On the syntax of object pronouns in Old English and Early Middle English

The paper investigates the conditions that determine the distribution of object pronouns in the period before the latter become fixed to the post-verbal position. In the transition between Old English and Early Middle English (henceforth OE and EME), object pronouns surface to the left of T, to the left of V and to the right of V. It will be shown that the distribution in the pre-T and post-V positions in OE is due to information structural and prosodic interface conditions, whereas the position at the left of V is analysed as unmarked. In EME, the pre-T position continues to be linked to topicalization, whereas the pre-V position shows a progressively less frequent distribution. The post-verbal mapping of object pronouns can still be linked to prosodic 1 and information structural (IS) mapping conditions, but the data show that the post-verbal position was on the way to being reanalysed as the unmarked position. The change will be attributed to two factors: a) the new option of de-accentuation in situ, leading pronouns to choose the non-finite verb as their phonological host, and b) the grammaticalization of the definite determiner that renders light (non-branching and left-branching) objects into right-branching, heavy objects at the end of the OE period. We will provide evidence that it is the latter factor that leads to the new option of de-accenting pronouns in situ (cf. Hinterhölzl 2017).


Introduction
VO word order is grammaticalized in the transition from OE to ME, but as Pintzuk and Taylor (2012 a;b;2015) have pointed out, pronouns and light elements retain a pre-verbal position for a longer period than object DPs or PPs.We noticed in a preliminary study of a sample of subordinate clauses from the EME texts of the Continuations of Peterborough, Holy Maidenhood and the Kentish Homilies, that pronouns can occur in three different positions in one and the same text.The following examples are taken from the Continuations of Peterborough: ( 'That he should obtain for him the monastery of Peterborough.'Moreover, in a thorough investigation of object placement in different EME texts, Kroch and Taylor (2000) observe that the Lambeth Homilies and the Katherine Group, both coming from the West Midlands, show a similar ratio of underlying VO word order,2 but a different ratio of pronoun scrambling, with pronouns being found in the post-verbal position in the majority of cases in the Katherine group.This piece of evidence is left unaccounted for in Kroch and Taylor (2000).In this paper, a uniform account will be proposed for the growing numbers of object pronouns in the post-verbal position, and for the simultaneous presence of object pronouns at the left of the finite verb and at the left of the non-finite verb.
If the post-verbal mapping of a pronoun is attributed to the grammaticalization of the VO word order, in fact, the question arises how the simultaneous presence of pronouns in pre-verbal and pre-T position can be explained.If object pronouns can occur in three different positions in the same text, the syntactic and interpretive conditions that determine their distribution need to be identified.
The paper is organized as follows.In Section 2, basic assumptions and the theoretical framework are introduced.In Section 3, the sample and the methodology are described.Section 4 and Section 5 present the quantitative and qualitative findings for the OE and EME data, respectively.Section 6 summarizes the empirical observations and Section 7 presents the diachronic account of the data.

