Dative Subjects in Gothic. Evidence from Word Order

Dative Subjects in Gothic: Evidence from Word Order* Giacomo Bucci & Jóhanna Barðdal Ghent University Abstract This article is devoted to the study of potential dative subjects in Gothic, the earliest attested Germanic language, focusing in particular on word order distribution in the Gothic Bible, the Skeireins and the Bologna Fragment. This entails a comparison between affirmative clauses, negated clauses and interrogative clauses, contrasting nominative subjects with potential dative subjects across both translated and native Gothic passages. We expand Ebel’s (1978) methodology to the Bologna Fragment in order to confirm which syntactic structures are native to Gothic. A comparison of the word order found with nominative subjects and potential dative subjects in native Gothic passages reveals that potential dative subjects pattern unambiguously with nominative subjects in several respects. Earlier research has documented that potential nonnominative subjects in Gothic may pass the control infinitive test (Barðdal & Eythórsson 2012). Here we adduce further evidence for their subject status based on word order and a hitherto unknown example of long-distance reflexivization. Keywords: argument structure, dative subjects, subject tests, long-distance reflexivization, word order, clause structure, negation, native texts, translated texts 1. Introduction Almost 50 years have passed since it was first established that there are languages with syntactic subjects that are not in the nominative case, but show up in the oblique cases instead, like accusative, dative or genitive (see 1a–b below). This was documented simultaneously for Modern Icelandic (Andrews 1976, Thráinsson 1979) and South Asian languages (Masica 1976, Kachru, Kachru & Bhatia’s 1976, Klaiman 1980) and has since been established for several modern languages belonging to one language family after the other. For corpus languages, research on the category of subject is considerably more strenuous, especially given the fact that some of the established subject tests are not very common in texts, for instance control infinitives (see Section 3 below). Yet, research on nonnominative subjects in the early Germanic languages started to proliferate in the first part of the 1990s, with work on both Old Norse-Icelandic (Rögnvaldsson 1991, 1995, 1996) and on Old and Early Middle English (Allen 1995), later to be followed with studies of Old Swedish and Old Danish (Barðdal 2000a, 2000b). A series of comparative Germanic research has since then seen the light of day, where data from additional early Germanic languages have been brought to the fore, including Old Saxon, Old High German and even Gothic (cf. Barðdal & Eythórsson 2012, Barðdal 2023). * We thank Anne Breitbarth, Karen de Clercq, and Carla Falluomini for comments and discussions. This research was supported with two generous research grants, the first one to Giacomo Bucci from the Research Foundation – Flanders (FWO, fundamental research, grant nr. 11B7423N) and the second one to Jóhanna Barðdal (PI) from Ghent University's Special Research Fund (BOF-STA grant nr. 01N02116). 2 The relevant subject tests shown to be valid for the early Germanic languages are raising-to-subject, raising-to-object and control infinitives, in addition to conjunction reduction for Old and Early Middle English and long-distance reflexivization for Old Norse-Icelandic and Gothic (for references, see Section 3 below). One of the subject tests applied in several modern Germanic languages, i.e. word order, has also been investigated in both WestGermanic and North-Germanic, more specifically in Old and Early Middle English (Allen 1995) and the history of Icelandic, including Old Norse-Icelandic (Barðdal & Eythórsson 2012). Both of these studies support an analysis of potential non-nominative subjects as behaving syntactically in the same manner as nominative subjects do with regard to position in the clause. Still, what is sorely lacking in the studies of subjects in the Early Germanic languages in general, and of potential non-nominative subjects in particular, is an investigation of East Germanic, i.e. Gothic. Existing knowledge of the nature of word order in Gothic is indirectly based on Wulfila’s translation of the Greek New Testament, in addition to the native passages in the Skeireins, which by some is also considered to be a translation (Schäferdiek 1981), and on the native Gothic text in the Bologna Fragment. Thus, an investigation of the word order distribution of non-nominative subjects in Gothic is not an easy task, given that Gothic is not only a small-corpus language (G. Kleinkorpussprache), but also because most of the Gothic material only exists in translations. Despite that, the goal of this article is to address this particular lacuna in the field, with as rigorous an enquiry of word order in both translated and native Gothic texts as possible. For this purpose, we pair, whenever possible, the Gothic evidence with its source text. This study is confined to verbs and compositional predicates selecting for potential dative subjects in the Gothic language, leaving the study of verbs occurring with potential accusative subjects for a later investigation. The relevant structures are shown in (1) below, where the potential dative subject is marked in bold, selected by the predicates agljan ‘be injured’ and wan wisan ‘lack’, respectively. (1a) ni þauh im agljai not though them.DAT would.be.in.pain.3SG ‘though they will not be in pain’ (Mark 16:18) (1b) nauh ainis þus yet one.GEN.SG you.DAT ‘yet you lack one thing’ (Luke 18:22) wan lack ist is.3SG We start in Section 2 with presenting research on the category of subject in the modern linguistic literature beginning in the 1970s, including work on potential non-nominative subjects. From there we proceed to a proper overview of previous studies of syntactic subjects in both the early Germanic languages and the early Indo-European languages. In Section 3 we provide a discussion of the subject tests available for the early Germanic languages and we show that both nominative subjects and to some degree, potential non-nominative subjects, pass the relevant subject tests in Gothic. In Section 4 we turn to the present study, starting with a discussion of the Gothic material, including what is already known about word order in Gothic, before we describe our methodology. Section 5 contains a description of our dataset, divided into sections on the basis of clause type. We first discuss affirmative clauses, then negated clauses and, finally, interrogative clauses, focusing on their distribution across native and translated texts, including a discussion of SV and VS orders with both nominative and potential dative subjects. In Section 6 we focus in particular on the native Gothic material in our dataset, in order to describe the properties of the three clause types, preparing for a discussion of the word order properties found in clauses containing potential oblique subjects in the biblical material. In Section 7 we pay special attention to Gothic examples with potential dative subjects which deviate from the Greek original, either because the word order is different or because the subject is pro-dropped in the Greek text, but is expressed in the Gothic text. In Section 8 we summarize our main findings and conclusions. 3 2. Earlier Research Through the emergence of modern linguistics, the syntacticians of the 1970s, with their emphasis on structure, adopted a more advanced approach to grammatical relations (cf. Comrie 1973, Anderson 1976, Craig 1976, Li 1976, Keenan 1976, Schachter 1976, Sasse 1978) than that found in traditional Latin school grammar, where the subject was regarded as being in the nominative case and agreeing with the verb. The work at this time culminated in a seminal article by Keenan (1976), where the widely-used distinction between coding and behavioral properties of subjects was introduced to the field. Coding properties consist of case marking and agreement, while behavioral properties involve syntactic behavior. This includes the syntactic behavior of subjects vis-à-vis several different structures found in language, like for instance a subject’s ability to be left unexpressed in control infinitives and “raised” to subject and object in raising-to-subject and raising-to-object constructions, to mention only a few. For a further discussion of behavioral properties, see Section 3 below. All this attention directed towards grammatical relations also raised the question in the literature as to which degree potential non-nominative subjects might behave as syntactic subjects, despite not being in the nominative case and not showing subject-verb agreement. Such verbal arguments were often referred to as psychological subjects or logical subjects in the traditional literature. The first scholars to deal with this issue within modern linguistics were Andrews (1976) and Thráinsson (1979) for Modern Icelandic and Masica (1976) Kachru, Kachru & Bhatia (1976) and Klaiman (1980) for South Asian languages, in particular HindiUrdu. Their analyses are based on Keenan’s (1976) distinction between coding and behavioral properties of subject, conclusively showing that there are languages which allow syntactic subjects to occur in other cases than the nominative, for instance in the accusative or the dative. Since it was first determined that there are non-nominative subjects in Modern Icelandic and Hindi-Urdu, the existence of non-nominative subjects has been firmly established in language after language around the globe, including Quechuan languages (Hermon 1985), Dravidian (Verma & Mohanan 1990), Dardic (Steever 1998), Japanese (Shibatani 1999), Greek (Anagnostopoulou 1999), Russian (Moore & Perlmutter 2000), Tibeto-Burman (Bickel 2004), Korean (Yoon 2004), Semitic (Landau 2009, Pat-El 2018), Cariban (Castro Alves 2018), Nakh-Daghestani (Comrie, Forker & Khalilova 2018), Romanian (Ilioaia 2021, Ilioaia & Van Peteghem 2021), in addition, of course, to several supplemental South-Asian languages (cf. references in Hock & Bashir 2016). There is no doubt that the category of subject is not only language-specific, but also construction-specific, displaying different behaviors both across languages and also within a given language (Dryer 1997, Croft 2001, Culicover & Jackendoff 2005, Van Valin 2005, Barðdal 2006, Bickel 2011, Barðdal, Eythórsson & Dewey 2019, Witzlack-Makarevich 2019). Still, the fact remains that genealogically related languages share structures which may be relevant for subject behavior. Or, in other words, certain subject tests may be common for related languages (cf. Barðdal 2006, Barðdal, Eythórsson & Dewey 2019, Barðdal et al. 2023). This, in turn, may motivate subject tests from one language being taken as a point of departure for research on subjecthood in related languages, provided of course that the validity of these subject tests may be established. While the status of non-nominative subjects is undisputed in several modern languages (see references above), including Modern Icelandic and Modern Faroese (Barnes 1986), recent research on Modern German also corroborates a subject analysis for potential nonnominative subjects in that language (Barðdal 2006, Barðdal, Eythórsson & Dewey 2019, Barðdal 2023: Ch. 6), even though the data are more controversial for Modern German than for the other modern Germanic languages. This raises the question of the status of potential non-nominative subjects in the early Germanic and Indo-European languages. In an influential study on early English, Allen (1995) argues that potential nonnominative subjects behave syntactically as subjects in Old and Early Middle English, based on two subject tests, conjunction reduction and word order. Before that, examples had been presented of raising-to-subject, raising-to-object and control infinitives (Seefranz-Montag 4 1983, 1984) in Old and Early Middle English. In a series of articles, Rögnvaldsson (1991, 1995, 1996), and following him, Barðdal and Eythórsson argue that Old Norse-Icelandic also had non-nominative subjects (Barðdal 2000a, 2000b, Barðdal & Eythórsson 2003, Eythórsson & Barðdal 2005). This research is based on a host of subject properties, including long-distance reflexivization, raising-to-subject, raising-to-object, control infinitives and word order. For the early and archaic Indo-European languages, research on behavioral properties of subjects only started around a decade and a half ago. Fedriani (2009, 2014) argues that potential non-nominative subjects in Latin show behavioral properties of subjects. Conti (2010), Danesi (2015), Danesi, Johnson & Barðdal (2018) and Benedetti & Gianollo (2020) discuss a subset of the relevant subject properties in Ancient Greek, of which at least Danesi (2015) and Danesi, Johnson & Barðdal (2018) present evidence for a subject analysis of potential nonnominative subjects in Ancient Greek. Le Mair et al. (2017) focus in particular on word order in Old Irish, due to the fact that Old Irish is a very strict VSO language. They demonstrate that potential oblique subjects in Old Irish display the same word order distribution as nominative subjects, while objects show a different word order. Moreover, in a brand-new comparative study of Classical Latin and Ancient Greek, Barðdal et al. (2023) establish, based on six different subject tests, that potential non-nominative subjects in Latin and Ancient Greek clearly behave syntactically as subjects in these languages. The six relevant subject tests are conjunction reduction, long-distance reflexivization, raising-to-subject, raising-to-object, control infinitives and word order. Turning closer to home, in two pan-Germanic studies Barðdal & Eythórsson (2012) and Barðdal (2023) discuss a variety of subject tests that have been shown to be valid for several modern Germanic languages, applying them on early Germanic material, including Gothic. In particular, they discuss long-distance reflexivization, raising-to-subject, raising-to-object and control infinitives. Barðdal (2023) presents new evidence from Old Saxon, Old Norse-Icelandic, Old Swedish and Old Danish involving control infinitives which clearly speak for a subject analysis of potential oblique subjects in these languages, while an object analysis is excluded. Barðdal & Eythórsson (2012) briefly touch upon the issue of word order in Gothic, although a systematic investigation of word order distribution of subjects in that language still remains a desideratum. The goal of this article is to address that gap in the field. 