 0_00, -p-c, f_VS] 'Terwijl ik hem uitdrukkelijk gevraagd had thuis te blijven is hij toch komen opdagen.' ('Whereas/ although I had explicitly asked him to stay home, he showed up anyway.')

terwijl3: "since, while" (EXO, 0_00, -p-c, f_%SV] 'Terwijl je hier nu toch bent, we kunnen net zo goed gaan zwemmen.' ('Since you are here now anyway, we can just as well go for a swim.')

toen: "when" [ENR, 1_00, +p+c, f_VS] 'Toen hij aan de macht kwam, verbeterde de toestand.' ('When he came into power, the situation improved.')

tot(dat)1: "until" [ENR, 1_11, +p+c, f_!VS] 'Tot(dat) hij de verkiezingen verloor verbeterde de toestand.' ('Until he lost the elections, the situation improved.')

tot(dat)2: "until" [ENR, 0_11, +p+c, f_VS] 'We stellen de vergadering uit tot(dat) ze terug is.' ('We postpone the meeting until she gets back.')

tot(dat)3: "until" [ENV, 1_00, +p-c, f_DM] 'Het duurde nogal tot(dat) hij hier geraakt was.' ('It took forever until he had gotten here.')

van te: "by, from" [ENV, 1_11, +p+c, f_VS] 'Van hard te werken word je niet ziek.' ('One does not get ill from working hard.')

vanaf dat1: "from the moment that" [ENR, 1_11, +p+c, f_VS] 'Ik werk daar al vanaf dat ik terug in Belgie ben komen wonen.' ('I've been working there from the moment that I moved back to Belgium.')

vanaf dat2: "from the moment that" [ENR, 0_11, +p+c, f_VS] 'Vanaf dat je achttien jaar bent kun je een rijbewijs halen.' ('From the moment you turn eighteen, you can get a driver's licence.')

vermits1: "since" [EPI, 0_00, -p-c, f_VS] 'Vermits hij ziek was is hij moeten thuisblijven.' ('Since he was ill he has had to stay home.')

vermits2: "since" [EXO, 0_00, -p-c, f_%SV] 'Vermits je bet me vraagt, hij is moeten thuisblijven.' ('Since you ask me, he has had to stay home.')

voor het geval (dat)1: "in case" [EPI, 0_10, +p+c, f_VS] 'Hij heeft zijn paraplu meegebracht voor het geval dat het regent.' ('He has brought his umbrella [just] in case it rains.')

voor het geval (dat)2: "in case" [EXO, 0_00, -p-c, f_%SV] 'Voor het geval hij erom vraagt, ik heb het gedaan.' ('In case he askes about it, I have done it.')

voor zover(re) (dat)1: "as far as" [EPI, 0_00, -p-c, f__VS] 'Voor zover ik me herinner had het iets met syntaxis te maken.' ('As far as I remember, it had something to do with syntax.')

voor zover(re) (dat)2: "as far as, in so far as" [EXO, 0_00, -p-c, f_%SV] 'Voor zover het nog niet duidelijk is, de zaken staan er niet goed voor.' ('In so far as it is not clear already, business does not look very promising.')

voor(dat)1: "before" [ENR, 1_10, +p+c, f_VS] 'Voor(dat) ik vertrok heb ik alle ramen dicht gedaan.' ('Before I left, I have closed all windows.')

voor(dat)2: "before" [EPI, 0_00, -p+c, f_VS] 'Voor(dat) ik begreep water gaande was, was hij alweer vertrokken.' ('Before I understood what was going on, he had already left again.')

voor(dat)3: "before" [EPI, 0_00, -p-c, f_VS] 'Voor(dat) je een oplossing vindt, moet je soms flink zoeken.' ('Before you find a solution, you sometimes need to search hard.')

voor(dat)4: "before" [EXO, 0_00, -p-c, f_% SV] 'Voor(dat) ik het vergeet, oma komt morgen op bezoek.' ('Before I forget, grandma is visiting tomorrow.')

voor(dat)5: "before" [ENV, 1_00, +p-c, f_DM] 'Het duurde een tijd voor(dat) ik het begreep.' ('It took a while before I understood.')

