SELIM06.pdf Javier Martín Arista, Selim 6 (1998): 32—50 EXTERNAL CONTROL IN FUNCTIONAL SYNTAX: FORMULATING LME CONSTITUENT ORDER RULES1 0. INTRODUCTION Following Dik (1978: 20; 1989: 359) as regards functional patterns, Con- nolly (1991) has explained ME constituent order by means of expression rules of the kind used by Functional Grammar that are based on: syntactic templates, which specify the number of empty positions in a given structure; placement rules, which determine the insertion of major clause constituents in the syntactic slots of the template; and functional patterns, which describe the linearization of the major constituents of the clause. Let us offer a simple illustration of Connolly’s model. Given a syntactic template such as (1.a), the placement rules in (1.b) turn out the functional pattern in (1.c): (1) a. Px-Py -Pz b. Insert constituent A into position Px if condition Ci holds Insert constituent B into position Py if condition Cj holds Insert constituent C into position Pz if condition Ck holds c. A -B-C Connolly’s model is not only ingenious and elegant but also consistent with the philosophic underpinnings of the functional paradigm; moreover, it 1 The research resulting in this paper has been funded by the Vicerrectorado de Inves - tigación de la Universidad de La Rioja through the research project 96PYB33JMA, entitled Aspectos metodológicos de la investigación lingüística en un paradigma funcional: diacronía y sincronía (1996). ____________________________________________________________________ 33 is based on an extensive quantitative analysis and has been rigurously tested through computer implementation. However, Connolly’s approach is mainly syntactic: the stronghold of his study is syntactic description and explanation. This is no wonder since linearization is, along with constituency, the essence of syntax. Our point is that a functionally -oriented syntax should be guided by the principle of external control, in the sense given by Kuno (1980: 126): syntactic description should be based on non-syntactic explanation. Connolly’s order rules, in contrast, are dependable almost exclusively on syntactic factors, for two reasons: firstly, because he attempts to provide an explanation for the drift SOV-SVO; and, secondly, because his main purpose is to be able to offer an implementation of the rules on the computer. In this paper we try to offer a more functionally -oriented study than Con- nolly’s in two respects: in offering an external explanation of syntactic phe- nomena relating to the order of constituent in LME and in dealing with syn- tactic constructions with which Connolly does not cope: passives, duplica- tion, discontinuity, coordination and subordination. This paper is organized as follows 1: in section 1 we put forward the data yielded by our study of a corpus of one thousand LME examples. In section 2 we offer a sketchy view of the theoretical model we espouse in our discus- sion and we concentrate on syntactic discontinuity, relative order and abso- lute order problems. In section 3 we summarize the main points of the article. 1. THE DATA 1 The following abbreviations appear in this paper: FG (functional grammar), S (subject), O (object), Od (direct object), Oi (indirect object), AUX (auxiliary), V (verb) Vf (finite verb), Vn (non-finite verb), Ag (semantic function agent), Man (semantic function manner), Loc (semantic function locative), Temp (semantic function time), VF (verb in final position), X (a major constituent of the clause other than subject, object, verb, negative morpheme and auxiliary), NEG (negative morpheme), TOP (topic), FOC (focus), m (marked), um (unmarked), PX (syntactic position number x), PF (clause-final position), VX (verb in syntactic position num- ber x), DECL (clause operator declarative), INT (clause operator interrogative), NEG (clause operator negative), IMP (clause operator imperative) and OPT (clause operator optative). ____________________________________________________________________ 34 From the methodological point of view, we suscribe to the view (found, among others, in Givón (1984/1990, 1995)) that qualitative conclusions, that is, the explanation of linguistic phenomena in the synchronic and the di- achronic dimensions, must be based on a quantitative study, that is, on a rig - orous description of the data under scrutiny.1 The corpus of this study consists of one thousand LME (1300-1450) ex- amples: one hundred instances have been extracted from Chaucer’s