3. The Category of Subject in Gothic We define subject as being the first argument of the argument structure of verbs, following Eythórsson & Barðdal (2005: 827–832, and subsequent work). This is a working definition of subject, which we believe captures the empirical core of the subject concept, to be recast in each reader’ favorite theoretical framework. This definition originates in the fact that when abstracting over the relevant subject properties, there is a generalization to be made, namely that the subject tests all single out the first argument of a verb’s argument structure. This means, simply put, that the established subject behaviors all target the first argument of a verb’s argument structure. This, in turn, raises the follow-up question of which factors impact the order of the arguments of the argument structure in the first place. We take the order of the arguments to be a derivative of event structure and the force dynamics holding between the participants of the event (cf. Talmy 1985, 1988, Croft 1998, 2012, Barðdal 2001, Barðdal, Eythórsson & Dewey 2019, Barðdal 2023, Ch. 2–3, Barðdal et al. 2023, etc.). Events, again, are structured through their causal properties, being causatives, inchoatives, statives, etc. For a verb like kill, there is an event of killing, which is causative in nature, there is an initiator of this event, the killer, and there is an endpoint, the animate entity being killed. And since the initiator is the entity exerting force upon another entity, the endpoint, the initiator is assigned to the first argument of the argument structure and the endpoint to the second argument of the argument structure. As such, there is an iconic relation between the participants of the event structure and the arguments of the argument structure. 5 Yet, events denoted by verbs and predicates selecting for potential non-nominative subjects do not conform to the event type found with verbs like kill, as described above. Instead, these predicates are typically force-dynamically neutral, mostly stative but also inchoative in some cases, expressing change of state. For psychological predicates this means that they allow for two construals, involving an experiencer and a stimulus, i.e. the entity being perceived or experienced. These two construals are Exp-Stim and Stim-Exp. In the first case, the experiencer directs attention to a stimulus, while in the second construal the stimulus affects an experiencer. The relevant construal typically found with potential dative subjects is of the first type, where the experiencer of a two-place predicate directs his or her attention to a stimulus. Also, one-place predicates exist where the experiencer is the first and only argument of the argument structure. Most of the potential dative subject examples discussed in this article are of that type. Observe that the first argument of the argument structure does not necessarily equate the first argument in the linear order in every Gothic clause. That is, the linear order of the arguments does not necessarily have to reflect the order of the arguments in the argument structure. For Gothic clauses representing neutral word order, one would indeed expect the first argument in the linear order to correspond to the first argument of the argument structure, and thus be the subject. By neutral word order we mean the word order where no information structural considerations may have affected the order of the constituents of a clause (SOV and SVO for Gothic). For clauses where information structure has affected the word order (OVS, VSO, VOS), the linear order of the arguments will not reflect the order of the arguments in the argument structure. The behavioral subject tests that have been applied on the Germanic languages are the following. ● ● ● ● ● ● Clause-bound reflexivization Long-distance reflexivization Raising-to-Subject Raising-to-Object Control Infinitives Word Order The remainder of this section is devoted to a discussion of the first five behavioral properties and to which degree they are also valid for Gothic. Starting with clause-bound reflexivization, Harbert (1978) argues that only subjects in Gothic may bind reflexives within their minimal clause. Such an example is given in (2) below, where the reflexive seinai ‘self’ is bound by the pro-dropped nominative subject ‘he’. Clause-bound reflexivization Translation of NTG καὶ ἔλεγεν αὐτοῖς ἐν τῇ διδαχῇ αὐτοῦ (2) jah [isi] qaþ im in laiseinai seinaii and he.NOM said them.DAT in teaching his.REFL ‘and he said to them during his teaching’ (Mark 4:2, from Harbert 1991: 34) However, as noted by Barðdal & Eythórsson (2012: 373–375), there is a great variation across the early Germanic languages with regard to i) whether or not subjects must bind reflexives, ii) whether subjects may also bind personal pronouns and iii) whether or not objects may bind reflexives. While Harbert may, of course, be right about the situation in Gothic, we will not build a case on the basis of clause-bound reflexivization, due to the variation both within and across the early Germanic languages. For long-distance reflexivization, however, the situation is very different, as there is no doubt that only subjects may bind reflexives across clause boundaries in the languages where long-distance reflexivization is found. For the Germanic languages, long-distance reflexivization exists in Modern Icelandic (Thráinsson 1976), Modern Faroese, and some WestNorwegian dialects (Strahan 2003). For the early Germanic languages, it has so far only been 6 documented in Old Norse-Icelandic (Rögnvaldsson 1991, 1995, 2007) and Gothic (Harbert 1978). To this end, Harbert (1978) claims that the example in (3) below is the only example of long-distance reflexivization that he has come across. Long-distance reflexivization (3) akei [isi] was kunnands þatei swaleikamma and he.NOM was knowing that such.DAT waldufnja mahtais seinaizosi nauþs ustaiknida wesi authority.DAT power.GEN REFL.GEN force.NOM shown would.be ‘He knew, nevertheless, that by such authority, the force of His power would be revealed’ (Skeireins 1:4, cited from Harbert 1978: 38) In this example, the reflexive sei[na]izos ‘self’ in the subordinate clause, headed by þatei ‘that’, is bound by the pro-dropped nominative subject ‘he’ in the main clause. There is no doubt that this example involves long-distance reflexivization since the reflexive is found in the subordinate clause and its controller is the prodropped nominative subject of the main clause. As a matter of fact, we have come across a parallel example of long-distance reflexivization involving a potential non-nominative subject, i.e. im ‘them’, selected by the verb þugkeiþ ‘thinks’. As is evident from (4) below, the reflexive seinai ‘selves’ in the subordinate clause, headed by ei ‘that’, is bound by the potential dative subject im ‘them’ in the matrix clause. On such an analysis, the example in (4) indeed speaks for a subject analysis of the dative, exactly as it speaks for a subject analysis of the nominative in (3) above.1 Long-distance reflexivization Translation of NTG δοκοῦσιν γὰρ ὅτι ἐν τῇ πολυλογίᾳ αὐτῶν εἰσακουσθήσονται (4) þugkeiþ imi auk ei in filuwaurdein seinaii thinks.3SG them.DAT then that in many.words their.REFL andhausjaindau. would.be.heard.MP.3PL ‘then they think that due to their extensive speech, they would be heard.’ (Mt. 6:7) Before leaving this issue, it might be added here that long-distance reflexivization is also found in Latin and Ancient Greek (Kühner & Gerth 1955, Kühner & Stegmann 1955), in addition to Gothic and the West-Nordic languages. Not unexpectedly, examples involving potential oblique subjects binding reflexives across clause boundaries, i.e. long-distance reflexivization, have very recently been documented in both Latin and Ancient Greek (see Barðdal et al. 2023) Turning to raising-to-subject, there are ample examples of such structures in Gothic with nominative subjects. One such is given in (5) below: Raising-to-subject Translation of NTG οὐ δύναται δένδρον ἀγαθὸν καρποὺς πονηροὺς ποιεῖν (5) ni mag bagms þiuþeigs akrana ubila gataujan not may.3SG tree.NOM good.NOM fruit.ACC evil.ACC bring.forth.INF ‘a good tree may not bear bad fruit.’ (Matthew 7:18) 1 As for Proto-Indo-European, Kiparsky (2011, as quoted in Lundquist & Yates 2017) claims that PIE likely had no reflexive pronoun, as *swe- was an adjective, meaning ‘own’, and *se- was originally a demonstrative used as a 3rd person pronoun (see also Orqueda 2019 about the heterogenous situation in Vedic Sanskrit). In contrast, Lundquist & Yates (2017: 2096) propose that *swe- was originally a reflexive pronoun which later gave rise to the reflexive adjective *swo- ‘one’s own’. Dunkel (2014: 2751– 2768) argues that the root forms, (1) *swe-, (2a) *se-, (2b) *su- ‘(one)self’, represents one pronominal stem, which he deems as reflexive. Yet, Miller (2019: 549) suggests that the reflexive use of *swe-/*se/*su- was a later development. However, in Gothic, there is no doubt that seins had a reflexive function (cf. Harbert 1978, 1991). 7 In this particular example, the nominative subject is bagms þiuþeigs ‘a good tree’ which stands in a nexus relation with the modal auxiliary magen ‘may’, while the main verb gataujan ‘bring forth’ is in the infinitive. In other words, the nominative subject, bagms þiuþeigs ‘a good tree’, selected by the main verb gataujan ‘bring forth’, behaves syntactically as the subject of the modal auxiliary magen ‘may’. This type of structure, often labeled raising-to-subject, is typically found with modal auxiliaries, aspectual auxiliaries, and others like that. What the example in (5) shows is that the nominative, bagms þiuþeigs ‘a good tree’, selected for by the lower verb, takes on the subject properties of the finite verb, thus behaving syntactically as an unequivocal subject. Structures involving raising-to-object, a.k.a. AcI, are typically found with causative verbs like ‘let’, verbs of perception like ‘see’ and ‘hear’, verbs of saying and verbs expressing intention in Gothic (Harbert 2007: 262–263). The same is true for native Old Norse-Icelandic texts (Kristoffersen 1996: 128). We let it suffice to demonstrate the existence of such structures in Gothic with the example in (6) below, containing the causative verb letan ‘let’: Raising-to-object Translation of NTG ἄφετε τούτους ὑπάγειν (6) letiþ þans gaggan. let.2PL these.ones.ACC go.INF ‘let these go away.’ (John 18:8) The matrix verb in (6) is ‘let’, here occurring in the 2nd person imperative, selecting for the verb gaggan ‘go, walk’, in the infinitive. What characterizes structures of this type is that the argument corresponding to the subject of the lower verb, gaggan ‘go, walk’, in finite clauses here takes on the behavioral properties of the object of the matrix verb. This is evident from the fact that þans ‘these ones’, which equates the subject of ‘go’ in finite clauses, occurs here in the accusative case. This accusative may be analyzed as coming from the matrix verb or coming from the construction itself, depending on one’s theoretical preferences (cf. Barðdal et al. 2023). Before we turn to the last subject test to be discussed here, control infinitives, it should be mentioned that no examples of raising-to-subject or raising-to-object are documented with potential non-nominative subjects in Gothic. This, however, does not, in and of itself, support a non-subject analysis of potential non-nominative subjects in Gothic, it only does not corroborate a subject analysis. In all likelihood, this lack of raising-to-subject and raising-toobject examples is simply a reflection of the small size of the Gothic corpus, which counts only ca. 68,000 words (Eythórsson & Barðdal 2005: 833, Snædal 2009: 148). Clearly, this is an exceptionally small corpus. We believe that if this corpus had been larger, the chances of retrieving such examples would have been considerably greater. Apart from that, both raising-to-subject and raising-to-object are found in the other early Germanic languages with potential non-nominative subjects, including Old Saxon and Old High German, Late Middle English, Old Norse-Icelandic, Old Swedish and Old Danish (see Barðdal & Eythórsson 2012: 380–385, Barðdal 2023: Ch. 4). It should also be mentioned here that such structures are found in both Latin and Ancient Greek, also with non-nominative subjects (Barðdal et al. 2023). Given the existence of raising-to-subject and raising-to-object constructions in both West and North Germanic, and given the existence of such structures with nominative subjects in Gothic, we are inclined to interpret the non-attestations of raising-to-subject and raising-toobject examples with potential non-nominative subjects in Gothic as representing a natural lacuna in the dataset, due to the small size of the corpus. That is, we take the comparative evidence from West- and North-Germanic, in addition to the evidence from Latin and Ancient Greek, to speak for the acceptability of such structures also in the Gothic language, despite their lacking attestations. For a reconstruction of such structures for Proto-Germanic, see Barðdal (2023: Ch. 5). Turning now to control infinitives, such examples involving nominative subjects are readily found in the Gothic material, including the example in (7) below: 8 Control infinitive Translation of NTG μετέβη ἐκεῖθεν τοῦ διδάσκειν καὶ κηρύσσειν ἐν ταῖς πόλεσιν αὐτῶν (7) [is] ushof sik jainþro du ____ laisjan jah he.NOM moved.3SG self.REFL thereto to PRO.NOM teach.INF and ____ merjan and baurgs ize PRO.NOM preach.INF in cities their ‘[He] left for there to teach and preach in their cities’ (Matthews 11:1) Here the argument corresponding to the nominative subject of ‘teach’ and ‘preach’ in finite clauses is left unexpressed on identity with the pro-dropped nominative subject of ‘move’ in the matrix clause in both Gothic and Ancient Greek. This behavior, to be left unexpressed in control infinitives, is confined to subjects and does not apply to objects. The example in (7) therefore counts as clear evidence for the subject status of the nominative selected for by the verbs ‘teach’ and ‘preach’. One parallel example involving potential non-nominative subjects has already been documented in Gothic (cf. Barðdal & Eythórsson 2012: 386), reproduced here as (8) below: Control infinitive Translation of NTG πᾶς ὁ βλέπων γυναῖκα πρὸς τὸ ἐπιθυμῆσαι αὐτὴν (8) hvazuh saei saihviþ qinon du ____ luston whoever who.NOM sees.3SG woman.ACC to PRO.ACC lust.INF ‘whoever looks at a woman in order to lust for her’ (Matthew 5:28) izos her.GEN The lower verb is luston ‘lust’, which we assume occurs in the Acc-Gen case frame in Gothic, although unfortunately this is the only example of luston ‘lust’ in the existing Gothic texts. However, we base our assumption on the comparative material from the other early Germanic languages, where the cognates of luston systematically occurs with an Acc-Gen frame, as is shown in (9a–e) below: (9a) so thíh es wola lústit so you.ACC it.GEN well pleases.3SG ‘So that it will please you well’ (Evangelienbuch 1, 1, 14) (9b) that ina bigan bi thero mennisko that he.ACC began.3SG because.of the.DAT humanity.DAT môses lustean meat.GEN desire.INF ‘that because of his humanity, he began to desire meat.’ (Heliand 1060) (9c) Hine nanes þinges ne lyste on ðisse him.ACC no.GEN thing.GEN not desired.3SG on this ‘He desired nothing in this world.’ (Boethius Cons.Phil. 35,6) (9d) Mich nelusted niewehtes me.ACC not.lusted.3SG nothing.GEN ‘I desired nothing’ (LW 85,8, here cited from van der Horst 2008: 241) (9e) er þig lysti þessa when you.ACC that.GEN desired.3SG ‘when you desired that.’ (Ljósvetningasaga, ch. 19) worulde world 9 These examples, stemming from Old High German, Old Saxon, Old English, Old Dutch and Old Norse-Icelandic, are uniform in the sense that they all instantiate the same case frame, Acc-Gen, shown above in bold. Thus, it is uncontroversial to assume that it is the accusative of luston which is left unexpressed in the control infinitive in (8), on identity with the nominative subject hvazuh saei ‘whoever who’ in the matrix clause, while the genitive object izos ‘her’ is expressed. The Greek text presents a Nom-Dat verb here, ἐπιθυμέω ‘desire’, with the nominative being left unexpressed and the dative being expressed. When translating this example, Wulfila clearly equated the potential accusative subject of luston ‘lust’ with the nominative subject of the Greek ἐπιθυμέω ‘desire’, indeed speaking for the subject behavior of potential non-nominative subjects in the Gothic language. Now that we have clarified what we mean by syntactic subject, we turn to the sixth behavioral property of subjects, namely word order in the Gothic language. 4. Word Order in Gothic In this section we start with a short description of the Gothic material available to us (4.1) before we report on what is already known about word order in Gothic (4.2). We then give a brief account of our methodological procedure, including how we deal with the translational nature of our source material. 4.1 The Gothic Material There are three main sources for the Gothic language which will be used here. The first and largest one is the Gothic Bible. It is mostly preserved in the Codex Argenteus, a lavish codex produced in the beginning of the 6th c., consisting of 187 leaves, in addition to one leaf known as the Speyer fragment. This leaf only contains limited sections of Mark (16:9-20), but several other parts of the epistles are preserved as palimpsests in other codices. These are the Codices Ambrosiani A, parts of which are now in the Codex Taurinensis, B, C, D in the Codex Carolinus, and in the Codex Gissensis (cf. Miller 2019: 9–11). The various codices and fragments contain a vast portion of the New Testament, although with some major lacunas, as well as a few fragments from the Old Testament book of Nehemiah (see Snædal & Petersen 2012). Our second source is the so-called Skeireins, a learned commentary of the Gospel of John (first edition by Maßmann 1834). The text, which was probably composed between the second half of the 4th and the beginning of the 5th c., is preserved in a 6th c. copy. It consists of eight leaves incorporated in two different manuscripts (leaves 1, 2, 5, 6, 7 in the Codex Ambrosianus E, leaves 3, 4, 8 in the Codex Vaticanus Latinus 5750), all transmitted as palimpsests. While the text contains a commentary to 37 biblical verses, which unfortunately do not always correspond to the ones found in the Gothic Bible (cf. Falluomini 2016), it has been debated whether the remainder of the Skeireins is a translation of an unknown Greek source or whether it is an original Gothic composition (Schäferdiek 1981, Miller 2019: 10–11). Our third and last source, the Codex Bononiensis (Falluomini 2014), also known as Gothica Bononiensia (cf. Finazzi & Tornaghi 2013), is a palimpsest on Augustine’s De civitate Dei. This source consists of two leaves, discovered in 2010 in the archive of the fabriceria of the Basilica of San Petronio in Bologna in Italy. The text, whose manuscript stems from the first half of the 6th c., presents several biblical verses from both the Old and the New Testament. Some of these biblical verses are not preserved in the direct tradition, but are instead embedded in what seems to constitute a rhetorically elaborate sermon by an unknown, yet a learned, author (cf. Falluomini 2017: 284–285, 2018: 162). 10 4.2 Gothic word order Sentence topology is an important topic when analyzing word order in Gothic. However, as is already well known to scholars of Gothic, such a task is fraught with caveats. This is because the translation process for the main source for the Gothic language, the Gothic Bible, yielded a text where the word order of every element from the Greek source roughly corresponds oneto-one to the Gothic text, as is shown in (10) below: (10) þarei auk ist huzd izwar, þaruh ὅπου γάρ ἐστιν ὁ θησαυρός σου, ἐκεῖ for also is.3SG treasure.NOM your.NOM.PL there ist jah hairto izwar ἔσται καὶ ἡ καρδία σου is.3SG also heart.NOM your.NOM.PL ‘For where your treasure is, there will your heart also be’ (Matthew 6:21) The only difference between the two texts is that New Testament Greek (NTG) σου is singular, while Gothic izwar is plural in both cases. Despite this, both language internal and external comparisons allow us to extract some relevant information from the Gothic material on word order, which we use below as a baseline for establishing grammatical relations for potential dative subjects. The examples in (11a–b) below show that the unmarked order of the subject and the verb is SV in main clauses. The evidence comes from super- and subscriptions in the Gothic manuscripts: Translation of NTG εὐαγγέλιον κατὰ Λουκάν ‘Gospel according to Luke’ (11a) aiwaggeljo þairh lukan anastodeiþ gospel.NOM by Luke begins.3SG ‘the gospel of Luke begins’ (Codex Argenteus, f. 118r) Translation of NTG εὐαγγέλιον κατὰ Μᾶρκον ‘Gospel according to Mark’ (11b) aiwaggeljo þairh marku ustauh gospel.NOM by Mark has.ended.3SG ‘the gospel of Mark has ended’ (Speyer fragment)2 We analyze þairh lukan ‘through Luke’ and þairh marku ‘through Mark’ as appositions to the nominative subject aiwaggeljo ‘gospel’ in (11a–b). Falluomini (2018: 164) shows that the word order in (11) differs from both possible Greek and possible Latin sources. While the Greek text in (11) is in fact verbless, a potential Latin source would represent verb-first (V1) order, as is shown in (12) below: Latin (12) incipit evangelium begins.3SG gospel.NOM ‘the gospel by Luke begins’ secundum according.to lucam Luke Turning to two-place predicates, Eythórsson (1995), in his study of word order in the early Germanic languages, suggests that the unmarked word order in Gothic was SOV (cf. Lehmann 2005, Barðdal & Eythórsson 2012: 386). For Gothic, this had already been suggested by Koppitz (1900, 1901) and Fourquet (1938, here quoted from Falluomini 2018). However, Eythórsson’s hypothesis is based not only on a comparison with the other early Germanic languages, but also, and more importantly, on the translation of morphologically synthetic 2 Cf. also the explicit to the First letter to the Corinthians: Du Kaurinþium ·a· ustauh ‘To the Corinthians I have ended.’ 11 verbs in Greek into two elements in Gothic. Two such examples are given in (13) below, where a simple Greek verb is translated with an adjective and a finite verb. Translation of NTG συμφέρει (13a) batizo ist better is.3SG ‘it is better’ (Matthew 5:29, cited from Eythórsson 1995: 20) Translation of NTG ἐμώρανεν (13b) dwala gatawida foolish made.3SG ‘made foolish’ (Cor. I 1:20, cited from Eythórsson 1995: 20) In addition, Ebel (1978) shows that the word order in finite clauses in the commentary parts of the Skeireins is overwhelmingly OV, as opposed to the biblical quotations found in the same text which mostly present VO order (Ebel 1978: 80). One such example is presented in (14) below, where the finite verb, fragaf ‘gave’ occurs clause-finally, with both the direct object, ganauhan þaurftais ‘sufficient need’, and the indirect object im ‘them’, preceding the verb: (14) ni þatainei ganauhan þaurftais im fragaf not only sufficiency.ACC need.GEN them.DAT gave.3SG ‘[He] didn’t only give them the sufficient need [but much more]’ (Sk. 7, l. 13-14, cited from Ebel 1978: 62) At the same time, there are indications that the Gothic language was developing from verbfinal to verb-second (V2) (cf. Delbrück 1911, Kotin 2012: 374, Falluomini 2018: 164–165). This is especially evident in affirmative main clauses in the Bologna fragment (see Falluomini 2018: 166–167, Miller 2019: 513–514). The following example documenting this is a marginal gloss in Gotica Veronensia, which is a translation from Latin. Translation of Latin: dē septem pānibus populum saturāverit (15) us z hlaibam gasoþida manageins out.of 7 loaves.DAT satiated.3SG many.ACC ‘out of seven loaves of bread [he] satiated the crowd’ (here cited from Falluomini 2018: 165 and Miller 2019: 514) In the Latin original the verb, saturāverit ‘satiate’, is in clause-final position, while gasoþida ‘satiate’ is in second position in the Gothic translation in (15) above. Eythórsson (1995: 28) also observes that in the Gothic Bible V2 order is common when the complement is definite or focused, like in (16) where in bokom libainais ‘the book of life’ is a definite complement. Note also that the Greek original is verbless: Translation of NTG ὧν τὰ ὀνόματα ἐν βίβλῳ ζωῆς (16) þizeei namna sind in bokom libainais whose.GEN names.NOM.PL are.3PL in book life.GEN ‘whose names are in the book of life’ (Philippians 4:3, cited from Eythórsson 1995: 28) Unless preceded by a topicalized NP, V1 order represents unmarked word order in imperative and negated sentences (cf. Eythórsson 1995, Falluomini 2018, Miller 2019: 516, and literature cited therein). The sentential negator ni ‘not’, for example, is found immediately in front of the finite verb in both the Gothic Bible (17a) and in the Bologna fragment (17b): 12 Translation of NTG οὐ γὰρ μὴ κληρονομήσῃ (17a) unte ni nimiþ arbi until not receives inheritance.ACC ‘for [he] will not receive the inheritance’ (Gal. 4:30, cited from Eythórsson 1995: 24) (17b) akei ni wilda galaubjan skapa jah dagand but not will believe.INF creator and illuminator allaizo wiste all.GEN creatures.GEN ‘but [he] will not believe in the creator and illuminator of all creatures’ (BF, 2v, l. 11-3, cited from Falluomini 2018: 170) Finally, subject-verb inversion may also be found in main clauses in Gothic, resulting in V1 word orders. This fronting of the verb is found in questions and structures involving the socalled narrative inversion, a distinctive Germanic construction which is typically used to signal textual cohesion (Sigurðsson 1993, Eythórsson 1995, Dewey 2006) and in some early Germanic varieties, it also assigns a perfective meaning to the verb (cf. Leiss 2007: 76–77). 4.3 Methodology It should be evident by now that word order in Gothic is far from being fixed, but instead that each linearization pattern must be analyzed in itself, by singling out the specific clause types found for each relevant linearization pattern. On the basis of these premises, we apply the word order test on the appropriate clause types (see below) in order to compare the word order distribution of potential dative subjects with ordinary nominative subjects. The word order test may be divided into two subtests: ● ● First position in declarative clauses Subject-verb inversion We base our study on three different clause types for which the word order is known on the basis of both internal and external comparison, namely affirmatives, negatives, and interrogatives, with one caveat. Since the Gothic Bible is a translation, subject position cannot be extracted directly from that text, as that might result in a distorted picture of Gothic syntax, due to possible translation effects. Instead, we compare the position of potential nonnominative subjects in the Gothic Bible with the position of nominative subjects in the text parts that are not direct biblical quotations in the Skeireins and in the Bologna Fragment. The “native” quality of these particular texts has already been investigated in some detail (cf. Ebel 1978, Falluomini 2018). Nevertheless, in order to obtain as many insights as possible regarding the order of the constituents, data from both texts will be reanalyzed and systematized as well, and eventually compared with the word order in biblical passages taken from the entire Gothic corpus. Finally, we pair, whenever relevant, the Gothic evidence with its source text. This aspect of the work cannot be disregarded, given the nature of most of the surviving Gothic material. Still, the issue of the source text must be addressed explicitly. The Greek text that is usually employed for such a comparison was devised by Streitberg (2000). Recent research, however, has shown that the version of the text provided by Streitberg is not entirely adequate and that the translation of the Gothic Bible was most probably based on a byzantine translation of the Bible, today collected in the so-called Majority Text (cf. Robinson & Pierpont 2005). This text has shown overall more textual contiguity to the Gothic Bible than the Greek text devised by Streitberg in his critical edition (cf. Ratkus 2011, Falluomini 2015). For reasons of space, we provide examples from the alleged Greek Vorlage only if necessary, that is, only if the two versions differ in some meaningful detail. In all other cases, no deviations from the Greek text are presented. 13 5. The dataset We start with a general introduction of our dataset in 5.1, before discussing affirmative clauses, negated clauses and interrogative clauses, each in a subsection of its own, including their distribution across native and translated texts. We also present the main tendencies for nominative subjects across SV and VS word orders. Our findings will then be used as a baseline for establishing the main similarities and differences in word order between biblical and native Gothic in Section 6. We then apply the word order test on potential dative subjects, making use of this baseline, in Section 7. 