Philologica Canariensia 30 (2024), pp. 393-423
DOI: https://doi.org/10.20420/Phil.Can.2024.683
Received: July 18, 2023; revised version accepted: November 13, 2023
Published online: June 22, 2024
Are There Serial Verb Constructions in Old English? A New Perspective on the Changes in Verbal Complementation
¿Hay construcciones de verbos seriales en inglés antiguo?
Una nueva perspectiva sobre los cambios en la complementación verbal
Existe-t-il des constructions de verbes en série en vieil anglais ?
Une nouvelle perspective sur les changements dans la complémentation verbale
Ana Elvira Ojanguren López
Universidad de la Rioja
ORCID: 0000-0001-5356-7391
This article discusses the existence of serial verb constructions in Old English and explores the complementation of aspectual and control verbs as well as other related constructions. The discussion of the Old English data finds remarkable differences with respect to the syntactic configurations of Present-Day English that are described as serial by Roberts (2010). Although the range of serial constructions is more restrictive in Old English, at least with aspectual verbs, the analysis of verbal complementation from the perspective of serialisation allows us to bring other convergent constructions to the discussion, including configurations with linked infinitives and participles in predicative constructions and configurations with pre-modals and the causative verb dōn ‘to do’.
Keywords: Old English, Role and Reference Grammar, serial verbs, aspectual verbs, control verbs
|
Resumen |
Este artículo trata sobre la existencia de construcciones verbales seriales en inglés antiguo, centrándose en la complementación de verbos aspectuales y de control, así como en otras construcciones relacionadas. El análisis de los datos del inglés antiguo muestra diferencias notables con respecto a las configuraciones sintácticas del inglés contemporáneo que se describen como seriales en Roberts (2010). Aunque el rango de construcciones seriales es más restrictivo en inglés antiguo, al menos con verbos aspectuales, el análisis de la complementación verbal desde la perspectiva de la serialización nos permite incorporar otras construcciones convergentes a la discusión, incluyendo configuraciones con infinitivos y participios en construcciones predicativas vinculadas, así como configuraciones con premodales y el verbo causativo dōn (‘hacer’).
Palabras clave: inglés antiguo, Gramática del Papel y la Referencia, verbos seriales, verbos aspectuales, verbos de control
|
Résumé |
Cet article traite de l'existence de constructions verbales sérielles en vieil anglais, en se concentrant sur la complémentation des verbes aspectuels et de contrôle, ainsi que sur d’autres constructions connexes. L’analyse des données en vieux anglais révèle des différences remarquables par rapport aux configurations syntaxiques de l’anglais moderne qui sont décrites comme sérielles dans Roberts (2010). Bien que la gamme de constructions sérielles soit plus restrictive en vieux anglais, du moins avec les verbes aspectuels, l’analyse de la complémentation verbale du point de vue de la sérialisation nous permet d’intégrer d’autres constructions convergentes à la discussion, notamment des configurations avec des infinitifs liés et des participes dans des constructions prédicatives, ainsi que des configurations avec des pré-modaux et le verbe causatif dōn ‘faire’.
Mots-clés : vieux anglaise, Role and Reference Grammar, verbes sériels, verbes aspectuels, verbes de contrôle
1. Introduction
The aim of this article is to discuss the existence of serial verb constructions in Old English. This end calls for a synthetic definition of the typological concept of serial verb construction (SVC) and a description of the semantic-syntactic configurations of Old English that satisfy the requirements of SVCs. Such a description of SVCs in Old English intends to throw new light on the evolution of the verbal complementation of English, including finite and non-finite clausal complements.
This study is conducted within the framework of Role and Reference Grammar (hereafter RRG) because this theory of language (Foley & Van Valin, 1984; Van Valin & LaPolla, 1997; Van Valin, 2005; Van Valin 2023) is concerned with the pragmatic and semantic basis of syntactic structures and, moreover, because RRG posits an analysis of serial verb constructions based on the syntactic relation of cosubordination, or dependent coordination (Bohnemeyer & Van Valin, 2017; Van Valin, 2021, 2023). In line with the approach of RRG to the question, the term construction is used is this article with a descriptive value, without implying a schematic or productive character.
The article is structured as follows. Section 2 considers the SVC from a typological perspective. Section 3 reviews the proposal for the existence of SVCs in Present-Day English. A synthetic definition of SVC is given at this point. Section 4 presents the juncture-nexus configurations with Old English verbs of aspect and control and addresses the question of the status of tō. The next two sections are organised by verbal class. Thus, Section 5 deals with aspectual verbs, while Section 6 is devoted to control verbs. These sections consider the constituent projection (juncture level and nexus relation) and the specific aspect of negation in the operator projection. Section 7 discusses the advantages of the approach to verbal complementation based on serialisation and defines a serialisation cline whereby some non-serial configurations are developing serial properties while other serial configurations are losing or keeping their properties. Finally, Section 8 summarises the main points made in this article.
2. Serial Verb Constructions in Typological Perspective
This work considers a concatenation of verbs an SVC (Haspelmath, 1997) provided that it meets certain requirements discussed in sections 2 and 3. Aikhenvald (2006, p. 1) defines an SVC as “a sequence of verbs which act together as a single predicate, without any overt marker of coordination, subordination, or syntactic dependency of any other sort”. An SVC, according to Haspelmath (2016, p. 292), is “a monoclausal construction consisting of multiple independent verbs with no element linking them and with no predicate-argument relation between the verbs”. Aikhenvald (2006) and Haspelmath (2016) underline other characteristics of serial verb constructions, including that they are monoclausal and comprised of free forms, that they conceptualize a single event and that they may share the negation, the intonation contour as well as verbal operators of tense, aspect and modality and core arguments. Comrie (1995, as cited in Foley, 2010, p. 79) underlines the single intonation contour and eventhood of SVCs and also remarks that “there is no marking of most verb categories (person-number, tense-aspect-mood, polarity, etc.) except on the main verb, or all verbs must have the same categories as the main verb”.
With respect to the morpho-syntactic types of SVCs, Van Staden and Reesink (2008) distinguish four morpho-syntactic types of SVCs, namely, independent serialisation, dependent serialisation, co-dependent serialisation, and complex verb serialisation. In independent serialisation, all the verbs in the construction present full inflectional morphology (tense, aspect, mood and subject agreement). In dependent verb serialisation, one of the verbs carries all the inflections, while the others show “their ‘bare’ form or in a stripped-down form, possibly with an affix indicating, for instance, that the word is a predicate (despite not being inflected for subject, tense, etc.)” (Van Staden & Reesink, 2008, p. 24). While in dependent serialisation the shared argument is the subject, in co-dependent serialisation the object of the first clause is the subject of the second, without marks of raising or gapping. In complex verb serialisation, two or more verbs share one set of affixes, so that prefixes are attached to the first verb in the sequence and the suffixes to the last verb.[1]
As far as the diachrony of SVCs is concerned, Roberts (2010, p. 4) distinguishes between symmetrical and asymmetrical serial verb constructions. In symmetrical configurations, all the verbs have the same status whereas in asymmetrical configurations one verb modifies the meaning of the other verbs by expressing, for instance, the direction of motion, the result of an action. The verbs of symmetrical SVCs belong to open lexical classes and tend to lexicalise. The minor verbs in asymmetrical SVCs have a closed lexical membership and tend to grammaticalise (Aikhenvald, 2006, p. 36).[2] For authors such as Givón (1991; 2009b) and Rose (2009), serial verb constructions arise from complex sentences.