transla - tion of Boetius’ De Consolatione Philosophiæ, Book II (in Navarro et al. 1991: 101-126), quoted as CHB; and another nine hundred have been taken from The Wycclifite Sermons, 1-70 (Macintosh file by Professor González Fernández-Corugedo, Universidad de La Coruña), quoted as WS. Given that the purpose of this study is to account for such phenomena as syntactic discontinuity in LME, we have put aside active constructions, in which syntactic discontinuity is far less fre quent than in (prose) passives. In this sense, passive constructions are older than their active equivalents and accept embraciated constituents, which is not the case with their active coun- terparts.2 As is well known, the reliability of a linguistic corpus largely depends on the choice rather than on the number of examples one selects for his study. Consequently, we have selected examples that cover all the types found in Visser’s (1984: 2165ff) repertoire.3 The figures in table 1 show that a vast majorit y of examples display syn- tactic continuity: X constituents do not break into the Vf -Vn continuum. This generalization applies both to independent and dependent clauses. The fig - 1 See also Hopper and Traugott (1993: 32 ff) and Harris & Campbell (1995: 61 ff). 2 For a more detailed discussion, we refer the reader to Martín Arista (1995). 3 Our examples conform to the these patterns (following Visser’s own classification): (i) Type pær beop pa wanigendan welras gefylde; (ii) Type He wæs læred pis fram Silvestre; (iii) Type Eadwine wearp ofslagen; (iv) Type Ic geaf him a book; (v) Type To-morrow worp pe Marriage I-mad of Meede and of Fæls; (vi) Type Wearp geworden; (vii) Type She were worthy to be blamed; (viii) Type Be ruled by me; (ix) Type He was given a book; (x) Type Hanged worpe he on an hok!; (xi) Type Sleep could be had in the wash-house; (xii) Type It is said that John had an acci- dent; (xiii) Type There was quite a bit of fun poked at American scholars; (xiv) Type Gode be pancod; (xv) Type Hi wæron gederede; (xvi) Type He is found treacherous/a traitor; (xvii) Type Mon him ofteah para clapa; (xviii) Type He is wyrp eallra synna geclænsod and (xix) Type He waes heafde becorfen. ____________________________________________________________________ 35 ures in table 1 show that most examples display syntactic continuity: X constituents do not break into the Vf -Vn continuum. This statement is appli- cable both to independent and dependent clauses. Syntactic discontinuity usually turns up as discontinuity of degree 1, in other words, the verbal complex is only interrupted by one X constituent. It seems sensible, then, to describe discontinuous constructions as marked ones. If this remark is valid, several degrees of markedness might be distinguished, because, in the light of the statistical study, X constituents appear much more frequently between Vf and Vn than S. Syntactic discontinuity examples total % Continuous constructions (Vf-Vn) 871 1,000 87.1 Discontinuous constructions (Vf-X-Vn) 129 1,000 12.9 Degree of discontinuity 1 (Vf-X-Vn) 122 129 94.5 Degree of discontinuity 1 (Vf-S-Vn) 12 122 9.9 Degree of discontinuity 1 (Vf-X-Vn) 110 122 90.1 Degree of discontinuity 2 (Vf-X-Y-Vn) 6 129 4.6 Degree of discontinuity 3 (Vf-X-Y-Z-Vn) 1 129 0.7 Table 1: Syntactic discontinuity in the corpus. As is displayed in table 2, the order Vf -Vn (copular verb -past participle) in preferred in LME in more than 97% of the cases, to the exclusion of Vn -Vf order. Consequently, we describe Vn -Vf order, at least in DECL clauses, as marked, Vf-Vn being described as unmarked. This generalization hold s both for syntactic continuity and syntactic discontinuity. However, as we shall see in the discussion, the order Vn -Vf is not marked with clause opera tors other than DECL, such as OPT. Relative order of Vf and Vn examples total % Vf-Vn 972 1,000 97.2 Vf-Vn (syntactic continuity) 845 972 86.9 Vf-Vn (syntactic discontinuity) 127 972 13 Vn-Vf 28 1,000 2.8 Vn-Vf (syntactic continuity) 26 28 92.8 Vn-Vf (syntactic discontinuity) 2 28 7.1 Table 2: Relative order of Vf and Vn in the corpus. ____________________________________________________________________ 36 As table 3 shows, Vn ---P1 constructions are more frequent than Vf ---P1 ones. The figures demand a certain context since V1 constructions represent less than 5% of the clauses in the corpus. These figures evidence that V1 DECL clauses are marked, as we have just suggested. There remains the question whether V1 OPT, IMP