vooraleer1: "before" [ENR, 1_10, +p+c, f_VS] 'Vooraleer ik vertrok heb ik alle ramen dicht gedaan.' ('Before I left, I have closed all windows.')

vooraleer2: "before" [EPI, 000, -p+c, f_VS] 'Vooraleer ik begreep water gaande was, was hij alweer vertrokken.' ('Before I understood what was going on, he had already left again.')

vooraleer3: "before" [EPI, 0_00, -p-c, f_VS] 'Vooraleer je een oplossing vindt, moet je soms flink zoeken.' ('Before you find a solution, you sometimes need to search hard.')

vooraleer4: "before" [EXO, 0_00, -p-c, f_%SV] 'Vooraleer ik het vergeet, je komt toch ook naar het feest?' ('Before I forget, you are also coming to the party, aren't you?')

vooraleer5: "before" [ENV, 1_00, +p-c, f_DM] 'Het duurde een tijd vooraleer ik het begreep.' ('It took a while before I understood.')

wanneer1: "when/if" [ENR, 1_00, +p+c, f_VS] 'Wanneer je je best doet mag je mee op reis.' ('If you do your best, you can come along travelling.')

wanneer2: "when(ever)" [ENR, 100, +p+c, f_VS] '(Telkens) warmeer ik haar zie moet ik lachen.' ('When[ever] I see her, I have to laugh.')

wanneer3: "when" [ENV, 1_00, +p-c, f_DM] 'Het zal wel erg opvallen wanneer we nu weer te laat komen.' ('It will definitely get noticed when we are late once again.')

zo ... al: "if ... at all" [EXO, 000, -p-c, f_%SV] 'Zo hij het al wist, hij heeft niet gereageerd.' ('If in fact he did know at all, he certainly didn't respond.')

zo1: "if" [EPI, 0_00, -p-c, f_VS] 'Zo zij het wist zou zij het wel zeggen.' ('If in fact she did know at all, she would certainly say so.')

zo2: "as far as" [EPI, 0_00, -p-c, f_VS] 'Zo ik hoor moet hij vanavond nog vertrekken.' ('As far as I hear, he still has to leave this evening.')

zoals1: "like, the way that" [ENR, 1_00, +p+c, f_!VS] 'Doe het maar zoals je het geleerd hebt.' ('Please do it the way you have learned it.')

zoals2: "as" [EPI, 0_00, -p-c, f_VS] 'Zoals ik al voorspeld had is hij gezakt voor zijn examen.' ('As I had predicted before, he failed his exam.')

zoals3: "as" [EXO, 0_00, -p-c, f_%SV] 'Zoals ik al voorspeld had, hij is gezakt voor zijn examen.' ('As I had predicted before, he failed his exam.')

zodat: "so (that)" [EPI, 0_00, -p-c, f_no] 'Hij heeft niet ontbeten zodat hij nu al honger heeft.' ('He hasn't had breakfast, so that he is already hungry now.')

zodra: "as soon as, once" [ENR, 1_00, +p+c, f_VS] 'Zodra ik geld heb, betaal ik u mijn schulden terug.' ('As soon as I have some money, I'll pay you back my debts.')

zolang (als/dat)1: "as long as" [EPI, 0_00, -p-c, f_VS] 'Hiertegen bestaat geen bezwaar zolang (als) er maar niets gebeurt.' ('There is no objection to this as long as nothing happens.')

zolang (dat/als)2: "as long as" [ENR, 1_00, +p+c, f_VS] 'De kinderen speelden rustig zolang de hond buiten was.' ('The children were playing quietly as long as the dog was outside.')

zonder dat: "without" [EPI, 0_00, -p-c, f_VS] 'Hij heeft bet uurwerk gestolen zonder dat iemand iets merkte.' ('He stole the watch without anybody noticing anything.')

zonder te1: "without" [ENR, 1_00, +p+c, f_VS] 'Zonder het papier te scheuren heeft hij bet pakje opengemaakt.' ('Without tearing the wrapping, he opened the present.')

zonder te2: "without" [EPI, 0_00, -p-c, f_VS] 'Zonder te willen ontkennen dat hij zijn best gedaan heeft vind ik dat hij de prijs niet verdiend heeft.' ('Without denying that he has done his best, I think that he doesn't deserve the price.')

zonder te3: "without" [EXO, 0_00, -p-c, f_%SV] 'Zonder te willen ontkennen dat hij zijn best gedaan heeft, hij heeft de prijs niet verdiend.' ('Without denying that he has done his best, I think that he doesn't deserve the price.')