3. Serial Verb Constructions in Present-Day English
Roberts (2010, p. 26) points out that “a prototypical SVC contains two or more verbs that are fully lexical verbs.” This author remarks that the verbs of an SVC express a single complex event, have a unified intonation contour, contain no markers of coordination or subordination, share at least one semantic argument, display only one specification for tense, aspect, modality and negation, and contain only one grammatical subject. Roberts (2010, p. 7) stresses the incompatibility of SVCs and reflexive constructions by remarking that “a true SVC will not contain two overt NPs which refer to the same argument.”
Roberts (2010, p. 34) presents an inventory of constructions with phase verbs (PVCs) in English that includes the bare present participle (stopped crying), the from-present participle (prevented people from going to work), the to-infinitive (encourages people to stop smoking), the bare infinitive (helps him keep alert) and the past participle (got burned by the blast). Then, he compares SVCs and PVCs in English and draws the conclusion that “Phase verbs in English have the same semantic and syntactic properties as serial verb constructions found in Creole languages, West Africa, mainland Southeast Asia, New Guinea and the Pacific Islands. We can conclude therefore that phase verbs in English are a type of serial verb construction” (Roberts, 2010, p. 33).
In the remainder of this article, an SVC is understood, in line with Roberts (2010, p. 26), as comprised of two lexical verbs that take part in a single clause which expresses a single event and has no markers of subordination (Aikhenvald, 2006; Haspelmath, 2016). There is no predicate-argument relation between the verbs of an SVC (Haspelmath, 2016, p. 292), although they may share operators of negation or tense, aspect and modality. All the verbs involved in an SVC must be free forms, although their inflections may belong to both the finite and the non-finite sets of the paradigm. This definition coincides with Van Staden and Reesink’s (2008, p. 24) dependent verb serialisation, in which one of the verbs is fully inflected (the finite form), thus carrying the operators, whereas the other presents a simplified morphology and shares the operators (the non-finite form). In dependent serialisation the first argument (subject) is also shared by the verbs in the construction. Overall, this is an asymmetrical configuration in which the verbs in the SVC do not have the same status (Roberts, 2010, p. 4).
Section 6 and Section 7 discuss to what extent the Old English candidates for SVC abide by this definition. The discussion puts special emphasis on some syntactic differences between Old English and Present-Day English, particularly with respect to juncture level, nexus relation and negation. Before addressing this question, it is necessary to present the juncture-nexus configurations of Old English. This is done in Section 4.
4. Juncture and Nexus in Old English
The complex configurations of Old English under analysis are presented in this section. As the model of the complex sentence of RRG is adopted, the relevant aspects of the theory of juncture and nexus are reviewed in the first place. This includes the related question of clause-linkage markers (Van Valin & LaPolla, 1997, p. 466) and other linking elements.
In RRG (Van Valin & LaPolla, 1997; Van Valin, 2005; Van Valin, 2023), the association between semantics and syntax is called linking. The linking from semantic to syntax is necessary for language production, while the linking from syntax to semantics operates in language comprehension. In linking, the Completeness Constraint (Van Valin, 2023, p. 116) stipulates that all the arguments of the semantic representation of the sentence are realized in the syntax.
The sentence is a projection from the semantics of the logical structure onto a full representation of arguments and operators (morphological features conveying negation, tense, aspect, modality, illocutionary force, etc.). Logical structures are comprised of the Aktionsart (internal aspect) type and the arguments of the verbal predicate. Logical structures are expanded by means of the assignment of macroroles (generalized semantic roles) and syntactic functions (arguments, argument-adjuncts and adjuncts). All the elements that determine the realization of arguments in the layered structure of the clause of RRG, including case, agreement, government and the specific aspects of the construction, give rise to the linking algorithm (Van Valin, 2023, p. 116).[3]
The sentence is represented hierarchically by means of the layered structure of the clause, which can be broken down into several semantic-syntactic layers defined by the status of constituents (macrorole arguments vs. non-macrorole arguments, for instance), the association of peripheries (containing adjuncts) to cores (consisting of the verb and the arguments) and the scope of operators. Four layers are distinguished in the layered structure of the clause: the verbal nucleus, the core (formed by the nucleus and the arguments), the clause (consisting of a core with which a periphery can be associated) and the sentence (comprised of one or more clauses). Figure 1 illustrates the layered structure of the clause by means of a tree diagram. It corresponds to a labeled bracket annotation of the type [[[RP]Argument [Pred]Nucleus [RP]Argument]Core [PP]Periphery]Clause [PP]Periphery]Sentence.
Figure 1. The layered structure of the clause in English (Van Valin, 2023, p. 23)
The theory of the complex sentence of RRG distinguishes between the type of unit (called juncture) and the type of relation between the units of the sentence (called nexus). The various syntactic configurations of a language are described in terms of juncture-nexus types. Juncture types are defined into three levels of juncture depending on the structural complexity of the combining units: nuclear juncture, core juncture and clause juncture. The structurally simplest type is nuclear juncture, which is formed by two nuclei, such as forced and open in Sam forced open the car. According to Van Valin and LaPolla (1997, p. 445), the linked predication of a nuclear juncture can be adjacent to the matrix predication and must be intransitive. Core junctures contain two or more cores, as in I had Sam to force the car open. Core junctures, unlike nuclear junctures, may contain complementizers such as to, from, etc. The clause juncture is the most complex juncture type and can be illustrated by means of instances like John likes Mary and Susan likes her too.
Three types of nexus are defined, to wit, coordination, subordination and cosubordination. This typology is based on a distinction between independent coordination and dependent coordination or cosubordination (Van Valin 2023, p. 71). Independent coordination requires two different first arguments (the subject of traditional grammar). For instance, independent coordination holds in core junctures like I made the shop manager apologize and in clause junctures like I complained and the shop manager apologized. Subordination can be classified as daughter subordination or as peripheral subordination. In daughter subordination, the linked clause is an argument of the matrix clause, as in That the song was a hit surprised everyone. In peripheral subordination, the clause is an adjunct, as in The book came out before the readers expected. It is a requisite of subordination that the linked element can be clefted and passivised (Van Valin, 2023, p. 74). For example, the nexus type of Jack criticised Sue’s joining the firm is subordination because the cleft It was Sue’s joining the firm that Jack criticised and the passive That Sue left the firm was criticised by Jack are possible. In cosubordination (or dependent coordination), arguments and operators can be shared. In the constituent projection of an instance of cosubordination, the first argument must be shared by the matrix and the linked predication. For example, in The students were tired and left in a rush, the first argument (the students) is shared by the two clauses. In the operator projection of an instance of cosubordination, the combining units must share at least one operator at the relevant level of juncture (Van Valin, 2021, p. 248). For example, in Jack sat reading a book the operator of progressive aspect has scope over the matrix and the linked predication because it is possible to render this expression as Jack was sitting and reading a book. Bohnemeyer and Van Valin (2017, p. 142) remark that the nexus of cosubordination is used to describe single events. In Present-Day English, there is nuclear, core and clause cosubordination. In nuclear cosubordination, two adjacent nuclei belong in the same core, in such a way that the linked predication is intransitive, as in The customer left complaining. Core cosubordination can be illustrated through instances like I managed to arrive on time. Clause cosubordination in Present-Day English can be found only in coordinate subject constructions such as I got up late and had brunch.