and INT clauses are unmarked, as our rea- soning leads to hold, or not. Absolute order: V1 examples total % V1 43 1,000 4.3 Vf---P1 19 43 44.1 Vf---P1 (DECL) 17 19 89.4 Vf---P1 (NEG) 2 19 10.5 Vf---P1 (syntactic continuity) 4 19 21 Vf---P1 (syntactic discontinuity) 15 19 78.9 Vf---P1/Vn ---P4 4 19 21 Vn---P1 24 43 55.8 Vn---P1/Vf---P2 22 24 91.6 Vn---P1/Vf---P4 0 24 0 Vn---P1 (syntactic continuity) 22 24 91.6 Vf---P1 (syntactic discontinuity) 2 24 8.3 Table 3: V1 constructions in the corpus. As regards Vf and Vn in P2, almost 75% of LME passive clauses are V2 clauses. As is shown in table 4, among the V2 cases, Vf ---P2 cases constitute a great majority. V2, then, is the unmarked ordering. As we have already put forward, the unmarked order in DECL clauses is S-V (vs. V-S), Vf-Vn (vs. Vn - Vf) and Vf-Vn-X (vs. Vf-X-Vn). Among the V2 cases, Vf-2 clauses are almost 100%, which makes these instances unmarked with respect to Vn ---P2 ones, which are considered as marked. Table 4 also indicates the order of passive DECL dependent clauses does not differ from the order of independent ones in LME. We can also see that Vf ---P2 usually correlates with Vn ---P4, as a result, as our discussion will show, of the presence of X in P3. ____________________________________________________________________ 37 Absolute order: V2 examples total % Vf---P2/Vn ---P2 720 1,000 72.0 Vf---P2 711 720 98.7 Vn---P2 9 720 1.2 Vn---P2 (independent clauses) 4 9 44.4 Vn---P2 (dependent clauses) 5 9 55.5 Vf---P2/Vn ---P4 111 111 100 Vn---P2/Vf---P4 0 111 0 Table 4: V2 constructions in the corpus. With reference to VF, this construction is relatively infrequent in the cor- pus (about 20%). Practically all VF clauses are Vn ---P4 clauses, as table 5 displays. In P4 templates VF is, therefore, marked. In o ur discussion we shall try to demonstrate that what is marked in these clauses is not Vn: Vn ---P4 is the result of the displacement caused by X constituents with special prag- matic relevance. This happens both in dependent and independent clauses. Absolute order: VF examples total % Vf---P4/Vn ---P4 201 1,000 20 Vf---P4 1 201 0.5 Vn---P4 200 201 99.5 Table 5: VF constructions in the corpus. 2. DISCUSSION For the discussion of the data we have just offered, we adopt the theoreti- cal model of Functional Grammar as presented in Dik (1978, 1989). As regards the treatment of the assignment of pragmatic functions, we re - sort to the typology of pragmatic functions advanced by Dik et al. (1981: 41ff), de Jong (1981: 89ff) and Dik (1989: 268ff). We have suggested elsewhere (Martín Arista 1994a, 1994b), that the typology of syntactic functions put forward by Dik (1989) might be modified by following the proposals made in ____________________________________________________________________ 38 Dezsó (1978) and Halliday (1985).1 The model advanced in Martín Arista (1994a, 1994b) can be summarized as follows: we consider unmarked those constituents with special pragmatic relevance that are signalled by in tonation contour alone and marked those constituents with special pragmatic relevance that are signalled by the intonation contour of the clause plus a special position in the linear order of the clause. Pragmatic function marked- ness is thus defined in terms of clause position of the constituent to which a given function is assigned (plus sentence stress assignment, in order to avoid circularity).2 With reference to the form and function of placement rules, we follow, in the first place, Dik (1989). We also draw on Connolly (1991: 27ff), who has modified the status of the functional pattern proposed by Dik (1978: 20ff):3 Connolly (1991: 57 ff) regards functional patterns as having derived -rather than primary - status with respect to syntactic templates, which specify the number syntactic positions in the linear order of the clause. However, we do not take the template to be invariable, as Connolly does, but as offering a variable number of empty slots in such a way that no structure -changing op- eration is needed.4 1 The distinction drawn by Halliday (1985: 45) between marked and unmarked themes is of special interest for our purposes. In declarative clauses, Theme is nor- mally assigned to S. The construction in which the S bears the function Theme is the unmarked option; when Theme is assigned to constituents other than S in DECL clauses, we come across instances of Marked Theme. For an alternative view, we refer the reader to Hannay (1990) and Mackenzie and Keizer (1991). 