Theoretical framework
Our framework is based on the theoretical account laid out in detail in Hinterhölzl (2014;2017).In fact, the purpose of the paper is to verify (or falsify) some of the claims made in Hinterhölzl (2017) about the factors determining the word order change from OV to VO in the history of English.Hinterhölzl (2017) lays out a theoretical scenario based on interface conditions that closely matches the empirical findings reported in Kroch & Taylor (2000).The major claims of this paper are: A) the post-verbal placement of nominal objects is due to the grammaticalization of the definite determiner at the end of the OE period which makes these objects heavy from a prosodic point of view.In the interface account, given but heavy objects must be spelled-out in the post-verbal domain in a prosodic position that counts as weak.B) This process leads to the option of de-accenting arguments in situ which constitutes the basis for the post-verbal placement of pronouns in the successive stage.
While the scenario devised in Hinterhölzl (2017) is based on quantitative data concerning the distribution of (referential) nominal objects and pronominal objects in Kroch & Taylor (2000) alone, the present paper sets out to test the second claim (claim B) and its theoretical underpinnings against the background of a thorough qualitative analysis of the IS-structural role and the prosodic profile of the distribution of object pronouns in both OE and ME. 3 On the basis of this qualitative analysis it can be shown that the surplus of post-verbal pronouns in EME texts corresponds to weak, unstressed pronouns constituting familiar topics in the terminology of Frascarelli and Hinterhölzl (2007).
The reader who is not familiar with the interface account employed here is referred to Hinterhölzl (2017), where the theoretical framework and its motivation are described in detail.In this paper, only the most important assumptions will be summarized for the convenience of the reader and focus will be on outlining in greater detail the informational structural categories that constitute the basis of our qualitative analysis of the data and the ways in which they can be identified in the actual text material.
In particular, the account aims to derive the pre-and post-verbal position of constituents within an anti-symmetric approach by assuming obligatory leftward movement operations of constituents from their post-verbal base position and interface conditions which determine the spell out of constituents either in their base position or in their derived position.The individual movement operations are listed in (2).The interface conditions that constrain the spell-out of copies are listed (but not motivated) in ( 3) and ( 4).While the conditions in (3) describe the impact of information structure on the spell-out of arguments and adjuncts, constituting restrictions on the LF-side of the computation, (4) specifies a prosodic interface condition that requires heavy constituents to be predominantly spelled out in a post-verbal position (cf.4a).
To achieve this effect, the condition is taken to apply in the I-domain, but does not apply to constituents in their base-position, that is to say, in the v-domain.In other words, this interface condition is phase-based.For the working of the prosodic condition, non-branching and left-branching constituents count as light and heavy constituents are defined as specified in (4b).The linearization effects summarized in (5) are derived from the syntactic structure outlined in (2), and the interface conditions given in (3)-(4). 44) a. Within the I-domain, heavy phrases must be mapped on a strong branch in prosodic structure (the condition does not apply in the v-domain) b.A syntactic phrase XP counts as heavy if both its head and the complement of X contain lexical material (5) Licensing movement of the object to the [Spec, CASEP] position, leaving a copy in VP: a Spell-out of a given object due to the Givenness Transparency condition: Spell-out of a focused object due to the Focus Transparency condition: Spell-out of a heavy object due to the prosodic mapping condition: Furthermore, it is assumed that prosodic structure is built in parallel with the syntactic derivation initiating with the verb and forming prosodic constituents in a successive cyclic way, according to the heading procedure given in ( 6).
(6) a. Extrinsic heading: (default) In prosodic composition, the right-hand member is metrically stronger than its sister constituent b.Intrinsic heading: (language particular) A weak element may project its inherent metrical value w independently of its position in prosodic structure (6a) ensures that constituents that occupy a right branch with respect to the verb will be mapped onto a stronger position than the verb, while constituents that occupy a left branch with respect to the verb will be mapped onto a weak(er) position in prosodic structure.Consequently, heavy and focussed elements are expected to appear in a post-verbal position, while light and given elements are predicted to appear in a pre-verbal position.This constitutes the default option.Notice that this option is not only derived from the theory, but reflects general observations on linear word order in Earlier Germanic (cf. Behaghel 1932;Hróarsdóttir 2000;Hinterhölzl 2017).
In addition, it is submitted that a language may use a costlier option in which the head of a prosodic constituent is determined by the metrical properties of its members, as specified in (6b).This option allows for the de-accentuation of a given constituent in its (post-verbal) base position.In addition, it is submitted that the C-domain is an LF transparent domain, where topicalized or focalised material can be spelled out without any weight restrictions.The I-domain in a VO-language like modern English is assumed to be a PF-transparent domain, while the I-domain of an OV-language like German is an LF-transparent domain (cf.Hinterhölzl 2017 for the details).What is important for the purposes of our account is that this difference amounts to the application of the prosodic condition in (4) in the English I-domain, but not in the German I-domain.
As far as the identification of information structural (IS-)categories is concerned, we take recourse to the notions and empirical observations in Frascarelli and Hinterhölzl (2007).In general, pronouns can appear as strong accented elements, for instance, when they are (contrastively) focused, as in (7), or they can occur as weak deaccented elements, typically when they function as continuing aboutness-topics (A-topics) or as familiar topics (F-topics) in the discourse.Frascarelli & Hinterhölzl (2007) distinguish between four types of topics.A) A shifted aboutness topic (SA-topic) refers to a referent that the sentence is about (Reinhart 1981;Lambrecht 1994); constitutes a constituent that is newly introduced, newly changed or returned to (Givòn 1983:8); which is a matter of current interest or concern (Strawson 1964), as is illustrated in (8).
(8) I want to organize a party next week.a. John already agreed to come.Mary said she was not sure.b.John already agreed to come.He said he would bring a salad.
In (8a), John introduces a new referent that the first sentence can be taken to be about (a newly introduced aboutness topic).Mary in (8a) introduces a shift to a new A-topic.(B) A contrastive topic (C-topic) is an element which induces alternatives that have no impact on the focus value but creates oppositional pairs with respect to other topics (Kuno 1976;Büring 1997), as is illustrated in (9).( 9) What are your siblings doing for Christmas?My brother is going to New York, while my sister is staying home with Mum.
C) A familiar topic (F-topic) is a given, d-linked constituent which is typically destressed and realized in pronominal form (Pesetsky 1987) and is generally used for topic continuity (Givòn 1983).D) Henceforth, a continuing aboutness topic is a subtype of an F-topic.
While several nominal constituents can be given or d-linked in a specific sentence in the discourse, only one of them can refer to the established aboutness topic in the discourse.
For instance, the occurrences of he in (8b) are continuing aboutness topics, while in (10), the occurrence of him counts as a plain F-topic, given that the pronoun she in subject position constitutes a continuing aboutness topic.
(10) I met John and Mary yesterday.She wants to marry him.
Frascarelli & Hinterhölzl (2007) also show that shifting A-topics and C-topics receive a specific accent and occur in a higher position in the C-domain (above the position of focus) in Italian.F-topics are destressed and occur in a position below focused elements in the C-domain in Italian.Continuing A-topics are typically realized in pronominal form as well and occur in the position of F-topics in Italian, where they enter into an Agree relation with the higher Aboutness head.As far as we can see, however, nothing excludes the possibility that they are realized in the Aboutness head as a destressed pronoun in other languages.
From this characterization, it follows that pronouns are primarily used as familiar topics and continuing A-topics, though it is not excluded that they serve as shifting A-topics or C-topics as well.In the latter case however, they will receive some accent and cannot be considered to be weak or clitic elements.
Weak or clitic elements instead are in need of a host.While special clitics in the terminology of Zwicky (1977) target a specific head, simple clitics form a prosodic constituent with an adjacent constituent as a phonological host.Selkirk (1995) distinguishes between affixal and free clitics.While an affixal clitic forms a prosodic word with its host, a free clitic is adjoined to the phonological phrase of its host, as is illustrated in (11).For the present paper, it is proposed without any argumentation that affixal clitics are derived via local head movement to the host, while free clitics only require to be moved (via XP-movement) into a position adjacent to the host.Given these general considerations of topic types and the syntactic and prosodic characteristics of pronouns, the aim is to investigate the distribution of pronouns in the history of English under the hypothesis that OE was a discourse configurational language where many if not all positions individuated for Italian and German were syntactically represented in this language stage.Given that the aim is to relate the different mappings of pronouns to the ongoing language change from linear OV to linear VO, it is appropriate to briefly give some background on the matter.According to Hinterhölzl (2017), Struik andvan Kemenade (2018) andDe Bastiani (2019), the mapping of constituents in OE was driven by IS and prosodic factors.Given (nominal and pronominal) objects are typically mapped in pre-verbal position, but a progressively more frequent spell-out of non-pronominal objects in post-verbal position can be noticed already in the OE period.Following the scenario in Hinterhölzl (2017), this empirical evidence will be linked to the grammaticalization of the definite determiner and shown that it is this factor that eventually leads to the option of de-accenting in situ of given constituents in general and of weak pronouns in particular (for more details, cf.Section 5), confirming in this way a major prediction made in Hinterhölzl (2017).
Summarizing the framework, it is predicted that at the outset pre-T pronominal elements will function foremost as shifting A-topics, C-topics or possibly also as continuing A-topics, while F-topics should appear in the unmarked pre-verbal position.Furthermore, the post-verbal position should be restricted to (contrastively) focused pronouns.
Moreover, the hypothesis will be tested that OE special clitics, that is, elements that target a specific syntactic position via head movement and are analysed as affixal clitics prosodically, undergo a reanalysis into simple clitics, which are to be analysed as free clitics prosodically (cf.Hinterhölzl 2017).According to Pintzuk (1999) object pronouns in OE can optionally cliticize to the left or to the right edge of the IP, in her terminology.In a more recent account, van Kemenade (2009) proposes the presence of a ΣP above the TP, which licenses subject or object pronouns at the left of a discourse marker preceding the finite verb. 5The hypothesis about the change in clitic status made in Hinterhölzl (2017) needs to be qualified against the background of the results of this study.While there is no doubt that ME pronouns are simple clitics, which are to be analysed as free clitics prosodically, the presupposition made in the scenario, namely that OE pronouns were special clitics targeting a specific functional head above TP needs rethinking on the basis of the observation that pronouns in the post-verbal domain in OE are not restricted to strong (stressed) pronouns.The latter fact indicates that weak/clitic pronouns can be licensed by the non-finite verb in OE and do not require to be moved to the so-called Wackernagel position, as they arguably did in OHG, possibly pointing to an interesting difference in the pronominal system between English and German very early on.
Finally, a uniform account for the attested variation and the pertinent change in word order will be devised in Section 7.