5.1 Introduction For the exploratory part of this research, we have collected data involving the following clause types: 38 affirmative declarative clauses, 17 negative declarative clauses, and 16 interrogative clauses, a total of 72 different clauses each with their own subject (see Table 1 and the Appendices). Their distribution across native and Biblical Gothic is given in Table 2. Table 1: Number of nominative and potential dative subjects across clause types NOM SUBJECT DAT SUBJECT TOTAL AFF 17 21 38 NEG 15 3 18 INT 12 4 16 44 29 72 Below we present the word order facts in more detail, starting with a comparison of the three types of clauses, affirmatives, negatives and interrogatives (5.2–5.4). We then pay special heed to the distributional differences between native and biblical passages (Section 6). Particular attention is given also to those biblical passages whose word order actually differs from the Greek source. A more in-depth discussion regarding the specific clause types, their linearization, and the comparison with potential dative subjects then follows (Section 7). Table 2: Distribution of clause types across native and translated texts AFF NEG INT BIBLICAL PASSAGES NATIVE GOTHIC TOTAL 25 7 10 13 11 6a) 38 18 16 72 a) We have excluded the verbless and subjectless interrogative in ƕis ‘for what (reason)?’ (BF., 1r, l. 17). The verbs that have been collected for the present study are given below. There are, in total, 14 baseline verbs, i.e. verbs selecting for nominative subjects, and 14 predicates selecting for potential dative subjects (the number of tokens extracted for each predicate is given in brackets): Nominative subject verbs galagjan ‘lay down’ (1), galaubjan ‘believe’ (1), ganiman ‘bear (a child)’ (1), fraþjan ‘think, understand’ (1), insakan ‘advise’ (1), lausjan ‘save, free’ (1), nauþjan ‘need’ (1) qiman ‘come’ (2), qiþan ‘say’ (5), usluton ‘tempt’ (1), staujan ‘judge’ (4), taujan ‘do’ (2) ushneiwan ‘bow down’ (1), wisan ‘be’ (13) 14 Potential dative subject verbs aglu wisan ‘have difficulty’ (2), azitizo wisan ‘be easier’ (1), batizo wisan ‘be better’ (2), fraqistnan ‘die, be lost’ (1), galeikan ‘like, please’ (4), ganauhan ‘find sufficient’ (1), gatiman ‘be fit/appropriate’ (1), goþ wisan ‘be good’ (6), kunþs wisan ‘be known’ (1), raþizo wisan ‘be known’ (1), skuld wisan ‘be permitted’ (1), þugkjan ‘seem’ (3), waila wairþan ‘recover’ (1), wan wisan ‘lack’ (2) The verbs selecting for potential dative subject were retrieved from the NonCanCase database (https://www.evalisa.ugent.be/noncancase), while the nominative subject verbs were collected independently in the remainder of the Gothic material (for ushneiwan ‘bow down’, see Falluomini 2017: 292). 5.2 Affirmative clauses We have collected a total amount of 38 affirmative clauses, 17 of which occur with a nominative subject, and 21 with a potential dative subject. Out of the 17 nominative subject clauses, four are biblical passages and 13 are native Gothic material. All the clauses containing potential dative subjects are found only in biblical passages. Table 3: Word order distribution of nominative and potential dative subjects across text types in affirmative clauses AFFIRMATIVE CLAUSES BIBLICAL TEXTS NATIVE TEXTS SV VS SV VS TOTAL NOM 2 2 7 6 17 DAT 7 14 21 9 16 7 6 38 Starting with SV word order, this is found 16 times in the dataset, of which nine examples are from biblical texts, while the remaining seven constitute native Gothic examples (cf. Table 3). One such example from the Bologna fragment is shown in (18), where the matrix verb is followed by an infinitive: (18) ak silba frauja qam du nasjan but self God.NOM came.3SG to save.INF ‘but God himself came to save us’ (BF, 1r, l. 13) unsis us.DAT There are 22 examples sporting a VS order in our dataset, of which six stem from native Gothic texts, while the remaining 16 sentences are all from biblical passages. One such example, involving a potential dative subject from a biblical source, is given in (19) below:3 3 From a modern Germanic perspective it might seem reasonable to analyze the ei ‘that’ clause as the subject of batizo ist ‘is better’, rather than the dative. This would involve the postulation of a null correlate of the ‘that’ clause, located in the matrix clause. However, structures of this type still exist in Modern Icelandic, as is shown in (i) below, where honum ‘him’ is the dative subject of betra er ‘is better’, also selecting for a ‘that’ clause. Such examples do not call for a null correlate in Modern Icelandic: (i) og betra er honum að þau heldur sjeu færri … and better is.3SG him.DAT that they.NOM rather would.be.3PL fewer ‘And it is better for him that they rather be fewer ...’ (Sæmundsson 1839: 213) 15 Translation of NTG συμφέρει ὑμῖν ἵνα ἐγὼ ἀπέλθω (19) batizo ist izwis ei ik better is.3SG you.DAT that I.NOM ‘It is better for you that I go away’ (Jh. 16:7) galeiþau go.OPT.1SG Observe that the word order of the Greek source text is copied in all the translated examples in the biblical texts, except for one. This example, given in (20) below, shows a VS word order in Gothic, on the assumption that the dative is S, while the Greek original has SV word order. This appears to be an example of predicate focus, where the whole predicate, goþ ist ‘is good’ has been fronted to initial position to emphasize new or contrastive information (cf. Barðdal et al. 2013: 355–358, based on Lambrecht’s 1994, 2000, 2001 and Dahlström’s 2003 ideas of focus fronting). As a consequence, the subject inverts with the predicate. Translation of NTG καλὸν αὐτοῖς ἐστιν lit. ‘good them is’ (20) goþ ist im jabei good is.3SG they.DAT if ‘[I say therefore to the unmarried and widows], it is good for them if ...’ (Cor. I 7:8) Apart from that, the alignment of the elements in the Greek and the Gothic texts correspond perfectly with one another in this example. It is only the position of the potential subject and the predicate that is inverted here. 5.3 Negated clauses There are 18 negated clauses in our dataset, of which 15 occur with a nominative subject, and three with a potential dative subject. While all the examples of negated clauses containing potential dative subjects are retrieved from biblical texts, negated clauses with nominative subjects are more evenly distributed across text types with ten examples stemming from native Gothic texts and four from biblical passages. In the following, we divide negated clauses up into classes on the basis of the position of the negation, which yields three different word orders: (Neg)VS, (Neg)SV, and S(Neg)V. In the final analysis, however, we regard both (Neg)SV and S(Neg)V as having the same SV word order. These two word orders are expected because a negation with scope over the whole sentence generally triggers subject-verb inversion in the Germanic languages, while a negation only with scope over the verb does not affect word order. This has been documented for subordinate clauses in Swedish, starting in the 14th century (cf. Falk 1993: Ch. 6). Table 4: Word order distribution of nominative and potential dative subjects across text types in clauses with negation NEGATED CLAUSES BIBLICAL TEXTS NATIVE TEXTS NEGVS SNEGV NEGSV NEGVS SNEGV NEGSV TOTAL NOM 1 3 0 9 1 1 15 DAT 0 2 1 0 0 0 3 1 5 1 9 1 1 18 Starting with examples where the negation occurs in first position, thus taking scope over the whole clause and triggering subject-verb inversion (NegVS), we have found 10 such examples in our dataset (cf. Table 4), of which nine represent native Gothic, whereas there is only one Biblical passage. All ten examples present nominative subjects, of which one is shown below. 16 (21) n=ist sa=ei nasjai ufar þuk not=is.3SG this.NOM=that.CONJ saves.OPT.3SG over you.ACC ‘there is not [anyone] who saves more than you, Lord’ (BF, 1r, l. 11) frauja Lord Turning to the examples where the subject precedes the verb (SNegV), six in total, of which four occur with a nominative subject and two with a potential dative subject. We give one example of each of these below, with a nominative subject in (22a)4 and a potential dative one in (22b). Note that (22a) is the only example of this type in native Gothic, as the remaining five stem from Biblical passages. (22a) jabai auk diabulau fram anastodeinai nih nauþjandin if also devil.DAT from beginning not forcing.DAT.SG ak uslutondin mannan but deceiving.DAT.SG man.ACC.SG ‘if then the devil from the beginning not forcing but deceiving the man’ (Sk., 1, l. 13-14) Note also that the accusative object ‘man’ is located after the verb. Translation of NTG ἵνα μή τι ἀπόληται (22b) þei waihtai ni fraqistnai that.CONJ thing.DAT not loses.OPT.3SG ‘[gather the remaining fragments], that nothing be lost’ (Jh. 6:12) Turning to (22b), note that the negation immediately follows the potential dative subject ‘thing’, while in the Greek original the negation actually precedes the non-cliticized pronominal subject. The reason for this discrepancy between the two languages is that the negation is generally cliticized to the verb in the early Germanic languages.5 Of clauses with negation, the only type left to be discussed is the order NegSV, shown in (23) with a potential dative subject. Translation of NTG [κἂν θανάσιμόν τι πίωσιν], οὐ μὴ αὐτοὺς βλάψῃ (23) ni þauh im agljai not then them.DAT would.be.in.pain.3SG ‘[and if they drink something deadly], then they will not be in pain’ (Mk. 16:18) Note that the verb is intransitive, agljai ‘be in pain, a deadjectival verbalization in *-ijan, only selecting for the potential dative subject. In contrast, the Greek source has an accusative pronoun, αὐτοὺς ‘them’, here, while the Gothic translation uses a dative form, im ‘them’. 5.4 Interrogative clauses Our dataset contains 16 interrogative clauses in total, 12 of which occur with a nominative subject and four with a potential dative subject. As mentioned above, all the examples of potential dative subjects stem from biblical passages, like half of the examples with nominative subjects. This includes six examples from native Gothic passages and six examples from biblical passages. 4 In this example it is the absolute construction that triggers dative case assignment of the subject argument ‘devil’ and of the present participles ‘forcing’ and ‘deceiving (cf. Dewey & Syed 2009). This verb selects for a nominative subject in general in Gothic, including in negated existentials. Thus, we categorize this example as belonging with nominative subject verbs in our dataset. 5 Anne Breitbarth, p.c. 17 Table 5: Word order distribution of nominative and potential dative subjects across text types in interrogative clauses INTERROGATIVE CLAUSES BIBLICAL TEXTS NATIVE TEXTS SV VS SV VS TOTAL NOM 4 2 3 3 12 DAT 1 3 4 5 5 3 3 16 There are in total eight occurrences of SV word order in interrogative clauses, of which five stem from biblical passages, while the remaining three are native Gothic sentences. Below, two SV examples are shown, both from the Bible, with a nominative subject in (24a) and a potential dative subject in (24b): Translation of NTG Μὴ καὶ ὑμεῖς πεπλάνησθε (24a) ibai jah jus afairzidai not.at.all also you.NOM.PL deceived.NOM.PL ‘Aren’t you also deceived? (John 7:47) siuþ? are.2PL Translation of NTG τί ὑμῖν δοκεῖ (24b) ƕa izwis þugkeiþ? what you.DAT.PL seems.3SG ‘what do you think?’ (Matthew 26:66) Observe that in (24b) above, the direct object ƕa ‘what’ is found in first position, immediately followed by the potential dative subject and then the finite verb. This word order is not the expected SV word order, as a wh-phrase in initial position should trigger subject-verb inversion (Eythórsson 1995: 104). Instead, the word order in this example directly reflects the word order in the Greek original. Turning to VS word order, there are eight examples with that word order in our dataset, four of which stem from biblical passages and four from native Gothic texts. In three of the four native examples, there is a fronted prepositional object in first position triggering subject-verb inversion. This is shown in (25a), where bi ƕana ‘about whom’ occupies the first position. (25a) bi ƕana qaþ praufetus about who.ACC said.3SG prophet.NOM.SG ‘About whom said the prophet’ (BF, 2r, l. 26) Translation of NTG πάλαι δοκεῖτε ὅτι ὑμῖν ἀπολογούμεθα (25b) aftra þugkeiþ izwis ei sunjoma uns wiþra again think.2PL you.DAT.PL that would.excuse.1PL us.ACC.PL with izwis? you.ACC.PL Again, think ye that we would excuse ourselves unto you? (Cor. II 12:19) The example in (25b) exemplifies a potential dative subject in an interrogative clause, with VS word order. Note that this example deviates from the Greek original, as the subject of δοκεῖτε ‘think’ is pro-dropped in the Greek text. In one of the native Gothic examples the subject and the finite verb occupy the penultimate and the final position in the clause, with the whole verb phrase being topicalized to the beginning of the clause, following the interrogative pronoun. Such a topicalization should result in subject-verb inversion, which does not occur here. 18 (26) ƕaiwa how stojan judge.INF jah and ni NEG stojan sa sama mahtedi? judge.INF the.NOM.SG same.NOM.S could.OPT.3SG G ‘How could the same [one] judge and not judge?’ (Sk. 5, l. 14) This particular order in an interrogative clause seems to be found only in (26) and is, to our knowledge, not found elsewhere in the native Gothic corpus. We now turn to a discussion of differences and similarities across biblical and native Gothic passages in the next section, before we discuss the word order test for potential dative subjects in Section 7. 6. Word-order distribution in biblical passages vs. native Gothic passages The question that now arises is whether there is a clear distinction in the word order between biblical passages and native Gothic, which may be used in a comparison with potential dative subjects. In order to address that issue, and as a part of our methodology outlined in Section 4.3 above, we first discuss word order in the three clause types of negative, interrogative and affirmative clauses in native Gothic passages in Section 6.1, before we proceed to the word order comparison with biblical passages in 6.2. 