2 This treatment is eclectic in the sense that it is coherent with the proposals by Halli- day (1985), Cruttenden (1986), Dik (1989) and Bossong (1989). We have followed Bossong as regards the existence of a markedness hierarchy but not as regards par- tial marking as a result of TOPm. Another proposal on which we have drawn is made in Bolkestein (1987). According to Bolkestein (1987: 167) there is no incom- patibility for clauses between offering focal information and being expressed hy- potactically. We also follow Bolkestein (1985: 1 ff) as regards the theoretical justifi- cation for a clausal treatment of TOP and FOC, although we do not deny the exis - tence of the TOP continuum advanced by Givón (1983) and followed, to some ex- tent, by Dik (1989: 263 ff). A similar treatment is offered by Mackenzie and Hannay (1982: 43ff) and Siewierska (1987: 147ff). 3 Connolly’s proposal has been made after Connolly (1983) and in a coherent way with the distinction he draws between -purely syntactic- Od and Oi. Connolly’s re- vision of Dik seems to be based on the idea that placement rules and functional patterns do not occur satisfactorily for free-order languages like Latin (Connolly 1991: 50). 4 We follow de Groot (1990: 189) as regards the difference between trigger (primary and secondary) rules and placement rules. A rather different proposal for the form ____________________________________________________________________ 39 As for the constituent X of functional patterns of the kind S-O-V-X, we have already stressed its importance for diachronic explanation in Martín Arista (forthcoming): given VN and X, X is the highlighted element in se- mantic, syntactic, pragmatic and phonological terms.1 And regarding markedness we suscribe to Dik (1989), Croft (1990) and Givón (1995) the view that there is a correlation between structural (qualitative) and statistical (quantitative) markedness. We also follow Bossong (1989: 27ff) for the hierarchy of markedness.2 Given the evidence provided by the corpus, we propose, for LME, an as - signment of unmarked pragmatic functions as follows: TOPum---P1 and FOCum---PF. This assignment, which corresponds to DECL passive clauses, will be tested against the different clause operators and all the possibilities of constituent ordering. This proposal implies: (i) a displacement of FOCum- bearing constituents towards the final position of the clause (with respect to the description we have advanced for OE pragmatic function assignment in Martín Arista (1995)); and (ii) a grammaticalization of TOP in clause-initial position when TOP=S. This explanation, which suits the pragmatic -rule-first principle that should govern a functional syntax, is coherent with the expla - nation for markedness shift that is generally accepted: Marked > Un marked > Grammaticalized. Let us concentrate on the first place on syntactic continuity. In example (2.a), we assign TOPum in clause-initial position and FOCum in clause-final position. This pragmatic explanation is compatible with the data yielded by the corpus, according to which examples like those under (2) are statistically unmarked. (2) of placement rules (although it is also based on the description of syntactic tem- plates put forward in Connolly (1983)) is found in Bakker (1990: 237). 1 This proposal is by no means new. Its contribution may lie in the matching of all levels of linguistic description and in the functionalist methodology. We may refer the reader, for instance, to Dik (1989: 345 ff). For an alternative view, see Lightfoot (1991: 42 ff) and Denison (1993: 25 ff). 2 Bossong (1989: 27 ff) has dealt with the (morphemic) marking of TOP and FOC and has drawn the conclusion that partial marking means that only TOP is marked whereas total marking covers both TOP and FOC. For more detailed information, see Gunkel et al. (1988: 285 ff) and Andrews (1990: 9 ff). ____________________________________________________________________ 40 a. They schall wille pat it were destruyed (WS13) b. He was condempned to be deuoured with feers bestis (CHB) c. In depe preson he was commaundyde to Iy (CHB) d. For avaryce maketh alwey mokereres to ben hated (CHB) It is worth mentioning that all infinitive constructions, such as (2.b), (2.c) and (2.d), in our corpus display syntactic continuity. When dealing with marked constructions, it is necessary to resort to a markedness hierarchy (Martín Arista 1995) that accounts, at least, for two degrees of markedness, marked and heavily marked. We