Sample and methodology
Sentences with a complex verbal form with a subject and at least one object were collected, in order to control for V-to-T movement.The sample of OE main clauses was gathered by querying the whole YCOE Taylor et al. (2003), queried through the Corpus Studio Suite (Komen 2011) corpus and setting the number of examples to 100 for each time frame, obtaining a total amount of 395 clauses.The sample of OE subordinate clauses was obtained in a similar way, but keeping the number of examples lower, obtaining a total amount of 161 clauses.Among the clauses obtained, the ones containing an object pronoun were filtered out, resulting in a total number of 105 main clauses and 36 subordinate clauses.The sentences present Aux -V as well as V -Aux order.The OE data serve as a starting point for the analysis of EME.
The EME data were collected by querying the PPCME2 files of the following texts: Kentish Homilies, Kentish Sermons (Kentish), The Peterborough Chronicle (East Midlands), The Trinity Homilies, Vices and Virtues (South-East Midlands), the Lambeth Homilies, the Life of Saint Juliana, the Guardianship of the Soul and Holy Maidenhood (West Midlands).Each text was queried for the total amount of sentences with a complex verbal form, a subject and at least one object, both main and subordinate.The results were then filtered by selecting the sentences containing an object pronoun.A total amount of 238 main clauses and 237 subordinate clauses with the given features was obtained.
While the OE sample serves to determine the base line of the ongoing change, the focus here will be on an in-depth investigation of the great variation found in EME texts.Thus, it is important to provide some more information about the dialectal6 provenance and transmission history of the EME texts chosen. 7The texts representing the Kentish dialect are the Kentish Homilies and the Kentish Sermons.The Kentish Homilies consist of two homilies, one of which defined as the "earliest ME document".8Their composition is dated between 1108-1122 A.D. The Kentish Sermons were presumably composed before 1250 A.D., but the manuscript is dated around 1275 A.D. They consist of five sermons, all translated from French.Hall (1972) observes that the influence of the French is remarkable, permeating also the syntax of the work.As far as was possible,9 the influence of the French version was controlled for, and as far as could be seen, the mapping of pronouns in this text is not influenced by the French version.
The texts of the South-East Midlands dialects consist in the dialogue Vices and virtues and the Trinity Homilies.The Trinity Homilies are dated around 1225, five sermons are shared with the Lambeth Homilies.10Despite being composed in the EME period, the text is characterized by archaisms and constructions inspired by the OE Blickling Homilies, and from Bede's and AElfric's works.
According to Hall (1972), the text of Vices and Virtues was composed in the northern border of the South-East Midlands area, probably from a copy from the Middle or Western South.Archaic forms are used, which indicate a stylistic choice, rather than an OE source exemplar.
The Peterborough Chronicle was copied and composed in the monastery of Peterborough (East Midlands), around 1150.For the analysis, only the sentences composed after 1121 (when the so called "Continuations of Peterborough" begin) are analysed.According to Clark (1970), the syntax of this work is very modern, and displays all the changes which would then affect the ME language.
The Lambeth Homilies are classified in the PPCME2 under the West Midlands dialect, but the decision was made to treat them separately from the other West Midlands texts of the Katherine Group, since a substantial part of the text was copied from OE exemplars (cf.Sisam 1951).Moreover, as seen above, five homilies overlap with the Trinity Homilies, and the other two homilies are a transliteration of two OE homilies by AElfric.The work is dated around 1200 A.D. and Kroch and Taylor (2000) report that the homilies were composed in the same area as the Katherine Group.
Holy Maidenhood, the Life and Passion of Saint Juliana and the Guardianship of the Soul belong to the Katherine Group.The texts are dated between 1200-1225.These texts are written in a literary language, which is supposed to be the "direct descendant of the OE prose".D'Ardenne (1961) describes the language as a "living local speech, with a relatively unbroken spoken tradition from Old English".In a more recent edition of the Katherine Group, Huber and Robertson (2016) report that the works were written in a literary standard, which stems from a language community in which influences from five different languages can be found: English, Scandinavian, French, Welsh and Latin.
Summarizing, the ME sample consists of a selection of texts which cover different dialectal areas; a further categorization of texts can be made, by considering their transmission history.On the one hand, there are texts which were directly composed in the ME period, i.e.The Peterborough Chronicle, the Katherine Group and the Kentish Sermons, the latter being a translation.These texts are mainly narrative, except for Holy Maidenhood, which is a religious treaty.On the other hand, we have texts which are manipulations or copies of older material, i.e. the Kentish Homilies, the Trinity Homilies, the Lambeth Homilies and Vices and Virtues.These are religious texts, where the narration of episodes from the Bible or the Gospel is combined with commentaries given by the authors of the text.
Within the wider scenario of the language change from a linear OV to a VO order in the transition between OE and ME, considering the dialectal provenance of the texts is an important task in ascertaining possible external influences on the English language.In fact, in accounts such as Trips (2002) and Kroch and Taylor (2000), the more innovative character of the texts from the Northern East Midlands area is ascribed to the external influence of the language spoken by the Scandinavian settlers in the Danelaw.For reasons of space, these data cannot be reviewed in full, but it will be argued that dialectal provenance alone cannot be considered as the decisive element responsible for a different mapping of constituents in the texts, since other factors can lead to a more innovative or conservative flavour in the grammar observed.We propose, in fact, that the transmission history of a text may also have bearings on the syntax.
The information structural analysis was conducted by analysing the context in which the resultant sentences are embedded.It was tested whether the IS distinctions of pronouns as A-topics, both shifting and continuing, C-topics or F-topics as defined above, or as (contrastively) focused11 correlate with the licensing conditions governing their spellout.For each sentence, the type of auxiliary, the respective order of auxiliary and verb, the position of the pronoun, the presence of adverbials, the IS value of the pronoun and the type of clause (for the subordinate ones) were annotated.In this way, the aim was to obtain a complete IS, prosodic, and syntactic picture of each sentence.The results of this analysis are presented in the Sections 4.2 and 5.2.
The type of auxiliary was annotated in order to control for different distributions due to the different grammaticalization stages of these verbs (cf.Romèro 2005); in fact, a grammaticalized auxiliary, with reduced inflectional endings, would not represent a good phonological host for the pronouns.Moreover, the type of subordinate clause is annotated, since it is reasonable to assume that a complement clause, which usually has a declarative illocutionary force, would present a distribution more similar to main clauses, i.e. potentially non V-End orders.
The IS categories are determined by taking into account the context surrounding the sentence obtained with the corpus query; reflexive pronouns are not assigned an IS category.To provide two examples for illustration, the label continuing A-Topic is assigned to the pronouns him and he in (12), since the pertinent referent is introduced in the first sentence and continued in the following discourse that revolves around King Egbert. (

Old English
Before going into the distributional and IS analysis, a quantitative overview of the OE sample is given.