Received 23 July 2002

Revised version received

11 September 2003

Katholieke Universiteit Leuven

F. W. O.-Vlaanderen

Notes

(1.) This research was supported by grants from the F.W.O.-Vlaanderen to Hans Smessaert (post-doctoral research fellow) and Dagmar Divjak (research assistant). We are greatly indebted to Jose Tummers for his "critico-constructive" participation in the research project leading up to Cornillie et al. (2001). Furthermore, we would like to express our gratitude to Pieter Seuren, Claire Blanche-Benveniste, and Jan Scharpe, as well as to the anonymous referees of the journal for their pertinent remarks on an earlier version of the present article. Correspondence address: K. U. Leuven, Department of Linguistics, Blijde-Inkomststraat 21, 3000 Leuven, Belgium. E-mail: Hans.Smessaert@arts.kuleuven.ac.be.

(2.) A detailed introduction to the framework is presented in Eynde and BlancheBenveniste (1978), Blanche-Benveniste et al. (1987 [1984]), Eynde et al. (1998) and Eynde et al. (2002). The special issue of the International Journal of Lexicography (volume 7) dealt with "A pronominal approach to valency dictionaries" (Melis and Eggermont 1994). A list of publications can be found at http://bach.arts.kuleuven.ac.be/PA/.

(3.) The term "referentials" is borrowed from Hiz (1970). The more general notion of minimal referentials allows an analogous treatment of those languages whose referential skeleton is not expressed by means of pronominal paradigms.

(4.) A related group of assertive elements are the PARANOUNS, such as someone, something, everywhere, or never, which lack the above mentioned features but instead involve extra features of quantification, as is illustrated for French in Eynde et al. (1988).

(5.) The two paradigms in (lb), for instance, differ from one another in that the subject paradigm of betwijfelen ('doubt') cannot contain elements such as dat/wat/iets/... ('that/ what/something') whereas its object paradigm cannot contain elements such as hem/ haar / wie / iemand /... ('him/her/who/somebody/...').

(6.) The traditional position can be found in the ANS (Haeseryn et al. 1997: 458-459). In Paardekooper (n.d.: 551-552), however, the fundamental relationship between the traditional pronouns and these so-called adverbial elements is reflected in the terminological distinction between vn and 'vn' (where vn stands for voornaamwoord, i.e., pronoun). Whereas Paardekooper's 'vn' category already includes expressions of time, place, and manner, the definition of minimal referential adopted in the pronominal approach allows the extension to paradigms of cause or reason, such as daarom/waarom ... ('therefore/why'), and of degree or measure, such as zozeer/hoezeer ... ('to that/which extent'), which exhibit exactly the same proportionality relations.

(7.) The "phrasal conjunction" in het besef dat (the Dutch "voegwoordelijke uitdrukking'" [lit. "conjunctional expression"] Haeseryn et al. 1997: 543) relates to the "phrasal preposition" in het besef van (the Dutch "voorzetsel-uitdrukking" [lit. "prepositional expression"] Haeseryn et al. 1997: 530-538). Not all phrasal conjunctions have a prepositional counterpart: in de wetenschap dat ('in the knowledge that') versus *in de wetenschap van ('*in the knowledge of').