define this marked- ness hierarchy in the following way: (3) Markedness Hierarchy Heavily marked = TOPm+FOCm Marked = TOPm+FOCum/TOPum+FOCm Unmarked = TOPum+FOCum Our point is that this hierarchy, is applicable to the constructions we are commenting on. This is illustrated by example (4). We hold that the marked character of these constructions, which the corpus proves to be statistically marked, is due to the assignment of TOPm in interverbal position, FOCum being aligned in clause-final position. (4) a. … in whiche is pis horn picchid (WS27) b. pus was Steuene martirud (WS40) c. pus, by monye resownys, was Crist clepud of Nazareth (WS42) So clauses, both DECL like (5.a) and NEG like (5.b), offer the same pattern as the examples in (4), thus constituting a sub-group with stable ordering, as is evidenced by the following examples: (5) a. So is it byfallen that thou art a litil departed fro the pees of thi thought (CHB) b. So was not a ston left vpon anopur vndestruyed (WS19) ____________________________________________________________________ 41 It should be noted that, in stating that these clauses are marked, we agree with Connolly (1991: 150). We do not follow Connolly, however, as regards the description of clauses that qualify as VF -S-VN in which XLoc is inserted into P1 as unmarked. Our point is that XLoc---P1 constructions were un- marked in OE, as a result of TOPum---P1 and FOCum---P3, but that they are marked in LME and ModE. It does not seem advisable to describe LME XLoc- --P1 clauses as unmarked, given the statistical data drawn from the corpus and the assignment of pragmatic functions in similar clauses in OE and ModE. To go on, we discuss relative order problems. We present two cases of Vf-Vn order, of which (6.a) is a correlate of syntactic continuity and (6.b) of syntactic discontinuity: (6) a. Monye myraclis weron byfallen abowte pe byrpe of pis Iohn (WS51) b. It is wel seyd pow mayst not see pis poynt of byleue (WS1) We explain the order of (6.a) as a result of the assignment of TOPum---P1 and FOCum---P4, which reflects the unmarked character of the clause. In (6.b), where Vf-Vn and syntactic discontinuity coexist, XMan breaks into the Vf -Vn continuum. The marked character of (6.b) is not due to the Vf -Vn order, which is unmarked, but to the as signment of FOCm--XMan in P3. As regards the order Vn -Vf, such linearization is not marked when clause operators other than DECL are involved. There follow two examples of OPT: (7) a. Blessud be men of clene herte, for pei schal se God (WS68) b. Blessud be mercyful men, for pei schal suwe mercy pat schal be comun to al pe Chirche (WS68) Examples (7.a) and (7.b) are described as unmarked, which we explain in terms of the assignment FOCum---P1 and TOPum---P3. In the discussion of absolute order problems we see, in the first place, to the effects of coordination. In example (8) the presence of Vf in P1 is brought about by the coordinative structure in which the clause belongs: ____________________________________________________________________ 42 (8) And is al maat and overcomen by wepynge and sorwe for desir of the. In example (8) S, which is assigned TOPm, causes discontinuity of Vf and Vn, thus making the clause marked, FOCum is assigned to XReason in P5. A similar explanation can be put forward in cases like the following: (9) For yit ben ther thynges dwelled to the-ward that n o man douteth that they ne be more derworthe to the than thyn so owene lif (CHB) The instances of coordination of infinitives also qualify as V-initial clauses: (10) a. Crist hap ordeyned hise preestis bope to teche and preche his gospel, and not for to preye pus, and to be hyd in suche closettis (WS14) b. … and to ben helude of syknesse pat pei weron inne (WS14) In examples (8)-(10) V-initial has shown up as Vf -initial. Vn -initial clauses with clause operator OPT are not marked in LME, as we have re marked with regard to example (7). Example (11) is an instance of marked order in an OPT clause: (11) Be he kyld of iuste men (WS5) Since the clause operator involved here is OPT, we have regarded S in P2 as TOPm and XAg in P4 as FOCm, which accounts in a satisfactory way for the highly marked character of this clause: it shows syntactic discontinuity of the type Vf -S-Vn and Vf---P1. XAg, on the other hand, follows Vf, S and Vn, which can be considered the straight order. As for V2, Vf---P2 is the unmarked option, as the corpus evidences. We explain the statistically unmarked character of (12) as