Quantitative overview
Table 1 shows that a small percentage (≈10%) of pronouns can appear in post-verbal position already in the OE period (see also De Bastiani 2019). 12The fact that a small percentage of post-verbal pronouns can be found in OE signals that the changes we notice in the EME period had already started during the OE period.Interestingly, the majority of the OE texts come from the West Saxon dialect, which was the least affected by contact with the Scandinavian settlers.If the tendencies noted in the OE period show a linear continuation in the subsequent period, then the influence of the Scandinavian language on the English syntax should be reconsidered.

Qualitative analysis 1314
In Section 4.2, the aim is to determine whether the mapping of pronouns can be motivated by IS and prosodic constraints on their spell-out. 15In particular, the aim is to investigate whether (contrastive) focus can motivate the post-verbal mapping of pronouns, in line with the Focus Transparency condition, and whether their status as shifting aboutness or contrastive topics can motivate their mapping into the left periphery.The syntactic constraints we take into account include the type of auxiliary, the presence of an adverbial and the type of subordinate clause.They will be dealt with separately in Section 4.2.1.

Distributional analysis
In order to obtain a complete picture of the descriptive distribution of pronouns, it was investigated whether the mapping of pronouns correlates with the type of auxiliary found in the sentence.The auxiliaries found are beon 'to be', weorðan 'to become' (used to form the passive), habban 'to have', onginnan 'to begin', magan 'may' and gewitan 'to know'.The most commonly used are beon, weorðan, habban, and onginnan.The only ones that show an asymmetry are beon and weorðan, which are used mostly in passive constructions.The pronoun is never found in post-verbal position in main or embedded clauses, when the auxiliaries are beon or weorðan.Often the subject is post-verbal, as is illustrated in ( 14): (14) (coaelhom,AEHom_4:1.516) him waes sođlice benaemed his gesihđ and spraec.him.datwas truly deprived his sight.nomand speech.nom'In truth, he was deprived of his sight and faculty of speaking.' 13 Each sentence has one pronoun; if a sentence has two pronouns, it is counted twice. 14Since the inflected verb is found at the end of the clause, it is of course not possible to determine whether the pronoun has been scrambled in the left periphery, so the label pre-T was not given in the V -Aux sentences. 15The observations given in Section 4.2 refer to both Aux -V and V -Aux clauses.As can be seen in Table 1, however, V -Aux clauses do not present any post-verbal pronouns, so if differences are observed, these apply to Aux -V sentences only.In sentence ( 14), a new referent is introduced in the previous line and the following discourse then revolves around him.The pronoun anchors the topic in the narration and can be considered as a continuing A-topic, whereas the new element, which is the grammatical subject of the clause, is mapped onto the post-verbal position.Moreover, it was investigated whether other structural requirements demand their postverbal mapping.In this framework, light elements are predicted to be spelled out per default in pre-verbal position, unless IS conditions apply and require their spell out in a post-verbal position.According to this framework, the expected spell-out of a (contrastively) focused pronoun is in a post-verbal position.
Leaving aside IS conditions, which will be analysed in the next sub-section, we investigate whether other conditions can motivate the post-verbal mapping of pronouns.Among the OE main clauses, it is found that 11/101 pronouns are mapped into post-verbal position.Of these, 10 cases occur with a second argument of the verb in post-verbal position, 16an example is the following: The verb biddan requires two arguments: the object of the request and the addressee of the request, which are both found in post-verbal position.Furthermore, among the post-verbal pronouns followed by a second argument of the verb, two are followed by the reflexive reinforcer sylfe 'self'.Only in one instance is there a post-verbal pronoun which is not followed by a heavier second argument of the verb, but it is only followed by sylfe 'self', indicating a focus interpretation.According to Mitchell (1985), the pronoun self was not fundamental in OE to convey the reflexive reading of the pronoun but was rather used as a (emphatic) reinforcer.
(16) (cowulf,WHom_6:48.273)And sona swa þaet waes þaet hi swa gedon haefdon, þa haefdon And soon so comp was comp they.nomso done Had then had hy forworhte hy sylfe.they.nomruined t hem.acc selves.acc'And as soon that they had done so, they had ruined themselves' In the subordinate clauses, there are two instances in which the pronoun is mapped in post-verbal position.In both cases the pronoun is followed by the second argument of the verb.
So far, it has been shown that in OE post-verbal pronouns occupy the post-verbal field with a second argument of the verb.However, this piece of evidence alone cannot motivate their post-verbal mapping, since the reverse is not true.In other words, two arguments of the same verb do not have to be mapped together in post-verbal position.Among the 50 pre-verbal pronouns in main clauses only seven depend on a verb taking two objects.In four of these seven cases also the second argument is mapped in pre-verbal position.In the three remaining cases, the second argument, which is mapped in post-verbal position, is a complement clause (two instances), or a coordination of PPs.Given this variation, the presence of a second argument of the verb in post-verbal position cannot be analysed as the necessary trigger for the post-verbal spell-out of the pronoun.
Finally, the type of subordinate clause was also annotated.The most frequent ones are the adverbial clauses, but also complement and relative clauses as well as indirect questions occur.All the different types of subordinate clauses display the three different mappings of pronouns.Concluding this subsection, it has been shown that some structural properties common to post-verbal pronouns can be identified.
Analysing the pre-verbal position as the default unmarked one, then both the position of the pronoun in the left periphery as well as the post-verbal position are to be analysed as derived and driven by some constraint.The reason for analysing the pre-verbal spell-out as the unmarked option, is determined by the theoretical framework which follows from the empirical observation that light elements surface in pre-verbal position in Early Germanic languages (cf. Behaghel 1932;Hróarsdóttir 2000;Hinterhölzl 2017).The mapping of a given element in a position higher than the vP shell is analysed as scrambling.The next subsection will analyse whether the position in the left periphery can be motivated by information structural properties.In particular, the aim is to investigate whether these pronouns can be analysed as A-and C-topics.Furthermore, a more detailed investigation of information structural and prosodic cues for the post-verbal mapping of pronouns will be carried out.