(8.) Phrasal conjunctions such as naargelang dat ('according as'), in afwachting dat ('in expectation of'), or in plaats dat ('instead of') lack a determiner position and, hence, lack the prototypical internal proportionality discussed so far. Nevertheless, they do exhibit an alternative form of assertive internal proportionality: naargelang DAARvan ('according thereto/to that'), in afwachting DAARvan ('in expectation thereof/of that'), or in plaats DAARvan ('instead thereof/of that'). These forms are related to their phrasal prepositional counterpart, namely, naargelang van ('according to'), in afwachting van ('in expectation of'), or in plaats van ('instead of'). However, the presence of a determiner, as with in het besef dat, need not exclude the alternative type: in het besef DAARvan Occurs side by side with in dat besef

(9.) With a number of prepositions, the internal proportional counterparts do not use this standard conversion mechanism, but reflect archaic morphological processes. This is the case, for instance, with endotactic sindsDIEN ('since that one' [lit.]) or epitactic DESondanks ('that in spite of' [lit.]).

(10.) A similar distinction holds between the abstract spatial paradigm hier/daar/waar ('here/there/where') on the one hand, and more specific paradigms such as hierop/ daarop/waarop ('on this/that/what') or hierin/daarin/waarin ('in this/that/what') on the other.

(11.) As was pointed out in Section 3.1, the two logically possible combinations [101] and [0_01], which have suspensive but no assertive internal proportionality, are excluded.

(12.) Clefting can be seen as a "binarization" mechanism (Blanche-Benveniste et al. 1984: 37ff.), that is, a mechanism which sets apart one term (syntagm) from the rest of the construction. As a result, both components of this opposition relation can be independently modalized (Eynde and Van Durme 1998). The indeterminacy of the scope with the adverbial vermoedelijk ('presumably') in (ia) is resolved in (ib).

(i) a. Dirk heeft vermoedelijk aan Francine een ties jenever gegeven. 'Dirk presumably gave a bottle of gin to Francine.'

b. Het was (vermoedelijk) aan Francine dat Dirk (vermoedelijk) een ties jenever gegeven heeft. 'It was presumably to Francine that Dirk (presumably) gave a bottle of gin.'

Analogous cases of indeterminacy arise with the epistemic modal auxiliary zou ('would/is said to').

(13.) In the case of the indirect object in (14a) alternative clefting constructions are available:

(i) Het was aan Francine dat Dirk gisteren een ties jenever gegeven heeft. 'It was to Francine that Dirk gave a bottle of gin yesterday.'

(ii) Het was Francine aan wie/waar(aan) Dirk gisteren een ties jenever (aan) heeft gegeven.

'It was Francine to whom/that Dirk gave a bottle of gin (to) yesterday.'

In (i) the prepositional phrase aan Francine ('to Francine') is clefted, that is, the whole syntagm combining referent and syntactic function marker aan ('to'), whereas in (ii) only the referent noun phrase Francine is extracted and the syntactic function marker is either fronted with wie, incorporated into waaraan, or stranded with the predicator. This distinction between SYNTAGM CLEFTING in (i) and e,.EVERENT CLEFTING in (ii) will not be elaborated further in this contribution, however (see also Haeseryn et al. 1997: 1391).

(14.) In many cases, the addition of focus particles such as alleen/pas ('only') or modal adjuncts renders clefting more fluent. The restrictive nature of these elements is in line with the "restriction of scope" associated with the binarization mechanism discussed above.

(15.) In (25) and (26), the states of affairs referred to in the subclause and the main clause depend on one another. Since they are conceived of as developing in parallel ways, both the subordinate nature of the former and the independent nature of the latter seem to be weakened. The fact that clausal clefting is allowed in (25b) and (26b) may well be related to this more balanced configuration. Welschen (1999: 8) introduces the term "dual syntax" for this more coordinate type of configuration.

(16.) In the ANS (Haeseryn et al. 1997: 1228) this left-dislocation position is called "aanloop" ('introductory position') as opposed to the "eerste zinsplaats" ('sentence initial position') inside the proper sentence structure. As in (27d), the latter is standardly occupied by the pronominal element in the focus position of the cleft, triggering inversion of the subject bet ('it') and the finite verb was ('was'). However, the mechanism of leftdislocation is independent of the possible anteposition of the pronominal constituent, as demonstrated in the alternative pattern of (i):

(i) Alsof hij me niet begreep, het was zo dat hij me aankeek. 'As if he didn't understand me, it was that way that he looked at me.'