The application of RRG to Old English needs an explanation for the status of linkage elements and the resulting levels of juncture. On the one hand, as far as the linking words in complex configurations are concerned, the status of the subordinating conjunction ðæt ‘that’ is not problematic as it is a clause-linkage word. On the other hand, the inflected infinitive deserves more attention. According to Ringe and Taylor (2014, p. 483), Old English has two infinitives, the bare infinitive, as in wyrcan ‘to work’, and the inflected infinitive, which has its origin in the dative case of a neuter verbal noun governed by the preposition tō, as in tō witanne ‘to know’. For Molencki (1991), Kageyama (1992) and Fischer (1996), tō is not a complementiser but forms a unit with the inflected infinitive (as in to leornianne) because tō and the inflected infinitive are always adjacent. As Fischer (1996, p. 109) remarks, the Old English inflected infinitive cannot be split (as in to duly perform) or stranded (as in You may go if you want to). Kageyama (1992, p. 101) also argues that the Old English inflected infinitive cannot be verbal but nominal because it can be coordinated with a prepositional phrase, as in Ut eode to his gebede oððe to leornianne mid his geferum ‘He went out to say his prayers or to study with his friends’ (Bede 162.7; Fischer, 1996, p. 110). For these reasons, the inflective infinitive is considered in this work a verbal form that takes a prefixal and a suffixal inflectional morpheme.
That said, Old English clause junctures require a clause-linkage marker between the matrix clause and the linked clause, which has a finite form of the verb, usually in the subjunctive. Nuclear junctures consist of two adjacent nuclei, in such a way that the first nucleus is a non-finite form of the verb and that the linked nucleus is intransitive. In an Old English core juncture, the verb in the linked predication must be inflected for a non-finite form and may be transitive. The core junctures of Old English cannot take clause-linkage markers, but can include pre-verbal inflectional morphemes, typically tō in the inflected infinitive, so that the adjacency of the two nuclei is undermined.
5. Sources and Data
The source of the data of this study is The York-Toronto-Helsinki Parsed Corpus of Old English Prose (Taylor et al., 2003), which contains approximately 1.5 million words of Old English prose. Text numbers have been taken from The Dictionary of Old English Corpus (Healey et al., 2004), except the ones discussed by the authors cited in the article, which are kept as in the original. The inventory of Old English verbs of control and aspect draws on Faber and Mairal’s (1999) lexical domains of English. Within the lexical domain “Action”, the verbs presented in Figure 2 have been found in the dictionaries of Old English (Clark-Hall & Meritt, 1996; Bosworth & Toller, 1973; and Healey, 2018) with the relevant meaning components.
|
Not to do something [fail]: fail; neglect, omit; give up. forgān, forsittan, linnan, mīðan, oferāhebban, ofergīman, oferhealdan, oferhebban, ofersittan, oflinnan, (ge)sparian, (ge)trucian, wandian. |
|
To stop doing something [end]: end, finish; cease, stop; desist, relinquish. āblinnan, ætstandan, blinnan, for(e)sacan, geblinnan, linnan, ofersittan, oflinnan, oðstillan. |
|
To make an effort in order to be able to do something [try]: try, attempt; strive, struggle, endeavour. (ge)cneordlǣcan, (ge)fandian, fundian, hīgian, onginnan, (ge)tilian. |
|
To cause somebody not to do something [prevent]: prevent, restrain; constrain, impede; forbid, prohibit. āgǣlan, belēan, bewerian, foresacan, forbēodan, forfōn, forhabban, forwiernan, gǣlan, tōcweðan. |
Figure 2. Old English verbs of inaction: control and aspect
6. Aspectual Verbs
Old English aspectual verbs are verbs of inception, continuation, termination and attempt that convey meaning components corresponding to [fail], including forgān, forsittan, linnan, mīðan, oferāhebban, ofergīman, oferhealdan, oferhebban, ofersittan, oflinnan, (ge)sparian, (ge)trucian, wandian; [end], like āblinnan, ætstandan, blinnan, for(e)sacan, geblinnan, linnan, ofersittan, oflinnan, oðstillan; and [try], such as (ge)cneordlǣcan, (ge)fandian, fundian, hīgian, onginnan, (ge)tilian.[4]
In this section, aspectual verbs are considered both in the constituent projection and in the operator projection. As regards the constituent projection, the questions of juncture level and nexus relation are addressed. With respect to the operator projection, the operator of negation is taken into account.
Some aspectual verbs can be found in complex syntactic configurations based on a nexus relation of cosubordination. For instance, in (1) the first argument Romane ‘the Romans’ is shared by the matrix verb ongunnon and the linked verb aræran ‘to raise’. The operators of tense, aspect and modality are also shared at the level of juncture.
(1) [Or 5 011800 (9.122.22)]
Þa ongunnon Romane þa mæstan sace him betweonum up aræran.
|
Þa |
ongunnon |
Romane |
þa |
|
then-ADV |
begin-PST.3PL |
Roman-NOM.PL |
the-ACC.SG |
|
Mæstan |
sace |
him |
betweonum |
|
great-SUPERL |
strife-ACC.SG |
he-DAT.PL |
between-PREP |
|
Up |
aræran |
|
|
|
up-PREP |
raise-INF |
|
|
‘Then the Romans began to raise the greatest strife among themselves.’
The level of juncture of aspectual verbs can vary. The verb blinnan appears in a nuclear juncture in (2), in which the two verbal nuclei are adjacent and the linked predicate is intransitive, thus blunnun ricsian.
(2) [Bede 1 9.44.2]
Of þære tide Romane blunnun ricsian on Breotene.
|
Of |
þære |
Tide |
Romane |
|
from-PREP |
that-DAT.SG |
time-DAT.SG |
Roman-NOM.PL |
|
Blunnun |
ricsian |
On |
Breoten |
|
cease-PST.3PL |
rule-INF |
in-PREP |
Britain-DAT.SG |
‘From that time the Romans ceased to have dominion in Britain.’
The tree diagram in Figure 3 represents (2) as nuclear cosubordination. Notice that the node NUCLEUS is duplicated in order to mark dependent coordination.
Figure 3. Nuclear cosubordination
The verbs blinnan and onginnan can also partake in core junctures with a plain infinitive that is either non-adjacent to the matrix verb or complemented. For instance, the linked infinitive afeohton ‘to attack’ takes the second argument þa burg ‘the city’ in (3).
(3) [Bede 3 14.202.20]
& heo ealle afyrhte onweg flugon & blunnon þa burg afeohton.
|
& |
Heo |
ealle |
afyrhte |
|
and-CONJ |
he-NOM.3PL |
all-ACC.PL |
frightened-NOM.PL |
|
flugon |
& |
blunnon |
þa |
|
flee-PST.3PL |
and-CONJ |
cease-PST.3PL |
the-ACC.SG |
|
burg |
afeohton |
|
|
|
city-ACC.SG |
attack-INF |
|
|
‘And all fled away in alarm and ceased to attack the city.’
The verbs āblinnan, fundian, hīgian and wandian take part in core junctures in which the pre-verbal inflectional morpheme of the inflective infinitive tō precludes the adjacency of the two verbal nuclei. The linked core can be intransitive, as is the case with wislice to sprecanne ‘to speak wisely’ in (4a) or transitive, such as to asendenne bydelas and lareowas ‘to send messengers and teachers’ in (4b).