Information structural analysis
The information structural properties of pronominal constituents in their syntactic and discourse environment were investigated.In Tables 2 and 3, the quantitative distribution of type of pronouns with respect to their syntactic position in Aux-V in main and subordinate clauses is given: 17  Of the 40 pronouns in the left periphery of the Aux -V main clauses, for instance, 39 denote either an A-topic or a C-topics.The only case that has not been labelled as topic is a contrastively focused element.Below, example (12), repeated as (17) provides an example of A-topic, followed by an example of a C-topic and by the example of the contrastively focused pronoun in the left periphery. 17These tables report the type of topics which are found in the syntactic position given in the left column.
For instance, in the first line of the table, we find the pre-T pronouns, which amount to 40.Of these, 24 constitute A topics.In example ( 17), the entire year entry revolves around King Egbert.After his introduction, topic continuity is achieved by the accusative pronoun hyne in pre-T position, whereas the new information of the sentence (namely that it was Offa and Bertric who had banished him) is found after the finite verb.The remaining part of the year entry still revolves around this king, which is referred to with a demonstrative pronoun in the subsequent lines.This referent is labelled as a continuing A-Topic, since it is the referent about which the entire passage is constructed, ruling out an analysis as F-topic, since it is not merely an active element in discourse (cf.also example 21).
(18) (cochronA-2a,ChronA_[Plummer]:894.55.1060) and he him haefde geseald aþas and gislas, and he.nom him.dathad given oaths.accand hostages.acc7 se cyng him eac wel feoh sealde.and det.nom.king.nomhim.datalso well reward.accgave 'and he (Haesten) had given him (the king) oaths and hostages, and the king gave him (Haesten) also a bountiful reward.' In example (18), both the referent of the subject pronoun, Haesten, and the referent of the object pronoun, the king, are given in the context under examination.These two pronouns are analysed as C-topics.In fact, they correspond to the given alternatives in the proposition under consideration, but they do not have an impact on the focus of the sentence constituted by the direct object after the non-finite verb.In the sentence following the one under scrutiny, the subject and the indirect object roles are reversed.The referents of the subject and the object pronouns are set as the alternatives in the utterance.The contrastively focused pronoun in the left periphery is given in (19):18 (19) (coaelhom,AEHom_9:158.1371)and us he waes geswutelod; na eallum folce, ac we and us.dathe.nom was manifested not all.datpeople.datbut we.nom đe aeton mid him.rel ate with him.dat'And it was to us that he was manifested, not to all the people, but to us, who ate with him.' The object pronoun in pre-T position refers to the disciples of Christ.The sentence informs that Christ decided to reveal himself after his death only to them and not to all people.In this example, unlike in the previous one, the object pronoun is not presented as one of the alternatives active in discourse in opposition to other topics but is contrasted to the nonspecific indirect object after the non-finite verb.In this case, it can be hypothesized that the contrastively focused pronoun is hosted in FocP in the left periphery, as Petrova and Speyer (2011) demonstrate that contrastive focus can be realised in the OE left periphery.It is moreover interesting to note that, when object pronouns constituting C-topics are found in the position preceding the finite verb, these always follow the subject pronoun.
In the case at hand, however, the contrastively focused object pronoun precedes the subject pronoun, which constitutes a continuing A-topic.
Also all the pronouns in the left periphery of subclauses are to be analysed as continuing A-or C-topics.The assumption that the position in the left periphery is a derived position is supported by the examples in which the pronoun has been extracted from a PP and moved higher up in the clause (there are four such examples in the records for the main clauses): 19(20) 19 (cochronC,ChronC_ [Rositzke]:905.1.8.1033)  and vii aerendracan he him haefde to asend and seven messengers.acc he.nom him.dathad to sent 'And he had sent to him seven messengers.'So far, the predictions are met.Pronouns in the left periphery signal continuing A-and C-topics and the sentences hosting more than one pre-finite pronoun can be analysed as Topic > Focus structures.
Let us now turn to pre-verbal pronouns.Since it has been predicted that the pre-verbal mapping of pronouns is default, it would be expected that the pronouns in this position only refer to entities which are already activated, but which are neither contrasted nor constitute the A-topic of the passage.In the majority of the main clauses, this prediction is met.& on Sancte Martines maessedaeg heo weard him mid and on saint.genmartin.gen.mass-day she.nom.was him.datwith mycelan weorðscipe forgifen on Westmynstre.great dignity given on westminster.'And on Saint Martin's day she was given him with great dignity at Westminster.' In the sentence preceding this passage, the princess is introduced and becomes the A-topic of the subsequent passage consisting of sentences in which new information is added about this referent.The pronoun him refers to an activated referent, which unlike the object pronouns found in the left periphery is not the A-topic of the passage, but is part of the comment being predicated about the topic.
However, among the pre-verbal object pronouns, there are 12 in main clauses that are coreferential with the continuing A-topic of the sentence but are not moved to the left periphery of the clause.Moreover, there are 3 cases among the pre-verbal object pronouns in the main clauses constituting cases of C-topics, whereas one is contrastively focused.Finally, 12 of these pronouns are reflexive, and are thus not analysed for their IS value.In subordinate clauses, two pre-verbal object pronouns were found that are contrastively focused, as Table 2 shows.
De Bastiani and Hinterhölzl: On the syntax of object pronouns in Old English and Early Middle English Art.43, page 15 of 29 Summarizing, so far, it has been seen that pronouns in the left periphery clearly correlate with A-and C-topics and that post-verbal pronouns are typically followed by the second argument of the verb or a reinforcer.The pre-verbal position is the unmarked position in the majority of cases, even though continuing A-topics and cases of contrastive focus are not excluded.
As Pintzuk (1999)  The pronoun -even though it does not denote the aboutness topic -refers to a referent that is highly active in the context.The active referent is separated from the adverbial that modifies the VP.These sentences are compatible with an interpretation in which the VP is focussed, whereas the pronoun is part of the background of the sentence.These are opposed to other sentences in which the pronoun is mapped after an adverbial at the left edge of vP: (23) (coaelive,AELS_[Forty_Soldiers]:55.2504)þa began se dema eft hi herigan.then began det.nom judge.nomagain them.accinterrogate 'Then the judge began to question them again.
In this example, the forty Christian soldiers are prisoners of the heathen captain.They are brought before the heathen judge who begins to interrogate them again.The reference of the soldiers is highly active and is analysed as an F-Topic.However, the action described constitutes a new passage and hence new information in the narration.For this reason, the focus domain can be labelled as comprising the post-T domain.In conclusion, four positions can be identified in which object pronouns appear in the OE sample, as specified in (24). (24) Pron > aux > pron > adv > pron > V > pron According to the empirical investigation conducted, the position at the immediate left of the verb shows a mixed composition, but F-Topics comprise the majority of the referents found, whereas the position in the left periphery and at the left of an adverbial are due to more specific information structural requirements.
Let us now turn to post-verbal pronouns.The sentences with a post-verbal pronoun turn out to have one property in common.It was investigated whether the denotation of the verb can be labelled as given or inferred in the context under scrutiny. 21here are eleven sentences in the sample of main clauses, and two sentences in the sample of embedded clauses presenting a post-verbal pronoun.Ten out of the eleven main clauses under scrutiny present a second argument of the verb in post-verbal position, the same is valid also for the subordinate clauses (but we have already determined in Section 4.2.1 that this factor cannot be decisive in governing the post-verbal spell-out of constituents).Further structural features that these main clauses share is the presence of the temporal adverbial þa 'then' and the semi-modal onginnan 'to begin' (these features are shared by eight out of the eleven sentences under examination).
However, the denotation of the verb in these main clauses can be argued to be inferable or given, as in example ( 16), repeated here as (25).
(25) (cowulf,WHom_6:48.273)And sona swa þaet waes þaet hi swa gedon haefdon, þa haefdon and soon so comp was comp they.nomso done had then had hy forworhte hy sylfe.they.nomruined them.accselves.acc'And as soon that they had done so, they had ruined themselves' In this passage, the author explains how God created Adam and Eve, and it is stated that they had to save their kin in heaven, which the devil had ruined (forwyrcan) through his pride.After that, the author describes the temptation of Eve and Adam.This sentence is embedded in this context and it is said that, after Adam and Eve had done as the devil advised them, they had ruined themselves.The meaning of the verb is given at this point in the narration.The pronoun is not contrasted, but the reinforcing element (sylfe) may point at an emphatic interpretation (with their own hands), explaining the use of a strong pronoun in post-verbal position.In the remaining cases, however, a (contrastive) focus interpretation is not feasible.
Among the eleven post-verbal pronouns in the main clauses the meaning of the verb is given or inferable in eight cases, whereas the prediction is not met for three remaining examples and for the two post-verbal pronouns in the subordinate clauses.What is the impact of the second post-verbal element?It allows the pronoun to occupy a weak branch in prosodic structure, as required by the interface condition.This position counts as stronger than the position of the verb, however.We may surmise that this is exceptionally allowed if the verb is itself given, otherwise the verb has to be mapped onto a stronger position, that is, to the right of the pronoun.