6.1 Native Gothic The numbers in Table 6 show that the distribution between SV and VS is fairly even in affirmative and interrogative clauses in native Gothic texts. Table 6: The distribution of SV and VS across the three clausal types in native Gothic SV VS N f N f Total AFF 7 54% 6 46% 13 NEG 2 18% 9 82% 11 INT 3 50% 3 50% 6 11 19 30 In contrast, in negated clauses the distribution is heavily skewed in favor of the VS word order. This is, of course expected given that the preferred word order in negated clauses is with the verb immediately following the negation, causing subject-verb inversion. The only example where the subject, and the negation, precede the verb in clauses with negation is found in (22a) above from the Skeireins, repeated here as (26): (26) jabai auk diabulau fram anastodeinai nih nauþjandin if also devil.DAT from beginning not forcing.DAT.SG ‘if then the devil from the beginning not forcing [but deceiving the man]’ (Sk., 1, l. 13-14) In this particular example, the dative is motivated by the fact that this is an absolute construction, which means this is not a dative subject verb in Gothic. Observe that in (26), the negation does not take scope over the clause, but only over the verb, not causing subject-verb inversion. Turning now to the word order found in interrogative clauses in native Gothic texts, the three documented SV clauses are examples where the subject itself is represented by the interrogative pronoun ƕas ‘who’, as is shown in (28) below. 19 (28) ƕas gakannida þus who.NOM made.known.3SG you.DAT.SG ‘who made known to you’ (BF, 2r, l. 23-24) In the three VS examples found in interrogative clauses, prepositional objects have been fronted to first position, resulting in subject-verb inversion. One such example is given in (25a) above, repeated here as (29): (29) bi ƕana qaþ praufetus about who.ACC said.3SG prophet.NOM.SG ‘About whom said the prophet’ (BF, 2r, l. 26) Finally, turning to affirmative clauses, such structures allow both SVO and SOV word orders, as should already be clear from the discussion in Section 4.2 above. Yet, clauses including direct objects are quite rare in our native Gothic dataset; our only example of SOV is given in (30) below, where the accusative object þata ‘that’ immediately precedes the clause-final verb gatawida ‘had done’: (30) bi þamma gamelidin fijands manna þata gatawida through that.DAT.SG written.DAT.SG hating man.NOM that.ACC had.done ‘As is said, the hating man has done that ...’ (BF, 2r, l. 21-23) In addition, both SV and VS orders are found in affirmative clauses in general. The default word order is SV, although VS is also found in Gothic in certain types of structures (cf. Eythórsson 1995 and Falluomini 2018). There is no doubt that VS order is more marked than SV order in the early Germanic varieties (Smith 1971, Hopper 1975, Lehmann 2005), since as we discuss below, VS typically occurs when another constituent is fronted, for instance, due to focus (Barðdal et al. 2013) or narrative inversion (Sigurðsson 1983, 2019, Platzack 1985). As a matter of fact, Barðdal et al. (2013) go as far as to reconstruct a focus position as a clause-initial position in Proto-Indo-European on the basis of data from five different IndoEuropean branches involving predicate focus. Also, as is discussed by Falluomini (2018: 166–167), VS word orders involving V2 appear three times in the Bologna Fragment, in all cases within the same syntactic context, shown in (31a) below: (31a) swaleiks was sa unselja kaein such.NOM.MASC was.3SG the.NOM.MASC evil.NOM Cain.NOM ‘such was the evil Cain’ (BF, 2v, l. 18) Corresponding structures are readily found in Old Norse-Icelandic texts, like for instance in (31b) below: Old Norse-Icelandic (31b) slíkt var þeirra gaman such.NOM.NEUT was.3SG their.NOM amusement.NOM.NEUT ‘such was their amusement’ (Völsunga saga, Ch. 32) Note that Gothic swaleiks and Old Norse-Icelandic slíkt are cognates. Both occur in clauseinitial position, triggering subject-verb inversion. Also Chicosz et al. (2016: 64–75) document that complements may cause subject-verb inversion in translated texts in other early Germanic languages, for instance in Tatian and in OE translations of the gospel of Luke. 20 There is further evidence elsewhere in the Bologna Fragment that various elements naturally occurring in first position give rise to subject-verb inversion. Two such examples from our dataset are the following consecutive clauses (32a–b), introduced by the subjunction inuh þis ‘therefore’: (32a) Inuh þis qiþiþ praufetus without that.GEN says.3SG prophet.NOM ‘Therefore, says the prophet about him’ (BF, 2r, l. 13-14) bi about (32b) Inuh þis qam gamains without that.GEN came.3SG common.NOM ‘therefore, came a common savio of allr’ (Sk. 1, l. 3). allaize all.GEN ina he.ACC nasjands savior.NOM Axel (2007) and Chicosz (2010) also observe that subject-verb inversion is common with verbs of saying in the early Germanic languages, of which one example is given in (33a) below from Old High German: Old High German (33a) quad tho maria zi themo engile said.3SG then Mary.NOM to the.DAT angel.DAT ‘then Mary said to the the angel’ (Tatian, cited from Chicosz 2010: 83) (33b) þu nu þan qaþ you.NOM now then said.3SG ‘now you then said’ (BF., 1v, l. 17) Note also that in both examples in (33), there is a temporal adverb, tho, þan ‘then’, which is well known for also be found in initial position in V2 clauses in other early Germanic languages, like Old English (van Kemenade & Los 2006) and Old Norse-Icelandic (Sigurðsson 2018). However, it is more of a tendency that various elements in first position cause subject-verb inversion, rather than a rule, as is evident from the example in the Bologna Fragment, given in (33b) above, where the subject þu ‘you’ occurs in first position with the same verb, qaþ ‘said’. The last Gothic example to be discussed here involving subject-verb inversion is (34) below, occurring with an adverb in first position. (34) swa wesun sumai so was.3PL some.NOM.PL ‘So there were some (who had a form of godliness)’ (BF, 2v, l. 14-15)6 This type of subject-verb inversion has been documented cross-linguistically with SVO languages and has been subsumed under the broader term of syntactic inversion, for instance by Lambrecht & Polinsky (1997: 8–10). It has also been observed for Old English, for instance, that existential and presentational constructions, like (35) above, are often found with locative inversion (Chicosz et al. 2016: 122–124). 6 Falluomini (2017) observes that this passage contains a quotation from the epistle to Timothy (Tim. II 3:5 ἔχοντες μόρφωσιν εὐσεβείας ‘having a form of godliness’ in the Byzantine Majority Text, also found in the Gothic Bible as habandans hiwi gagudeins ‘id.’). The entire string still counts as native Gothic text, since neither the main clause nor the relative clause are directly translated from the Greek source. 21 6.2 Comparison with Biblical Gothic To summarize our findings so far, the word order patterns observed in the native Gothic material confirm the general view of Gothic word order for the three different clause types being considered here. Both negated and interrogative clauses consistently place the verb before the subject, except of course in cases of interrogative clauses where the interrogative pronoun is also the subject. The default word order in affirmative clauses is SV, except for in configurations where subject-verb inversion is expected anyway, such as when an element is fronted due to focus or when preceded by a subjunction or a sentence adverbial. Let us now compare these findings with word order counts from biblical passages, as shown in Table 7. Interrogative clauses show a fairly even distribution between SV and VS, as is expected, while affirmative clauses deviate from the pattern found in native Gothic texts with an overwhelming number of VS word orders instead of the default SV order. Moreover, the word order in negated clauses in the Bible does not seem to conform to the pattern found in native Gothic, as the lion’s share of the biblical examples show SV word order, as opposed to the expected VS order. In Section 5.3 above, we argue that the SV word order where the negation intervenes between the subject and the verb is most likely native to Gothic, as one example from the Skeireins, a native passage, shows this word order. Also, it turns out that another example from the Gothic Bible, which also shows this word order, deviates from the Greek original in this respect. In addition, all the other older Germanic languages behave in the same manner, since negation is mostly considered a clitic on the finite verb in early Germanic. While there is no doubt that the high proportion of SV order in clauses with negation in the Bible is due to translation effects, the fact still remains that this word order is also native to Gothic. Table 7: The distribution of SV and VS across the three clausal types in in native and biblical passages NATIVE GOTHIC BIBLICAL PASSAGES SV VS SV VS AFF 7 (54%) 6 (46%) 9 (36%) 16 (64%) NEG 2 (18%) 9 (82%) 6 (86%) 1 (14%) INT 3 (50%) 3 (50%) 6 (60%) 4 (40%) These facts warrant two main conclusions. First of all, the observations made in the previous literature were correct in identifying the main trends in Gothic word order. And, second, the syntax of the two sets of examples show clear similarities in their word order distribution (with the exception of a high percentage of SV in negated clauses in the Bible). This suggests, in turn, that the set of native Gothic examples gives an approximate yet realistic insight into actual Gothic syntax. This also means that the order between subjects and verbs found in the Bible counts as legitimate native Gothic syntax, irrespective of the word order in the source text. We realize, of course, that our claims are based on only 72 example sentences and clearly more extensive research is needed to corroborate our assertions. After having now established that the distribution of subjects and verbs found in the Gothic biblical quotations are native to Gothic syntax, the remainder of this article will now deal with the word order distribution of potential dative subjects, in order to assess whether there are any notable differences in terms of word order with nominative subjects. 7 The position of potential dative subjects when deviating from Greek We now turn to the word order properties of potential dative subjects in the Bible. We first discuss, in Section 7.1, a few examples where the order between the dative subject and the verb differs from the Greek original. We then compare pro-dropped subjects in the Greek 22 original with overt dative subjects in the Gothic translation in 7.2. And finally, in Section 7.3 we present one example of a pro-dropped dative subject from the Skeireins. 7.1 Specular position vis-à-vis the Greek source As has already been mentioned above, the main problem when studying the word order distribution of potential dative subjects in Gothic is that they are not found in so-called native Gothic texts, but only within translated biblical passages. Despite this apparent limitation, it is still possible to retrieve some direct evidence of the position of potential dative subjects. Through a comparison between the Gothic text and the Greek source, the original position of Gothic potential dative subject may in fact be observed, i.e. by comparing the word order of the Gothic Bible with that of its assumed source text. In (35a–c) below, the position of the potential dative subject and the verb is the opposite of the one found in the Greek source: Translation of NTG καλὸν αὐτοῖς ἐστιν (35a) goþ ist im good.NEUT is.3SG they.DAT.PL It is good for them [if they remain as I am]’ (Cor. I 7:8) Translation of NTG ἵνα μή τι ἀπόληται (35b) þei waihtai ni fraqistnai that thing.DAT.SG not would.be.lost.3SG ‘[Gather the remaining fragments], that nothing be lost’ (Jh. 6:12) Translation of NTG καλόν ἐστίν σοι (35c) goþ þus ist galeiþan good.NEUT you.DAT.SG is.3SG to.enter.INF ‘it is better for you to enter [in life lame]’ (Mk. 9:45) Starting with (35a), the verbal construction goþ wisan ‘to be good (for sb.)’ is fronted to clauseinitial position, most likely due to predicate focus, as is discussed to some length in Section 6.1 above, resulting in subject-verb inversion. In contrast, in (35b), the potential dative subject waihtai ‘nothing’ occupies the subject position, immediately following the conjunction þei ‘that’. Finally, the position of the potential dative subject in (35c) may be best explained through textual reasons. The expression goþ þus ist occurs three times in three very close passages, namely Mark 9:43, 9:45 and 9:47, as is shown in Table 8 below: Table 8: A comparison between the NTG original and its translation of ‘it is good for you’ in three paragraphs in the Gospel of Mark NTG original Gothic translation (Mk. 9:43) καλόν σοι ἐστὶν goþ þus ist (Mk. 9:45) καλόν ἐστίν σοι goþ þus ist (Mk. 9:47) καλόν σοι ἐστὶν goþ þus ist ‘It is good for you’ ‘It is good for you’ Notice that the Greek word order in Mark 9.45 deviates from the word order in Mark 9:43, which is only two paragraphs above. Thus, it is actually the Greek text in Mark 9:45 that presents a discrepancy in the order of its elements. It is reasonable to assume that Wulfila would have tried to normalize the translation by uniformizing the order of the elements throughout the three paragraphs. 23 To conclude, in two of the examples above, (36a–b), the potential dative subject occupies, on the one hand, the inverted subject position, and on the other hand, the canonical subject position in a subordinate clause, preceding the verb. 7.2 Null subjects in Greek vs. overt dative subjects in Gothic Moving further, another possible way to determine the position of potential dative subjects is to examine in particular the Gothic translations of Greek examples with no overt subject, as in the four examples below. In the first example in (36), the potential dative subject precedes the verb, while in the remaining three examples in (37–39), the order between the subject and the verb is inverted. This is due to clause-initial negation in (38) and narrative inversion in (39). Translation of NTG διὸ [Ø] εὐδοκῶ ἐν ἀσθενείαις (36) in þizei mis galeikaiþ in siukeim ... in that.GEN I.DAT likes.3SG in maladies ‘Because of this I am content in sickness [... for Christ’s sake]’ (Cor. II, 12:10) In (36) the SV order is expected for affirmatives, the Greek verb εὐδοκῶ ‘I take/find pleasure in sth’ in the source texts generally selects for a nominative subject which has been prodropped. Wulfila translated the text with the verb galeikan, here meaning ‘like’, occurring with a potential dative subject and a PP (cf. Eythórsson & Barðdal 2005: 831–832 for an overview of the different argument realizations of galeikan in Gothic). Translation of NTG ... δοκοῦσιν [Ø] γὰρ ὅτι ... (37) swaswe þai þiudo, þugkeiþ im like the pagans seems.3SG they.DAT.PL ‘like the pagans, for they think that …’ (Mt. 6:7) auk therefore ei ... that The verb δοκοῦσιν in (37) is a variant of the contracted transitive verb δοκῶ ‘think’ which only occurs with a nominative subject in Greek, here pro-dropped in the Greek original. The potential dative subject of þugkeiþ, im ‘they’, refers back to þiodo ‘pagans’ in the previous clause, which is one of the properties of narrative inversion, i.e. that the inverted subject is topical and has been mentioned in the previous discourse (cf. Sigurðsson 2019: 263–264). Narrative inversion is further confined to main clauses and is most commonly found with 1p pronouns, then 3p pronouns, while it is least common with non-pronominal arguments. Translation of NTG ... εὐδοκήσαμεν [Ø] καταλειφθῆναι ἐν ἀθήναις (38) in þizei ju ni usþulandans þanamais, in that.GEN yet not tolerating.NOM.PL anymore galeikaida uns ei biliþanai weseima in Aþeinim liked.3SG us.DAT that left.behind were.1PL in Athens ‘since we couldn’t tolerate it anymore, we were content that we were left behind in Athens’ (Thess. I 3:1) In (38) the inverted dative subject uns ‘we’ is again topical, referring to a participant implicitly mentioned in the previous clause, in this case through the present participle usþulandans ‘tolerating’ which is in the 1p plural. The nominative subject of the verb εὐδοκήσαμεν ‘find good’ is also left unexpressed in the Greek original, corresponding here to the inverted dative subject uns ‘us’. 