(4)
a. [CP 15.93.23]
Se ðonne se ðe fundige wislice to sprecanne, ondræde he him suiðlice, ðylæs his spræc gescynde ða anmodnesse ðara ðe ðærto hlystað.
|
Se |
Ðonne |
se |
|
that-NOM.SG |
therefore-ADV |
that-NOM.SG |
|
ðe |
fundige |
wislice |
|
who-REL |
try-PRS.3SG.SUBJV |
wisely-ADV |
|
to sprecanne |
ondræde |
he |
|
speak-INF.INFL |
fear-PRS.3SG.SUBJV |
he-NOM.3SG |
|
him |
suiðlice |
ðylæs |
|
he-DAT.3SG |
greatly-ADV |
lest-CONJ |
|
his |
Spræc |
gescynde |
|
he-GEN.3SG |
speech-NOM.SG |
disturb-PRS.3SG.SUBJV |
|
ða |
anmodnesse |
ðara |
|
the-ACC.SG |
confidence-ACC.SG |
those-GEN.PL |
|
ðe |
Ðærto |
hlystað |
|
who-REL |
thereto-ADV |
listen-PRS.3PL |
‘He therefore who hastens to speak wisely ought to fear greatly, lest his speech disturb the confidence of the hearers.’
b. [ÆCHom II, 5 43.53]
He fram frymðe middaneardes oð his geendunge. ne ablinð to asendenne bydelas and lareowas to lærenne his folc.
|
He |
Fram |
frymðe |
middaneardes |
|
he-NOM.3SG |
from-PREP |
beginning-DAT.SG |
world-GEN.SG |
|
oð |
His |
geendunge |
ne |
|
until-PREP |
he-GEN.3SG |
ending-DAT.SG |
not-NEG |
|
ablinð |
to asendenne |
bydelas |
and |
|
cease-PRS.3SG |
send-INF.INFL |
messenger-ACC.PL |
and-CONJ |
|
lareowas |
to lærenne |
his |
folc |
|
teacher-ACC.PL |
teach-INF.INFL |
he-GEN.3SG |
people-ACC.SG |
‘Because from the beginning of the world till its ending, he ceases not to send messengers and teachers to teach his people.’
The verbs āblinnan, blinnan, fundian, hīgian, onginnan and wandian appear in clause junctures linked by the clause-linkage marker þæt ‘that’, so that the verb in the linked clause is a morphological subjunctive, such as cwemon ‘please’ in (5).
(5) [HomS 14 141]
Ne ablinnan we, manna bearn, þæt we Gode cwemon, & deofol tynan, dæges & nihtes.
|
Ne |
Ablinnan |
we |
manna |
|
not-NEG |
cease-INF |
I-NOM.1PL |
man-GEN.PL |
|
bearn |
Þæt |
we |
Gode |
|
child-NOM.PL |
that-CONJ |
I-NOM.1PL |
God-ACC.SG |
|
cwemon |
& |
deofol |
tynan |
|
please-PRS.1PL.SUBJV |
and-CONJ |
devil-ACC.SG |
annoy-PRS.1PL.SUBJV |
|
dæges |
& |
nihtes |
|
|
day-GEN.SG |
and-CONJ |
night-GEN.SG |
|
‘Let us, the children of men, not cease to please God and annoy the devil day and night.’
The tree diagram in Figure 4 represents the first part of (5) as clause cosubordination. The clause linkage marker (CLM) þæt ‘that’ introduces the linked predication. The node CLAUSE is duplicated because the nexus of cosubordination takes place at the juncture level of the clause.
Figure 4. Clause cosubordination
In core junctures, the negative word ne ‘not’ consistently belongs to the matrix clause, so that the operator of negation is not repeated in the linked core. This is the case with (6).
(6) [ChrodR 1 79.39]
Forþi þonne swa miclan swa ge magon, mid worde and mid bysne, swa we bufan sædon, ne ablynnon ge to myngyenne þa eow betæhtan sceap.
|
Forþi |
þonne |
swa |
miclan |
|
therefore-ADV |
then-ADV |
as-CONJ |
much-DAT.SG |
|
swa |
Ge |
magon |
mid |
|
as-CONJ |
you-NOM.2PL |
can-PRS.2PL |
with-PREP |
|
worde |
And |
mid |
bysne |
|
word-DAT.SG |
and-CONJ |
with-PREP |
example-DAT.SG |
|
swa |
We |
bufan |
sædon |
|
as-CONJ |
I-NOM.1PL |
before-ADV |
say-PRS.1PL |
|
ne |
ablynnon |
ge |
to myngyenne |
|
not-NEG |
cease-INF |
you-NOM.2PL |
take care-INF.INFL |
|
þa |
eow |
betæhtan |
sceap |
|
that-ACC.PL |
you-ACC.PL |
take care-INF |
sheep-ACC.PL |
‘Therefore, as much as you can, in word as well as through example, as we said before, do not cease to take care of the sheep.’
In Figure 5, a segment of example (6) is represented by means of a tree diagram of core cosubordination. The nexus relation of cosubordination holds at the level of juncture of the core, which is represented by means of the repetition of the node CORE. The inflective infinitive is represented as a sequence of CLM (clause linkage marker) and the plain infinitive (see Section 8). The predication linked to the inflective infinitive to myngyenne ‘to be aware’ has been left unspecified, although it is worth pointing out that it results in a cosubordination inside another cosubordination (literally, you do not cease that you are aware that you take care).
Figure 5. Core cosubordination
In clause junctures with verbs of aspect, the operator of negation can be inserted in the matrix clause only or both in the matrix and the linked clause. In (7), the operator of negation appears in the matrix clause ne ablinnan we ‘let us not cease’.
(7) [HomS 14 141]
Ne ablinnan we, manna bearn, þæt we Gode cwemon, & deofol tynan, dæges & nihtes.
|
Ne |
Ablinnan |
we |
manna |
|
not-NEG |
cease-INF |
I-NOM.1PL |
man-GEN.PL |
|
bearn |
Þæt |
we |
Gode |
|
child-NOM.PL |
that-CONJ |
I-NOM.1PL |
God-ACC.SG |
|
cwemon |
& |
deofol |
tynan |
|
please-PRS.1PL.SUBJV |
and-CONJ |
devil-ACC.SG |
annoy-PRS.1PL.SUBJV |
|
dæges |
& |
nihtes |
|
|
day-GEN.SG |
and-CONJ |
night-GEN.SG |
|
‘Let us, the children of men, not cease to please God and annoy the devil day and night.’
Interestingly, it is possible for a given verb to present a different configuration, in which the operator of negation is inserted in the linked clause. This is the case with (8), in which the matrix verb is āblinnan, too.
(8) [GD 2 (C) 31.163.14]
Þa sona swa se Zalla gehyrde þone Godes wer næmnian, he ablan, þæt he na ma ne swencte þone ceorl mid þam tintregum.
|
Þa |
Sona |
swa |
|
then-ADV |
soon-ADV |
as-CONJ |
|
Se |
Zalla |
gehyrde |
|
the-NOM.SG |
Zalla-NOM.SG |
hear-PRS.3SG.SUBJV |
|
Þone |
Godes |
wer |
|
the-ACC.SG |
God-GENSG |
man-ACC.SG |
|
næmnian |
He |
ablan |
|
name-ACC.SG |
he-NOM.3SG |
cease-PST.3SG |
|
Þæt |
He |
na |
|
that-CONJ |
he-NOM.3SG |
not-NEG |
|
Ma |
Ne |
swencte |
|
more-ADV |
not-NEG |
torment-PRS.3SG.SUBJV |
|
Þone |
Ceorl |
mid |
|
the-ACC.SG |
servant-ACC.SG |
with-PREP |
|
Þam |
Tintregum |
|
|
that-DAT.PL |
torture-DAT.PL |
|
‘As soon as this Zalla heard the name of the man of God he ceased to torment him with those punishments.’
A configuration with negation duplicated in the matrix and the linked clause is also possible with āblinnan. This can be illustrated by means of ne ablan Romanus na forþon þæt he him ne þegnode ‘Romanus did not cease to serve him’ in (9), which also has double negation in the matrix verb.
(9) [GD 2 (C) 1.99.10]
& þonne hwæþre ne ablan Romanus na forþon þæt he him ne þegnode mid gerisenlicum gemetum.
|
& |
Þonne |
hwæþre |
ne |
|
and-CONJ |
then-ADV |
yet-ADV |
not-NEG |
|
Ablan |
Romanus |
na |
forþon |
|
cease-PST.3SG |
Romanus-NOM.SG |
not-NEG |
therefore-ADV |
|
Þæt |
He |
him |
ne |
|
that-CONJ |
he-NOM.3SG |
he-GEN.3SG |
not-NEG |
|
þegnode |
Mid |
gerisenlicum |
gemetum |
|
serve-PST.3SG |
with-PREP |
honourable-DAT.PL |
way-DAT.PL |
‘Yet, for all that, Romanus did not cease to serve him by all the possible means.’