Early Middle English
As described in Section 3, whereas the OE sample was not meant to be exhaustive, the sample we collected for EME aims at giving a comprehensive account of the different dialectal areas.In the transition from OE to EME, case declensions underwent significant levelling of forms so that accusative and dative cases were fused together into one oblique case (cf.Fisiak 1968, Roseborough 1970, Wright and Wright 1967).This, it can be assumed, has significant consequences for the analysis of pronouns as special clitics targeting a specific head high up in the clausal structure.Loss of case reduces the possibility of scrambling a DP to a high position in the middle field -potentially crossing another argument, from which it can be taken to undergo local head movement to adjoin to its licensing head.
In the texts examined, it was also checked whether remains of the OE verbal inflection can be detected.In the Kentish Homilies, in the Trinity Homilies, in Vices and Virtues and in the Lambeth Homilies, the OE infinitive ending -an and some inflected forms of the participle can still be found.In the Lambeth Homilies, the OE ending -an is recorded on 26 verbs pertaining to the sample.In the texts of the South East Midlands, there are some inflected participles (a total of only four cases among more than 100 sentences).In the texts of the Peterborough Chronicle and in the texts of the Katherine Group, there are no inflected participles nor the OE ending -an on infinitives.This could indicate that the prosodic word in ME was reduced to a maximal template of a weak and strong syllable, excluding affixal clitics in proclitic or enclitic position.
Since one of the goals of this study is to relate the growing post-verbal spell-out of object pronouns to the ongoing change from an OV to a VO surface word order, a brief overview of the spell-out of non-pronominal objects in the sample under scrutiny will be given.In this framework, it is predicted that given DP objects are spelled-out on a weak branch in prosodic structure, i.e. in pre-verbal position, whereas new and/or heavy objects are predicted to spell-out in the post-verbal position, due to the F-Transparency condition or due to the Heaviness condition given above.
During OE, the determiners se, seo and þaet, which would later grammaticalize into the invariant definite determiner þe, constitute an ambiguous category: they display typical demonstrative uses, but they can also be used to signal definiteness on a noun (cf.Crisma 2011;Breban 2012;Allen 2016;De Bastiani 2019).
In the transition to EME, these determiners evolved into the invariant definite determiner, which had become a functional element realised in the head of the DP (cf.Breban 2012; Allen 2016).Recall from the definition of prosodic heaviness above that a heavy phrase is defined as a phrase having both its head and its complement filled with lexical material.From this definition, it follows that a DP with a definite determiner is analysed as heavy and the framework predicts that these heavy but definite constituents will be mapped onto a post-verbal position.The rising post-verbal mapping of definite DPs leads to a mixed system, in which all types of non-pronominal objects are spelled-out in post-verbal position, whereas pronouns are still spelled-out either in pre-verbal or in pre-T position.Struik andvan Kemenade (2018) andDe Bastiani (2019) show that the predictions are met: in OE, the givenness of non-pronominal object prompts their mapping into a preverbal position, even in cases where the object is branching.However, it can be noticed that an increasing post-verbal spell-out of non-pronominal objects due to their heaviness is already at work.It is submitted that this correlates with the deictic and non-deictic use of the determiner in the individual text and its consequential analysis as a specifier or head of the nominal constituent.
In the EME texts under examination, definite DPs are predominantly spelled-out in postverbal position, even though some residual pre-verbal DPs can be found in the more conservative texts.In the texts of the Katherine Group, the Kentish Sermons and the Peterborough Chronicle, on the other hand, the only residual elements that are spelledout in the pre-verbal position are object pronouns or non-branching objects. 22 In this scenario, it is interesting to compare the texts from the Katherine Group with the older texts coming from the same area, namely with the Lambeth Homilies.In fact, when the non-pronominal objects are considered, it can be noted that their distribution is very similar.Moreover, most definite DPs are spelled-out in post-verbal position in the Lambeth Homilies.What differentiates the texts is the distribution of pronouns in the post-verbal position; in fact, the Lambeth Homilies -which, as stated in Section 3, mostly consist of elaborations of older material -show a very low rate of post-verbal pronouns, in a similar way to the OE dataset.The Katherine Group, which consists of original ME compositions shows a much higher number of post-verbal pronouns.
These observations clearly indicate that the change in DP object placement paves the way for object pronoun placement.Here the effect of the prosodic condition in (4) can be seen at work: when the grammaticalization of the definite determiner is completed and the latter is reanalysed as the head of DP the resultant phrase counts as a heavy right-branching element and is spelled-out in the post-verbal domain following (4).For given object DPs this is congruent with the IS-conditions in (3), only if the given DP object is de-accented in its post-verbal base position.This in turn implies that intrinsic heading and de-accentuation in situ becomes a grammatical option that is subsequently used by pronouns as well.
De Bastiani and Hinterhölzl: On the syntax of object pronouns in Old English and Early Middle English Art.43, page 18 of 29