24 Translation of NTG ... καὶ καλῶς ἕξουσιν [Ø] (39) ana unhailans handuns us-lagjand, on the.sick.ACC hands.ACC on-lay.3PL jah waila wairþiþ im and well becomes.3SG them.DAT ‘they shall lay hands on the sick and they shall recover’ (Mk. 16:18) The last example to be discussed here, (39), may be understood as an exhortative clause, functioning as an imperative, which in turn may prompt V1 word order (Miller 2019: 516). In addition to that, the potential dative subject is also topical, while the predicate waila wairþiþ ‘become well’ occurs in clause-initial position. As such, the order between the potential dative subject and the predicate may also be analyzed as an example of narrative inversion. 7.3 Potential dative subjects: Interim Summary and conclusion Table 9 summarizes the frequencies found for potential dative subjects in the examples discussed in this section, involving deviations from the expected word order pattern. The relevant examples are only found in affirmative and negative clauses, thus any comparison with interrogative clauses is left out. Of all seven examples, only one contains a full NP, while the remaining potential dative subjects are pronouns. Of these seven examples, five instantiate the SV word order, while three are found with VS order. Of the VS examples, one is a hortative clause which is known for displaying V1 word order and two exemplify narrative inversion. Four of the SV examples are found in ordinary affirmative clauses where the default is that the subject precedes the verb, while one is in a negated clause (see the discussion of 22b above). Table 9: Word order distribution of nominative and potential dative subjects across affirmative and negated clauses in native Gothic SV NOMINATIVE DATIVE VS AFF 7 NEG 2 AFF 6 NEG 9 TOTAL 24 4 1 3 0 7 Furthermore, Table 9 also includes corresponding numbers for nominative subjects in native Gothic, i.e. from the non-Biblical passages in the Skeireins and the Bologna Fragment. These are 24 in total, with nine examples found with SV word order and 15 with VS word order. We realize, of course, that the numbers for both nominative and potential dative subjects are small in size, but we believe that the quality of the examples outwins their scarcity. Since we have shown above that the examples involving word order distribution of potential dative subjects all deviate from the Greek original, there can be no doubt that they represent native Gothic. There is one clear difference between nominative subjects and potential dative subjects in our dataset, having to do with occurrences in negated clauses. There are proportionally more examples of nominative subjects with VS order in negated clauses than in affirmative clauses, while the numbers are the opposite for potential dative subjects. In order to shed some light on this difference, we would first like to point out that there are only three VS examples with potential dative subjects in our native Gothic material, while the corresponding number for nominative subjects is 15. Negated clauses are, of course, marked structures compared to affirmative clauses and VS word order in affirmative clauses is more marked than SV in affirmatives. Clearly, there are proportionally fewer examples of negated clauses with potential dative subjects than with nominative subjects, something we take to be an artifact of the small size of the corpus and the scarcity of examples with potential dative subjects in 25 general and in the native Gothic material in particular. Therefore, we do not regard this onesidedness of the distribution to be meaningful in any way. In contrast, when it comes to the distribution of SV and VS across affirmative clauses, we believe that matters are different. Of these, four potential dative subjects are SV, while three are VS. Exactly the same tendency is found for nominative subjects, of which seven examples are SV and six are VS. Even though the numbers are small, we still take the distribution of these numbers to suggest that potential dative subjects are of the same syntactic category as nominative subjects, first and foremost because if potential dative subjects were objects, one would not expect to find the same word order distribution for these, as we find for nominative subjects across SV and VS in affirmative clauses. On an object analysis of potential dative subjects, one would certainly not expect 57% topicalizations, as topicalizations are an optional device motivated by information structure. However, on a subject analysis of the dative, its behavior is expected as it mirrors the behavior of nominative subjects in affirmative clauses. In order to corroborate or disconfirm our claims, we have run a chi-square test of the frequencies found for nominative and potential dative subjects in affirmative clauses, in particular their distribution across SV and VS word orders. The results reveal a chi-square statistic of 0.02, with a p-value of .887593, which is far from being significant at the p-level of .05. This lack of significance suggests that there are no greater differences between the distribution of nominative subjects and potential dative subjects, corroborating our claim that potential dative subjects are not syntactic objects, but are instead syntactic subjects. It is well known in statistics that a certain minimum number of observations must be obtained in order for the results to reach the threshold level of significance. Thus, one might now object that, of course, the results are not significant, since they are based on very small numbers, seven in total for potential dative subjects and 13 for nominative subjects, and that one would need a much larger sample for the results to be significant. In order to investigate whether our lack of significance is due to the smallness of the dataset, we have run another chi-square test, where we add one zero to our numbers, computing 70 vs. 60 observations for nominative subjects and 40 vs. 30 for potential dative subjects. The results are quite similar with a chi-square statistic of 0.1998 and a p-value of .654882, which is far from being significant at the p-level of .05. Despite this, one might still object that raising the numbers of observations from 20 to 200 is not enough to yield significance anyway given these proportions. Therefore, we have run the chi-square test for the third time, this time adding two zeros to our numbers, resulting in 2,000 observations being calculated instead of 200. Once more, the chi-square statistic is 1.998 with a p-value of .157507, which is far from being significant at the p-level of .05. Thus, we conclude that the reason that our first chi-square test yielded non-significant results is not because of the low number of observations, i.e. 13 for nominative subjects and seven for potential oblique subjects. The null hypothesis here is that word order variation and variation in subject case marking are not interrelated. The lack of significant difference, documented above, regarding the word order distributions of nominative and potential dative subjects in affirmative clauses means that the null hypothesis cannot be rejected. In other words, there is no relation between variation in case marking and variation in word order. This suggests, as a matter of fact, that potential dative subjects behave syntactically as ordinary nominative subjects do with regard to word order distribution in affirmative clauses in the native Gothic material. 8 Summary and conclusions Earlier research into potential non-nominative subjects in Gothic, involving five subject tests, has in part confirmed their subject behavior. These five tests are clause-bound reflexivization, long-distance reflexivization, raising-to-subject, raising-to-object and control infinitives. Of these tests, no examples involving potential non-nominative subjects are found controlling clause-bound reflexives, or being “raised” to subject or object in raising-to-subject and raising- 26 to-object constructions in Gothic. This is not very surprising given that Gothic is a small-corpus language, with a total number of running text being less than 70,000 words. In any case, potential non-nominative subjects are not found in any abundance in the Gothic material. The absence of raising-to-subject and raising-to-object examples, therefore, does not in and of itself confirm a non-subject analysis of potential non-nominative subjects in Gothic. Instead, it only does not confirm a subject analysis. However, it should be mentioned here that examples of raising-to-subject and raising-to-object are documented in Old English, Old Saxon, Old High German, Old Norse-Icelandic, Old Swedish and Old Danish with potential non-nominative subjects, indeed supporting a subject analysis of potential non-nominative subjects in both West- and North-Germanic. Raising-to-subject and raising-to-object examples involving non-nominative subjects have also been documented in Latin and Ancient Greek, supporting a subject analysis of these in a wider early Indo-European context. In contrast, one example of a potential non-nominative subject being left unexpressed in control infinitives in Gothic, with the verb ‘lust’, has been documented in the earlier literature, showing beyond doubt that potential non-nominative subjects behaved syntactically as subjects in control infinitives. Here we have also presented an example of long-distance reflexivization, where a potential dative subject controls a reflexive in a subordinate clause. Both of these are clear-cut subject properties, speaking for a subject analysis of potential nonnominative subjects in Gothic, and against an object analysis. Turning to our Gothic dataset, it consists of 72 example sentences of which 44 contain verbs selecting for nominative subjects, while 28 select for potential dative subjects. These are divided across three types of clauses, affirmative clauses, negative clauses and interrogative clauses, distributed across both native Gothic passages and translated Biblical passages. The nominative subject verbs, the control set, consists of 14 types, while the dative subject verbs are 14 in total. Affirmative clauses may be either SV or VS, depending on whether the subject occurs in first position, yielding SV, or whether an object or a sentence adverbial occurs in first position, yielding VS. The same is true for interrogative clauses. Negated clauses where the negation is found in clause-initial position generally yield subject-verb inversion, i.e. VS. More than 45 years ago, Ebel (1978) developed a methodology to establish which syntactic structures in the Skeireins represent native syntax and which do not. This methodology consists in contrasting the non-translated parts of the Skeireins with the Bible quotations in the Skeireins. We expand Ebel's (1978) methodology to the Bologna Fragment and we compare the similarities and differences in word order found in the native passages in the Skeireins and in the Bologna Fragment to the Gothic Bible text. Not only does this comparison confirm the tendencies observed in the earlier literature on the order of the subject and the verb, it also shows that the word order constellations found in the Gothic Bible are also found in native Gothic, even though disproportional frequencies and individual word order choices may be due to a translational effect. In other words, the word order in the Gothic Bible represents grammatical choices in the Gothic language. After having established which parts of the Skeireins and the Bologna Fragment qualify as native Gothic, we conduct a comparison between the behavior of nominative subjects in native Gothic with the behavior of dative subjects in the Bible in contexts which count as native for the Gothic language. For this purpose, we scrutinize four examples involving potential dative subjects where the word order in the Gothic Bible is specular to the word order in the Greek Bible and four examples where the subject is pro-dropped in the Greek original, while being expressed in the Gothic text. Thus, these eight examples of potential dative subjects represent clear instances of native Gothic. A further comparison of the behavior of nominative subjects in the Skeireins and the Bologna Fragment with the behavior of potential dative subjects in native Gothic, in particular with regard to word order in affirmative clauses, reveals no statistical differences between the two. Instead, the results of a chi-square test corroborate our claim that potential dative subjects indeed behave syntactically in the same way as nominative subjects do. We thus conclude that our application of the word order test on Gothic material for the first time in the literature indeed corroborates earlier claims that there are non-nominative subjects in the Gothic language. In addition to our scrutiny of Gothic word order, we have also 27 introduced an example of long-distance reflexivization where a dative subject in a main clause binds a reflexive in a subordinate clause, a behavior also confined to subjects in Gothic. These facts, together with the existence of a control infinitive where an accusative subject is left unexpressed, show beyond doubt that there are non-nominative subjects in the Gothic language and that an object analysis of the relevant dative arguments is excluded. References Primary sources BF = Bologna fragment, in Falluomini (2017) GB = Gothic Bible, in Streitberg (1919 [2000]) Sk = Skeireins, in Streitberg (1919 [2000]) Secondary sources Allen, Cynthia L. 1995. Case Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Oxford University Press. Anagnostopoulou, Elena. 