(17.) For the definition and use of so-called "double marquage" in French, see BlancheBenveniste et al. (1987 [1984]: 66ff., 83ff.). The same mechanism of double marking occurs in constructions where the pronominal paradigm is not lexicalized as a clause but as a noun phrase or prepositional phrase. Furthermore, double marking interacts with the distinction between syntagm clefting and referent clefting introduced above in connection with (14).

(18.) In the previous paragraphs on types [-p,-c] and [-p,+c], judgements on the unacceptability of the respective clefting constructions were unanimous and confirmed by negative results for Internet searches with Google. With the exception of the [1_00] type in (27), however, judgements on the (un)acceptability of clausal clefting in the [+p, c] type were far less consistent, which is reflected in the ?? assignment instead of the asterisk. In particular, much apparently depends on the presence of particular adverbs, modals, or other verbal adjuncts, or even on particular tenses. Only the simplest constructions are judged here. Further research is needed to clarify the impact of these factors.

(19.) Notice that the analysis of the clefting constellation in (28) constitutes a refinement of the basic distinction in (22) and (23) between two types of [0_10] elements, namely those which are both clausally and pronominally cleftable [+p,+c], such as in de veronderstelling dat ('on the assumption that'), and those which are neither clausally nor pronominally cleftable [-p,-c], such as ondanks dat ('despite the fact that'). With the former, the availability of pronominal clefting automatically entails the possibility of double-marked clefting, whereas with the latter the absence of pronominal clefting renders double-marked clefting impossible by definition.

(20.) Notice that with the predicator interpreteren ('interpret') in (28) the hoe/zo ('how/that way') paradigm has valency status and not its more common rectional status (of manner adjunct).

(21.) This corresponds to the "resumptive" strategy in Konig and van der Auwera's (1988: 103) analysis of conditionals and concessives.

(22.) Notice that even the insertion of the proportional proform toen ('then') in its default sentence-medial position in (32d) does not render the left-dislocation of the subclause acceptable:

(32d') *Vooraleer we vertrokken, we hadden toen alle ramen dicht gedaan. '*Before we left, we had then closed all the windows.'

(23.) This particular example becomes more acceptable with the so-called afterthought pattern: the main clause bears the sentence-final intonation pattern and is followed by a considerable pause introducing the all-low intonation pattern of the subclause. However, with conjunctions such as omdat, one reading of which exhibits the constellation in (34), the application of the afterthought pattern does not improve the acceptability to the same extent:

(i) Omdat je het toch wil weten, ik heb bet gedaan. 'Because you really want to know it, I have done it.'

(ii) ??Ik heb het gedaan, omdat je bet toch wil weten.

(24.) In this construction, inversion in the main clause may occur, but in that case it is due to other factors such as imperative or interrogative sentence type in (i) or independent fronting of a nonsubject constituent in (ii):

(i) Vooraleer ik bet vergeet, komt Jan ook naar bet feest? 'Before I forget, is John coming to the party too?'

(ii) Vooraleer ik het vergeet, morgen komt Jan ook naar het feest. 'Before I forget, tomorrow John is coming to the party too.'

(25.) Concessive conjunctions such as (ook) al ('[even] though') in (i) exhibit a very peculiar syntactic constellation:

(i) a. Ze leest elke dag de krant, (ook) al heeft ze niet veel tijd. 'She reads the paper every day, (even) though she does not have much time.'

b. *(Ook) al heeft ze niet veel tijd, leest ze elke dag de krant. '*(Even) though she does not have much time, does she read the paper every day.'

c. (Ook) al heeft ze niet veel tijd, toch leest ze elke dag de krant. '(Even) though she does not have much time, still does she read the paper every day.'

d. (Ook) al heeft ze niet veel tijd, ze leest (toch) elke dag de krant. '(Even) though she does not have much time, she reads the paper every day (anyway).'

On the one hand, they resemble the exotactic distribution in (34) in that the (b) example is excluded but the (d) example is allowed. In the latter case, however, 'her not having time' in the a/clause does express a property of the state of affairs ('her reading the paper') in the main clause, as presented from the speaker's point of view. The (a) example, on the other hand, is possible as with the endotactic conjunction in (32). Finally, the (c) example superficially resembles the double-marking construction, but the toch does not fit into any pronominal paradigm. In other words, the main clause in (ic) should be considered as the fronted variant of the one in (id). The combination of ook al and toch is similar to coordinative pairs such as enerzijds ... anderzijds ('on the one hand ... on the other hand').