Quantitative overview
In Table 4, a quantitative overview for the whole EME sample is given (see also De Bastiani 2019).Looking at the entire set of EME data, it can only be observed that the percentage of post-verbal pronouns has risen -from 10.9% to 29.8% -and that the EME main clauses present a slightly higher percentage of pronouns in the left periphery, while the ratio of pronouns in the pre-verbal position has lowered with respect to the OE period, as can be seen from the comparison in Table 5.The percentage points of the subordinate clauses exhibit both a higher ratio of pre-T pronouns and post-verbal pronouns. 2324 Since the EME texts span from the years 1150 to 1275 and cover different dialectal areas and genres, each dialectal area will be treated separately in what follows.It will be seen that the percentage of post-verbal pronouns differs significantly in the different dialectal areas. 25

Kentish
Since the Kentish Homilies and the Kentish Sermons were composed within a time span of ca. 100 years, it is appropriate to treat them separately, even though the number of resultant sentences for each text is not high.The distribution of pronouns in the Kentish Texts is reported in Table 6.
Given the size of the texts, it is not surprising that such a low record is obtained.The raw data for these texts show a marked preference for the mapping of pronouns in the left periphery in both main and subclauses.Remarkably, the Kentish Homilies present no post-verbal pronouns.Recall that this text is described as the "earliest Middle English document" and that pronouns in OE texts are rarely attested after the non-finite verb, according to Fischer et al. (2000).

South East Midlands
The percentages in Table 7 show a continuity with OE, with the majority of the pronouns distributed in the derived position in the left periphery, even though post-verbal pronouns in main clauses have risen to 22.9% vs. the 10.9% in the OE sample.Given the archaic style of these works, it is not surprising that the composition of sentences is not dramatically different from the OE record, however, at the same time an ongoing progressively higher post-verbal spell-out of pronouns can be noticed. 25Since V -Aux clauses only display pre-verbal pronouns, in the following only Aux -V clauses are treated.

Lambeth Homilies
The distribution is very similar to OE, with almost half of the pronouns in the derived position in the left periphery, and with around 10% pronouns appearing in post-verbal position as can be observed in Table 8.

The Peterborough Chronicle
In this text, it can be noticed that more than half of the pronouns found in the subordinate clauses are mapped onto post-verbal position as Table 9 shows.A clear difference from the OE grammar can also be noticed in the next group of texts.

The Katherine Group
Table 10 shows that this group of texts presents a different grammar with respect to the OE sample, and with respect to the Lambeth Homilies, Kentish Homilies and the texts

Subordinate clauses 9
Pre-T pron 1 Pre-V pron 3 Post-V pron 5 Table 10: Object pronouns in the Katherine Group.

Subordinate clauses 32
Pre-T pron 9; 28.1% Pre-V pron 9; 28.1% Post-V pron 14; 43.8% of the South East Midlands.Given that the language has been described as genuinely reflecting the language of the higher classes of the time, the finding is qualitatively significant.Concluding this quantitative investigation, the EME sample shows different patterns and distributions, once the results are divided into the different texts and dialectal areas.It is true that the texts belonging to the South Eastern area (with the inclusion of the Lambeth Homilies) show results more similar to the OE sample, with respect to the Peterborough Chronicle (Northern East Midlands) and the Katherine Group (West Midlands), where there is a break with respect to the OE period.However, if the philological considerations reported in the editions are correct, it is thought that the different ratio of post-verbal spell-out of pronouns in the data follows from the transmission history of the works where the more conservative texts adhere to the common practice of copying from OE material, or deliberately using OE archaic forms.
The two types of texts that show a break with the OE syntax are the Peterborough Chronicle and the texts of the Katherine Group.As stated in Section 3, both the Peterborough Chronicle and the Katherine Group are original ME compositions.These texts show a diachronic development with respect to OE.Moreover, it is interesting to note that what distinguishes the text of the Lambeth Homilies from the Katherine Group -which, as stated in Section 3, were composed in the same area -is the mapping of pronouns, while they exhibit similar ratios of post-verbal non-pronominal objects (cf.Kroch and Taylor 2000;De Bastiani 2019).

Qualitative analysis
In this section, we consider the same structural and information structural features we have applied to the OE sample in order to determine whether there are qualitative differences in the data.