1999. On experiencers. In Artemis Alexiadou, Geoffrey Horrocks & Melita Stavrou (eds.), Studies in Greek Syntax, 67–93. Dordrecht: Springer. Anderson, Stephen R. 1976. On the Notion of Subject in Ergative Languages. Charles N. Li (ed.), 1–23. Andrews, Avery. 1976. The VP Complement Analysis in Modern Icelandic. North Eastern Linguistic Society 6. 1–21. Axel, Katrin. 2007. Studies on Old High German syntax: left sentence periphery, verb placement and verb-second. Amsterdam: John Benjamins. Barðdal, Jóhanna. 2000a. Oblique Subjects in Old Scandinavian. NOWELE: North-Western European Language Evolution 37: 25–51. Barðdal, Jóhanna. 2000b. The Subject is Nominative! On Obsolete Axioms and Their DeepRootedness. 17th Scandinavian Conference of Linguistics, ed. by Carl-Erik Lindberg & Steffen Nordahl Lund, 93–117. Odense: Institute of Language and Communication. Barðdal, Jóhanna. 2001. The perplexity of Dat-Nom verbs in Icelandic. Nordic Journal of Linguistics 24: 47–70. Barðdal, Jóhanna. 2006. Construction-Specific Properties of Syntactic Subjects in Icelandic and German. Cognitive Linguistics 17(1): 39–106. Barðdal, Jóhanna. 2023. Oblique Subjects in Germanic: Their Status, History and Reconstruction. Berlin: Mouton de Gruyter. Barðdal, Jóhanna, Valgerður Bjarnadóttir, Serena Danesi, Tonya Kim Dewey, Thórhallur Eythórsson, Chiara Fedriani & Thomas Smitherman. 2013. The Story of ‘Woe’. Journal of Indo-European Studies 41(3–4): 321–377. Barðdal, Jóhanna, Eleonora Cattafi, Serena Danesi, Laura Bruno & Leonardo Biondo. 2023. Non-Nominative Subjects in Latin and Ancient Greek: Applying the Subject Tests on Early Indo-European Languages. Indogermanische Forschungen 128. Barðdal, Jóhanna & Thórhallur Eythórsson. 2003. The Change that Never Happened: The Story of Oblique Subjects. Journal of Linguistics 39(3): 439–472. Barðdal, Jóhanna & Thórhallur Eythórsson. 2012. ‘Hungering and lusting for women and fleshly delicacies’: Reconstructing grammatical relations for Proto-Germanic. Transactions of the Philological Society 110(3): 363–393. Barðdal Jóhanna, Thórhallur Eythórsson & Tonya Kim Dewey. 2019. Alternating predicates in Icelandic and German: A sign-based construction grammar account. Constructions and Frames 11(1): 107–170. Barnes, Michael. 1986. Subject, Nominative and Oblique Case in Faroese. Scripta Islandica 38: 3–35. Benedetti, Marina & Chiara Gianollo. 2020. Criteria for Subjecthood and Non-Canonical Subjects in Ancient Greek. Historical Linguistics 2017: Selected Papers from the 23rd 28 International Conference on Historical Linguistics, San Antonio, Texas, 31 July – 4 August 2017, ed. by Bridget Drinka, 30–48. Amsterdam: John Benjamins. Bennett, William Holmes. 1960. The Gothic commentary on the Gospel of John: skeireins aiwaggeljons þairh iohannen; A Decipherment, Edition, and Translation. New York: Modern Language Association of America. Bickel, Balthasar. 2004. The Syntax of Experiencers in the Himalayas. Non-nominative subjects I, ed. by Peri Bhaskararao & Karumuri Venkata Subbarao, 77–111. Amsterdam: John Benjamins. Bickel, Balthasar. 2011. Grammatical Relations Typology. The Oxford Handbook of Language Typology, ed. by Jae Jung Song, 399–444. Oxford: Oxford University Press. Castro Alves, Flavia de. 2018. Sujeito dativo em Canela [Dative Subjects in Canela]. Boletim do Museu Paraense Emílio Goeldi: Ciências Humanas 13(2): 377–403. Chicosz, Anna. 2010. The Influence of Text Type on Word Order of Old Germanic Languages: A Corpus-Based Contrastive Study of Old English and Old High German, Frankfurt: Peter Lang. Chicosz, Anna, Jerzy Gaszewsky & Piotr Pęzik. 2016. Element Order in Old English and Old High German Translations. Amsterdam: John Benjamins. Comrie, Bernard. 1973. The Ergative: Variations on a Theme. Lingua 32: 239–253. Comrie, Bernard, Diana Forker & Zaira Khalilova. 2018. Affective Constructions in Tsezic Languages. Non-Canonically Case-Marked Subjects: The Reykjavík–Eyjafjallajökull Papers, ed. by Jóhanna Barðdal, Na’ama Pat-El, & Stephen Mark Carey, 55–82. Amsterdam: John Benjamins. Conti, Luz. 2010. Análisis del dativo en construcciones impersonales: los conceptos de sujeto y de semisujeto en griego antiguo [Analysis of the dative in impersonal constructions: The concepts of subject and semi subject in Ancient Greek]. EMERITA Revista de Lingüística y Filología Clásica (EM) LXXVIII(2): 249–273. Craig, Colette G. 1976. Properties of Basic and Derived Subjects in Jacaltec. Charles N. Li (ed.), 99–124. Croft, William. 1998. Event Structure in Argument Linking. The Projection of Arguments: Lexical and Compositional Factors, ed. by Miriam Butt & Wilhelm Geuder, 1–43. Stanford: CSLI Publications. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William. 2012. Verbs: Aspect and Clausal Structure. Oxford: Oxford University Press. Culicover, Peter W. & Ray Jackendoff. 2005. Simpler Syntax. Oxford: Oxford University Press. Dahlström, Amy. 2003. Focus constructions in Meskwaki (Fox). Proceedings of the LFG03 Conference, ed. by Miriam Butt & Tracy H. King, 144–163. Stanford: Center for the Study of Language and Information Publications. Danesi, Serena. 2015. Modality, Subjecthood and Semantic Change: A Case Study on Modal Verbs in Ancient Greek. A talk presented at ICHL 22, Naples. Danesi, Serena, Cynthia A. Johnson & Jóhanna Barðdal. 2018. Where Does the Modality of Ancient Greek Modal Verbs Come From? The Relation between Modality and Oblique Case Marking. Journal of Greek Linguistics 18(1): 45–92. Delbrück, Berthold G. G. 1911. Germanische Syntax II. Zur Stellung des Verbums, Leipzig: Teubner. Dewey, Tonya Kim. 2006. The Origins and Development of Germanic V2. University of California, Berkeley, PhD Dissertation. Dewey, Tonya Kim & Yasmin Syed. 2009. Case variation in Gothic absolute constructions. The Role of Semantic, Pragmatic, and Discourse Factors in the Development of Case, ed. by Jóhanna Barðdal & Shobhana L. Chelliah, 3–21. Amsterdam: John Benjamins. Dunkel, George Eugene. 2014. Lexikon der indogermanischen Partikeln und Pronominalstämme, 2 vols. Heidelberg: Winter. Dryer, Matthew S. 1997. Are Grammatical Relations Universal? Essays on Language Function and Language Type Dedicated to Talmy Givón, ed. by Joan Bybee, John Haiman & Sandra A. Thompson, 117–143. Amsterdam: John Benjamins. 29 Ebel, Else. 1978. Zur Folge SOV in der Skeireins. Sprachwissenschaft 3: 48–82. Eythórsson, Thórhallur. 1995. A verbal syntax in the Early Germanic languages. Cornell University PhD Dissertation. Eythórsson, Thórhallur & Jóhanna Barðdal. 2005. Oblique Subjects: A common Germanic inheritance. Language 81(4): 824–881. Falk, Cecilia. 1993. Non-referential subjects in the history of Swedish. Lund University Doctoral Dissertation. Falluomini, Carla. 2014. Zum gotischen Fragment aus Bologna. Zeitschrift für deutsches Altertum und deutsche Literatur 143: 281–305. Falluomini, Carla. 2015. The Gothic Version of the Gospels and Pauline Epistles: Cultural Background, Transmission and Character. Berlin: De Gruyter. Falluomini, Carla. 2016. Biblical quotations in the Gothic commentary on the gospel of John (Skeireins). Commentaries, Catenae and Biblical Tradition, ed. by Hugh A. G. Houghton, 277–293. Piscataway, NJ: Gorgias Press. Falluomini, Carla. 2017. Zum gotischen Fragment aus Bologna II: Berichtigungen und neue Lesungen. Zeitschrift für deutsches Altertum und deutsche Literatur 146(3): 284–294. Falluomini, Carla. 2018. The position of the verb in Gothic: The contribution of the Bologna fragment. NOWELE: North-Western European Language Evolution 71(2): 161–175. Fedriani, Chiara. 2009. The “Behavior-Before-Coding-Principle”: Further Evidence from Latin. Archivio Glottologico Italiano XCIV: 156–184. Fedriani, Chiara. 2014. Experiential Constructions in Latin. Leiden: Brill. Ferraresi, Gisella. 2005. Word Order and Phrase Structure in Gothic. Leuven: Peeters. Finazzi, Rosa Bianca & Paula Tornaghi. 2013. Gothica Bononiensia: Analisi linguistica e filologica di un nuovo documento. Aevum 87: 113–153. Fourquet, Jean. 1938. L’Ordre des éléments de la phrase en germanique ancien. Paris: Belles Lettres. Harbert, Wayne. 1978. Gothic Syntax: A Relational Grammar. University of Illinois Doctoral Dissertation. Harbert, Wayne. 1991. Binding, SUBJECT, and Accessibility. Principles and Parameters in Comparative Grammar, ed. by Robert Freidin, 29–55. Cambridge, Ma: MIT Press. Harbert, Wayne. 2007. The Germanic Languages. Cambridge: Cambridge University Press. Hermon, Gabriella. 1985. Syntactic Modularity. Dordrecht: Foris. Hock, Hans Henrich & Elena Bashir. 2016. The Languages and Linguistics of South Asia: A Comprehensive Guide. Berlin: De Gruyter Mouton. Hopper, Paul J. 1975. The Syntax of the Simple Sentence in Proto-Germanic. The Hague: Mouton. van der Horst, Joop. 2008. Geschiedenis van de Nederlandse syntaxis. Leuven: Universitaire Pers. Ilioaia, Mihaela. 2021. Non-Canonical Subjects in Romanian: The Status and Evolution of the mihi est Construction. Ghent University Doctoral Dissertation. Ilioaia, Mihaela & Marleen Van Peteghem. 2021. Dative experiencers with nominal predicates in Romanian: A synchronic and diachronic study. Folia Linguistica Historica 42(2): 255–290. Kachru, Yamuna, Braj B. Kachru & Tej K. Bhatia. 1976. The Notion 'Subject': A Note on HindiUrdu, Kashmiri and Punjabi. The Notion of Subject in South Asian Languages, ed. by M. K. Verma, 79–108. Madison: University of Wisconsin. Keenan, Edward L. 1976. Towards a universal definition of subject. Subject and Topic, ed. by Charles N. Li, 303–333. New York: Academic Press. Van Kemenade, Ans & Bettelou Los. 2006. Discourse Adverbs and Clausal Syntax in Old and Middle English. The Handbook of the History of English, ed. by Ans van Kemenade, Bettelou Los, 224–248. Oxford: Blackwell. Kiparsky, Paul. 2011. Did Indo-European have reflexive pronouns? Paper presented at the 23rd Annual UCLA Indo-European conference, Los Angeles, 28–29 October. Klaiman, M. H. 1980. Bengali Dative Subjects. Lingua 51(4): 275–295. Koppitz, Alfred. 1900. Gotische Wortstellung. Zeitschrift für deutsche Philologie 32: 433–463. Koppitz, Alfred. 1901. Gotische Wortstellung. Zeitschrift für deutsche Philologie 33: 7–45. 30 Kotin, Michail L. 2012. Gotisch im (diachronischen und typologischen) Vergleich. Heidelberg: Universitätsverlag Winter. Kristoffersen, Kristian E. 1996. Infinitival Phrases in Old Norse: Aspects of their Syntax and semantics. University of Oslo PhD Dissertation. Kühner, Raphael & Bernard Gerth. 1955. Ausführliche Grammatik der griechischen Sprache: Satzlehre, 4th ed. Hannover: Hahnsche Buchhandlung. Kühner, Raphael & Carl Stegmann. 1955. Ausführliche Grammatik der lateinischen Sprache: Satzlehre, 3rd ed. Leverkusen: Gottschalk. Lambrecht, Knud. 1994. Information Structure and Sentence Form. Cambridge: Cambridge University Press. Lambrecht, Knud. 2000. When Subjects Behave Like Objects: An Analysis of the Merging of S and O in Sentence-Focus Constructions Across Languages. Studies in Language 24: 611–682. Lambrecht, Knud. 2001. A framework for the analysis of cleft constructions. Linguistics 39: 463–516. Lambrecht, Knud & Maria Polinsky. 1997. Typological Variation in Sentence-Focus Constructions. Chicago Linguistics Society 33: 189–206. Landau, Idan. 2009. The Locative Syntax of Experiencers. Cambridge, Ma: MIT Press. Lehmann, Winfred P. 2005. A Grammar of Proto-Germanic, ed. by Jonathan Slocum. Linguistics Research Center, University of Texas at Austin. Books Online: https://lrc.la.utexas.edu/books/pgmc/index. Leiss, Elisabeth. 2007. Covert patterns of definiteness/indefiniteness and aspectuality in Old Icelandic, Gothic, and Old High German. Nominal Determination: Typology, context constraints, and historical emergence, ed. by Elisabeth Stark, Elisabeth Leiss & Werner Abraham, 73–102. Amsterdam: John Benjamins. Le Mair, Esther, Cynthia A. Johnson, Michael Frotcher, Thórhallur Eythórsson & Jóhanna Barðdal. 2017. Position as a behavioral property of subjects: the case of Old Irish. Indogermanische Forschungen 122: 111–142. Li, Charles N. (ed.). 1976. Subject and Topic. New York: Academic Press. Lundquist, Jesse & Anthony D. Yates. 2017. The Morphology of Proto-Indo-European. Handbook of Comparative and Historical Indo-European Linguistics, ed. by Jared Klein, Brian Joseph, Matthias Fritz & Mark Wenthe, 2079–2194. Berlin: De Gruyter Mouton. Masica, Colin. 1976. Defining a linguistic area: South Asia. Chicago: University of Chicago Press. Maßmann, H. F. 1834. Skeireins Aiwaggelions þairh Ïohannen: Auslegung des Evangelii Johannis in Gothischer Sprache: Aus römischen und mayländischen Handschriften nebst lateinischer Uebersetzung, belegenden Anmerkungen, geschichtlicher Untersuchung, gothisch-lateinischem Wörterbuche und Schriftproben. Munich: Verlag von Georg Jaquet. Miller, D. Gary. 2019. The Oxford Gothic Grammar. Oxford: Oxford University Press. Moore, John & David M. Perlmutter. 2000. What Does it Take to Be a Dative Subject? Natural Language and Linguistic Theory 18: 373–416. Orqueda, Verónica. 2019. Reflexivity in Vedic. Leiden: Brill. Pat-El, Na’ama. 2018. The Diachrony of Non-Canonical Subjects in Northwest Semitic. NonCanonical Subjects: The Reykjavík-Eyjafjallajökull Papers, ed. by Jóhanna Barðdal, Na’ama Pat-El & Stephen Mark Carey, 159–184. Amsterdam: John Benjamins. Piras, Antonio. 2007. Manuale di Gotico – Avviamento alla lettura della versione gotica del Nuovo Testamento. Roma: PFTS University Press. Platzack, Christer. 1985. Narrative Inversion in Old Icelandic. Íslenskt mál og almenn málfræði 7: 127–144. Ratkus, Arturo. 2011. The Adjective Inflection in Gothic and Early Germanic: Structure and Development. University of Cambridge PhD Dissertation. Robinson, Maurice A. & William G. Pierpont (eds.). 2005. The New Testament in the Original Greek: Byzantine Textform. Southborough, Ma: Chilton Book Publishing. 31 Rögnvaldsson, Eiríkur. 1991. Quirky subjects in Old Icelandic. Papers from the Twelfth Scandinavian Conference of Linguistics, ed. by Halldór Ármann Sigurðsson, 369–387. Reykjavík: Institute of Linguistics, University of Iceland. Rögnvaldsson, Eiríkur. 1995. Old Icelandic: A Non-Configurational Language?” NOWELE: North-Western European Language Evolution 26: 3–29. Rögnvaldsson, Eiríkur. 1996. Frumlag og fall að fornu [Subject and Case in Old Icelandic]. Íslenskt mál og almenn málfræði 18: 37–69. Rögnvaldsson, Eiríkur. 2007. Reflexives in Older Icelandic. Unpublished manuscript, University of Iceland. Sasse, Hans-Jürgen. 