a result of the assig ment TOPum---S in P1 and FOCum---XLoc in P4: (12) Cresus was lad to the fyer to ben brend (CHB) The following example also qualifies as V2, although what appears in P2 is Vn: (13) Ryght swich was sche whan sche flateryd the (CHB) ____________________________________________________________________ 43 This clause is highly marked: XMan is placed in P1, followed by Vn in P2, Vf in P3 and S in P4 (therefore, we have Vn -Vf and S-final). This is accounted for as a consequence of the assignment TOPm---S in P4 and FOCm---XMan in P1. The following example is also marked: we find Vn ---P2 in a Vn -Vf clause: (14) Sai we pat he riuen es wit beistes wild (CHB) We explain this ordering by resorting to the assignment FOCm---Vn. Consequently, it is not necessary to explain the Vn -Vf order in terms of de- pendent/independent clauses, which seems out of place in LME. As far as VF is concerned, VF is marked in P4 templates. This markedness is coherent with the assigment of FOCm to XLoc, XTemp and XMan in P3, on which we have already commented. Let us see an example: (15) By sixe or sevene he shulde soone delyvered be (CHB) The marked character of (15) is explained by means the presence of S in P2, which is assigmed TOPm, and XAg in P1, which bears FOCm. Moreover, Vf follows Vn in final position and XTemp follows AUX, which dis places YTemp , Vn and Vf one position to the right. The next example also qualifies as VF. It is Vn that comes in clause-final position this time. The construction is also marked: (16) a. By pis foule eresye is pe Chirche disseyued (WS45) b. In pis prefold disseit ben monye men blyndud (WS45) c. By pis secounde lesyng is pe Chirche disseyued (WS45) Indeed, a focalised XAg introduces the clause in (16.a), which undergoes syntactic discontinuity of the type Vf -S-Vn; and S is assigned TOPm. The same explanation can account for (16.b) and (16.c), in which FOCm is as - signed to XLoc in (16.a) and XAg in (16.b). By stressing this we mean that Vn (and less frequently Vf) in final position is not itself marked; on the contrary, it is the result of the displacement caused by focalised and topicalised constituents that push Vn (or Vf) towards the end of the clause. This state- ment also applies to dependent clauses. Otherwise, orderings like the one in (17) would have to be explained in clause-dependency terms, which does not seem advisable in the LME period: (17) ____________________________________________________________________ 44 a. … how pei weron specially doon (WS37) b. … in whiche eche part of pis rewme is monye weyes contenyd (WS11) c. such a man louep more godus of pis world pan he louep his God, for on hem his wylle is more set (WS22) We put down the markedness of (17.a) to the assignment of FOCm to XMan in P3; as a result, Vn is displaced towards P4. TOPum is assigned to S in P1. A similar explanation can be provided for examples (17.b) and (17.c). In (17.b) the presence of XLoc in P1 can be attributed to the fact that this is a relative clause; Vn is displaced to P5 by YMan, which appears in P4; YMan bears FOCm. In (17.c) all the constituents are displaced towards the right by XLoc, which is topicalised and occupies P1, Man occupying interverbal po- sition P4 and bearing FOCm. In general, V2 equals Vf ---P2 unless it is displaced by AUX: (18) a. Special knowyng schulde ben hyd (WS19) b. Owre iugement schal ben hool (WS19) c. pei schal be reprouede (WS20) AUX displaces Vf, Vn and X towards the right of the clause. This is the case with continuous and discontinuous passives, as the following examples illustrate: (19) a. … that may ben talcen im any wise (CHB) b. … for panne schal Cristus be reryd (WS12) As is shown in (19.a), P2, which is occupied by AUX, is not vacant and VF is placed in P3, VN in P4 and XMan in P5. The effect of AUX is similar in the case of syntactic discontinuity, as is displayed in (19.b). In (19.b) we take TOPum to have being assigned to XMan since this constituent is a pro noun, which suggests second reference; and S to bear FOCm in interverbal position. It should also be noted that by syntactic discontinuity we mean not only that Vf and Vn but also that AUX and V are cut off by one or more con- stituents. In this sense, the example that follows is not marked with respect to ____________________________________________________________________ 45 syntactic continuity but it does show markedness as regards S position, which is clause-final: (20) Whereuere his body were, schulden