The operator of negation in the matrix and the linked clause is also found in configurations involving the verb blinnan, such as Ne blan he þæt he his geongran ne manode ‘He did not cease to admonish his scholars’ in (10).
(10) [GD 1 (C) 4.27.4]
Ne blan he hwæðre, þæt he his geongran ne manode, þæt hi næfre gelyfdon heom sylfum to swiðe in þissere wisan.
|
Ne |
Blan |
he |
hwæðre |
|
not-NEG |
cease-PST.3SG |
he-NOM.3SG |
however-CONJ |
|
þæt |
He |
his |
geongran |
|
that-CONJ |
he-NOM.3SG |
he-GEN.3SG |
scholar-ACC.PL |
|
ne |
Manode |
þæt |
hi |
|
not-NEG |
admonish-PST.3SG |
that-CONJ |
he-NOM.3PL |
|
næfre |
Gelyfdon |
heom |
sylfum |
|
never-ADV |
believe-PST.3PL |
he-DAT.PL |
self-DAT.PL |
|
to |
Swiðe |
in |
þissere |
|
too-ADV |
right-ACC.SG |
in-PREP |
this-GEN.PL |
|
wisan |
|
|
|
|
wisdom-GEN.SG |
|
|
|
‘He did not cease to admonish his scholars not be proud of themselves for their wisdom.’
A convergent type of double negation is represented by a configuration with the verb blinnan and a negative adverb with scope over the matrix clause, as in þe ne blinnað næfre ‘who do not cease never’ in (11).
(11) [GD 4 (C) 46.335.4]
Hi wilniað, þæt hi lifigan aa in þære synne butan ænde, þa þe ne blinnað næfre, þæt hi syngian þa hwile þe hi lifgiað.
|
Hi |
wilniað |
þæt |
hi |
|
they-NOM.3PL |
wish-PRS.3PL |
that-CONJ |
they-NOM.3PL |
|
lifigan |
Aa |
in |
þære |
|
live-PRS.3PL |
always-ADV |
in-PREP |
this-DAT.SG |
|
synne |
butan |
ænde |
þa |
|
sin-DAT.SG |
without-PREP |
end-DAT.SG |
so-CONJ |
|
þe |
Ne |
blinnað |
næfre |
|
that-REL |
not-NEG |
cease-PRS.3SG |
never |
|
þæt |
Hi |
syngian |
þa |
|
that-CONJ |
they-NOM.3PL |
sin-PRS.3PL |
then-CONJ |
|
hwile |
Þe |
hi |
lifgiað |
|
while-PREP |
that-REL |
they-NOM.3PL |
live-PRS.3PL |
‘They wish to live always in this sin without end, so that they never cease to sin while they live.’
7. Control Verbs
Old English control verbs fall into two categories, as has been presented in Section 1: manipulative verbs [forbid], including āgǣlan, belēan, foresacan, forfōn, forhabban and gǣlan; and inverse control verbs [prevent], comprising bewerian, forbēodan, forwiernan and tōcweðan. In the remainder of Section 6, the constituent projection and the operator projection of control verbs are discussed, including the questions of juncture level, nexus relation and the operator of negation.
The first aspect which is relevant for the constituent projection of control verbs is that the verbs bewerian, forbēodan and forhabban can be passivised. Therefore, they hold the nexus relation of subordination with their linked units. With these verbs, the argument performing the thematic roles Patient and Theme can be the target of passivisation. If the Patient is the privileged syntactic argument[5] of the construction, it can be case-marked nominative, like he ‘he’ in (12a), or dative, like him ‘him’ in (12b). In instances like (12b), the dative case is preserved from the corresponding active.
(12)
a. [Bede 4 17.300.29]
Ne meahte he hwæðre from þære þegnunge beon bewered godspel to lærenne.
|
Ne |
Meahte |
he |
hwæðre |
|
not-NEG |
can-PST.3SG |
he-NOM.3SG |
still-CONJ |
|
From |
Þære |
þegnunge |
beon |
|
from-PREP |
the-DAT.SG |
service-DAT.SG |
be-INF |
|
Bowered |
Godspel |
to lærenne |
|
|
prevent-PST.PTCP |
gospel-ACC.SG |
teach-INF.INFL |
|
‘Still he could not be prevented from the service of teaching the gospel.’
b. [CP 11.73.15]
Sua hwelc ðonne sua ðissa uncysta hwelcre underðieded bið, him bið forboden ðæt he offrige Gode hlaf.
|
Sua |
Hwelc |
ðonne |
sua |
|
So-CONJ |
whoever-NOM.SG |
then-CONJ |
as-CONJ |
|
ðissa |
Uncysta |
hwelcre |
underðieded |
|
this-GEN.PL |
vice-ACC.PL |
whichever-GEN.PL |
subject-PST.PTCP |
|
bið |
Him |
bið |
forboden |
|
be-PRS.3SG |
he-DAT.3SG |
be-PRS.3SG |
forbid-PST.PTCP |
|
ðæt |
He |
offrige |
Gode |
|
that-CONJ |
he-NOM.3SG |
offer |
God-DAT.SG |
|
hlaf |
|
|
|
|
bread-ACC.SG |
|
|
|
‘Whoever, then, is subject to one of these vices is forbidden to offer bread to God.’
In (12a) and (12b), the linked predication is an argument of the matrix predication. In (12a), the juncture takes place at core level, while (12b) is an instance of clause subordination. The core juncture in (12a) has a linked core with a verb in the inflected infinitive (tō lærenne ‘to teach’). The clause junctures comprise the clause-linkage marker ðæt ‘that’ and a verb in the subjunctive. The tree diagram representation of (12b) is presented in Figure 6. Being a nexus of subordination, the core ðæt he offrige Gode hlaf ‘that he offers bread to God’ is an argument of the predicate bið forboden ‘is forbidden’. The first argument is the referential phrase in the dative case him ‘he’.
Figure 6. Subordination with control verbs
The verbs forwiernan and gǣlan can be found in configurations of core coordination. In terms of juncture, these verbs are complemented by linked cores with an inflective infinitive, such as tō specende ‘to speak’ in (13). From the point of view of nexus, the first argument of the matrix clause is not shared with the linked clause.
(13) [BenRW 7.39.26]
þe nygeðe eadmodnysse stæpe is, gef þeo mynecena hyre tunga forwyrnoð to specende.
|
Þe |
Nygeðe |
eadmodnysse |
stæpe |
|
then-ADV |
ninth-NOM.SG |
humility-NOM.SG |
step-NOM.SG |
|
Is |
Gef |
þeo |
mynecena |
|
be-PRS.3SG |
if-CONJ |
the-NOM.SG |
nun-NOM.SG |
|
Hyre |
Tunga |
forwyrnoð |
to |
|
she-GEN.SG |
tongue-ACC.SG |
restrain-PRS.3SG |
from-PREP |
|
Specende |
|
|
|
|
speak-PRS.PTCP |
|
|
|
‘If the female munuc restrains her tongue from speaking.’
The tree diagram for (13) can be seen in Figure 7. With control verbs that do not passivise, the linked predication cannot be an argument of the matrix predication. The nexus relation, therefore, is coordination. The shared argument hyre tunga ‘her tongue’ has been included in the matrix predication.