Distributional analysis
An examination was made of the distribution of pronouns with respect to the type of auxiliary used.Subordinate clauses with the auxiliary beon 'to be' do not show the postverbal mapping of pronouns, while main clauses with beon show all three distributions.The auxiliaries used most often are beon 'to be ', habban 'to have', sculan 'shall', magan 'may',  and willan 'will'. 26Similarly for the findings in OE, there is no correlation between the auxiliary used and the mapping of pronouns.
It was also checked whether the post-verbal pronouns are followed by a second argument of the verb.Among the whole EME sample of main and subordinate clauses, this is observed in 56 cases, whereas in 29 further cases the object pronoun is either followed by an adjunct, a complement clause, or an adverbial.Finally, there are 17 main sentences presenting an object pronoun as the only element in the post-verbal domain, the majority of which come from the Katherine Group.There are eight subordinate clauses presenting a post-verbal pronoun as the only element in the post-verbal domain.The only texts not presenting a pronoun as the only element in the post-verbal domain are the Kentish Homilies, which present no post-verbal pronouns, and the Lambeth Homilies, in which the post-verbal pronouns are followed by the second argument of the verb, or a heavier adjunct.
As was observed also for the OE sample, it is not the case that all verbs requiring two arguments map their pronouns into the post-verbal domain.Among the 141 preverbal pronouns in main and subordinate clauses, 35 are governed by a verb requiring two arguments.When looking at the proportion of post-verbal pronouns followed by a presence of a strong accented element in clause-final position.It has been seen that in these cases the verb is typically given which exempts it from being necessarily mapped onto a stronger branch than the pronoun is -licensing the spell-out of the pronoun in its base position.
We also investigated whether other structural constraints can be identified.However, the position of pronouns does not seem to be correlated to the type of auxiliary or modal verb used, nor to the type of subordinate clause.Finally, while it is true that post-verbal pronouns in the OE sample co-occur with the second argument of the verb, the reverse, namely that all ditransitive verbs involving one object pronoun realize it in post-verbal position is not true.Verbal structures with two arguments were found, only the second of which is mapped onto a post-verbal position.The post-verbal element in these cases is right-branching and can thus be considered to be heavy.
In Section 4, the mapping of pronouns in the EME period is analysed.As in the OE data, the position of pronouns does not correlate with the type of auxiliary or subordinate clause, nor with the argument structure of the verb.From a quantitative point of view, the data as a whole show an increase in the number of post-verbal pronouns, and a decrease in the number of pre-verbal pronouns, but when the results are divided according to specific dialectal areas, more striking qualitative and quantitative differences arise.
The EME sample shows differences that can be ascribed to the dialect as well as to the transmission history of the text and the frequent practice of copying or adapting older material.The texts that are analysed as genuinely representing the EME language, however, show that, whereas the position in the left periphery continues to host topicalized material (even though with a lower frequency), the pre-verbal position has undergone a reduction in the proportion of weak pronouns, which are predominantly mapped into a post-verbal position.
Closer attention has been given to those sentences presenting only one pronoun in the post-verbal domain and it emerges that most of them simply represent F-topics, while only a few present a (contrastively) focused pronoun.It has been seen that the texts presenting more unmarked pronouns in the post-verbal domain are the texts from the Katherine Group and the Peterborough Chronicle, which are not adaptations of older material, and which do not present traces of older inflection on the non-finite verb.
In conclusion, it is precisely these texts in which the beginning of the change in pronoun placement in the history of English can be observed.This is a change that will lead to a reanalysis of the post-verbal object position as an information-structurally unmarked position, a scenario for which an account will be sketched in Section 7.
From the analysis presented so far, it emerges, moreover, that the progressive unmarked spell-out of object pronouns in post-verbal position is related to an autochthonous language change scenario, starting in the OE period and progressively extending over time.The inclusion of the transmission history of the texts in the analysis shows that the hypothesis of the language contact with the Scandinavian settlers should be reconsidered, in view of the fact that it is not surprising that elaborations of older material show a more conservative grammar.

Diachronic scenario
The spell-out of pronouns constituting unmarked F-topics in the post-verbal position in these texts is made possible by the option of de-accentuation in situ that is brought about by the placement of given DPs.At the point at which the grammaticalization of the definite determiner is complete, a given DP argument has to be spelled-out on a strong branch or in a domain that is not weight-sensitive, like the v-domain, leading to the adoption of the more costly option of de-accenting in situ.While de-accenting in situ is necessary with DP objects, it becomes an option for pronominal objects.The unmarked procedure of extrinsic heading will map them into the preverbal domain, the initially more marked option of intrinsic heading allows them to be spelled-out post-verbally.
What eventually drove speakers to prefer the spell-out of pronouns in the base position?The answer is that de-accenting in-situ is prone to become the unmarked option in a language in which information-structural distinctions like given and new are only signalled prosodically.At the point at which both given and focused objects DPs are placed postverbally, their IS-status -which was formally signalled via syntactic position -can only be signalled via metrical prominence in the same position, making stress displacement and accenting and de-accenting in situ a default option.This arguably constitutes a case of change from a language with flexible word order (to signal IS-distinctions) and fixed accent to a language with flexible accent and fixed word order.
It is submitted that the grammatical basis of flexible accent is that the head of prosodic constituents is determined by intrinsic properties.If intrinsic heading as an unmarked option is in place, post-verbal placement of object pronouns becomes the most economic option according to the general interface condition proposed in Hinterhölzl (2014) given in (32). (32) A constituent is spelled-out in the smallest domain in which its PF-properties are satisfied in an unmarked fashion When new and given DP objects as well as strong and weak pronominal objects are spelledout in the post-verbal domain, the original mapping in terms of prosodic weight and information structural value (given -new) is lost and all objects will start to be placed in a post-verbal position, unless they contain a semantic feature like scope or negation that will require their spell-out in the I-domain.This holds for quantified and negative objects, which continue to be placed in the pre-verbal domain till the end of the ME-period.
movement of arguments into a Case phrase 4 b.Licensing movement of verb particles into the specifier of a low Asp phrase c.Licensing movements of predicative elements into a Predicative phrase (3) a. G(ivenness)-Transparency: A given constituent must occupy a weak position in prosodic structure.b.F(ocus)-Transparency: A constituent representing new information must occupy a strong position in prosodic structure.
In this year king Egbert died, and Offa king of the Mercians and Bertric king of the West Saxons had banished him for three years from England to France, before he became king.'Asstatedabove,C-Topics are those topics that belong to an oppositional pairs that does not have an impact on the focus domain of the clause.In (13), the pronouns ich and him constitute such topics with the expressions his iuel and mid werse and the PP mid werse representing the respective focus values of the contrastive relation.

Table 1 :
The distribution of object pronouns in the OE sample.
Then the recruited workers immediately started to ask them for food.'

Table 2 :
Distribution of topic types in OE main clauses.

Table 3 :
Distribution of topic types in OE subordinate clauses.In this year king Egbert died, and Offa king of the Mercians and Bertric king of the West Saxons had banished him for three years from England to France, before he became king.' already noticed, there is another derived position in the Old English middle field, namely a scrambled position at the left of vP adverbs, but under TP: 20

Table 4 :
The distribution of object pronouns in the EME sample.

Table 5 :
Comparison between the OE and the EME Aux -V clauses.

Table 6 :
Object pronouns in the Kentish texts.

Table 7 :
Object pronouns in the South-East Midlands texts.

Table 8 :
Object pronouns in the Lambeth Homilies.

Table 9 :
Object pronouns in the Peterborough Chronicle.