be gederude men (WS69) It is worth noticing that the whole Loc clause is under focus in this utter- ance. Thus, FOCm is assigned to XLoc in P1 and TOPm to S in P5. After revising the role of AUX in DECL clauses, let us discuss the role of AUX in clauses with clause operator NEG. Again, we draw our attention to syntactic continuity and discontinuity. In (21) NEG and AUX enter the clause structure: (21) Ne shal the corn in his berne ben eten wid no muis (CHB) In (21), NEG occupies P1 and is assigned FOCum. As a result, AUX is placed in P2 and S, which bears TOPum, in P3, which causes discontinuity in the verbal complex; XAg follows Vf and Vn in P6. What follows from this explanation is that the assignment of pragmatic functions in LME NEG pas- sive clauses containing AUX is the same as in NEG OPT and NEG IMP, that is, TOPum is assigned in P3 and FOCum in P1. This is tantamount to saying that clauses like the following are still marked in LME: (22) a. … pat ne it schal be schewed panne (WS17) b. … pat ne it schal be knowe panne (WS17) Although both (22.a) and (22.b) are dependent clauses, we do not imply that this order is the result of subordination. Drawing on Dik (1989: 353) and Givón (1993 vol I: 207) we suggest that the evolution of NEG in terms of markedness from LME to ModE might have been as follows: (23) a. LME NEG-AUX-S-Vf-Vn (unmarked) NEG-S-AUX-Vf-Vn (marked) S-AUX-NEG-Vf-VN (marked) S-MOD-NEG-Vf-Vn (unmarked) b. ModE S-MOD-NEG-Vf-Vn (grammaticalized) The hypothesis tentatively presented in (23) has two advantages: in the first place, it is statistically justified by the data extracted from our corpus; ____________________________________________________________________ 46 and, in the second place, it is compatible with the evolution Marked-Un- marked-Grammaticalized on which we have already commented. Let us turn to double NEG clauses now. As is well known, double nega- tives, either nominal-verbal or verbal-verbal, were a common device in LME. Our figures indicate that double negatives were statistically marked in LME. This explanation, however, could hardly resist diachronic comparison since there is no point in arguing for an evolution Marked-Unmarked-Gram- maticalized in double negatives, as these ModE examples evidence: (24) a. *He was not given nothing b. *He not was not given anything Neither does it seem sensible to state that the double negative was gram- maticalized in LME clauses because, as a matter of fact, single negatives oc- curred more frequently than double negatives. The solution that we propose is to consider NEG in P3 as a duplication of NEG in P1, thus taking up no position in the clause and leaving vacant P3 for S. Let us see an example: (25) Ne wol noght ben cast thow with the lowde blastes of the wynd Eurus (CHB) In (25) we speak of markedness not because of the presence of double NEG but because of the assignment of TOPm to S, which is placed in P5. 3. CONCLUDING REMARKS In this article we have formulated a number of placement rules that satisfy the requirements of -at least- LME passive clauses: (26) Given P1-P4 and DECL or NEG TOPum---P1 FOCum---P4 Given P1-P4 and OPT, NEG OPT, IMP or NEG IMP TOPum---P3 FOCum---P1 ____________________________________________________________________ 47 Given P1-P4 and OP INT TOPum---P2 FOCum---P1 Very briefly, what this proposal argues for is a grammaticalization of TOP in clause-initial position when TOP=S and a displacement of the constituents bearing FOCum towards clause-final position. There remains to demonstrate that these rules can explain the order of ac- tive clauses in an adequate way. Our position on this respect is that they can since active clauses exhibit a lower degree of syntactic complexity in the sense that syntactic discontinuity is not a frequent phenomenon in active clauses. Apart from their applicability, these rules have the advantage of being consistent with the basic tenet of a functionally -oriented syntax: syntactic rules must be based on external, i.e. non-syntactic, factors such as the as - signment of semantic and pragmatic functions. Javier Martín Arista Universidad de La Rioja REFERENCES Andrews, E. 1990: Markedness Theory. Durham: Duke University Press. Bakker, D. 1990: A formalism for Functional Grammar expression rules. > Connolly & Dik eds.: 45-63. Bolkenstein, A. M. 1985: Cohesiveness and syntactic variation: Quantitative vs. qualitative grammar. > Bolkestein et al. eds.: 1-14. Bolkenstein, A. M. 1987: Discourse functions of