Figure 7. Coordination with control verbs
Only one instance of core coordination with the plain infinitive has been found, in particular, with the verb bewerian. Even though there is no clause-linkage marker, the two verbal nuclei are not adjacent and the linked infinitive is transitive (þam halgan geryne onfōn ‘receive the holy sacrament’). This is shown in (14).
(14) [Bede 1 16.82.23]
We him ne sculon biwerigan þam halgan geryne onfon.
|
We |
Him |
Ne |
sculon |
|
I-NOM.1PL |
he-DAT.3SG |
not-NEG |
shall-PRS.1PL |
|
biwerigan |
Þam |
halgan |
geryne |
|
prohibit-INF |
the-DAT.SG |
holy-DAT.SG |
sacrament-DAT.SG |
|
onfon |
|
|
|
|
receive-INF |
|
|
|
‘We shall not prohibit him from receiving the holy sacrament.’
The verbs forwiernan and forfōn can be found in complex syntactic configurations of clause coordination. This is illustrated, respectively, by (15a) and (15b). In (15a) and (15b) the first argument is not shared. The juncture takes place at the level of the clause because the linked unit comprises the clause-linkage marker þæt and a finite form of the verb, thus the subjunctive geweorðe ‘become’ in (15a), and the indicative cymð ‘comes’ in (15b).
(15)
a. [Bo 41.144.26]
Nat he hit no forðyþe he wille þæt hit geweorðe, ac forðy þe he wile forwernan þæt hit ne geweorðe.
|
Nat |
He |
hit |
|
know-PRS.3SG.NEG |
he-NOM.3SG |
it-ACC.3SG |
|
No |
forðyþe |
he |
|
not-NEG |
because-CONJ |
he-NOM.3SG |
|
Wille |
Þæt |
hit |
|
wish-PRS.3SG.SUBJV |
that-CONJ |
it-NOM.3SG |
|
Geweorðe |
Ac |
forðy þe |
|
happen-PRS.3SG.SUBJV |
but-CONJ |
because-CONJ |
|
He |
wile |
forwernan |
|
he-NOM.3SG |
wish-PRS.3SG.SUBJV |
prevent-INF |
|
Þæt |
Hit |
ne |
|
that-CONJ |
it-NOM.3SG |
not-NEG |
|
Geweorðe |
|
|
|
happen-PRS.3SG.SUBJV |
|
|
‘He does not know it because he wishes that it happens, but because he would like to prevent that it happens.’
b. [LawCn 1020 5]
& þæt hæbbe [ic] mid Godes fultume forene forfangen, þæt eow næfre heononforð þanon nan unfrið to ne cymð, þa hwile þe ge me rihtlice healdað & min lif byð.
|
& |
Þæt |
hæbbe |
ic |
|
and-CONJ |
that-CONJ |
have-PRS.1SG |
I-NOM.1SG |
|
Mid |
Godes |
fultume |
forene |
|
with-PREP |
God-GEN.SG |
help-DAT.SG |
before-ADV |
|
Forfangen |
Þæt |
eow |
næfre |
|
prevent-PST.PTCP |
that-CONJ |
you-ACC.PL |
never-ADV |
|
Heononforð |
Þanon |
nan |
unfrið |
|
from this time-ADV |
forth-ADV |
no-NOM.SG |
hostility-NOM.SG |
|
To |
Ne |
cymð |
þa hwile þe |
|
upon-PREP |
not-NEG |
come-PRS.3SG |
as long as-ADV |
|
Ge |
Me |
rihtlice |
healdað |
|
you-NOM.2PL |
I-ACC.1SG |
loyaly-ADV |
support-PRS.2PL |
|
& |
Min |
lif |
byð |
|
and-CONJ |
I-Gen.1SG |
life-NOM.SG |
be-PRS.3SG |
‘And with the help of God, I have taken measures to prevent hostility ever from this time forth coming upon you from that quarter, as long as you support me loyaly and my life lasts.’
In the operator projection, the negative word is inserted in the matrix clause of core junctures with the verbs forbēodan, forwiernan and gǣlan, so that the negation operator does not have scope over the core. This is the case with he us ne forbēode yfel tō donne ‘he does not forbid us to do evil’ in (16).
(16) [ÆLet 6 210]
And he ne abyhð na us, he us ne forbeode ealle unrihtwisnyssa and yfel to donne.
|
And |
He |
ne |
|
and-CONJ |
he-NOM.3SG |
not-NEG |
|
Abyhð |
Na |
us |
|
bow.PRS.3SG |
not-NEG |
I-DAT.1PL |
|
Þæt |
He |
us |
|
that-CONJ |
he-NOM.3SG |
I-DAT.1PL |
|
Ne |
forbeode |
ealle |
|
not-NEG |
prohibit-PRS.3SG.SUBJV |
all-ACC.PL |
|
Unrihtwisnyssa |
And |
yfel |
|
unrighteousness-ACC.PL |
and-CONJ |
evil-ACC.SG |
|
to donne |
|
|
|
do-INF.INFL |
|
|
‘And he does not bow to us, so that he does not prohibit us to do all follies and evil.’
The verbs forbēodan and forwiernan behave exactly in the same way if the juncture takes place at the level of the clause: the negative word is inserted in the matrix clause only. The operator does not have scope over the linked core. The verb bewerian also follows this pattern. This is clearly the preferred configuration when the matrix verb belongs to the set of pre-auxiliaries, including willan ‘will’, magan ‘may, might’ and mōt ‘must’, illustrated in (17).
(17) [Bo 21.49.22]
Se ilca forwyrnð þæræ sæ þæt heo ne mot þone þeorscwold oferstæppan þære eorþan.
|
Se |
Ilca |
forwyrnð |
þæræ |
|
the-NOM.SG |
same-NOM.SG |
hinder-PRS.3SG |
the-DAT.SG |
|
sæ |
þæt |
heo |
ne |
|
sea-DAT.SG |
that-CONJ |
she-NOM.3SG |
not-NEG |
|
mot |
þone |
þeorscwold |
oferstæppan |
|
may-PRS.3SG |
the-ACC.SG |
threshold-ACC.SG |
overstep-INF |
|
þære |
eorþan |
|
|
|
the-GEN.SG |
earth-GEN.SG |
|
|
‘The same warns the sea that it may not overstep the threshold of the earth.’
When the matrix verb is not a pre-auxiliary and the juncture takes place at clause level, the negative word is inserted in the linked clause, which is introduced by the clause-linkage marker þæt ‘that’ and presents a verb in the morphological subjunctive. This happens with the verbs āgǣlan, bewerian, forbēodan, forhabban and forwiernan. This is illustrated with respect to āgǣlan in (18).
(18) [Or 3 9.73.14]
& swa eall þæt folc wearð mid him anum agæled þæt hie þæs wealles nane gieman ne dydon, oð Alexandres þegnas toemnes him þone weall abræcon & þær in coman.
|
& |
Swa |
eall |
þæt |
|
and-CONJ |
as-CONJ |
all-NOM.SG |
the-NOM.SG |
|
Folc |
Wearð |
mid |
him |
|
people- NOM.SG |
be-PST.3SG |
with-PREP |
he-DAT.3SG |
|
Anum |
Agæled |
þæt |
hie |
|
alone-DAT.SG |
preoccupy-PST.PTCP |
that-CONJ |
he-NOM.3PL |
|
Þæs |
Wealles |
nane |
gieman |
|
the-GEN.SG |
wall-GEN.SG |
none-ACC.SG |
take care-INF |
|
Ne |
Dydon |
oð |
Alexandres |
|
not-NEG |
do-PST.3PL |
until-CONJ |
Alexander-GEN.SG |
|
Þegnas |
Toemnes |
him |
þone |
|
thane-NOM.PL |
alongside-PREP |
he-DAT.3PL |
the-ACC.SG |
|
Weall |
Abræcon |
& |
þær |
|
wall-ACC.SG |
breach-PST.3PL |
and-CONJ |
there-ADV |
|
In |
Coman |
|
|
|
in-PREP |
come-PST.3PL |
|
|
‘In this way all the people were preoccupied with him alone so that none of them paid attention to the wall until Alexander’s thanes breached it alongside them and so entered there.’