(26.) This type of interpretation of subelauses has been discussed for a number of languages, not only in the realm of conditionals (Auwera 1986; Sweetser 1990; Van Belle 1997; Declerck and Reed 2001) but also in that of temporal, causal, and concessive constructions (Rutherford 1970; Kac 1972; Thompson and Longacre 1985; Verstraete 1998, 2002).

(27.) Two types of (assertive) sentence-final intonation can be distinguished: in (i) an extralow tonal morpheme falls on the last accented mora and the rest of the section remains extra-low, in (ii) a high tonal morpheme falls on the last accented mora whereas the rest of the section is extra-low. In the representation below, the three basic tone levels are indicated as H (high), L (low), and LL (extra low), whereas the actual stressed syllable is indicated in caps:
(i) L daar Nadat hij
 LL NA heb ik het gedaan. WEG was heb ik bet gedaan
 'After that I did it' 'After he left I did it'

(ii) H NA WEG
 L daar Nadat hij
 LL heb ik bet gedaan, was heb ik het gedaan
 'After that I did it' 'After he left I did it'


(28.) Clefting necessarily has this pattern with sentence-final intonation on or inside the focus constituent and the extra-low basic tone level on the rest of the clefted construction:

(i) a. Het was toen hij al !weg was dat ik het heb gezien 'It was when he had already left that I noticed.'

b. Het was !toen dat ik bet heb gezien 'It was then that I noticed.'

(29.) At least three mechanisms need to be distinguished since they can be applied simultaneously in one and the same sentence, namely (1) clefting of the toen-clause, (2) fronting of the focus constituent of the cleft, and (3) location of the sentence-final intonation morpheme on any of the stressable syllables of the different components within the fronted constituent:

(i) a. Toen !hij al weg was, was bet dat ik bet heb gezien. 'It was when he had already left that I noticed.' [lit. When he had already left was it that ...]

b. !Toen was het dat ik het heb gezien. 'It was then that I noticed.' [lit. Then was it that ... ]

(30.) When the tot(dat)-clause is left-dislocated with double-marking of zolang in sentenceinitial position, the sentence-final intonation pattern can no longer fall within the subclause, but may be located anywhere in the main clause:

(i) Tot hij een nieuwe heeft gekocht, (!)zolang heb (!)ik (!)Jan (!)mijn (!)flets geleend. 'Until he has bought a new one, so long did I lend John my bike.'

(31.) Furthermore, when the subclause occurs in sentence-final position it is necessarily preceded by the intonation break (represented as % above). This holds both for the subordinators of comparison [als.sub.6] ('as if') and [alsof.sub.2] ('as if') and those of consequence [dat.sub.2] ('so that'), [dat.sub.3] ('since, so that'), met het/als gevolg dat ('with the/as consequence that'), met het/als resultaat dat ('with the/as result that'), and zodat ('so that').

(32.) It is important to observe that in those cases where a given conjunction appears twice in a certain row, the different indices are accounted for by their proportionality to different pronominal paradigms. First of all, the valency elements [als.sub.3] and [als.sub.5] in (i) are proportional to dat/wat ('that/what') and zo/hoe ('that/what way'), respectively:

(i) a. Ze zouden het betreuren [als.sub.3] Pieter opstapte. 'They would regret it if Peter were to go away.'

b. Ze doet als5 hoort ze me niet. 'She acts as if she doesn't hear me.'

Secondly, the rectional elements als1 and als2 in (ii) can be distinguished in terms of the internal structure of the paradigm they are proportional to. Although both are proportional to dan/wanneer ('then/when') only in the second case this paradigm also contains the so-called paranouns (n)ooit/altijd ('[n]ever/always'):

(ii) a. Alsj je je best doet mag je mee op reis. 'If you do your best, you can come along travelling.'

b. [Als.sub.2] ik haar zie moet ik lachen. 'When I see her, I have to laugh.'