predications: The back- ground/foreground distinction of tense and voice in Latin main and subordinate clauses. > Nuyts & Schutter eds.: 163-178. Bolkenstein , A. M., Vet, C. & Hannay, M. eds. 1985: Syntax and Pragmatics in Functional Grammar. Dordrecht: Foris. ____________________________________________________________________ 48 Bossong, G. 1989: Morphemic marking of topic and focus. Belgian Journal of Linguistics 4: 27-51. Connolly , J. 1983: Placement Rules & Syntatic Templates. > Dik: 247- 266. Connolly, J. 1991: Constituent order in Functional Grammar: Synchronic and diachronic perspectives. Berlin: Foris. Connolly , J. & Dik, S. eds. 1990: Functional Grammar and the Computer. Dordrecht: Foris. Conte, M. et al. eds. 1978: Wortstellung & Bedeutung. Tübingen: Niemeyer. Croft, W. 1990: Typology and Universals. Cambridge University Press. Cruttenden, A. 1986: Intonation. Cambridge: Cambridge University Press. Denison, D. 1993: English Historical Syntax. London: Longman. Dezsó, L. 1978: Towards a Typology of Theme and Rheme: SOV Languages. > Conte et al. eds.: 3-11. Dik, S. 1978: Functional Grammar. Dordrecht: Foris. Dik, S. 1989: The Theory of Functional Grammar I: The Structure of the Clause. Dordrecht: Foris. Dik, S. et al. 1981: On the Typology of Focus Phenomena. > Hoekstra et al. eds.: 41-74. Dik, S. ed. 1983: Advances in Functional Grammar. Dordrecht: Foris. Givón, T. 1984/1990: Syntax: A Functional-Typological Introduction (2 vols.). Amsterdam: John Benjamins. Givón, T. 1993: English Grammar: A Function-Based Introduction (2 vols.). Amsterdam: John Benjamins. Givón, T. 1995: Functionalism and Grammar. Amsterdam: John Benjamins. Givón, T. ed. 1983: Topic Continuity in Discourse. Amsterdam: Benjamins. Groot, C. de 1990: Morphology and the typology of expression rules. > Han- nay & Vester eds.: 187-201. ____________________________________________________________________ 49 Gunkel, J. et al. 1988: On the functions of marked and unmarked terms. > Hammond et al. eds.: 285-301. Halliday, M. 1985: An Introduction to Functional Grammar. London: Arnold. Hammond, M. et al. eds. 1988: Studies in Syntactic Typology. Amsterdam: John Benjamins. Hannay, M. 1990: Pragmatic function assignment and word order variation in a Functional Grammar of English. Working Papers in Functional Grammar 38. Amsterdam: University of Amsterdam. Hannay, M. & Vester, E. eds. 1990: Working with Functional Grammar: De - scriptive and Computational Applications. Dordrecht: Foris. Harris, A. & Campbell, L. 1995: Historical syntax in cross-linguistic per- spective. Cambridge: Cambridge University Press. Hoekstra , T. et al. eds. 1981: Perspectives on Functional Grammar. Dor- drecht: Foris. Hopper, P. & Traugott, E. C. 1993: Grammaticalization. Cambridge: CUP. Jong, J. DE. 1981: On the Treatment of Focus Phenomena in Functional Grammar. > Hoekstra et al. eds.: 89-115. Kuno, S. 1980: Functional Syntax. > Moravcsik & Wirth eds.: 117-135. Lightfoot, D. 1991: How to Set Parameters. Cambridge, Mass.: MIT Press. Mackenzie, J. & Keizer, E. 1991: On assigning pragmatic functions in English. Pragmatics 1: 169-215. Mackenzie , J. & Hannay, M. 1982: Prepositional predicates and focus con- structions in a functional grammar of English. Lingua 56: 43-57. Martín Arista, J. 1994a: Funciones pragmáticas marcadas y no marcadas. Miscelanea 15: 391-404. Martín Arista, J. 1994b: Aspectos semánticos y pragmáticos de la operación de las reglas de expresión. > Martín Arista ed.: 193-238. Martín Arista, J. 1995: The prefield -postfield drift and the evolution of the English passive. Revista Canaria de Estudios Ingleses 30-31. ____________________________________________________________________ 50 Martín Arista, J.(forthcoming): The role of VN in functio nal syntax. Martín Arista, J. ed. 1994: Estudios de Gramática Funcional. Zaragoza: Mira Eds. Moravcsik, E. & Wirth, J. eds. 1980: Syntax and Semantics 13: Current AVnroaches to Syntax. New York: Academic Press. Navarro Errasti, M. P. et al. 1991: Setting up a Presence-and-Permanence Glossary of Native and Non-Native Terms in Medieval English I. Uni- versidad de Zaragoza: Servicio de Publicaciones. Nuyts , J. & Schutter, G. DE. eds. 1987: Getting One’s Words into Line. Dor- drecht: Foris. Siewierska, A. 1987: Postverbal Subject pronouns in Polish in the light of topic continuity and the Topic/Focus distinction. > Nuyts & Schutter eds.: 147-162. Visser, F. 1984 (1963/1973): A Historical Syntax of the English Language (4 vols.). Leiden: Brill. * † *