8. Discussion
SVCs have been defined, in line with previous research by Aikhenvald (2006), Van Staden and Reesink (2008), Roberts (2010) and Haspelmath (2016), as consisting of two lexical verbs that express a single event and share the first argument. The verbs in an SVC do not hold a predicate-argument relation and there cannot be any subordination marker between them. The verbs in an SVC may share operators of negation as well as tense, aspect and modality.
In order to decide whether aspectual and control verbs can be said to partake in SVCs in Old English or not, it is necessary to look at the constituent projection and at the operator projection of configurations with verbs from these classes and to check on the above mentioned characteristics of SVCs.
In the constituent projection, Old English aspectual verbs share the first argument, not only at the nucleus and the core levels but also at the clause level of juncture. This reinforces the semantic unity of the event. No subordination markers are found at the nucleus or core levels. In contrast, subordination markers are compulsory at the level of the clause. To continue with the constituent projection, Old English verbs of control do not share the first argument. Moreover, they turn up in configurations in which the linked predication is an argument of the matrix predication, including some instances of coordination and subordination. In complex syntactic configurations with aspectual verbs, the linked predication is not an argument of the matrix predication.
In the operator projection, the operator of negation is shared by the matrix and the linked predication with aspectual verbs. This is consistent with the nexus relation of cosubordination, which is defined on the grounds of shared arguments and operators. It is possible, therefore, for syntactic configurations with aspectual verbs to mark the negation in the matrix predication, in the linked predication or both in the matrix and the linked predication. Given the unity of the event, it does not make any semantic difference to insert the negative word once or twice, given that the negation of one of the verbs turns the whole construction negative. On the other hand, negation does not have scope over the two predications of complex configurations with control verbs. This means that the negative word can be inserted in the matrix predication or in the linked predication, but not in both of them. With control verbs, clause junctures admit negation in the matrix or negation in the linked unit, whereas core junctures consistently mark the negation in the linked unit. The semantic unity of the event is clearly undermined if the operator of negation cannot be shared. Put differently, coordination and subordination are looser syntactic configurations and, as such, more adequate for the expression of relatively independent events.
To summarise, aspectual verbs share the first argument and the operator of negation, whereas control verbs share neither. This is represented in the theory of the complex clause of RRG by means of the nexus relation of cosubordination, which holds with aspectual verbs; and by means of the relations of subordination and coordination, which can be identified with control verbs. The linked predication of aspectual verbs is not an argument of the matrix predication. We can reach the conclusion, therefore, that Old English aspectual verbs take part in SVCs while control verbs do not.
This raises the additional question of the juncture level at which SVCs can be identified. Nuclear junctures are the best candidates for SVCs because they meet all the requirements listed above. Core junctures can also be considered SVCs because the pre-verbal inflectional morpheme tō is not a clause-linkage marker but rather belongs in the inflectional paradigm of the verb, like the post-verbal inflectional morpheme of the infinitive, -enne. Clause junctures display a clause-linkage marker that introduces the linked clause, in such a way that they cannot be described as SVCs.
This conclusion restricts the range of SVCs to cosubordination. For this reason, aspectual verbs rather than control verbs can be said to give rise to SVCs. In these syntactic configurations, aspectual verbs are monoeventual, unmarked for coordination or subordination, and share one argument, the subject. Furthermore, the non-finite verb cannot be said to be an argument of the finite one, both belonging in an only clause and intonation group. In this sense, Van Valin (2005, p. 206) defines the interclausal semantic relation ‘phase’ as holding if “a separate verb describes a facet of the temporal envelope of a state of affairs, specifically its onset, its termination, or its continuation”. This study concurs with this view rather than with Roberts (2010), which includes control verb within Phase. At least in Old English, control verbs differ both semantically and syntactically from aspectual verbs. Aspectual verbs, at least in nuclear and core junctures, comply with the requirements of SVCs whereas control verbs do not.
This poses two further questions. Firstly, are there other areas of Old English syntax where serial configurations can be found? And, secondly, does the discussion of serialisation improve our understanding of the syntax of Old English? These questions are addressed in the remainder of this section with respect to pre-modal verbs, the verb dōn ‘to do’ and the predicative construction.
Beginning with pre-modal verbs, Old English preterite-present verbs include āh ‘to possess’, ann ‘to grant’, cann ‘can, dēag ‘to avail’, dearr ‘dare’, geman ‘remember’, geneah ‘be enough’, mag ‘may’, mōt ‘must’, sceal ‘shall’, wāt ‘to know’, þearf ‘need’ (Campbell, 1964, pp. 343-345; Hogg & Fulk, 2011, p. 299). The modal auxiliaries of Present-Day English, as is well-known, develop from Old English Preterite-present verbs. Mitchell (1985, ∫∫990-992) excludes the preterite-present verbs āh ‘to possess’, ann ‘to grant’, geman ‘remember’, geneah ‘be enough’ and wāt ‘to know’ from the inventory of modal auxiliaries, while he includes the anomalous verb wile ‘to wish’.
Ringe and Taylor (2014, p. 427) remark that “PDE modals were not distinct, syntactically and semantically, from full verbs”, thus instances like forðan ðe he symble wyle god and næfre nan yfel ‘because he always desires good and never evil’ (coaelive,+ALS_[Sebastian]:49.36). Denison (1993), Mitchell (1985, §990-1024) and Lowrey (2012, p. 8), among others, note that not all preterite-present verbs can be used as full lexical verbs in Old English. Lowrey (2012, p. 8) underlines some modal properties that distinguish preterite-present verbs form fully lexical verbs, including defective morphological paradigms (in which the infinitive is frequently unattested) and syntactic ellipsis, as in Ic worda gespræc ma þonne ic sceolde ‘I have spoken more words than I should’ (O2 St. Andrew, 28: 956). Ringe and Taylor (2014, p. 426), quoting Warner (1993, p. 111), also note the property of pseudo-gapping, as in he sceall hyran feondan, gyf he nele freondan ‘he will be subject to enemies if he will not be [subject to] friends’ (cowulf,WHom_17:50.1398).
According to Denison (1993, p. 308), mag ‘may’ and wile ‘to wish’ can govern a þæt-clause instead of an infinitive, as can be seen in Ac þæt hie magon þæt hie þas tida leahtrien ‘But all they can do is blame the times (Or 74.25; Denison, 1993, p. 308); and in Ic wolde þæt þa ongeaten þe ... hwelc mildsung siþþan wæs, siþþan se cristendom wæs ‘I would like those who ... to understand what kind of mercy there was after the coming of Christianity’ (Or 38.10; Denison, 1993, p. 308). The acquisition of auxiliary properties has been paralleled by the loss of the complementation with a þæt-clause, although there is not necessarily a causal relationship between these aspects. The replacement of þæt-clauses with plain infinitives, as in & he næfre hine oferswiðan meahte ‘and he was never able to overcome him’ (Mart 3 178.41; Denison, 1993, p. 303), favours the adjacency of the two verbal nuclei, which also share the first argument and do not hold a predicate-argument relationship. However, the auxiliarisation of preterite-present and anomalous verbs leading to the modal auxiliaries of Present-Day English (Lowrey, 2012) has resulted in the loss of a critical property, namely that all the verbs in an SVC must be lexical verbs.