Finally, [dat.sub.2] and [dat.sub.3], or [zo.sub.1] and [zo.sub.2], for which no proportionality holds, enter into different implicational relations with other conjunctions. A subclause introduced by [zo.sub.1] implies the corresponding subclause introduced by indien ('if') but not naar ('[as far] as'). The subclause with zo2 implies naar but not indien:

(iii) a. Zo zij het wist zou zij het wel zeggen. 'If she knew at all, she would certainly say so.'

b. Zo ik hoor moet hij vanavond nog vertrekken. 'As far as I hear, he still has to leave this evening.'

(33.) Taking the different clefting constellations as the starting point would yield anything from one proportionality subclass (for [-p,+c] clefting) to six proportionality subclasses (for [+p,+c] clefting). The present proportionality--driven ordering is to be preferred in that it yields a more balanced subclassification consisting of six cases of oneto-one correspondence, two bipartitions, and one tripartition.

(34.) The labels CORRELATIVE, INTERPRETATIVE, and COMMENTATIVE are intended to reflect a step-wise increasing degree of speaker involvement or subjectivity. Correlative expressions, on the one hand, still basically establish a factual relationship between the main and subordinate clauses, whereas with commentative expressions, on the other hand, a subjective relationship is imposed upon the states of affairs by the speaker. Interpretative expressions can then be seen as an intermediate stage.

References

Auwera, Johan van der (1986). Conditionals and speech acts. In On Conditionals, Elizabeth Traugott, Alice ter Meulen, Judith Snitzer Reilly, and Charles A. Ferguson (eds.), 197214. Cambridge: Cambridge University Press.

Blanche-Benveniste, Claire; Deulofeu, Jose; Stefanini, Jean; and Eynde, Karel van den (1987 [1984]). Pronom et syntaxe: l'approche pronominale et son application au francais, 2nd ed. Paris: SELAF.

Cornillie, Bert; Divjak, Dagmar; Eynde, Karel van den; Smessaert, Hans; and Tummers, Jose (2001). From endotactic to epitactic subordination in Dutch. Preprint 179, Department Linguistiek, K. U. Leuven.

Declerck, Renaat; and Reed, Susan (2001). Conditionals." A Comprehensive Empirical Analysis. Berlin and New York: Mouton de Gruyter.

Dooren, Karine van; and Eynde, Karel van den (1982). A structure for the intonation of Dutch. Linguistics 20, 203-235.

Eynde, Karel van den; and Blanche-Benveniste, Claire (1978). Syntaxe et mecanismes descriptifs: presentation de l'approche pronominale. Cahiers de lexicologie 32, 3-27.

--; Broeders, Erik; Eggermont, Carmen; and Vangilbergen, Ludo (1988). The pronominal approach in NLP: a pronominal feature analysis of coordination in French. Computers and Translation 3, 177-213.

--; and Dooren, Karine van (1983). Intonation and syntactic structure in Dutch. ITL Review of Applied Linguistics 79, 27-42.

--; Kirchmeier-Andersen, Sabine; Mertens, Piet; and Schoesler, Lene (2002). Distributional syntactic analysis and valency. Basic notions, procedures and applications of the pronominal approach. In The Legacy of Zellig Harris: Language and Information into the 21st Century. Volume 2: Computability of Language and Computer Applications, Bruce Nevin (ed.), 163 202. Amsterdam and Philadelphia: John Benjamins.

--; Mertens, Piet; and Swiggers, Pierre (1998). Structuration segmentale et suprasegmentale en syntaxe: vers un module integrationniste de l'ecrit et de l'oral. In Analyse linguistique et approches de l'oral: recueil d'etudes offert en hommage h Claire Blanche-Benveniste, Mireille Bilger, Karel van den Eynde, and Francoise Gadet (eds.), 33-57. Leuven and Paris: Peeters.

--; and Van Durme, Karen (1998). A verb typology on a distributional basis II: classification of modals. In Studies in Valency IV." Valency and Verb Typology, Karen Van Durme; and Lene Schoesler (eds.), 20-32. Odense: Odense University Press.

Foley, William A.; and Van Valin, Robert D. (1984). Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press.