1978. Subjekt und Ergativ: Zur pragmatischen Grundlage primärer grammatischer Relationen. Folia Linguistica 12: 219–252. Sæmundsson, Tómas. 1839. Um bigging jarða, meðferð og úttektir [On the development of landholding, handling and evaluation]. Búnaðarrit Suðuramtsins Húss- og bústjórnar félags 1(1): 139–238. Schachter, Paul. 1976. The subject in Philippine Languages: Topic, Actor, Actor-Topic, or None of the Above. Charles N. Li (ed.), 491–518. Schäferdiek, Knut. 1981. “Die Fragmente der 'Skeireins' und der Johanneskommentar des Theodor von Herakleia”. Zeitschrift für deutsches Altertum und deutsche Literatur 110(3): 175–193. Seefranz-Montag, Ariane von. 1983. Syntaktische Funktionen und Wortstellungsveränderung: die Entwicklung ’’subjektloser‘‘ Konstruktionen in einigen Sprachen. Munich: Wilhelm Fink Verlag. Seefranz-Montag, Ariane von. 1984. ‘Subjectless’ Constructions and Syntactic Change. Historical Syntax, ed. by Jacek Fisiak, 521–553. Berlin: Mouton de Gruyter. Shibatani, Masayoshi. 1999. Dative Subject Constructions Twenty-Two Years Later. Studies in the Linguistic Sciences 29(2): 45–76. Sigurðsson, Halldór Ármann. 1983. Um frásagnarumröðun og grundvallarorðaröð í forníslensku ásamt nokkrum samanburði við nútímamál [Narrative Inversion and Basic Word Order in Old Icelandic with some Comparison with Modern Icelandic]. Reykjavík: University of Iceland Master's Thesis. Sigurðsson Halldór Ármann. 1993. Argument-drop in Old Icelandic. Lingua 89: 247–280. Sigurðsson, Halldór Ármann. 2018. Icelandic declarative V1: a brief overview. Working Papers in Scandinavian Syntax 101: 49–55. Sigurðsson Halldór Ármann. 2019. Topicality in Icelandic: Null arguments and Narrative Inversion. Architecture of Topic, ed. by Valéria Molnár, Verner Egerland & Susanne Winkler, 249–272. Berlin: De Gruyter Mouton. Smith, Jesse Robert. 1971. Word Order in the Older Germanic Dialects. University of Illinois PhD Dissertation. Steever, Sanford B. (ed.) 1998. The Dravidian Languages. London: Routledge. Snædal, Magnús. 2009. Ostgermanische Morphologie / East-Germanic morphology. International conference on morphology and digitisation, ed. by Matthias Fritz & Jost Gippert, 147–167. Prague: Charles University Press. Snædal, Magnús & Christian T. Petersen. 2012. A Gothic fragment of the Old Testament reidentified: Landau vs. Kauffmann. Zeitschrift für deutsches Altertum und deutsche Literatur 141: 434–443. Strahan, Tania E. 2003. Long-Distance Reflexives in Norwegian: A Quantitative Study. Munich: LINCOM. Streitberg, Wilhelm. 1920(2000). Die gotische Bibel, Bd. I: Der gotische Text und seine griechische Vorlage; Mit Einleitung, Lesarten und Quellennachweisen sowie den kleineren Denkmälern als Anhang, 7. Auflage. Mit einem Nachtr. von Piergiuseppe Scardigli. Heidelberg: Universitätsverlag Winter. Talmy, Leonard. 1985. Force Dynamics in Language and Thought. Chicago Linguistic Society 21: 293–337. Talmy, Leonard. 1988. Force Dynamics in Language and Cognition. Cognitive Science 12(1): 49–100. 32 Thráinsson, Höskuldur. 1976. Reflexives and Subjunctives in Icelandic. NELS 6: 225–239. Thráinsson, Höskuldur. 1979. On Complementation in Icelandic. New York: Garland. Van Valin, Robert. 2005. Exploring the Syntax–Semantics Interface. Cambridge: Cambridge University Press. Verma, M. K. & K. P. Mohanan (eds.). 1990. Experiencer Subjects in South Asian Languages. Stanford: CSLI Publications. Witzlack-Makarevich, Alena. 2019. Argument Selectors: A New Perspective on Grammatical Relations: An Introduction. Argument Selectors: A New perspective on Grammatical Relations, ed. by Alena Witzlack-Makarevich & Balthasar Bickel, 1–38. Amsterdam: John Benjamins. Yoon, James H. 2004. Non-Nominative (Major) Subjects and Case Stacking in Korean. NonNominative Subjects 2, ed. by Peri Bhaskararao & Karumuri Venkata Subbarao, 265–314. Amsterdam: John Benjamins. Appendix A: Affirmatives Example Translation Subj. case Passage But God himself ak silba frauja qam came to save du nasjan unsis us NOM BF, 1r, l. 13 swa auk jah pawlus So speaks also qiþiþ Paul NOM BF, 1r, l. 13-14 Allai ushniwun samana Inuh þis qiþiþ praufetus bi ina All turned aside Because of this I then spoke about all of his good deeds This one also then said [he] shall save his people from their sins You then said [...] And Luke advised about this [...] For this reason speaks the prophet about him qaþ unfroþs in hairtin seinamma sa sama ist jah unsibjis Said the fool in his heart it is the same as the impious Inuh þis ik þan qaþ in allaim wailadede is sa auk þan qaþ ganasjiþ managein seina af frawaurhtim ize þu nu þan qaþ bi þanei jah lukas insok Biblical quotation Word-Order n/a SV NOM BF, 1r, l. 22 n/a SV πάντες ἐξέκλιναν (Psalm 14:3, Rom. 3:12) SV NOM BF, 1r, l. 23 n/a SVO NOM BF, 1r-1v, 26-1 n/a SV NOM BF, 1v, l. 17 n/a SV NOM BF, 1v, 24 n/a OSV NOM BF, 2r, l. 13-14 VSO NOM BF, 2r, 18 n/a Εἶπεν ἄφρων ἐν καρδίᾳ αὐτοῦ (Psalm 14:1) NOM BF, 2r, l. 19 n/a PredVS VS 33 bi þamma gamelidin, fijands manna þata gatawida þinsan du diabaulu As it is said, the hating man has done that, [to] attract the devil NOM BF, 2r, l. 21-23 jah gatawida guþ þana mannan And God created the man NOM BF, 2r, l. 26 NOM BF, 2v, l. 7 n/a SOV καὶ ἐποίησεν ὁ θεὸς τὸν ἄνθρωπον (Gen. 1:27) VSO ἡ παρθένος ἐν γαστρὶ ἕξει (Mt. 1:23) SV NOM BF, 2v, l. 18 n/a PredVS NOM BF, 2v, l. 19-20 n/a PredVS NOM BF, 2v, l. 14-15 n/a VS NOM Sk. 1, l. 3 VS DAT Lk. 18:22 DAT Mk. 10:21 azitizo ist ulbandau þairh þairko neþlos galeiþan It is easier for a camel to go through the eye of a needle DAT Mk. 10:25 in þizei mis galeikaiþ in siukeim, in anamahtim, in nauþim, in wrekeim, in þreihslam faur Xristu galeikaida guda þairh þo dwaliþa þizos Because of this I am content in sickness, in reproaches, in necessities, in persecutions, in distresses for Christ's sake DAT God liked the foolishness of the message to DAT See, the maiden bears in the whomb Such was the evil Cain It was similar to galeiks was the devil the diabulau sa afguda ungodly farao pharaoh So there were swa wesun sumai some who had þaiei habaidedun a form of hiwi gagudeins godliness For this reason inuh þis qam came a gamains allaize common nasjands saviour of all nauh ainis þus wan and yet you ist lack one thing you lack one ainis þus wan ist thing Sai, magaþs in kilþein ganimiþ swaleiks was sa unselja kaein Cor. II 12:10 Cor. I 1:21 n/a ἔτι ἕν σοι λείπει ἕν σοι ὑστερεῖ εὐκοπώτερ όν ἐστιν κάμηλον διὰ τῆς τρυμαλιᾶς τῆς ῥαφίδος εἰσελθεῖν διὸ εὐδοκῶ ἐν ἀσθενείαις, ἐν ὕβρεσιν, ἐν ἀνάγκαις, ἐν διωγμοῖς καὶ στενοχωρία ις, ὑπὲρ χριστοῦ εὐδόκησεν ὁ θεὸς διὰ τῆς μωρίας OSV OSV VS SV VSO 34 wailamereinais ganasjan þans galaubjandans save the believer τοῦ κηρύγματο ς σῶσαι τοὺς πιστεύοντα ς galeikaida jah mis jah ahmin weihamma [...] þus meljan both I and the Holy Spirit were content [...] to write to you DAT Lk. 1:3 In þizei ju ni usþulandans þanamais, galeikaida uns ei biliþanai weseima in Aþeinim ainai Since we couldn’t resist anymore, we were content that we were left behind alone in Athens DAT Thess. I 3:1 þatuh ganah unsis that pleases us DAT Jh. 14:8 goþ ist unsis her wisan It is good for us to be here DAT Mk. 9:5 goþ þus ist it is better for hamfamma in libain thee to enter galeiþan into life maimed DAT It is good for goþ ist im, jabai them, if they sind swe ik. remain as I do DAT goþ ist imma mais ei galagjaidau asiluqairnus ana halsaggan is it is better for him that a millstone were hanged about his neck Cor. I 7:8 DAT Mk. 9:42 DAT Mk. 9:45 it is good for you to enter the kingdom of God one-eyed DAT Mk. 9:47 it is better for goþ þus ist galeiþan you to enter in libain haltamma lame into life goþ þus ist haihamma galeiþan in þiudangardja gudis Mk. 9:43 ἔδοξεν κἀμοὶ [...] σοι γράψαι διὸ μηκέτι στέγοντες εὐδοκήσαμ εν καταλειφθῆ ναι ἐν ἀθήναις μόνοι καὶ ἀρκεῖ ἡμῖν καλόν ἐστιν ἡμᾶς ὧδε εἶναι καλόν σοι ἐστίν κυλλὸν εἰσελθεῖν εἰς τὴν ζωὴν καλὸν αὐτοῖς ἐστιν καλόν ἐστιν αὐτῷ μᾶλλον εἰ περίκειται λίθος μυλικός περὶ τὸν τράχηλον αὐτοῦ καλόν ἐστίν σοι εἰσελθεῖν εἰς τὴν ζωὴν χωλὸν καλόν σοι ἐστιν μονόφθαλμ ον εἰσελθεῖν εἰς τὴν VS VS SVO VS SV VS VS SV SV 35 batizo ist auk þus it is also better for you DAT Mt. 5:29 batizo ist izwis ei ik galeiþau it is better for you that I go away DAT Jh. 16:7 sah þan siponeis was kunþs þamma gudjin and at that time the disciple was known to the priest DAT Jh. 18:15 raþizo allis ist ulbandau þairh þairko neþlos þairhleiþan For it is easier for a camel to go through a needle's eye DAT Jh. 18:25 ana unhailans handuns uslagjand, jah waila wairþiþ im they shall lay hands on the sick, and they shall recover DAT Mk. 16:18 for they think that they shall þugkeiþ im auk ei in be heard for filuwaurdein seinai their many andhausjaindau. words DAT Mt. 6:7 Because [he] was close to bi þatei nehva Jerusalem, and Iairusalem was, jah it seemed to þuhta im ei suns them that the skulda wesi kingdom of God þiudangardi gudis should appear gaswikunþjan soon DAT Lk. 19:11 βασιλείαν τοῦ θεοῦ συμφέρει γάρ σοι συμφέρει ὑμῖν [dat] ἵνα ἐγὼ ἀπέλθω ὁ δὲ μαθητὴς ἐκεῖνος ἦν γνωστὸς τῷ ἀρχιερεῖ εὐκοπώτερ ον γάρ ἐστιν κάμηλον διὰ τρυμαλιᾶς ῥαφίδος εἰσελθεῖν ἐπὶ ἀρρώστους χεῖρας ἐπιθήσουσι ν καὶ καλῶς ἕξουσιν δοκοῦσιν γὰρ ὅτι ἐν τῇ πολυλογίᾳ αὐτῶν εἰσακουσθή σονται διὰ τὸ ἐγγὺς αὐτὸν εἴναι Ἱερουσαλή μ, καὶ δοκεῖν αὐτοὺς ὅτι παραχρῆμα μέλλει ἡ βασιλεία τοῦ θεοῦ ἀναφαίνεσθ αι VS VS VS VS VS VS VS 36 Appendix B: Negatives Example Passage Biblical passage Word-Order NOM BF, 1r, l. 11 n/a NegVS NOM BF, 1v, l. 5-6 n/a ComplNegVS NOM BF, 2r, 16-18 n/a NegVS NOM BF, 2r, l. 23 n/a NegVS NOM BF, 2r, l. 26 n/a NegVS NOM BF, 2v, l. 6 n/a NegVS NOM BF, 2v, 9-10 n/a NegVS NOM Sk., 1, l. 13-14 n/a NOM Sk. 2, l. 8-9 þammuh þan ni froþ Nekaudemus ni nauhþanuh galagiþs was in karkarai Iohannes iþ sa afar mis gagganda swinþoza ‘John was not yet cast into prison’ NOM ‘but the one who will come after NOM nasei mik frauja, unte nist saei nasjai ufar þuk, frauja þeinaizos naseinais nist wokrs nih fairlet þan qaþ þatei7 ni sijai guþ swe allai hausideduþ praufetu insakan jah qiþan jabai nist g(u)þ ƕas gataih þus þata namo jabai nist guþ bi ƕana [q]aþ airus jabai nist guþ bi ƕana qaþ esaeia? jabai nist guþ bi ƕana qaþ praufetus jabai auk diabulau fram anastodeinai nih nauþjandin, ak uslutondin mannan 7 Subj. case Translation ‘Save me, O Lord, for there is not who saves more than you, O Lord’ ‘of your salvation there is no usury nor desolation’ ‘Then said: it would not be God as you all have heard the prophet describe and say’ If there is no God, who announced the name to you? If there is no God, about whom talked the messenger? If there is no God, about whom talked Esaiah? ‘If there is no God, about whom talked the prophet’ ‘For inasmuch as the Devil from the beginning had not forced the man but had deceived him’ ‘Nicodemus did not understand this then’ SNegVO n/a ONegVS οὔπω γὰρ ἦν βεβλημένος εἰς τὴν φυλακὴν ὁ Ἰωάννης Sk. 3, l. 2 (Jh. 3:24) NegVS οὗ ἐγὼ οὐκ Sk. 3, l. 25-26 εἰμὶ ἄξιος SNegV The use of þatei as particle for introducing direct speech is also found in the Gothic Bible as a direct translation of the so-called ὅτι recitativum (Piras 2007: 246). 37 mis ist, þizei ik ni im wairþs ei anahneiwands andbindau skaudaraip skohis is me is mightier than me, of whom I am not worthy that I should stoop and unbind the lachet of his sandal’ ἵνα λύσω αὐτοῦ τὸν ἱμάντα τοῦ ὑποδήματος (Jh. 1:27) NOM οὐδὲ γὰρ ὁ Πατὴρ κρίνει οὐδένα (Jh. Sk. 5, l. 11-12 5:22) SNegVO τούτῳ ὑμεῖς οὐ πιστεύετε Sk. 6, l. 25 (Jh. 5:38) OSNegV NOM Sk. 5, l. 14-15 n/a ‘Neither does the nih þan atta ni stojiþ Father judge any ainnohun man’ NOM þammuh jus ni galaubeiþ ni auk þatainei namne inmaideins twaddje andwairþje anþarleikein bandweiþ, ak filaus mais fimf þusundjos waire inuh qinons jah barna swe at mikilamma nahtamata anakumbjandans at ni wisandin aljai waihtai ufar þans fimf hlaibans jah twans fiskans ‘And you don’t believe him’ For not only the change of names signifies the difference of the two persons, but much more five thousand men, as well as women and children - as if sitting down to a great supper, there being nothing but the five loaves and two fish ni þauh im agljai ‘they will not be injured’ þei waihtai ni fraqistnai jah þamma fairnjin ni gatimid þata af þamma niujin ‘that nothing be lost’ ‘and to the old [piece] is not appropriate that [piece] from the new [cloth]’ (my translation) DAT (absolute) [NOM] Sk. 7, l. 10-11 n/a οὐ μὴ αὐτοὺς βλάψῃ (Mk. DAT Mk. 16:18 16:18) ἵνα μή τι ἀπόληται DAT Jh. 6:12 (Jh. 6:12) καὶ τῷ παλαιῷ οὐ συμφωνήσε ι τὸ ἀπὸ τοῦ καινοῦ (Lk. DAT Lk. 5:36 5:36) NegSV NegVS NegConjSV SNegV SNegVO 38 Appendix C: Interrogatives Subj. case DAT Passage Mk. 10:24 Are we permitted to give payment to Caesar or not? what do you think? DAT Lk. 20:22 DAT Mt. 26:66 Again, think ye that we excuse ourselves unto you? Who shall free me from the body of this death? But who is the God who would pull you out of my hands? how could this same one judge and not judge? but what are they among so many? DAT Cor. II 12:19 πάλαι δοκεῖτε ὅτι ὑμῖν ἀπολογούμεθα; VS NOM BF, 1r, l. 1415/Rom. 7:24 VS NOM BF, 2v, l. 2425/Dan. 3:15 τίς με ῥύσεται ἐκ τοῦ σώματος τοῦ θανάτου τούτου; καὶ ποῖος θεὸς ἐξελεῖται ὑμᾶς ἐκ τῶν χειρῶν μου; NOM Sk. 5, l. 14 n/a SV NOM Jh. 6:9/Sk. 7, l. 6 SV Are you also seduced? Does our law judge a man [...]? NOM Jh. 7:47 NOM Sk. 8, l. 21/Jh. 7:51 Are you from Galilee, too? ‘Who announced to you the name’ ‘who made known to you’ ‘About whom said the messenger’ ‘About whom said Esaiah’ ‘About whom said the prophet’ NOM Sk. 8, l. 26/Jh. 7:52 BF, 2r, l. 23 ἀλλὰ ταῦτα τί ἐστιν εἰς τοσούτους; Μὴ καὶ ὑμεῖς πεπλάνησθε; μὴ ὁ νόμος ἡμῶν κρίνει τὸν ἄνθρωπον ἐὰν μὴ ἀκούσῃ παρ' αὐτοῦ καὶ γνῷ τί ποιεῖ; μὴ καὶ σὺ ἐκ τῆς γαλιλαίας εἶ; n/a n/a SV NOM BF, 2r, l. 2324 BF, 2r, l. 26 n/a VS NOM BF, 2r, l. 26 n/a VS NOM BF, 2r, l. 26 n/a VS Example hvaiwa aglu ist þaim hugjandam afar faihau in þiudangardja gudis galeiþan Translation how hard is it for them that trust in riches to enter into the kingdom of God skuldu ist unsis kaisara gild giban þau niu ƕa izwis þugkeiþ? aftra þugkeiþ izwis ei sunjoma uns wiþra izwis? ƕas mik lauseiþ us þamma leika dauþaus þis[?] Iþ ƕas ist g(u)þ saei usþinsai izwis us handum meinaim[?] ƕaiwa stojan jah ni stojan sa sama mahtedi? akei þata ƕa ist du swa managaim? ibai jah jus afairzidai siuþ? ibai witoþ unsar stojiþ mannan [...]? ibai jah þu us Galeilaia is? ƕas gataih þus þata namo ƕas gakannida þus bi ƕana qaþ airus bi ƕana qaþ esaeias bi ƕana qaþ praufetus View publication stats NOM NOM Biblical passage πῶς δύσκολόν ἐστιν τοὺς πεποιθότας [acc] ἐπὶ χρὴμασιν εἰς τὴν βασιλείαν τοῦ θεοῦ εἰσελθεῖν ἔξεστιν ἡμῖν καίσαρι φόρον δοῦναι, ἢ οὔ; τί ὑμῖν δοκεῖ; WordOrder VS VS SV VS SV SV SV SV