As for the verb dōn ‘to do’, Royster (1922, p. 337) finds 17 instances of causative dōn ‘to do’ plus the uninflected infinitive (mostly in poetry), and sixty-one of causative dōn ‘to do’ followed by a þæt-clause. Visser (1963-1973, p. 1345) provides the two glosses for Latin ‘ego uiuere faciam’: ic lifgan gedom and lifian ic do (Vespasian and Regius Psalters). Royster (1922, p. 335) gives examples from prose such as Matheum he gedyde gangan ‘Matthew made him go’ (Psalms 28, 8), in which the linked verb is the plain infinitive gangan ‘to go’; and Drihten us gedyde þæt we moston buian ‘The Lord made us dwell’ (Blicking Homilies, 239, 16), where the linked predication is the þæt-clause þæt we moston buian ‘that we must dwell’.
The acquisition of auxiliary properties by dōn ‘to do’ (Denison, 1993, p. 457) is the origin of expressions like I do not buy oranges, in which, if we assumed that there is a first argument of do, it would be shared with buy. It makes more sense, however, to consider do an auxiliary and to analyse it as an operator of negation. This is a consequence of the auxiliary status of do, with the associated loss of lexical meaning. As is the case with modal verbs, lack of lexical meaning is incompatible with serialisation.
The pre-modal verbs and the causative verb dōn ‘to do’ belong in the area of auxiliarisation. From a diachronic point of view, these verbs lose lexical meaning as they acquire grammatical meaning. Nevertheless, it has to be noted that, while they keep their full lexical meaning and satisfy other central requirements, they give rise to SVCs. This can be said of the two examples discussed above: & he næfre hine oferswiðan meahte and ic lifgan gedom. In these expressions, there are no markers of subordination, there is no verb-argument relation, the first argument is shared by the matrix and the linked predication and both verbs convey lexical meaning. The acquisition of auxiliary properties is parallel to the loss of serial status. A serialisation cline can be described, therefore, that distinguishes fully serial status from declining serial status and non-serial status. Along the serialisation cline, pre-modal verbs and the causative verb dōn ‘to do’ are at a stage of declining serial status because the alternative non-serial construction is possible (the þæt-clause) and because the lexical meaning of the verb is already yielding way to its grammatical meaning.
Whereas the acquisition of auxiliary properties is incompatible with serialisation, predicative constructions clearly have serial status. Ringe and Taylor (2014, pp. 494-495) describe adjunct participial clauses as “non-finite clauses headed by a participle, present or past. Syntactically, there is no connection between the participial clause and the matrix; semantically they play an adverbial function”. An instance of this type of clause is And ðær com ridende sum egeful ridda ‘and a terrible rider came riding there’ (coaelive, LS[Maccabees]:773.5334; Ringe and Taylor, 2014, p. 493). Martín Arista (2022) enlarges this construction in order to include adjunct infinitival clauses, which Callaway (1913, p. 197) and Traugott (1992, p. 249) explain in terms of competition with the participle, in pairs like Þa com þær færlice yrnan an þearle wod cu ‘Then by chance there came running a very mad cow’ (ÆLS (Martin) 1038); and Heo com þa yrnende mid egeslicum eagum ‘She came then running with terrifying eyes’ (ÆLS (Martin) 1043). Callaway (1913, p. 290), Ogura (2000, p. 70; 2002, p. 38) and Ringe and Taylor (2014, p. 488) have found evidence of this competition with verbs of motion and rest, thus instances such as Þa com ðær yrnan sum olbenda ‘Then a camel came running there’ (comart3,Mart5[Kotzor]…Se27, A.26.1887); and Þæt scræf ... þe ða seofon halgan lagon inne slapan ‘The cave ... that the seven saints lay sleeping in’ (cosevens,LS34[SevenSleepers]:375.278; Ringe & Taylor 2014, p. 488).
According to Martín Arista (2022, p. 102), the semantics of the matrix verb of the predicative construction is not restricted to motion and rest. Instead, it includes motion and rest (general motion, manner of motion, path of motion, non-translational motion and position) as well as action, causative, change of state, contact, perception, possession, speech and state verbs. This author states that the infinitive of the predicative construction is replaced by the participle because the semantics of the matrix verbs found in this construction is not tight enough for the syntactically tight nuclear juncture, which is comprised of two adjacent verbal nuclei.
Sections 5 and 6 have shown the complementary development. Verbs of control and aspect present competition between tighter (nuclear or core juncture) and looser syntax (clausal juncture). The development of the complementation of these verbs, noted by Molencki (1991), and described as competition by Los (2005), can be explained on the grounds of the replacement of clausal junctures as well as the maintenance of nuclear and core junctures. The preference for the tighter syntactic configuration found in nuclear and core junctures is a consequence of the semantics of phase or causative interclausal relations. This development is predicted by the Interclausal Relations Hierarchy (Van Valin & LaPolla, 1997, pp. 480-481) and by the degree of event integration (Givón, 2009a, p. 68). From the perspective of serialisation, aspectual verbs with plain infinitives, as is the case with the predicative infinitive and participle, fully qualify as SVCs.
On the serialisation cline, some Old English syntactic configurations can be classified as SVCs. Such configurations include linked infinitives and participles in predicative constructions. Other syntactic configurations can be considered SVCs in Old English, but they have lost this status in Present-Day English. This is the case with pre-modal verbs and the causative verb dōn ‘to do’. Finally, some non-serial configurations can be described as undergoing a change in the direction of serial verb constructions. Linked predications of the clausal level with finite forms of the verb, typically in the subjunctive, belong in this category. The resulting constructions are clearly serial with aspectual verbs, as they partake in nexus relations of cosubordination. In diachronic analysis, this is in keeping with Van Staden and Reesink’s (2008, p. 24) typological distinction between independent serialisation (with fully inflected verbs, independently of their status) and dependent verb serialisation (with a fully inflected verb and the others in a bare form). The complementation of Old English aspectual verbs with linked non-finite predications is a clear instance of dependent verb serialisation not only for morphological reasons but also for syntactic reasons. In dependent verb serialisation, the first argument is shared by the matrix and the linked predication, as is the case with aspectual verbs in Old English, including clause junctures.
It turns out that the two questions discussed in this section are related to each other. The existence of serial constructions in Old English is not restricted to aspectual verbs, which throws new light into the relation between several phenomena usually treated as independent; and allows us to describe a cline of serialisation along which certain constructions achieve, keep or lose the serial status. This adds a new perspective to the changes of verbal complementation throughout the history of English. These changes have been described in terms of the competition between finite clause complements and non-finite clause complements but the questions raised by the discussion based on SVCs have not been addressed, including the semantic unity of the event.
9. Conclusion
This article has discussed the existence of SVCs in Old English. It has been found that the matrix predication of aspectual verbs shares the first argument and the operator of negation with the linked predication. This is not the case with control verbs, which cannot be said to occur in SVCs. With aspectual verbs, nuclear junctures and core junctures can be considered serial, whereas clause junctures cannot because they involve a clause-linkage marker that separates the matrix verb from the linked verb.
This article has also related complementation to auxiliarisation through a cline of serialisation. Along the serialisation cline, syntactic configurations achieve, keep or lose their serial status. Predications of the clausal level of juncture with finite forms of aspectual verbs in Old English can be described as going up the cline of serialisation. The non-finite counterparts clearly enjoy serial status both in Old English and PDE. As for the other convergent constructions, linked infinitives and participles in predicative constructions have remained serial, while the configurations with pre-modals and the causative verb dōn ‘to do’ can be considered serial in Old English but they lose their serial status as they acquire auxiliary properties.
To conclude, this article has shown that serialisation is a result of a tendency towards synthetic configurations that are preferred over analytic configurations with verbs of various classes. Future research should explore other constructions that follow this tendency and explain them on typological grounds.
