Sergio López Martínez, Selim 21 (2015–2016): 167–178. ISSN 1132-631X Embedded topicalisation in Old English: Does it exist? Sergio López Martínez University of Oviedo, University of Kent The study of the left periphery of subordinate structures in Old English is not an exhaustively explored field, and it has often been inaccurately described by grammars and manuals of this language. It has been almost unanimously admitted that subordination and topicalisation are mutually exclusive in Old English. Only a few authors admit that topicalisation and subordination can coexist in Old English, and not a single systematic study of this issue has ever been provided. Therefore, the main objective of this paper is to present accurate statistical data about the combinatiorial potential and distribution of preverbal arguments in Old English, highlighting the existence of structures that were thought to be forbidden in Old English syntax. Not only pronominal elements were found among those topicalised structures in subordinate sentences, but also syntactically complex and highly informative arguments. Thus, it is demonstrated that topicalisation is possible in Old English subordination. Keywords: Old English; syntax; topicalisation; subordination; information structure 1. Introduction The left periphery of subordinate structures in Old English (OE) has often been inaccurately described in grammars and manuals. Thus, it has been generally admitted that subordination and topicalisation are mutually exclusive in OE. In most accounts, topicalisation is simply banned from subordination (Pintzuck 1991: 69), and in the few cases where it is considered acceptable (Allen 1980: 52) no quantitative data are presented. The main objective of this paper is to provide statistical data about the combinatorial possibilities and distribution of preverbal arguments in OE, focusing especially on topicalised embedded objects, and highlighting the 168 Sergio López Martínez existence of structures that were thought to be forbidden in OE syntax. A large corpus of prose texts from the OE period has been analysed, and the data have been mapped against linguistically relevant variables, such as type of topic (pronominal or fully nominal), type of subordinate clause, syntactic position of the elements following the topic (with special emphasis on V2-ing), and overall informational load of the elements involved in embedded topicalisation. 2. Embedded topicalisation in other Germanic languages Before analysing the question of embedded topicalisation in OE, it is necessary to consider the state of the matter in other Germanic languages. In Present- Day English (PDE), topicalisation is possible in main clauses (1b), while it is not acceptable in subordinate ones (1d): (1) a. I don’t drink wine. b. Wine, I don’t drink. c. He said that he doesn’t drink wine. d. *He said that wine he doesn’t drink. According to Haider (2010: 141), word order in modern German is more flexible than in other Germanic OV languages like Dutch. In order to get some insight concerning the possibility of embedded topicalisation in German, an informal survey with native speakers of German was carried out. Informants were asked to answer a grammaticality judgment test about the following three subordinate sentences: (2) a. …, dass der Arzt[NOM] den Patienten[ACC] besuchte. (SOV) b. …, dass den Arzt [ACC] der Patient [NOM] besuchte. (OSV) c. …, dass den Arzt [ACC] besuchte der Patient [NOM] . (OVS) As seen in (2a–c) above, informants where asked to judge the grammaticality of a subordinate sentence with canonical SOV order, another one with a topicalised object (OSV order), and a third one with a topicalised object and inversion (OVS order). Results are shown in Table 1. Embedded topicalisation in Old English 169 Table 1. Informal survey with native speakers of German: grammaticality judgement. ✓ ✓? X a 28 -- -- b 4 10 14 c -- -- 28 As we can observe, sentence (2a), with canonical SOV order, was considered fully grammatical by the totality of informants. Only four of the informants considered that a subordinate sentence with a topicalised object (2b) was grammatical, while ten judged it to be grammatical but only in certain colloquial contexts. The majority of informants considered it to be ungrammatical. Finally, all the informants found topicalisation and inversion (2c) ungrammatical. Thus, we can conclude that, while object topicalisation in subordination may be acceptable in German, topicalisation with verb inversion is disfavoured in this language. 3. Embedded topicalisation in Old English It is now necessary to consider if embedded topicalisation in OE is impossible (as in PDE), acceptable only in limited contexts (as in German), or whether it is a phenomenon that could occur naturally. Furthermore, the present section will shortly analyse the implications of the existence of embedded topicalisation for other processes of word order change in the history of English. 3.1. Implications of embedded topicalisation in OE According to Stockwell & Minkova (1991), subordination and inversion cannot occur at the same time in OE, since the complementiser (COMP) blocks the raising of the verb to Inflection (INFL) position. This is due to the fact that both INFL and COMP occupy the same node, as shown in Figure 1. 170 Sergio López Martínez Figure 1. Syntactic representation of subordinate structures in OE according to Stockwell & Minkova (1991) According to this theory, VP is a dense and cohesive package, from which nothing can be taken out. After the OV internal operation leaves the verb in the absolute left periphery of the VP, in an adjacent position to the subject which precedes it, providing robust SV input which will serve as the basis for the subsequent reanalysis XV main clauses (accomplished Figure 2. Syntactic representation of Stockwell & Minkova (1991). Under this perspective, the OV clauses around 1400 would be motivated by analogy from the structure of subordinate sentences. However, this theory is based on the fact that no Sergio López Martínez . Syntactic representation of subordinate structures in OE according to Minkova (1991). According to this theory, VP is a dense and cohesive package, from which nothing can be taken out. After the OV � VO reanalysis (c. 1200), this internal operation leaves the verb in the absolute left periphery of the VP, in sition to the subject which precedes it, providing robust SV input which will serve as the basis for the subsequent reanalysis XV � main clauses (accomplished c. 1400) (see Figure 2). . Syntactic representation of the OV � VO Reanalysis OE according to Stockwell & Minkova (1991). Under this perspective, the OV � VO reanalysis which took place in main clauses around 1400 would be motivated by analogy from the structure of subordinate sentences. However, this theory is based on the fact that no . Syntactic representation of subordinate structures in OE according to According to this theory, VP is a dense and cohesive package, from which 1200), this internal operation leaves the verb in the absolute left periphery of the VP, in sition to the subject which precedes it, providing robust SV � SV in VO Reanalysis OE according to VO reanalysis which took place in main clauses around 1400 would be motivated by analogy from the structure of subordinate sentences. However, this theory is based on the fact that no Embedded topicalisation in Old English 171 element can be extracted from VP, and it would not work if topicalisation and inversion were possible in OE subordinate structures. 3.2. State of the matter It has been generally admitted that topicalisation and subordination are mutually exclusive in OE. For instance, Pintzuck (1991: 69) categorically affirms that topicalisation is not possible in subordinate clauses, a view which has been widely accepted. Stockwell & Minkova (1991: 384–385) state that verb fronting and topicalisation rules cannot apply in subordinate clauses in OE, while Kroch & Taylor (1997: 309) consider that “topicalisation has a very weak discourse motivation” in subordinate sentences, and that “underlyingly I- final clauses” are not expected to exhibit V2 order. Other authors like Haeberli (2001: 213–214) simply acknowledge the need for more work on word order in OE subordinate clauses. Only Cynthia Allen recognises that “topicalisation could also take place within a subordinate clause” (1980: 52), and that “both OSV and OVS order 1 are found in embedded clauses” (1995: 46). However, she does not provide any kind of quantitative data which illustrate this phenomenon in order to support her claim. This lack of data has been the main motivation for the present study. 4. Data In order to carry out an accurate collection of data, a selection of texts from the York-Toronto-Helsinki Parsed Corpus of Old English Prose (2003) were analysed using Corpus Search 2. Since style, genre and text-type are very relevant variables for the present study, texts were selected from various and different genres and types. Thus, the selection includes the following texts: � Narrative/descriptive: The Anglo-Saxon Chronicle (A & E) (Chr. A/E), Orosius (Or.) � Narrative: Bede (Bed.), Ælfric’s Lives of Saints (Liv.), Ælfric’s Old Testament (O.T.) � Argumentative: ‘Preface’ Cura Pastoralis (C.P.), Boethius (Bo.) 1 That is, both topicalisation and topicalisation with inversion. 172 Sergio López Martínez � Technical: Herbarium (Her.), Medicina de Quadrupedibus (Med.) The nature of the texts as original OE ones or as Latin translations is also a relevant factor that must be taken in consideration. Namely, while The Anglo- Saxon Chronicle and the preface of the Cura Pastoralis are texts written originally in Old English, the rest of the works present in this selection have a Latin source. This could also have been a significant factor in the inclusion of topicalisation in subordinate sentences by the scribes. First of all, a considerable amount of examples of subordinate sentences with a topicalised object (i.e. OSV order) were found, as illustrated in examples (3a–b): 2 (3) a. On þæm dagum on Egyptan wæs þæs kyninges þeaw Bosiriðis þæt ealle þa cuman þe hine gesohton he to blote gedyde. (Or. 1:8.27.9.529) b. [...] ðonon gelomp þætte þa seolfan moldan, þær his lichoma gefeol, monige menneomende wæron (Bed. 3:7.178.5.1739) It is important to note that topicalisation does not only occur with light NP or pronominal objects, but also with heavy NP objects, like the ones illustrated in (3a–b). In those examples, it is possible to find relative clauses within the object (þe hine gesohton, þær his lichoma gefeol). Furthermore, in order to demonstrate that not only objects, but also other elements of the sentence could be topicalised in a subordinate clause, topicalised PPs were included in the search. Examples (4a–b) illustrate that what can be called XSV order is also possible in OE subordination. (4) a. Forþan þe on his dagan ælc riht afeoll. & ælcunriht for Gode & for worulde up aras. (Chr.E 1100.12.3324) b. æfter þæm wæs an ger full þæt ofer eall Romana rice seo eorþe wæs cwaciende & berstende. (Or. 2:6.50.6.958) One of the most interesting findings in this study is that not only are there examples of topicalisation in subordinate clauses in the corpus, but it is also 2 Italics added to signal the topicalised element. Embedded topicalisation in Old English 173 possible to find topicalisation and inversion in the same subordinate sentence, as illustrated in (5a–b). 3 (5) a. Witodlice Basilius […] awrat ealle ða þenunga þæra halgan mæssan, swa swa hit healdað Grecas. (Liv. 142.546) b. Hu Sardanopolus wæs se siþemesta cyning in Asiria, ond hu hiene beswac Arbatus his ealdormon; (Or. 1.12.13) Again, prepositional phrases were included in the search, and examples (6a–b) show that elements different from the object could also be topicalised in subordinate sentences with inversion. (6) a. Geðencað eac þæt on ðisum lytlan pearroce þe we ær ymb spræcon bugiað swiðe manega þeoda & swiðe mislica […](Bo. 18.42.21.765) b. Is ðæt ec sæd þætte in ðere stowe, þer hio ofslegne weran, weolle an welle (Bed. 11.418.19.4207) 5. Statistics In order to provide accurate data about embedded topicalisation in the selection of OE texts, a statistical analysis of all the occurrences of this phenomenon has been included in this section. Table 2 shows the totality of examples of each of the four word order patterns mentioned in the previous section. 3 The verb in subordinate sentences with inversion has been underlined. 174 Sergio López Martínez Table 2. Total number of examples of each word order pattern in subordinate sentences. 4 Chr. A Chr. E Bed. Or. Bo. Her. Med. Liv. O.T. OSV 4 11 41 75 31 12 2 20 11 XSV -- 1 18 13 6 2 -- 5 1 OVS -- -- 2 3 1 -- -- 2 4 XVS -- -- 5 3 6 -- -- 6 -- All the occurrences of embedded topicalisation have been mapped against a series of variables in order to find any patterns and to distinguish the differences between the four word order patterns included in the study. Thus, examples have been categorised attending to whether the topicalised object is a pronoun or a full NP, whether the subject is a full NP or the impersonal ‘man’, and whether the subordinate clause is nominal or adverbial. Table 3. Statistical data: OSV order. Chr. A Chr. E Bed. Or. Bo. Her. Med. Liv. O.T. Total 4 11 41 75 31 12 2 20 11 O. Pr. 4 10 33 73 27 11 2 19 11 O. NP -- 1 8 2 4 1 -- 1 -- S. ‘man’ 1 7 24 43 15 10 2 10 9 S. NP 3 4 17 32 16 2 -- 10 2 N. Cl. 1 6 21 27 7 2 -- 5 2 Adv. Cl. 3 5 20 48 24 10 2 15 9 4 Since no examples of embedded topicalisation were found in ‘Preface’ Cura Pastoralis, it has not been included in the tables. Embedded topicalisation in Old English 175 As shown in Table 3, the majority of topicalised objects in subordinate sentences with OSV order are pronouns. It can also be observed that there is a balanced proportion between NP and ‘man’ subjects. Concerning the type of clause, there is an abundance of adverbial clauses. Table 4 shows that, when the topicalised element is a PP, the vast majority of subjects are full NPs, with only two occurrences of the impersonal ‘man’. Table 4. Statistical data: XSV order. Chr. A Chr. E Bed. Or. Bo. Her. Med. Liv. O.T. Total -- 1 18 13 6 2 -- 5 1 S. ‘man’ -- -- -- 1 1 -- -- -- -- S. NP -- 1 18 12 5 2 -- 5 1 N. Cl. -- -- 9 7 1 1 -- 3 -- Adv. Cl. -- 1 9 6 5 1 -- 2 1 It can be observed that the number of occurrences of topicalisation with inversion is not as high as with OSV order. Nevertheless, the fact that they are present in five out of the ten texts analysed shows that it is not an isolated phenomenon. As seen in Table 5, the entirety of topicalised objects in embedded topicalisation with inversion are pronouns, and the totality of subjects are full NPs (there is a complete absence of NP objects or ‘man’ subjects). In this case, the vast majority of clauses are adverbial. 176 Sergio López Martínez Table 5. Statistical data: OVS order. Chr. A Chr. E Bed. Or. Bo. Her. Med. Liv. O.T. Total -- -- 2 3 1 -- -- 2 4 O. Pr. -- -- 2 3 1 -- -- 2 4 O. NP -- -- -- -- -- -- -- -- -- S. ‘man’ -- -- -- -- -- -- -- -- -- S. NP -- -- 2 3 3 -- -- 2 4 N. Cl. -- -- -- -- -- -- -- -- 1 Adv. Cl. -- -- 2 3 3 -- -- 3 3 Finally, Table 6 shows that, when the topicalised element in a clause with inversion is a PP, the totality of subjects is NPs. Again, there are no examples of ‘man’ subjects. Table 6. Statistical data: XVS order. Chr. A Chr. E Bed. Or. Bo. Her. Med. Liv. O.T. Total -- -- 5 3 6 -- -- 6 -- S. ‘man’ -- -- -- -- -- -- -- - -- S. NP -- -- 5 3 6 -- -- 6 -- N. Cl. -- -- 2 1 3 -- -- 3 -- Adv. Cl. -- -- 3 2 3 -- -- 3 -- Embedded topicalisation in Old English 177 6. Conclusions Analysing the statistical data provided in the previous section, it can be appreciated that there is a tendency towards the topicalisation of pronoun objects in OE (pronouns are usually thematic elements, that is, they refer to previously given information). Furthermore, the data show that topicalisation with inversion always takes place with full (and sometimes complex) NP subjects, which are rhematic elements, representing new information. These two facts are clearly connected with information structural factors, since the motivation to topicalise certain elements or to invert the verb and leave a heavy subject in the right periphery of the sentence may be their thematic or rhematic nature as well as their semantic load. Studying the connection between embedded topicalisation and information structure in Old English, however, is clearly out of the scope of this paper, so more work on this topic is needed. Nevertheless, it has been demonstrated that topicalisation with or without S-V inversion is definitely an available option in OE subordinate sentences, against widespread belief. As suggested in Section 3, the fact that this phenomenon is possible also has some important implications for other associated processes of word-order change in the history of English, such as the change XV � SV. That is clearly a matter for future research. References Allen, Cynthia. 1980. Topics in Diachronic English-Syntax. New York: Garland. Allen, Cynthia. 1995. Case Marking and Reanalysis. Oxford: Clarendon Press. Haeberli, Eric. 2001. ‘Speculations on the syntax of subordinate clauses in Old English’, in Reading Working Papers in Linguistics 5: 201–229. Haider, Hubert. 2010. The Syntax of German. Cambridge: Cambridge University Press. Kroch, Antony & Ann Taylor. 1997. ‘Verb Movement in Old and Middle English: Dialect Variation and Language Contact’. In Ans van Kemenade & Nigel Vincent (eds.) Parameters of Morphosyntactic Change. Cambridge: Cambridge University Press, 297–325. Pintzuck, Susan. 1991. Phrase Structures in Competition: Variation and Change in Old English Word Order. PhD Thesis. Department of Linguistics. University of Pennsylvania, Philadelphia. 178 Sergio López Martínez Stockwell, Robert P. & Donka Minkova. 1991. ‘Subordination and word order change in the history of English’. In Dieter Kastovsky (ed.) Historical English Syntax. Berlin: Mouton de Gruyter, 367–408. Taylor, Ann, Anthony Warner, Susan Pintzuk & Frank Beths. 2003. The York- Toronto-Helsinki Parsed Corpus of Old English. Department of Language and Linguistic Science, University of York. Author’s address University of Oviedo, University of Kent School of European Culture & Languages Canterbury, CT2 7N Kent, UK received: 13 January 2016 e-mail: lopez195@gmail.com revised version accepted: 23 September 2016