Haeseryn, Walter; Romijn, K.; Geerts, Guido; Rooij, J. de; and Toorn, M. C. van den (eds.) (1997). Algemene Nederlandse Spraakkunst, 2nd completely revised edition. Groningen and Leuven: Wolters-Noordhoff.

Hengeveld, Kees (1998). Adverbial clauses in the languages of Europe. In Adverbial Constructions in the Languages of Europe, Johan van der Auwera and Drnall P. O. Baoill (eds.), 335-419. Berlin and New York: Mouton de Gruyter. Hiz, Henry (1970). Referentials. Semiotica 1(2), 136-166.

Hopper, Paul J.; and Traugott, Elizabeth Closs (1993). Grammaticalization. Cambridge: Cambridge University Press.

Kac, M. B. (1972). Clauses of saying and the interpretation of because. Language 48, 626-632.

Konig, Ekkehard; and Auwera, Johan van der (1988). Clause integration in German and Dutch conditionals, concessive conditionals and concessives. In Clause Combining in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds.), 101-133. Amsterdam and Philadelphia: John Benjamins.

Kortmann, Bernd (1997). Adverbial Subordination. A Typology and History of Adverbial Subordinators Based on European Languages. Berlin and New York: Mouton de Gruyter.

--(1998). Adverbial subordinators in the languages of Europe. In Adverbial Constructions in the Languages of Europe, Johan van der Auwera and Donall P. O. Baoill (eds.), 457-561. Berlin and New York. Mouton de Gruyter.

Lehmann, Christian (1988). Towards a typology of clause linkage. In Clause Combining in Grammar and Discourse, John Haiman and Sandra A. Thompson (eds.), 181-225. Amsterdam and Philadelphia: John Benjamins.

Melis, Ludo; and Eggermont, Carmen (eds.) (1994). A pronominal approach to valency dictionaries. Special issue of International Journal of Lexicography 7.

Paardekooper, P. C. (n.d.). Beknopte ABN-syntaksis. Eindhoven: uitgave in eigen beheer.

Rutherford, W. E. (1970). Some observations concerning subordinate clauses in English. Language 46, 96-115.

Sleeman, Petra; Verheugd, Els; and Vlugter, Bep (2000). Informatiestructuur en syntaxis: overeenkomsten en verschillen tussen Frans en Nederlands. Leuvense Bijdragen 89, 289-304.

Sweetser, Eve (1990). From Etymology to Pragmatics. Cambridge: Cambridge University Press.

Thompson, Sandra; and Longacre, Robert E. (1985). Adverbial clauses. In Language Typology and Syntactic Description II, Timothy Shopen (ed.), 171-234. Cambridge: Cambridge University Press.

Van Belle, William (1997). Conditionele zinnen: een continuum van een dialogisch over een argumentatief naar een logisch verband. Taalbeheersing 19, 215-224.

van Dale Groot Woordenboek der Nederlandse Taal op cd-rom (2000). Utrecht/Antwerpen: van Dale Lexicografie bv.

Verstraete, Jean-Christophe (1998). A semiotic model for the description of levels in conjunction: external, internal-modal and internal-speech functional. Functions of Language 5, 179-211.

--(2002). Interpersonal grammar and clause combining in English. Unpublished doctoral dissertation, Katholieke Universiteit Leuven.

Welschen, Ad (1998). Duale syntaxis en polaire contractie. Negatief gebonden of-constructies in het Nederlands. Munster: Nodus.

Zwart, Jan-Wouter (1998). Where is syntax? Syntactic aspects of left dislocation in Dutch and English. In The Limits of Syntax, Syntax and Semantics 29, Peter Culicover and Louise McNally (eds.), 365-393. San Diego: Academic Press.
COPYRIGHT 2005 Walter de Gruyter GmbH & Co. KG
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2005 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Smessaert, Hans; Cornillie, Bert; Divjak, Dagmar; Van Den Eynde, Karel
Publication:Linguistics: an interdisciplinary journal of the language sciences
Date:May 1, 2005
Words:23871
Previous Article:Competing motivations for the ordering of main and adverbial clauses *.
Next Article:Ladies first? Phonology, frequency, and the naming conspiracy *.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |