ABSTRACT. The present study is motivated by significant changes in patterns of naming among the diasporic community of Indians in South Africa. This community is now into the 6th generation with different degrees of language maintenance and shift occurring in the sub-communities, whose ancestral languages include the Dravidian languages Tamil and Telugu and the Indic languages (aka Indo-Aryan) Bhojpuri-Hindi, Konkani, Meman, Gujarati and Urdu. In keeping with some of the themes of ICHL 23, changes in personal names will be shown to give a strong indication of the relative roles of prestige versus non-dominant speaker practices, and change at the individual versus systemic levels. The first part of the paper will give a brief overview of how descendants of labourers fashioned a naming system that mediated between the varied village-based practices of rural India of the 19th C and the bureaucratic expectations of a western colonial system. This gave rise to what may be thought of as ‘standard’ first name-surname sequences which characterised the community members of the 2nd and 3rd generation up to the 1980s: e.g. Namasivayan Govender (Tamil) or Biharilal Maharaj (Bhojpuri-Hindi). The present project examines subsequent changes in first names emanating subsequently. These changes involve (a) Sanskritisation; (b) prestige adoptions of names of Indian film stars and elites; (c) Westernisation; (d) adoption of names of global stars; (e) neologisms. The neologisms are of particular interest insofar as some involve a “Sanskrit-sounding” or “Indian-sounding” name (e.g. Krijay or Prenolan). These appear to result from contact between different sources and are giving rise to new suffixes which signify in these two cases ‘modern Indian South African of South Indian background’. An empirical analysis of over a thousand names culled from high school records in South Africa attempts to uncover the layerings of tradition, prestige, innovation and modernity. A variationist analysis using R is particularly useful in showing the differential effects of social factors like religion, class and ancestral language. More subtle effects pertaining to changing community identities and ideologies are explored by more qualitative means.

ABSTRACT. Language at the macro scale is a composite of the micro-scale choices made by individual speakers in production. Thus language change must reflect a shift in these choices. This talk presents a simple account of representational production based on past experience, and looks at forces which can lead to (a) production choices departing experience, and thus (b) change in the language. The discussion is based on a range of analyses and experiments undertaken within the field of Experimental Semiotics [1].

A Probabilistic Model of Language Use

A minimal stochastic model of language - with allowances for bilingualism - sees a language user’s experience as made of 4-tuples. Each tuple (l,s,f;p) identifies: the language l being used at the time, the semantics s being expressed, the construction f which encodes that aspect of the semantics, and the relative frequency p with which the l−s −f combination occurs. The model assumes that speakers have collected their entire linguistic experience, marked up with one or more tuples describing each linguistic event.

In production, speakers access this totality of experience and use it to determine the likelihood with which they should use a particular form f to express a meaning s when in a language context l. If the distribution over produced constructions is always the same as speakers have garnered from experience, then the language continues unchanged. Individual speaker does not change the distribution over language-meaning-form tuples. Where a consistent bias means that the forms are selected for production out of proportion to the frequency with which they occur in speaker experience, then language change is underway.

Trends vs Individual Differences

Biases may apply in individuals or in sub-populations without resulting in language change. In most linguistic communities, male adults speak with a substantially lower F0 than they hear (on average), and female adults speak with a substantially higher. However, their balanced numbers mean that the net effect on the language is minimal. Likewise, individual differences resulting from noise in exposure to linguistic input, or in their output mechanisms, may lead speakers to show quite distinct distributions in selected forms. However, if these differences vary from individual to individual randomly then they are likely to balance out, and so the net effect is again negligible.

Potential atoms of language change include biases that result in consistent misreflection of experienced inputs in constructed output. Experimental semioticians have sought to explore these biases through experiments.

Learnability Bias

In 2008, Kirby et al. [2] reported that - in the absence of a functional pressure for communicative distinctiveness - representational systems in an interated learning task shifted from being richly distinctive to systems with very few distinct forms. They argue that the systems were being selected to be more learnable, with the resulting lack of distinctiveness reflecting the systematic underspecification found in natural languages.

The important point here is that, in the absence of functional needs, representational systems shift towards more learnable systems at the expense of faithfully reproducing their input systems.

Useability Bias

In contrast, Fay & Ellison [3] show that useability can trump learnability as a bias when communicative functionality is important. Analysing the result of a pictionary-like communication identification game, they show that effort by the representation producer is minimised consistently, regardless of the fact that this makes signs more difficult to associate with their meaning. However, of two equally easy-to-produce forms, the more learnable one was prefered.

This drive for useability in micro-level selection mirrors the traditional linguistic notion that language change may be motivated to a wish to maximise ease of articulation.

Egocentric Bias

The experience bank of a language user likely contains more data from other people’s productions than from the user’s own productions. For example, language users get to overhear conversations in which they are not participants.

However, Tamariz et al. [4] have shown - in an experiment where the experience of each participant with the representational system could be followed - that producers strongly weigh their own productions beyond those of others. In other words, producers are more likely to copy themselves than someone else.

If this bias is active in natural language scenarios, we should expect that language changes propagate speaker-by-speaker. Speakers will more readily accept an innovatory form if they have never expressed that meaning in the past: if the concept is new, or if they are young. In contrast, speakers who have expressed the concept in the past are more likely to stick with the form that they used before, even if they have encountered others using a different form more often. This is the egocentric bias.

Content and Systematicity Biases

The egocentric bias can, however, be overridden by other biases. In the same paper [4], Tamariz et al. showed that departures from the egocentric bias could be explained by a content bias. When a participant encountered a representation intrinsically superior to what they had used in the past, they adopted the new representation as a replacement their own. Intrinsic superiority of representations for a concept was a parameter in the model. In this experiment, the content bias applied when a sign was both useable (easy to produce) and efficient (easy to learn). A further paper [5] argues that participants switched to signs that showed more retained iconicity - for no more effort, the sign was better at reflecting its meaning. In natural languages iconicity - a sign’s ability to be interpretable at first sight - reflects its embedding in a rich morphological system. While you have probably never encountered the word undevouredness before, but for English speakers its meaning is transparent. The bias identified by Tamariz et al. suggests that compositional multimorphemic forms will be prefered to a monomorphemic form of (crucially) equal complexity. This offers a possible explanation for the progressive shift of verbs in English from strong (irregular) to weak (regular) formations.

Support for this interpretation of the content bias can be found in an experiment looking at systematicity [6]. In this experiment, participants interacted representing six distinct concepts to each other (graphically). After a period, 6 more were introduced, each closely related to one element of the previous set, then 6 more. The results showed a strong bias to reuse components of the representation of one meaning to carry a related similar meaning with a suitable mutation of the representation to mark the distinction. The results of this experiment suggest that form selection is biased towards systematicity - similar meanings having similar forms, i.e. the development of regular morphology.

Task Improvement Bias

In work in preparation, Fay et al. [7] show that linguistic forms and structures can undergo cumulative cultural evolution to better solve particular problems, in this case, path reproduction on a map. Participants were organised in chains, with each participant being instructed to follow a path on the map from their predecessor in the chain, then instructing their success in following another path. The only thing shared between one participant and the next was their shared linguistic experience in solving the task once. When each route-interpreter became a route-producer, they did not merely reflect the frequencies of terms used in their past experience, but adapted those frequencies to improve their ability to solve the task with the new participant. This task adaptation bias affecting natural languages is seen in the universal tendency to create technical jargons.

The Anti-Doppel Bias

Finally, Ellison & Miceli [8] have shown a bias in bilinguals against doppels. In their terminology, a doppel is a word recognisably similar in form and meaning across two or more languages - and corresponds to the psycholinguistic and applied linguistic use of the term cognate. They show that, compared to monolinguals (who have no means of distinguishing doppels from non-doppels), bilinguals use doppels less frequently when there is an alternative. This bias is attributed to the action of a language monitor, which ensures that speakers only use word forms from the language they intend to use. Doppels result in an ambiguous response from the language monitor, which in turn leads speaks to (sometimes) choose an alternative if one exists. Over long periods, this bias alone could account for the rapid lexical divergence of closely related languages.

Conclusion

Consistent biases in how speakers select linguistic forms from their internalised linguistic experience will result in language change. Experimental semiotics, looking at biases in the development and use of communication systems, has identified - amongst others - the above biases in form selection. As unitary mechanisms at the level of individual sign selection, these biases are ideal atoms of language change.

References

[1] Bruno Galantucci and Simon Garrod. Experimental Semiotics: A Review. Front Hum Neurosci, 5, February 2011. [2] Simon Kirby, Hannah Cornish, and Kenny Smith. Cumulative cultural evolution in the laboratory: An experimental approach to the origins of structure in human language. PNAS, 105(31):10681–10686, August 2008. [3] Nicolas Fay and T. Mark Ellison. The Cultural Evolution of Human Communication Systems in Different Sized Populations: Usability Trumps Learnability. PLoS ONE, 8(8):e71781, August 2013. [4] Monica Tamariz, T Mark Ellison, Dale J Barr, and Nicolas Fay. Cultural selection drives the evolution of human communication systems. Proceedings of the Royal Society B: Biological Sciences, 281(1788):20140488, 2014. [5] Nicolas Fay, T. Mark Ellison, and Simon Garrod. Iconicity: From sign to system in human communication and language. Pragmatics & Cognition, 22(2):244–263, 2014. [6] T. Mark Ellison, Michael Wilson, and Nicolas Fay. From icon to system. in preparation. [7] Nicolas Fay, T. Mark Ellison, Riccardo Fusaroli, Kristian Tyĺén, and Simon Garrod. The cumulative cultural evolution of a language game. in preparation. [8] T. Mark Ellison and Luisa Miceli. Language monitoring in bilinguals as a mechanism for rapid lexical divergence. Language, forthcoming.

ABSTRACT. L2 acquisition of Old French structural dative as a trigger for the English recipient passive

This paper investigates the rise of the indirect or, following Allen (1995), “recipient” passive (RP) in English which emerged in the fourteenth century. An example in Present Day English is (1)a, where the recipient is moved into the subject position, as opposed to (1)b, where the Theme is moved.

(1) a. Mary was given presents (by Tom)

b. Presents were given to Mary (by Tom)

On the basis of Middle English data from the York-Helsinki Parsed Corpus of Early English Correspondence (PCEEC) it will be shown that the RP emerged first with ditransitive French verbs. The frequency of occurrence of RP with the French loan verbs pay, promise, offer, allow, deny, serve, fine is 11.4%, but only 0.13% with the native verbs send, give, tell, show. The puzzling fact is that Old French (OF) did not have the RP. This excludes simple grammatical replication (Heine and Kuteva, 2005) and calls for

a) a more theoretical account concerning the structure that was borrowed;

b) an explanation that places the competence of bilingual speakers at the centre of the analysis, given the fact that we can safely assume that large parts of the population spoke Anglo-French (also called “Anglo- Norman”) as well as Middle English (Ingham, 2010).

The first part of the talk presents the historical data in French and English, and develops the theoretical explanation of the “recipient dative puzzle”. It is based on the analysis of the dative as a mixed Case that crosslinguistically surfaces in the form of either lexical, inherent or structural Case (cf. e.g. Zaenen and Maling, 1990, Harley, 1995, Woolford, 2006). Only structural Case can be subject to the Dative-Nominative alternation yielding the RP.

In this perspective, the assumption of a new type of structural ‘French loan dative’ that developed under contact with French can account for the fact that French verbs appear in Middle English RP constructions long before the native verbs do. This led to a system in Middle English where the residues of the Old English lexical/inherent dative co-existed with the structural French ‘dative’. The fact that in Old English, the fronted dative argument retained its case marking is an indicator of lexical or inherent Case. In OE the to-dative construction already existed (cf. de Cuypere, 2014) but passivisation of active sentences with these datives never led to the DAT-NOM alternation (see also Allen, 1995).

As is well known, in ME, a large amount of lexical items were borrowed from OF, including many verbs of actual or future change of possession and transfer of communication. OF distinguished two morphologically marked cases: Nominative and Oblique. The former Latin dative was normally expressed by PPs with a governing an oblique DP. The datives of OF ditransitive verbs that were borrowed into English and express the RP are instances of structural Case.

A first argument supporting the assumption of a ‘French loan dative’ having properties of structural Case is the fact that Old French ‘dative’ a-PPs are pronominalised by dative clitics (e.g. OF li = ModF lui) as in (2)). This distinguishes a as a case marker from a as a locative preposition, the latter being pronominalised by i (see Zaring 1991 on this point in ModF).

(2) se li donnoit on plus tere (< se donnait on plus tere a home) so CLITDAT gave one more land (so gave one more land to man) So one gave him more land. (< So one gave more land to the man.) (La Conqueste de Constantinople de R. de Clari, p.103)

This argument is strengthened by Troberg’s (2008) diachronic analysis of OF three-place and two-place dative verbs. She claims that the relevant property of the dative arguments is the particular relation between the direct and the indirect object. The same relation holds between the arguments of two-place change predicates expressing change of location (go), relation (resemble) or psychological states (please) Troberg (2008: 166170). She considers these cases as ’“structurally” licensed’, and as different from reflection or interaction verbs (remédier ‘remedy, rectify’, aider ‘help’).

The first part will show that in the 14th c., when the influence of French was strong, and when ME had lost the morphological Accusative/Dative case distinction, French verbs prompted a reanalysis of “Datives” in terms of structural Case.

The second part of the talk will discuss the “recipient passive puzzle” in the context of theories of language contact and language acquisition. Current theories of language contact and change give a major role to change arising from imperfect L2 acquisition (McWhorter, 2007; Meisel, 2011; Trudgill, 2011), that is, where language is acquired in the post-critical period by demographically numerous or socially influential population groups. Johanson’s (2002) model of code copying provides an appropriate frame for this situation. In his terms, French is the “sociolinguistically-dominant secondary code”. The borrowing of French verbs can be seen as an instance of global copying, and the use of to modelled on à as a case marker as an instance of semantic-combinational copying: Johanson cites analogous cases of Guernsey English observed by Mari Jones. Johanson acknowledges “that typological distance between the two codes may favour restructuring [...]. However, this does not necessarily mean that the differences are the result of ’imperfect learning’ ” (ibid, 297).

In the second part, we will contend that speakers using French verbs were probably L2 speakers of French, and that they were aware of two facts: a) French verbs were foreign. This is the obvious point, since these speakers had a competence of French, and were the actors in the borrowing process.

b) The argument realisation of the borrowed verbs was not equivalent to native verbs, so that it could not be readily copied. One argument in favour of this point is that PPs with to that were copied from French à were perceived as different from “native” PPs. The second claim is based on the idea that L2 learners often encounter pronominal rather than lexical arguments. In this respect, ME is quite different from OF, where object pronouns are preverbal clitics. According to Haeberli (2014) the preverbal placement of object pronouns is frequent in ME until the middle of the 14th century (43-46%), but “drops abruptly at the end of the century”. In order to know which argument was internal, speakers were bound to distinguish between the two preverbal clitics, as in ModF Jean le lui paie, lit. ‘he it her/him pays’.

The L2 competence that best matches the data—and explains the recipient passive puzzle—is one that is good enough to borrow and use French verbs, and also to be aware of the differences between French and native constructions. This explains why the RP was formed first and predominantly with French verbs. The explanation of why the RP devoloped at all is less straightforward. The talk will discuss the following options: (a) The degree of perfection of L2 speakers was such that they wrongly assumed that French allowed a recipient passive, and therefore used it with French verbs (and only with those), (b) another possibility is that they confused the two objects of French ditransitives (and only of those), and therefore promoted the wrong argument to the subject position (work on L2 acquisition of clitics may indicate the source of this confusion). (c) At the time when these L2 learners were confronted with OF input including clitics they could still activate categories and properties non-existent in their L1 although we would expect to find inconsistent acquisition patterns for placement, accusative and dative Case (cf. Santoro 2007). Thus, the structural Case feature could have licensed RP constructions. This would imply a scenario where the use of French verbs is assimilated with the use of an L2.

Allen, C. 1995. Case Marking and Reanalysis: Grammatical Relations from Old to Early Modern English. Oxford: Oxford University Press.

de Cuypere, L. 2014. “The Old English to-dative construction.” English Language and Linguistics 19.1: 1–26.

Haeberli, E. 2014. “When English Meets French: A Case Study of Language Contact in Middle English”. In Papers Dedicated to J. Moeschler, J. Blochowiak et al. (eds), Geneva: Department of Linguistics.

Harley, H. 1995. Subjects, events and licensing. Ph.D. thesis, MIT, Cambridge, MA.

Heine, B. and Kuteva, T., (eds). 2005. Language Contact and Grammatical Change. Cambridge Approaches to Language Contact. Cambridge: Cambridge Univ. Press.

Ingham, R. 2010. “The Transmission of Later Anlo-Norman: Some Syntactic Evidence”. In The Anglo-Norman Language and its Contexts, R. Ingham (ed), 164–182. Woodbridge: Boydell and Brewer.

Johanson, L. 2002. “Contact-induced change in a code-copying framework”. In Language change: the interplay of internal, external and extra-linguistic factors, M. C. Jones and E. Esch (eds), 285–313. Berlin: de Gruyter.

McWhorter, J. H. 2007. Language interrupted: signs of non-native acquisition in standard language grammars. Oxford ;New York: Oxford University Press.

Meisel, J. 2011. “Bilingual language acquisition and theories of diachronic change: Bilingualism as cause and effect of grammatical change”. Bilingualism: Language and Cognition 14 (2): 121–145.

Santoro, M. 2007. “Second language acquisition of Italian accusative and dative clitics”. Second Language Research 23 (1): 37–50. Troberg, M. 2008. Dynamic Two-place Indirect Verbs in French: A Synchronic and Diachronic Study in Variation and Change of Valence. PHD Thesis: University of Toronto.

Trudgill, P. 2011. Sociolinguistic typology: social determinants of linguistic complexity. Oxford: Oxford University Press.

Woolford, E. 2006. “Lexical case, inherent case, and argument structure”. Linguistic Inquiry 37: 111–130.

Zaenen, A. and Maling, J. 1990. “Unaccusative, passive and quirky case”. In Modern Icelandic syntax, A. Zaenen and J. Maling (eds), 137–152. New York: Academic. Zaring, L. 1991. “On Prepositions and Case Marking in French”. Canadian Journal of Linguistics 36: 363–377.

ABSTRACT. One of the challenges that the field of historical (socio)linguistics faces is the so-called bad data problem (cf. Labov 1994; Nevalainen & Raumolin-Brunberg 2003), i.e. the lack of historical background information related to the speakers at the time, their social networks and communities, the scribal practices, as well as a bias towards formal writings by the upper layer of society (cf. Auer et al. 2015). The frequent lack of information regarding possible contact scenarios and possible bi- and multilingualism of speakers makes it difficult to explain change through language or dialect contact. Related to this, another important and often overlooked issue that should be considered is the relationship between oral and written language. Whereas dialect and language contact primarily takes place on an oral level (and existing contact models are primarily based on spoken language that is produced in a broad variety of communicative settings, cf. Van Coetsem 1988, 1995, 2000; Winford 2005, 2007), the sources at our disposal as historical linguists are of a written nature, i.e. manuscripts and printed texts, which were produced by a select group of literate people.

This raises the question of how we can best determine the occurrence of contact scenarios and thus how to provide evidence for contact-induced language change based only on written historical data. In order to answer this question, we apply an approach that

(a) scrutinises the socio-economic history and migration history of a specific place before considering the linguistic changes, and

(b) pays special attention to the relationship between oral and written language and to what extent oral language and contact-induced influences could be reflected in the written language.

The linguistic findings will then be interpreted after having determined the socio-economic and migration history of a specific place. In fact, in this study, we will illustrate the approach by taking a closer look at the linguistic realities (in as far as they can be reconstructed) of three English urban vernaculars, notably Bristol, Coventry and York, during the period 1400- 1700. In order to establish possible contact scenarios and the possible reflection of contact phenomena in written language, we address the following questions:

● What long- and short-term migration patterns can be discerned over the period 1400- 1700, and what effect did this have on society? Using existing databases on (trade) migration, we outline the migration histories of the respective urban centres.

● What text types were written over time, and what was the role of these texts within the urban community, i.e. who wrote them and who read them? Most importantly, how likely was it that migrants were directly or indirectly involved in the text production? By providing detailed background information about literacy and the production of the different text types, we aim to shed light on how and in what way migrants may have been involved in text production and could have affected the written language.

Case studies from the different urban centres will allow us to illustrate the reconstructed contact scenarios and shed some light on the possibilities of contact-induced language change in Late Medieval and Early Modern England. For the historical background information, we will investigate the England’s Immigrants 1330-1550 Database and socio-economic histories of the respective towns. As regards the linguistic data, the case studies will be based on self-compiled manuscript corpora of a variety of text types such as civic records and letters. Based on these data, and in particular the fact that bilingual (im)migrants were not necessarily literate, it may therefore not be possible to directly trace contact-induced language change in written records. Indirect influences may be possible, although these are often very difficult, if not impossible, to prove.

References:

Auer, Anita, Catharina Peersman, Simon Pickl, Gijsbert Rutten & Rik Vosters (2015). Historical Sociolinguistics: The Field and its Future. Journal of Historical Sociolinguistics Volume 1(1): 1–12.

Van Coetsem, Frans (1988). Loan Phonology and the Two Transfer Types in Language Contact. Dordrecht: Foris.

Van Coetsem, Frans (1995). Outlining a Model of the Transmission Phenomenon in Language Contact. Leuvense Bijdragen 84: 63–85.

Van Coetsem, Frans (2000). A General and Unified Theory of the Transmission Process in Language Contact. Heidelberg: Winter.

England’s Immigrants Database 1330-1550: https://www.englandsimmigrants.com/

Labov, William (1994). Principles of Linguistic Change. Volume 1: Internal Factors. Oxford: Blackwell.

Nevalainen, Terttu & Helena Raumolin-Brunberg (2003). Historical Sociolinguistics. Language Change in Tudor and Stuart England. London: Longman.

Winford, D. (2005). Contact-induced Changes: Classification and Processes. Diachronica 22(2): 373–427.

Winford, D. (2007). Some Issues in the Study of Language Contact. Journal of Language Contact 1(1): 22–40.

ABSTRACT. Markedness reversal between antipassive and transitive constructions as a possible diachronic process of alignment change

INTRODUCTION. It is well-known that the same event structure may be realized by more than one type of a construction. For instance, in some languages a bivalent event with two core referents can be expressed either by a transitive construction or by one of its variants: antipassive and passive, ex. (1).

Given that the valency-changing constructions are prone to language change, various linguists suggest that the shift from the morphologically accusative languages to morphologically ergative languages results from the reanalysis of a passive construction as the basic transitive one (cf. Glidea 1997; Dahl 2016). But this scenario, though plausible, remains speculative. Indeed, once the distinction between passive and resultative constructions is recognized, there are relatively no convincing arguments proving the passive origin of the transitive construction (cf. Creissels [to appear] for further discussion).

AIM. Taking into account that antipassive constructions of morphologically ergative languages reveal the coding pattern found in transitive constructions of morphologically accusative languages (both involve unflagged agent and flagged patient), the aim of this talk is to examine the reversal markedness between these two constructions in Eskaleut family as a possible diachronic process likely to shift these languages from ergative to accusative coding pattern. Another aim is to approach the possible link between the diachronic process and argument coding system within the typological parameter of Obligatory Coding Principle i.e. the reformulation of the traditional distinction between morphologically accusative and morphologically ergative languages (cf. Creissels [to appear]).

STATE OF ART. Building on the frequency of use, Carrier (2012) observes that in Itivimiut, an eastern dialect of Inuktitut, the antipassive construction occurs more often than the corresponding transitive one. In the narrative text, the author notes 12 occurrences of the transitive construction against 18 occurrences of the passive and 117 occurrences of the antipassive. Johns (2001, 2006) offers a similar observation to Labrador Inuttut and South Baffin, other eastern dialects of Inuktitut. This intra-linguistic distribution of transitive and antipassive constructions is further supported by direct evidence from the acquisition of Inuktitut itself (cf. Allen 2013). The findings from caregiver speech and spoken narratives show that not only young Inuktitut-speakers aged 2;0-3;6 avoid structures with ergative morphology but also older children and adults reveal this tendency, see Table 1.

Transitive Antipassive Passive Noun Incorporation Grade 3 0 47 15 1 Grade 8 0 49 12 9 Adult 0 84 15 25 Table 1. Frequency of use of four types of constructions to express bivalent clause with third person Agent (Allen 2013).

Thus, in this set of genealogically related languages, the construction that was considered to be a variant of the basic transitive one changed its status by gradually replacing the default transitive construction. Additionally, this tendency seems to be geographically motivated, since the distribution of the ergative pattern diminishes in prominence from western to eastern dialects (Johns 1999).

METHODOLOGY. I examine markedness reversal between transitive and antipassive constructions in some Inuktitut variants and beyond with special attention given to the morphosyntactic, behaviour and semantic variables (Table 2). Grammars, monographs and journals will constitute an important source of reference for data collection. I will argue that the predominance of the antipassive over the ergative construction can be best explained along with the asymmetry of frequency of occurrence. Following Carrier (2012), among others, I will claim that the replacement of the transitive construction by the antipassive one results from the fact that the latter became less marked hence more frequent.

CONCLUSION. Building on the preliminary observation, it is not unreasonable to think that the Eskaleut languages are currently undergoing the historical change from an ergative argument coding to the nominative one. The children’s language from Inuktitut reflects that change. Obviously more data are required to support this observation. This paper aims at filling up this gap. Given the distribution of the antipassive within the Inuktitut dialects, I will also search for the causes that push language change.

(1) Baffin Island Inuktitut (Spreng, 2005: 2-3) a. Anguti-up arnaq kunik-taa. Transitives construction man-ERG woman kiss-3SG.3SG ‘The man kissed the woman.’

b. Arnaq kunik-tau-juq anguti-mut Passive construction woman kiss-PASS-3SG man-ABL.SG ‘The woman was kissed by the man.’

c. Anguti kunik-si-vuq arna-mik. Antipassive construction man kiss-AP-3SG woman-MOD.SG ‘The man is kissing a woman.’

VARIABLES

MORPHOSYNTACTIC category of AGENT/PATIENT: nominal, pronominal, zero indexing of AGENT/PATIENT: verb agreement, cross-referencing, zero flagging of AGENT/PATIENT: case, adposition marking, zero

BEHAVIORAL realization of AGENT: over, non-overt realization of PATIENT: omission, demotion (obligatory/optional)

SEMANTIC number of PATIENT/AGENT: singular, plural, dual

Table 2. Morphosyntactic, behaviour and semantic variables.

Abbreviations ABL ablative, AP antipassive, ERG ergative, MOD modalis, PASS passive, SG singular

References Allen, Shanley. 2013. The acquisition of ergativity in Inuktitut. In Edith Laura Bavin & Sabine Stoll (eds.), The acquisition of ergativity. (Trends in Language Acquisition Research 9). Amsterdam: John Benjamins Publishing Company. 71–104.

Carrier, Julien. 2012. L’expression de la transitivité en itivimiut. Montréal: Université du Québec M.A. Dissertation.

Creissels, Denis. to appear. Obligatory Coding Principle in diachronic perspective. In Sonia Cristofaro & Fernando Zúñiga (eds.), Typological Hierarchies in Synchrony and Diachrony. Amsterdam: John Benjamins Publishing Company.

Dahl, Eystein 2016. ‘The origin and development of the Old Indo-Aryan predicated -tá construction.’ In Eystein Dahl and Krzysztof Stroński (eds.), Indo-Aryan Ergativity in Typological and Diachronic Perspective. Amsterdam: John Benjamins Publishing Company. 63-110.

Gildea, Spike. 1997. Evolution of grammatical relations in Cariban: How functional motivation precedes syntactic change. In T. Givón (ed.), Grammatical relations: a functionalist perspective. Amsterdam: John Benjamins Publishing Company. 155-198.

Johns, Alana. 1999. The decline of ergativity in Labrador Inuttut. In Papers from the Workshop on Structure and Constituency in Native American Languages. MITWPL 17. 73-90.

Johns, Alena. 2001. Ergative to Accusative: Comparing Evidence from Inuktitut». In Jan Tetje Faarlund (ed.), Changing Relations. Amsterdam: Johns Benjamins Publishing Company. 205-221.

Johns, Alana. 2006. Ergativity and change in Inuktitut. In Alana Johns, Diane Massam & Juvénal Ndayiragije (eds.), Ergativity: Emerging Issues. (Studies in Natural Language and Linguistic Theory 65). Dordrecht: Springer. 293–311.

Spreng, Bettina. 2005. Third person arguments in Inuktitut. In Solveiga Armoskaite and James J. Thompson (eds.), Workshop on the structure and constituency of the languages of the Americas. University of British Columbia: UBC Working Papers in Linguistics. 215-225.

ABSTRACT. Japanese is perhaps the best known example of a language where an exceptionless accusative alignment pattern has emerged through development of a genitive case marker to a nominative. In Modern Tokyo Japanese the subjects of both transitive intransitive verbs are marked by nominative =ga and the objects of transitive verbs are marked by accusative =o, while in Old Japanese (OJ; 8th century) =ga was a genitive particle also marking the subjects of “nominalized” subordinate clauses and =(w)o marked objects in a differential object marking pattern (Yanagida 2006). Previous research has claimed that the reanalysis of a major subclass of nominalized clauses as a matrix pattern led straightforwardly to reanalysis of genitive as nominative. Our purpose in this paper is to show that matters are more complex. First, Yanagida and Whitman (2009) show that the alternation between =ga and zero in OJ shows properties of active alignment, including the familiar syncretism between active/ergative and genitive. Similar properties can still be observed in some Ryūkyūan languages. The OJ to ModJ shift is thus not just a change in case marking, then, but a change in alignment. Second, more recent and detailed research shows that genitive subject =ga in earlier Japanese is not the direct source of nominative =ga in ModJ. Instead, a pattern that we call impersonal psych transitive constructions played the central role in the shift in Japanese case marking and alignment. The impersonal psych transitive construction in OJ involves an implicit first person experiencer object, as in (1):

Old Japanese (8th century) (1) [papa wo panarete yuku]=ga kanasi-sa (Man’yōshū 4338) mother obj part go.adn=gen sad-nmlz ‘(I am) sad about parting from Mother.’

Examples like (1) increased in frequency after OJ. The examples in (2-3) are cited from Ohno (1977:142). Ohno (1977, 1987) observes that in EMJ (9th-12th c.), adnominal clauses marked by =ga are used predominantly with psych predicates with a first person experiencer (2), as in OJ (1), but that they also begin to appear with non-psych intransitive verbs as in (3).

Early Middle Japanese (11th century) (2) [kokorobape wo mi-ru]=ga wokasiu mo (Genji monogatari, 11th c.) kindness acc see-adn=gen thankful excl ‘Seeing (someone’s) kindness makes (me) thankful.’ (3) [kumo no usuku watare-ru]=ga nibi iro naru wo (Genji monogatari, 11th c.) cloud gen shallow pass.away-adn agt red color become excl ‘the clouds passing thinly away become red’

In (3) the adnominal clause marked by =ga is the subject of a non-psych intransitive verb, and it involves no implicit first person experiencer. A further change in EMJ is that while this psych predicate construction was used only in nominalized clauses in OJ, it came to appear in non-nominalized main clauses. Based on Middle Japanese (MJ, 800-1600) data, we show that ModJ nominative =ga is descended from =ga marking the clausal complements of psychological adjectives (or experiencer adjectives). Based on Ohno’s observations and data collected from the NINJAL Diachronic Corpus of Japanese (http://historicalcorpus.jp/), we show that nominative =ga developed from a reanalysis of impersonal psych-transitives as true intransitives, where the =ga marked argument acquires the status of the subject argument of the predicate. As a consequence of this reanalysis, =ga reappears in Late Middle Japanese as a nominative postposition, first marking the theme subject of intransitives. It was extended to mark the subjects of transitive verbs in Early Modern Japanese. The Japanese example demonstrates the need to look behind apparently simple shifts in case marking or alignment, such as genitive/active > nominative. The path to realignment turns out to be not reanalysis in the classical sense, but rather extension of a fairly restricted pattern (impersonal psych intransitive to intransitive) followed by a syntactic readjustment (clausal/theme complement to subject).

ABSTRACT. A small number of Central Malayo-Polynesian languages of central Flores (e.g. Keo Rongga, Ngada, Nage, Lio, Ende, Palue) are extremely isolating languages, in contrast to the various degrees of affixation typical of Austronesian languages in general, as well as of Central Malayo-Polynesian languages and even those of western and eastern Flores. It has often been supposed that a language can lose all, or almost all, of its bound inflectional machinery via ordinary, step-wise grammar-internal change, such that radical analyticity is a natural state that a grammar might reach. It is also well-known that extensive non-native acquisition, such as pidginization and creolization, also has an extremely subtractive effect on inflectional affixation. However, it is assumed that a language can lose all or most of its inflection either via extensive second-language acquisition or grammar-internal change.

In fact, the idea that “drift” alone can eliminate a grammar’s bound inflectional morphology is based primarily on 1) the account of Egyptian in Hodge (1970), which actually describes modest, not total, loss of inflection, and 2) the difference between modern and Old English, Mainland Scandinavian languages and Old Norse, and the Romance languages and Latin. However, an increasing volume of research suggests that the above cases were due to extensive second-language acquisition by adults (Kusters 2003, McWhorter 2007, Trudgill 2011).

I suggest that the Flores situation suggests that these isolating languages’ typology is due not to “drift” but to adult acquisition in the past, as part of a general hypothesis that this is can be analyzed as true of all languages with little or no bound inflectional morphology. I will present a number of arguments that converge on a second-language acquisition account for Flores, including issues involving the difference between contextual and inherent inflection (Booij 1993); phonotactics; specific parallels between the central Flores languages and languages known to have been born as creoles; the geographical distribution of the isolating languages; and folkloric evidence that Flores was invaded by people from the nearby island of Sulawesi in the past.

Booij, Geert. 1993. Against split morphology. Yearbook of morphology 1993, ed. by Geert Booij and Jaap van Marle, 27-49. Dordrecht: Kluwer. Hodge, Carleton. 1970. The linguistic cycle. Language Sciences 13: 1-7. Kusters, Wouter. 2003. Linguistic complexity: the influence of social change on verbal inflection. Utrecht: Landelijke Onderzoekschool Taalwetenschaap. McWhorter, John H. 2007. Language interrupted: signs of non-native acquisition in standard language grammars. New York: Oxford. Trudgill, Peter. 2011. Sociolinguistic typology. Oxford: Oxford University Press.

ABSTRACT. Singapore English (SingE) is one of the most extensively studied New Englishes and has traditionally been considered a prototypical second-language variety. However, over the last few decades, it has experienced important changes in its sociolinguistic status as it has been gradually developing into a first language for an ever-growing number of children. This has often been noted in recent times (e.g. Bolton & Ng 2014; Lim 2007; Tan 2014) but has never been investigated in an empirical, comprehensive fashion. The present paper is part of a large-scale research project investigating such changes in Singapore. Pursuing an apparent-time approach, it adds a historical perspective to the investigation of SingE by comparing L2 adult with L1 child SingE. Drawing on a newly established child language corpus, the paper presents findings from an analysis of the acquisition of past tense marking (marked vs. unmarked verb forms) by Singaporean children. The data come from 30 children aged 2;5-12;1 and were elicited systematically in video-recorded task-directed dialogue between researcher and child, consisting of several parts: a grammar elicitation task, a story retelling task, elicited narratives, and free interaction. The data were coded for the absence and presence of past tense marking and results were interpreted with respect to the sociolinguistic variables ‘age’ and ‘ethnicity’ (as these have turned out to be significant factors in the acquisition of other linguistic phenomena of L1 SingE in earlier analyses; e.g. AUTHOR 201X). While British and American children produce bare verb forms in very early acquisitional stages only (e.g. Marchman & Bates 1994; Wexler 1998), the results of the present study show that Singaporean children continue to variably omit past tense marking even at more advanced stages of first language acquisition, i.e. that past tense marking is at least partially lost in L1 SingE. This is also true for L2 Singapore Colloquial English (of which bare verb forms are one of the pertinent characteristics), the variability in use in L1 SingE thus being indicative of the fact that both Singapore Standard and Singapore Colloquial English serve as input varieties in its acquisition. This variability can be observed both within the children’s idiolects and across speakers, specifically motivated by the children’s ethnicity (here Chinese and Indian). Comparing the findings to L2 data and results from the ICE-Singapore from the 1990s, the paper also offers a real-time dimension. The results are interpreted with respect to what they tell us about ongoing changes in SingE and directions of language change in English worldwide driven by child language acquisition.

AUTHOR. 201X. “English as a First Language in Singapore: A Corpus-based Analysis of Zero Subjects”. Presentation at the CONFERENCE. Bolton, K. & Ng, B. C. 2014. “The dynamics of multilingualism in contemporary Singapore”. World Englishes 33(3): 307-318. Lim, L. 2007. “Mergers and acquisitions: on the ages and origins of Singapore English particles”. World Englishes 26(4): 446-473. Marchman, V. A., & Bates, E. 1994. “Continuity in lexical and morphological development: A test of the critical mass hypothesis”. Journal of Child Language 21(2): 339-366. Tan, Y.-Y. 2014. “English as a ‘mother tongue’ in Singapore”. World Englishes 33(3): 319-339. Wexler, K. 1998. “Very early parameter setting and the unique checking constraint: A new explanation of the optional infinitive stage”. Lingua 106: 23-79.

ABSTRACT. Though the organizers seek presentations that are empirical in approach, the present proposal veers to the opposite extreme by adopting an approach that is essentially speculative. It is traditional to assume that grammatical resemblances between Germanic and Romance (if not in the final analysis merely coincidental) could only be created by contact between Germanic and Romance (with either as a possible source for innovations in the other), it is in theory possible that contact between 1) Germanic and Celtic, and 2) Romance and Celtic could produce grammatical resemblances between Germanic and Romance, which would then create the illusion that such resemblances were created by contact between Germanic and Romance, which are after all much better-known. For reasons of (pre-)history and geography, significant language contact between Germanic and Celtic would be unexpected outside of West Germanic, which apparently gained a lot of territory (southern and western Germany and most of Benelux)) by language shift from Celtic during the last centuries B.C. (Prokosch 1939: 31-2). There is at least one case where positing that parallel influences from Celtic have created grammatical resemblances between Germanic and Romance appears to be plausible: not having a reflexive/non-reflexive distinction in possessive pronouns. For convenience, this will from here on be called “possessive non-reflexivity”. A map (necessarily cobbled together using evidence from different periods) of where possessive non-reflexivity either certainly or probably occurred as of around 1000 would include the territories of Celtic (Lewis & Pedersen 1989[1961]: 195, 205), West Germanic (Prokosch 1939: 274, 280), contiguous Romance (Penny 1991: 127; Maiden 1995: 113; Pope 1934: 328), Balkan Romance (Augerot & Popescu: 1971: 63), Albanian (Orel 2000: 243-4), and Greek (Sihler 1995: 383). Thus we have in essence a western zone and a Balkan zone. The latter is separated both from the former and within itself, by South Slavic, which is of course intrusive, and where the original reflexivity of all IE has not been undone (Schmalstieg 1976: 68; Entwistle & Morris 1949: 389, 391). Possessive non-reflexivity does not occur in non-West Germanic or Latin. Since it is known that that possessive non-reflexivity in any IE language is an innovation, the areal evidence is broadly consistent with the proposition that possessive non-reflexivity had somehow became general in the later Roman Empire in Europe (despite not existing in Latin) and in nearby areas (Ireland, SW Germany), before being lost wherever Slavic intruded into its former domain. For the purposes of the present program, the Balkan part of the puzzle must be ignored. As for the rest, among well-attested languages of northern and western Europe, those with objective non-reflexivity (having no distinct reflexive pronouns used as objects), most notably Celtic, always have possessive non-reflexivity, whereas those with possessive non-reflexivity may or may not have objective non-reflexivity, which is typically associated with liberal use of reflexive verbs For example, French (of all periods) has objective reflexivity, whereas Old English does not. This subset relation seems to imply that possessive reflexivity is for some reason (probably functional) more difficult to sustain than is objective reflexivity. It would indeed be somewhat infelicitous to use one pronoun in “It’s her birthday” and another in “She is celebrating her birthday”, especially when this would involve using different pronouns to refer to the same person in consecutive (or near-consecutive) sentences. Be that as it may, the non-Balkan part of the areal pattern seems to work as a sort of ghost image of where Celtic was once spoken as of around 500 B.C., with some understandable overflow 1) along the North Sea coast and 2) into non-northern Italy. The proposed presentation will be largely concerned with how Celtic influence could work to explain the different types of possessive non-reflexivity seen in southerly West Germanic, where the reflexive form (e.g. NHG “sein”) has won out, and in northerly West Germanic, where the non-reflexive form (e.g. PDE “his”) has won out.

References: Augerot, James & Florin Popescu. 1971. Modern Romanian. Columbus: Slavica. Entwistle, W.J. & W.A. Morrison. 1949. Russian and the Slavonic Languages. London: Faber & Faber. Lewis, Henry & Holgar Pedersen. 1989 [1961]. A Concise Comparative Celtic Grammar; 3rd edition, with a supplement by Henry Lewis. Gōttingen: Vandenhoek & Ruprecht. Maiden, Martin. A Linguistic History of Italian. London: Longman. Mann, Stuart 1977. An Albanian Historical Grammar. Hamburg: Helmut Buske. Orel, Vladimir. 2000. A Concise Historical Grammar of the Albanian Language. Leiden: Brill. Penny, Ralph 1991. A History of the Spanish Language. Cambridge: University Press. Pope, Mildred.1934. From Latin to Modern French. Manchester: University Press. Prokosch, Eduard. 1939. A Comparative Germanic Grammar. Philadelphia: Linguistic Society of America. Schmalstieg, William. An Introduction to Old Church Slavonic. Columbus: Slavica. Sihler, Andrew. 1995. New Comparative Grammar of Greek and Latin. Oxford: University Press.

ABSTRACT. Straddling the boundary between lexicon and syntax, phrasal borrowing - the material borrowing of analysable multi-word units - is cross-linguistically well-attested (Gomez Rendon 2008; Onysko 2012). Such forms presumably start out as multi-word code-switches, a phenomenon widely studied from a synchronic perspective (Myers-Scotton 1997). The diachronic impact of such switches on grammar, however, has received little attention. In cases where phrasal borrowings encompass whole paradigms, they remain analysable even to monolingual speakers. Their internal syntax is thus integrated into all speakers' competence, becoming part of the grammar of the language. Yet the concept of phrasal borrowing is absent from the indexes of recent reference books on language contact such as Matras (2009) or Hickey (2010).

Phrasal borrowing is strikingly frequent among languages in contact with Arabic; in North Africa, as in East Asia, the borrowing of numerals is often accompanied by the borrowing of entire paradigms of numeral+noun phrases (Kossmann 2013:311). Borrowed numeral+noun phrases in this region are typically limited to the counting of a limited, specific set of nouns. In two cases, however - Jerusalem Domari (Matras 2012) and Beni-Snous Berber (Destaing 1907) - a more far-reaching phenomenon is described: for some (but not all) borrowed numerals, borrowed numeral+noun phrases are substituted no matter which noun is being counted. This phenomenon of grammatical rules in one language that induce the speaker to switch to another language, labelled ‘bilingual suppletion’ by Matras, so far seems to be described nowhere else in the world; communities bilingual in Arabic thus have a pivotal role to play in the study of this phenomenon.

A comparative study of paradigmatic phrasal borrowing across the Arabic contact zone reveals that the conditions governing phrasal borrowing within each language show remarkable cross-linguistic consistency. In numeral+noun phrases, all instances of phrasal borrowing seem to be triggered language-internally by one of two conditions: 1) numerals whose agreement requirements differ between Arabic and the minority language (Souag & Kherbache 2016), or 2) nouns which are measure words.

A cross-linguistic condition also applies to all the languages concerned: 3) the borrowing of numerals below 10.

Although the only test cases available for it seem to be in the Arabic contact zone, Condition 1) amounts to the diachronic manifestation of a rather cross-linguistically robust principle elaborated synchronically by Myers-Scotton (1997): the Embedded Language Island Hypothesis. Conditions 2) and 3) can more readily be tested across a wider area, and appear equally valid outside the Arab world. Together, these conditions make phrasal borrowing in the domain of numeral phrases relatively predictable.

References Destaing, Edmond. 1907. Etude sur le dialecte berbère des Beni-Snous. Paris: Ernest Leroux. Gomez Rendon, J. A. 2008. Typological and social constraints on language contact : Amerindian languages in contact with Spanish. Leiden: University of Leiden. Hickey, Raymond (ed.). 2010. The Handbook of Language Contact. Chichester: Wiley-Blackwell. Kossmann, Maarten. 2013. The Arabic Influence on Northern Berber. Leiden: Brill. Matras, Yaron. 2009. Language Contact. (Cambridge Textbooks in Linguistics). Cambridge, UK: Cambridge University Press. Matras, Yaron. 2012. A grammar of Domari. (Mouton Grammar Library 59). Berlin: De Gruyter. Myers-Scotton, Carol. 1997. Duelling Languages: Grammatical Structure in Codeswitching. Oxford: Oxford University Press. Onysko, Alexander. 2012. Anglicisms in German, Borrowing, Lexical Productivity, and Written Codeswitching. Berlin: De Gruyter. https://www.degruyter.com/view/product/175616 (23 November, 2016). Souag, Lameen & Fatma Kherbache. 2016. Syntactically conditioned code-switching? The syntax of numerals in Beni-Snous Berber. International Journal of Bilingualism 20(2). 97 –115.

ABSTRACT. Recent advances in developmental linguistics underscore the importance of identifying different types of bi-/multilingual speakers in determining the role they are likely to play in contact-induced language change (cf. Meisel 2011, Meisel et al 2013). Such insights come at a time when many contact linguists, too, are advocating speaker-based models of language contact in an attempt to bring clarity to results-focused claims that ‘anything can happen’ (cf. Winford 2005). Both perspectives, importantly, recognize that not all bilinguals play an equal role in facilitating language change. Van Coetsem’s (2000) model of language contact distinguishes three types of speaker agentivity according to the speaker’s linguistic dominance. In this framework the type of bilingual speaker implicated in structural transfer is dominant in the source language as opposed to the recipient language. Focusing specifically on core grammatical change, Meisel (2011) and Meisel et al (2013) propose that the likely agents of change are successive child and adult L2 bilinguals. Though the terminology used to describe different types of bilinguals may differ, both approaches single out a specific type of speaker as likely agent of structural change: the non-dominant bi-/multilingual.

Identifying the likely agents of change is essential in reconstructing plausible scenarios of contact-induced change; this underscores the importance of incorporating a more nuanced understanding of different types of bi-/multilingual speakers into historical accounts of contact-induced change. Even with this understanding, however, it can be one matter to make a plausible case for non-dominant speakers as agents of structural change in a given contact situation, and quite another to locate sources of data produced by such speakers in support of one’s claims. For example, Meisel (2011: 140) hypothesizes “that in larger and more complex societies, situations in which L2 learners exert a major influence on a language are most likely to emerge in periods of substantial demographic changes”. The early modern Dutch urban context is known to be one of extreme demographic instability. From the late sixteenth to the mid-seventeenth century in particular, several cities experienced a doubling or tripling of their population, fueled by individuals from a variety of linguistic backgrounds (cf. Lesger 2006, Van Nederveen Meerkerk 2009, Van Zanden and Prak 2012). The socio-historical context is one where non-dominant speakers could be argued to have played a significant role in ongoing structural changes, particularly in the urban centers where demographic change was most extreme. However, locating the linguistic evidence to support this scenario can, for a variety of reasons, prove nigh impossible.

In this talk I address several methodological challenges in determining the role that non-dominant multilingual speakers likely played as agents of change in the early modern Dutch context when it is well-documented that they made up a significant proportion of the growing urban population at this time. Dutch archives are teeming with manuscripts preserved from the early modern period, but these data sources are not necessarily random in terms of the types of speakers they represent. In municipal archives there is a bias towards preservation of materials relevant to the local town, city or province so that there is great potential for these sources to reflect stability and continuity rather than instability and change. Moreover, the types of sources of interest to contact linguists—those which most closely reflect the spoken language of time such as personal letters and diaries or ‘egodocuments’(cf. Koch and Oesterreicher 1985)—represent a minute fraction of the manuscripts preserved in Dutch archival collections. While it may be tempting to assume that egodocuments capture a wider variety of types of speakers known to have been present in this turbulent socio-historical context, this appears not to be the case (cf. Lindeman et al 1993). More problematic is the fact that even when we base our analyses of contact-induced change on egodocuments, we may not know enough background information about their authors to determine what types of speakers they might have been (monolingual, bilingual, non-dominant bilingual, etc.). This begs the question: If we don’t know much or anything about the individuals whose writings we’re using as evidence of continuity or change, how do we know what types of speakers we’re capturing in our analysis of linguistic data and whether they even include the likely agents of change?

I propose in this talk that one way to tackle these challenges is to incorporate a more nuanced understanding of different types of bi-/multilingual speakers into the linguistic analysis. To this end, I include data from non-dominant bilinguals in my analysis of two changes known to have occurred during the early modern Dutch period: the rise in the use of the distinct reflexive pronoun ‘zich’ and the loss of brace negation. The discussion of ‘zich’ highlights in particular the importance of knowing more about the linguistic repertoires of the individuals who produced the data we use to construct hypotheses about language change. While locating data from non-dominant bilinguals may be akin to finding needles in haystacks, by adopting an approach that pays attention to different types of speakers, I argue that we can better evaluate sociolinguistic biases in the historical evidence and so reconstruct more accurately the circumstances that favor the diffusion of innovations.

Coetsem, Frans van. 2000. A general and unified theory of the transmission process in language contact. Heidelberg: Universitätsverlag C. Winter. Koch, Peter & Wulf Oesterreicher. 1985. Sprache der Nähe – Sprache der Distanz: Mündlichkeit und Schriftlichkeit im Spannungsfeld von Sprachtheorie und Sprachgebrauch. Romanistisches Jahrbuch 36, 15-43. Lesger, Clé. 2006. Informatiestromen en de herkomstgebieden van migranten in Nederland in de vroegmoderne tijd. Tijdschrift voor sociale en economische geschiedenis 3, 3-23. Lindeman, R., Y. Scherf & R.M. Dekker. 1993. Egodocumenten van Noord-Nederlanders van de zestiende tot begin negentiende eeuw. Rotterdam: Erasmus Universiteit. Meisel, Jürgen M. 2011. Bilingual language acquisition and theories of diachronic change: bilingualism as cause and effect of grammatical change. Bilingualism: Language and Cognition 14(2), 121-45. Meisel, Jürgen M., Martin Elsig & Esther Rinke. 2013. Language acquisition and change. A morphosyntactic perspective. Edinburgh: Edinburgh University Press. Nederveen Meerkerk, Elise van. 2009. De economische aantrekkingskracht van vroegmoderne steden. In Leo Lucassen & Wim Willems (Eds.), Waarom mensen in de stad willen wonen 1200 – 2010, 103-123. Amsterdam: Uitgeverij Bert Bakker. Zanden, Jan Luiten van, and Maarten Prak. 2012. Demographic change and migration flows in Holland between 1500 and 1800. In Marcel van der Linden & Leo Lucassen (Eds.) Working on labor. Essays in honor of Jan Lucassen, 237-245. Leiden/Boston: Brill. Winford, Donald. 2005. Contact-induced change. Classification and processes. Diachronica 22(2), 373-427.

Unpublished archival sources: Bibliotheca Thysiana, Universiteitsbibliotheek Leiden, The Netherlands.

ABSTRACT. Introduction: The current study explores expression of future-time reference in Palestinian Arabic (PA). Different speakers of PA in Gaza City express future-time reference in morphologically distinct ways; furthermore, these differences correlate with speakers’ dialect background and age. The pilot study described here deals with the development of future marking in two urban PA dialects spoken in Gaza City which came into contact following the Palestinian-Israeli conflict in 1948 and the influx of thousands of Palestinian refugees from Jaffa to Gaza. The paper examines the development of the future marker in PA from the lexical verb ra:ḥ ‘go’. The analysis is framed in the context of the code-copying approach to contact-induced language change (Johanson 1999, 2002) and the principles of grammaticalization theory (Hopper and Traugott 2003, Bybee, Pagliuca, and Perkins 1994). Research Questions: The study is designed to answer the following research questions: does the variation in the expression of future time reference represent a case of ongoing grammatical change in PA? If so, is this change the outcome of internal factors and universal linguistic tendencies or do they involve multiple complex and correlated processes that call for a multiple-causation approach to language change? Methodology: To establish the range within which PA dialects spoken in Gaza City vary in their morphological expression of future time reference, data was collected using interviews from speakers of two urban dialects: Gazan and Jaffan. For the study, fourteen female speakers representing three different age groups were interviewed: seven speakers with Gazan dialect background and seven with Jaffan dialect background. The recordings were transcribed, and the verb forms appearing were organized into paradigms according to their morphological form and function. A control group of four Jaffa dialect speakers who still live in Jaffa, Israel, were also interviewed using the same questions. Findings: The observed variation correlates with speaker age and dialect background. For example, the morphology of the middle-aged and younger Jaffan and Gazan speakers living in Gaza exhibits verbal morphology properties with respect to the way they mark ‘future’ that are absent in the Jaffan dialect still spoken in Jaffa, Israel and are not characteristic of the speech of older Jaffan and Gazan speakers in Gaza. Data from Jaffan speakers still living in Jaffa show that future time reference is expressed outside the verbal paradigm by means of a particle ra:ḥ, or the quasi-verb bidd- ‘want’. However, middle-aged and younger Gazans and Jaffans living in Gaza primarily express ‘future’ within the verbal paradigm by means of a prefix ḥa- attached to the non-past stem. Based on comparison of the data, two possible scenarios of development are suggested. First, Gazan usage served as model for Jaffans who came into contact with Gazan speakers and copied the Gazan prefix ḥa- into their basic code. The second scenario is that Jaffan dialect served as a model code for the Gazan dialect. ra:ḥ is a lexical motion verb that had not yet developed into a grammaticalized future marker in the dialect of Gaza by the time of contact. In the Jaffan dialect, ra:ḥ was in the midst of a grammaticalization process by which it had become a particle marking ‘future’. Through a process of selective grammatical copying, Gazan speakers adopted the grammatical function of ra:ḥ as a future marker on the model of the Jaffan dialect and inserted it into their own basic code, followed by further grammaticalization and change in the syntactic behaviour of the verb. The second possibility is the one I adopt here and base my analysis on. This decision is based on the absence of ra:ḥ or ḥa- from the means available to the older Gazan speakers for the expression of future time reference. This finding suggests that it is the Gazan dialect that took Jaffan usage as a model at the time of initial contact and copied ra:ḥ into their basic code. Discussion: The results suggest that through the transmission mechanism (Van Coetsem 2000), Gazan speakers, the socially dominant but linguistically non-dominant group, reanalyzed and remodelled the function of the verb ra:ḥ and borrowed or adopted it as a future marker and inserted it into their primary code. This copying is motivated by the speaker need for communication. According to Johanson (2008: 63), the act of copying is best viewed as an innovative act and a creative technique that is used by individual speakers to enrich their language or variety and enhance its functionality. Thus, it is probable, that with the copying of the particle ra:ḥ, the Gazan speakers enriched their dialect by adding a new linguistic element whose only function was to express the future. This insertional copying increases the functionality of the Gaza dialect in certain respects. Originally, future time reference in Gazan dialect could only be expressed by bidd- ‘want’ or relevant adverbials. In addition to its function as a future marker, the semi-verb bidd- is mainly used to express volition and wishes in Gazan dialect. Therefore, copying an element whose primary function is to express the future satisfies the individual speaker’s need for using the language to express future time reference and also makes the system more efficient. As the particle ra:ḥ is used by Gazan speakers, and its frequency increased, it is further grammaticalized into the prefix ḥa- that is attached to the non-past form of the verb and exclusively used by younger speakers from both Jaffan and Gazan dialects to express the future. With respect to social dominance, Johanson (1999: 54) defines dominance relations in terms of social, economic and political relations and strength where the language of the immigrant community is usually the dominated code. Johanson suggests that the asymmetric dominance relations between the two codes cause “one-directional dynamics” where the dominated variety adopts new linguistic habits from the dominant variety (1999: 54). As Jaffan PA is a diaspora dialect, it would be plausible to think of Gazan PA as the dominant dialect and of Jaffan as the dominated dialect, and thus to expect the change to be unidirectional, i.e. copying from Gazan into Jaffan and other refugee dialects. However, the data collected in this pilot study suggests that copying is performed by speakers of the socially-dominant Gazan dialect to fulfill psycholinguistic needs.

References

Bybee, Joan, William Pagliuca, and Revere Perkins. 1994. The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. Hopper, Paul J., and Elizabeth Closs Traugott. 2003. Grammaticalization, 2nd ed. Cambridge: Cambridge University Press. Johanson, Lars. 1999. The dynamics of code-copying in language encounters. In: Bernt Brendemoen, Elizabeth Lanza and Else Ryen (eds.), Language encounters across time and space, 37-62. Oslo: Novus Press. Johanson, Lars. 2002. Contact-induced change in a code-copying framework. In Marie C. Jones and Edith Esch (eds.), Language change: The interplay of internal, external, and extra-linguistic factors, 285-309. Berlin: Mouton de Gruyter. Van Coetsem, Frans. 2000. A general and unified theory of the transmission process in language contact. Heidelberg: C. Winter.

ABSTRACT.  

 Language change in East African Bantu: multilingualism and its effects 

With some 350 languages spoken across much of Central, Eastern and Southern Africa, the Bantu language family provides an ideal lens for the examination of language change and linguistic variation. The Bantu-speaking area has long been characterised by high levels of bi- and multi-lingualism and a sustained history of language contact. East Africa in particular is a region of high linguistic diversity, with representatives from all four of Africa’s languages phyla found. This talk examines the role of bilingualism in processes of language change with a focus on structural innovation in the Bantu languages of East Africa and the role of second language speakers in these processes of change. 

Given the linguistic diversity found in the East African region, two broad categories of language contact can be identified: contact between Bantu languages, and contact between Bantu and non-Bantu languages. Swahili has a unique status across much of the region where it functions as an important lingua franca. However, the vast majority of present-day speakers of Swahili have another language as their first language, with bi-lingualism widespread. The increasing shift made by these speakers to using Swahili has been noted in a number of previous studies (Legère 1992; Yoneda 1996). However, less attention has been paid to the specific impact of this ‘shift’ as a form of language contact. One example comes from the Bantu language Matengo which is spoken in the southwest of Tanzania. The most notable influence of Swahili on Matengo is observable in the lexicon with extensive evidence of borrowing. However, there are also apparent influences in the realm of grammar. The ‘original’ Matengo construction shown in (1), is now increasingly replaced by that shown in (2) which employs a combination of the Matengo verb iti ‘do’ alongside the passive form of the Swahili verb, in this case kutegwa ‘to be cursed’. 

(1) Mundu dju-m-teg-iti Kinunda. 

1.Someone SM1-OM3SG-curse-PRF Kinunda 

‘Someone has cursed Mr. Kinunda.’ (Matengo, Yoneda 2016) 

(2) Kinunda dju-tend-iti ku-teg-w-a. 

Kinunda SM3SG-do-PRF INF-curse-PASS-FV 

‘Mr. Kinunda has been cursed.’ (Matengo, Yoneda 2016) 

Language contact scenarios in East Africa also have an impact on Swahili. At least two instances of influence of other (in this case, Bantu) languages on Swahili can be found in the realm of tense-aspect-mood marking. Bantu languages commonly employ a variant of the suffix -ag- to encode imperative, repetitive or habitual meanings (this has been reconstructed as *-ag- for Proto-Bantu). However, in Swahili this historic suffix has been lost and has instead been replaced with the formative hu- (3). However, the habitual -aga is found in many of the East African Bantu languages and its presence has been noted in colloquial varieties of Swahili spoken in mainland Tanzania (4). The suggestion is therefore that this marker has been introduced into Swahili by speakers who have another Bantu language (which does have the habitual marker -ag) as their first language.  

(3) Wewe hu-la wapi? 

2sg.PP HAB-eat-FV where 

‘Where do you (usually) eat?’ (Standard Swahili) 

(4) U-na-kul-aga wapi? 

SM2sg-PRES-eat-HAB where 

‘Where do you (usually) eat)? (Colloquial Swahili, Gibson & Marten 2016) 

The second type of language contact under examination in the present talk is contact between unrelated languages, in this case, with a focus on contact between Bantu and non-Bantu languages. The Bantu language Rangi, spoken in central Tanzania, has been noted to exhibit a number of features which cause it to stand out from a typological perspective (Dunham 2005, Gibson 2012, Gibson & Marten 2015). Rangi has an unusual word order in which the auxiliary appears after the verb in restricted syntactic contexts. Rangi also exhibits a number of other features which it has been proposed are the result of a high level of contact with neighbouring Cushitic languages spoken in the area, and in particular, adult second language acquisition of Rangi. These features include an inclusive/exclusive distinction in first person plural possessive pronouns, a clause-final negative marker which appears to be Cushitic in origin, and the presence of verb-internal deictic markers, all of which are non-canonical features in East African Bantu. 

The talk draws on examples such as these with a view to better understanding how second-language speakers impact processes of language change, with a focus on the Bantu languages and associated sociolinguistic contexts of East Africa. A related question is whether the effects of language contact are different in instances in which the contact is between related languages and in those in which it is between unrelated languages. This talk shows that both of these types of contact appear to be able to impact on all linguistic domains with no apparent difference between the two. The talk contributes to our understanding of the role of second language speakers in situations of language contact and historical processes of language change. 

References 

Dunham, M. (2005): Eléments de description du langi, langue Bantu F.33 de 

Tanzanie. Peeters, Louvain-Paris-Dudley MA. 

Gibson, H. 2012. Auxiliary placement in Rangi: A Dynamic Syntax perspective. PhD 

thesis. SOAS University of London. 

Gibson, H & L. Marten. 2015. ‘Contact and change in Bantu: the case of Rangi’ Paper 

presented at World Congress on African Languages, Kyoto Japan 20-24 August. 

Gibson, H. & L. Marten 2016. ‘Language contact and innovation: insights from 

colloquial varieties of Swahili, Sheng and African youth languages’ Paper presented at 2nd Swahili Baraza, SOAS University of London, 29 October 2016. 

Legère, K. 1992. Language shift in Tanzania. In (M. Brenzinger, ed.) Language 

Death: Factual and Theoretical Explorations with Special Reference to East Africa, 

pp. 99-115. 

Stegen, O. (2011): In quest of a vernacular writing style for the Rangi of Tanzania: 

Assumptions, processes, challenges. PhD thesis. University of Edinburgh. 

Yoneda, N. 1996. The impact of the Diffusion of Kiswahili on Ethnic Languages in Tanzania: 

A Case Study of Samatengo. In (S. Hino, ed). African Urban Studies IV, 29-37. 

Yoneda, N. 2016. ‘Swahilization of Bantu languages in Tanzania: The case of Matengo’. 

Paper presented at SOAS University of London. 20 May 2016.

ABSTRACT. Basque is the only ergative language in Western Europe. While the synchronic discussions on this (morphological) ergativity have abounded since the 70s (see Ortiz de Urbina 1989), Trask’s (1977) suggestion that Basque transitive verbs come from passive predicates has hardly received any evaluation, either in favor or against. No alternative account has been proposed either, with the exception of Aldai (2000), which entailed a long series of sinuous developments, and has come to a dead end. A complication of Trask’s hypothesis was the difficulty of finding any trace of a passivizing morpheme on the finite verb, presumably necessary to form active / passive oppositions. Trask’s proposal that the -n- of past/irrealis forms might be a remnant of it runs into a number of serious problems. I propose that these difficulties can be easily overcome by simply assuming that the roots of historical transitive verbs were originally unspecified for voice or diathesis, just as those of Chinese (as in yú chī-le ‘the fish has eaten ~ has been eaten’). Thus, the root kus of ikusi ‘see’ would have been ‘see’ or ‘be seen’ (depending on the context), and hence we would not need any passivizing morpheme. This assumption allows us to interpret historical present forms like n-a-kus-Ø P1sg-PRES-see-A.3sg ‘(s)he sees me’, as *ni da-kus S/A1sg ipfv-see~be.seen ‘I am seeing ~ being seen’, and couple it with intransitives like n-a-tor S1sg-PRES-come ‘I come’ < *ni da-tor (note that the reconstruction *ni da-kus, in application of Givón’s principle that “today’s morphology is yesterday’s syntax”, as well as the imperfective or progressive origin of the present morpheme, is commonly assumed in literature; see Gómez & Sainz 1995, Lakarra 2006, Ariztimuño 2013). The active or passive nature of predicates like *ni da-kus would be determined by the context. Thus, when followed by a direct object, they would be active (*ni da-kus Peru ‘I am seeing Peru’), whereas, when followed by an NP suffixed by the general locative *-ga functioning as an agentive complement, they would be passive (*ni da-kus Peru-ga ‘I am being seen by Peru’). Since all transitive finite forms of the present have the template P.person.marker-TAM.morpheme-verb.root-A.person.marker, they must all come from passive predicates which were later reanalyzed as active. The other alignment we find in the morphology of the Basque finite verb is the one present in the past/irrealis. In this morphological sector, we have P.person.marker+TAM-verb.root-A.person.marker(-PST.morpheme) when P = 1st or 2nd person (nen-karr-en ‘(s)he brought me’, zint-u-da-n ‘I had you’), but A.person.marker+TAM-verb.root(-PST.morpheme) when A = 1st or 2nd person and P = 3rd person (ne-karr-en ‘I brought it’). This distribution is expected according to the principles of typology, and, within the interpretation proposed, it allows to bring finite forms with P- in the first slot back to passive predicates, and those with A- in the first slot to active predicates, as in fact occurs in Nootka, Makah, Yurok and some other languages (see Trask 1979). What is puzzling is the fact that the only finite verbs with A- in the initial slot are found precisely in the past, more P-oriented, an anomaly already noted by Ortiz de Urbina (1989: 13-15). I suggest that the notion originally expressed by this morphological sector was the more A-oriented irrealis ((ba-)ne-kar ‘if I brought it’), and that later, in a period when the finite forms had already formed, the past —perhaps an aorist or a perfect in origin— was derived from it (→ ne-karr-en ‘I brought it’). This account also provides a pragmatic explanation for the word order change AVO > AOV, assumed by most scholars. The passive option of expressing predicates may have started as a marked construction restricted to cases where the agentive complement was inanimate. When this was generalized to any semantically transitive predicate, the semantic A stood at the end of the clause. Since A is typically animate, definite, and known information, it was promoted to the beginning of the clause (*Peru da-kus hi-ga > *hi-ga Peru da-kus ‘By you Peru is seen’). The subsequent active reanalysis of the clause brought about the fulfillment of the > AOV change (→ hi-k Peru dakus-k/n ‘you see Peru’; see Trask 1997: 247 for an embrionary version of this account). From a typological perspective, this account is important because, if accepted, it provides another instance where the ergative comes from a passive construction.

References

Aldai, G. (2000). “Split ergativity in Basque. The pre-Basque antipassive-imperfective hypothesis”. Folia Linguistica Historica 21. 31-98. Ariztimuño, B. (2013). “Finite Verbal Morphology”. In: Martínez-Areta, M. (ed.). 359-427. Gómez, R. & Sainz, K. (1995). “On the Origin of the Finite Forms of the Basque Verb”. In: Hualde, J. I. & Lakarra, J. A. & Trask, R. L. (eds.). Towards a History of Basque. Amsterdam: John Benjamins. 235-274. Lakarra, J.A. (2006). “Notas sobre iniciales, cambio tipológico y prehistoria del verbo vasco”. In: Lakarra, J. A. & Hualde, J. I. (eds.) (2006). R. L. Trasken oroitzapenetan ikerketak euskalaritzaz eta hizkuntzalaritza historikoak. Donostia/Bilbo: Gipuzkoako Foru Aldundia, EHU (= ASJU 40). 561-621. Martínez Areta, M. (ed.) (2013). Basque and Proto-Basque. Language Internal and Typological Approaches to Linguistic Reconstruction. Frankfurt am Mein: Peter Lang. Ortiz de Urbina, J. (1989). Parameters in the Grammar of Basque. Dordrecht: Foris Publications. Trask, R. L. (1977). “Hystorical Syntax and Basque Verbal Morphology: Two Hypotheses”. In: Douglass, W. A. & Etulain, R. W. (eds.). Anglo-American Contributions to Basque Studies: Essays in Honor of Jon Bilbao. Reno: University of Nevada. 203-217. — (1979). “On the Origins of Ergativity”. In: Plank (ed.). Ergativity. Towards a Theory of Grammatical Relations. London: Academic Press. 385-404. — (1997). The History of Basque. London: Routledge.

ABSTRACT. Mayan languages have undergone a variety of morphosyntactic splits in the alignment of person marking on verbs. The array of distinct alignment patterns that have emerged from the predominantly ergative alignment reconstructible for proto-Mayan is consistent with claims (e.g. Silverstein 1976) that ergative languages are particularly prone to splits in alignment. In this paper, I will describe four different split ergative alignment patterns found in Mayan languages—aspect-based splits, clausal dependency-based splits, person-based splits, and semantic agentivity-based splits—and the evidence for their development. Splits based on aspect are found in Yukatekan, Chol, Chontal, Ixil, and Poqomchi’. These languages have ergative alignment in the completive aspect and accusative alignment in the incompletive. In Ch’orti’, we find a modified ergative pattern, with ergative alignment in the completive, and a tripartite system in the incompletive. Another split alignment pattern, found particularly in Mamean and Q’anjob’alan languages, is an ergative-accusative split based on clause structure, in which dependent and other aspectless clauses follow an accusative pattern, and main clauses follow an ergative pattern. Yet another Mayan language, Mocho’, has an ergative split based on person, with ergative alignment for third person and accusative alignment for first and second person. Intriguingly, this same alignment has been proposed for some dependent clauses in Colonial Yukatek (Yasugi 2005). Finally, several Mayan languages, including Chol and Mopan (Gutiérrez Sánchez 2004, Danziger 1996), are split intransitive languages: patientive or unaccusative intransitives follow an ergative pattern (in the completive aspect) but agentive intransitives only occur in transitive light-verb constructions and arguably follow an accusative pattern of person marking, even in the completive aspect. This means, in practice, that ergative alignment in a language like Chol is evident only for patientive intransitives in the completive aspect. In these disparate changes, commonalities can be seen, particularly relating to the role of nominalization and subordination in their development (Larson and Norman 1979). However, I argue that grammatical details of split ergativity in each Mayan language reflect varied pathways of change, rather than a consistent evolutionary process. In addition, a large number of the changes in alignment patterns appear to have been spread through contact between Mayan languages, so that an understanding of the processes and pressures that gave rise to these changes requires attention to external, as well as internal motivations.

References Danziger, Eve, 1996. Split intransitivity and active-inactive patterning in mopan maya. IJAL, 62:379–414. Gutiérrez Sánchez, Pedro, 2004. Las clases de verbos intransitivos y el alineamento agentivo en el chol de Tila, Chiapas. Master’s thesis, CIESAS, México. Larsen, Thomas W., and William M. Norman. 1979. Correlates of ergativity in Mayan grammar. In Ergativity: Towards a theory of grammatical relations, ed. Frans Plank, 347–370. London/New York: Academic Press. Silverstein, Michael. 1976. Hierarchies of features and ergativity. In Grammatical categories in Australian languages, ed. R. M. W. Dixon, 112–171. New Jersey, NJ: Humanities Press. Yasugi, Yoshiho. 2005. Fronting of Nondirect Arguments and Adverbial Focus Marking on the Verb in Classical Yucatec. IJAL 71: 56-86.

ABSTRACT. The present paper concentrates on the diachronic aspects of the typology of transitivity oppositions and valency-changing categories. It also aims to draw attention to the regrettable imbalance of the synchronic and diachronic typological studies. Due to impressive achievements in the domain of synchronic typology of several linguistic categories, such as, in particular, alignment, voice and valency-changing categories (causative, reciprocal, reflexive), we have at our disposal full catalogues of possible alignment types and systems of valency-changing derivations attested in the languages of the world, their morphological, syntactic and semantic synchronic features and properties. By contrast, a systematic treatment of these categories in a diachronic perspective is lacking. Their rise, development and decline remain on the periphery of the typological interests. A powerful tool for such diachronic typological research is a diachronic typological questionnaire. Questionnaires are widely used in typology (particularly in the framework of the Leningrad/St.Petersburg Typological Group) for a synchronic study of valency-changing categories, such as causative, passive, reflexive, reciprocal etc. (e.g. Nedjalkov 2007; Geniušienė 1987). By contrast, my questionnaire (a preliminary version of which is offered in this paper) is diachronically-oriented, focusing on the most important features which determine the basic trends in the evolution of alignment and valency-changing categories. It seems advisable to start a diachronic typological research with collecting evidence from languages (or language groups) with a history well-documented in texts for a sufficiently long period of time (around 1000 years or more). On the basis of this evidence, it might be useful to outline a diachronic typological portrait of the category under study in the given language group, tracing it from the earliest attested texts in an ancient language (L0) onwards up to its reflexes in the daughter languages (L1, L2 etc.). Of particular interest would also be – if available – evidence from the sister languages of L0 (L´, L´´ etc.), which can serve as basis for a tentative reconstruction of the hypothetical history and possible sources of the category under study in proto-language *L. Such methodology can be applied, above all, to the study of alignment, voices and valency-changing categories in languages (language groups) with well-documented history, such as Indo-Aryan, Latin/Romance, or Greek. For instance, in the case of Indo-Aryan, the rich evidence collected by the Indo-European comparative linguistics creates a good basis for hypotheses about the origin and possible sources of the morphological categories (for instance, causative) attested in Old Indo-Aryan (Vedic Sanskrit) and thus provides material for a retrospective diachronic typological research. On the other hand, evidence from late Vedic and Middle Indo-Aryan texts, as well as from New Indo-Aryan languages, allows for a prospective diachronic study: how the Old Indo-Aryan categories develop into their reflexes in Middle and New Indo-Aryan). It is argued that on the basis of a diachronic typological analysis of historical evidence available from attested Indo-European languages and reconstructed for Proto-Indo-European, one may posit two basic types of the evolution of the encoding of transitivity oppositions. One major type (located in the West of the Indo-European area) includes Germanic, Romance, Greek and Slavic branches that share a number of parallel tendencies and can be qualified as instantiating a ‘syncretic’ type of evolution: (i) the middle voice (syncretically encoding several intransitivizing derivations such as passive, anticausative and reflexive) is either preserved (as in Greek), or replaced with a new middle, mostly going back to the Proto-Indo-European reflexive pronoun *s(u̯)e ; (ii) Proto-Indo-European morphological causatives (foremost, with the suffix *-éi̯e/o-), virtually disappear; (iii) labile verbs (of the type John opens the door / The door opens) become productive. By contrast, languages belonging to some Eastern branches of Indo-European radically abandon the syncretic strategy and develop special markers for voice and valency-changing derivations. These include, in particular, Indo-Aryan, Iranian and Armenian markers of the morphological (medio-)passive going back to Proto-Indo-European suffix *-i̯e/o-; Old Indo-Aryan reflexive pronouns tanū́- (originally meaning ‘body’) and ātmán- (‘breath’). Another such process, well-documented in Indic (Vedic) and Iranian (Avestan) is the grammaticalization of the reciprocal pronoun, Vedic anyó ... anyá- (anyonya-) and Avestan ańiiō ainīm ‘another ... another’ (Kulikov 2014). Furthermore, morphological causatives become productive in many Eastern branches, including Indo-Iranian, Tocharian and Armenian (causative marker uc‘anem based on the nasal present derived from a sigmatic aorist). One might call this type ‘antisyncretic’. Furthermore, constructions with non-canonical subjects, well-preserved in many Western Indo-European branches, become relatively rare in such Eastern branches as Indo-Aryan and Tocharian (Viti 2016), being replaced by patterns with canonical (nominative) subjects. An indirect trace of these reconstructable oblique subject patterns may be, however, the obligatory middle inflection, which becomes a frequent feature of the corresponding verbs in Vedic Sanskrit and Tocharian. By contrast, many members of the Eastern branches, in particular, Indo-Iranian, increasingly attest different object marking (see e.g. Dahl 2009). Some Eastern branches develop a number of features of ergative alignment. Typologically, the Eastern type, as attested in Indo-Aryan, shares more features with some non-¬Indo-European families, such as Turkic or Dravidian, rather than with the Western Indo-European type, as attested in Germanic or Greek. The East/West division within the Indo-European linguistic family provides us with important evidence for the existence of two important types of the evolution of the early Indo-European morphosyntax and can serve as a model for similar diachronic typological studies on alignment types and valency-changing categories in other language families.

References Dahl, E. 2009. Some semantic and pragmatic aspects of object alternation in Early Vedic. In: J. Barðdal & S.L. Chelliah (eds.), The role of semantic, pragmatic and discourse factors in the development of case, 23–55. Amsterdam: Benjamins. Geniušienė, E. 1987. The typology of reflexives. Berlin: Mouton de Gruyter. Kulikov, L. Grammaticalization of reciprocal pronouns in Indo-Aryan: Evidence from Sanskrit and Indo-European for a diachronic typology of reciprocal constructions. Journal of South Asian Languages and Linguistics 1.2: 117–156. Nedjalkov, V.P. (ed.) 2007. Reciprocal constructions. 4 vols. Amsterdam: Benjamins. Viti, C. 2016. The morphosyntax of experience predicates in Tocharian. Cahiers de Linguistique – Asie Orientale (CLAO) 45: 26–70.

ABSTRACT. While the lexicon of English has unmistakably been influenced by French in the course of its history (especially in the Middle English period), there is less agreement among linguists as to syntactic influence of French on English. Fischer (2013: 40), for instance, states that “syntactic borrowing is unlikely to have occurred as a result of contact with Latin and French”, but still allows for the existence of syntactic borrowing when an analogous construction was available in the target language (English). One such instance of an analogous construction is the absolute construction (AC), which has occurred across the various stages of the history of English as well as of French.

The absolute construction can be defined as a non-finite or verbless construction in which a (pro)nominal subject and a predicate (or head) are combined, and which often bears an adverbial relation to the matrix sentences (ex. 1 and 4). Absolutes can also be augmented (ex. 2 and 5), which means that they can be introduced by a preposition or conjunction. Typically, the head in past or present participle, but non-verbal heads are also possible (adjectival (ex. 3 and 6), adverbial, PP) (see van de Pol 2016: 1-2).

(1) Her mistress having died, she adopted us, not we her, about 5 weeks ago, .... (BNC, Personal letters, 1985-1994) (2) The Rebels win a pyrrhic victory, with the war destroying them almost beyond recognition. (http://www.cracked.com/article_22906_6-dumb-aspects-original-star-wars-trilogy-you-forgot.html, access 10-08-2015) (3) The mass of silver hair framed a perfectly formed face, though the lips seemed thinner, the eyes above the high cheek bones cold and unsmiling. (BNC, The prince of darkness, 1992)

(4) Sa mission accomplie, Lucien est pris en chasse par un autre tueur, Lenny. (corpus used by Télé Z, n° 489, 25.01.1992, example taken from Müller-Lancé 1994: 59) (5) Avec son mari buvant comme un trou, Bernadette est de plus en plus malheureuse. (example taken from Choi-Jonin 2007) (6) Gil [...]/Parvint, la lance haute et la visière basse, / Aux confins du pays. (Victor Hugo, Légende des siècles)

On the basis of parallel corpus data from the Penn-parsed corpora for Middle and Early Modern English (PPCME2, PPCEME), and the MCVF corpus for French, this paper examines contact influence of the AC in French on its English counterpart. In particular, it weighs different types of evidence of contact influence between French and English ACs, distinguishing between types of evidence for real-time influence in Medieval England (from 1100 up until about 1400) and post-contact effects.

Direct, real-time contact influence will be traced in two ways: 1. A contrastive diachronic analysis of ACs in the two languages in the period 1100-1500 will be carried out. Taking “synchronic snapshots” at 75/100-year intervals (intervals for English and French determined by the periodization in the PENN corpora), we will trace the various types of AC present at each of these moments, and the frequency distribution of the various types. If it can be demonstrated that particular types of AC in French precede (or occur more frequently at any given time than) their English counterpart, this is hypothesized to be suggestive of syntactic influence. In that case, the AC in English may still have developed independently from (but later than) the development in French, but, arguably, the fact that the AC was already present in English would have made it receptive to influence from French. French may therefore have reinforced the presence of ACs in Middle English. 2. Depending on the availability of relevant data, we will examine the typology and frequency of ACs in English texts translated from French (1100-1500) and compare with original English texts; to the extent that translated texts from French show a wider range of AC-types and a higher frequency of ACs than non-translated texts, contact influence may be (of will be argued to be) at play.

We will then examine post-contact effects (i.e., when English and French were no longer in direct contact). Tracing the afterlife of the constructions is likely to tell us more about the preceding period of direct contact: if both languages go their own way starting from a shared initial setting (the point reached after direct contact stopped or became less intense), this is additional evidence that there was contact in the period where they ran in parallel. An interesting issue in this respect is the development of the AC with quasi-coordinate (rather than adverbial) status in English.

References Choi-Jonin 2007. La construction en 'avec' en position détachée. In Nelly Flaux & Dejan Stosic (eds.) Les constructions détachées, Jun 2005, Timisoara, Roumanie, 57-74. Arras : Artois Presses Université. Available at https://halshs.archives-ouvertes.fr/hal-00497349/document Fischer, Olga. 2013. Contact and syntactic change in Old and Middle English. In Daniel Schreier & Marianne Hundt (eds.), English as a contact language, 18-40. Cambride: Cambridge University Press. Müller-Lancé, Johannes. 1994. Absolute Konstruktionen vom Altlatein biz zum Neufranzösischen. Ein Epochenvergleich unter Berücksichtigung von Mündlichkeit und Schriftlichkeit. Tübingen: Gunter Narr. van de Pol, Nikki. 2016. The development of the Absolute Construction in English. PhD Dissertation, KU Leuven.

CORPORA Corpus MCVF: Modéliser le changement : les voies du français, dir. France Martineau, Université d’Ottawa; http://continent.uottawa.ca/fr/corpus/corpusmcvf/ PPCEME: Penn-Helsinki Parsed Corpus of Early Modern English. 2004. Anthony Kroch, Beatrice Santorini, and Ariel Diertani. http://www.ling.upenn.edu/hist-corpora/. PPCME2: Penn-Helsinki Parsed Corpus of Middle English, 2nd edn. 2000. Anthony Kroch. http://www.ling.upenn.edu/hist-corpora/.

ABSTRACT. German sich (1, 2) and Spanish se (4, 5) can have a reflexive or an anticausative interpretation but only Spanish se (6) can have a passive reading while German sich cannot, as in (3). (1) Hans wäscht sich. (2) Die Tür öffnete sich. (3) *Die Wohnungen verkauften sich. Hans washes self the door opened self the flats sold selves “Hans washes himself.” “The door opened.” Intended: “The flats were sold.” (4) Juan se lava. (5) Se abrió la puerta. (6) Se vendieron los pisos.  John Reflse washes AntiCse opened the door Passse sold   the  flats “John washes himself.” “The door opened.” “The flats were sold.” We argue that Spanish Passse is the result of interaction between two linguistic cycles (Bahtchevanova & Van Gelderen 2016). We make two claims: i) pro merges in Spec,Voice in Passse, following MacDonald (to appear), due to the subject agreement cycle; and ii) se heads Voice due to the reflexive (se) cycle. Different se types are derived by the presence or absence of pro and se’s status as a head or a DP. The types of constructions a language has depends on whether it has pro and whether it has grammaticalized the reflexive pronoun as inflection. The reflexive cycle: Crosslinguistically, reflexive pronouns are grammaticalized as valency-marking morphology on the verb (Faltz 1985, Haspelmath 1990, i.a.). This cycle takes a full DP reflexive pronoun and turns it into a Voice head (Van Gelderen 2011, Maddox 2016). In German, this cycle has not taken place since, following Schäfer (2008), sich is an independent DP that has free word order status, abstract case, and selects the have auxiliary rather than be, showing that it is an argument in Spec,Voice. In Spanish, the cycle is complete. The pronoun se came from Latin. Maddox (2016) argues based on movement, coordination, modification, and auxiliary selection that se underwent grammaticalization whereby it changed from a full DP pronoun in Latin and Old Spanish (OS) to a D-head in Middle Spanish (MidS) to a Voice head in Modern Spanish (ModS). That se heads Voice or a similar projection is independently claimed by others such as Cuervo (2003, 2014), Folli & Harley (2005), i.a. Latin sē has the distribution of a DP: it can be coordinated, separated from the verb via XP-movement (7) and modified (8). (7) mē et   sē hīsce impedīvit nuptiīs! (8) sē ipse ... dēfenderet. me and Reflse this  shackled marriage Reflse very.M.S defended "He shackled me and himself in this marriage!" "He defended his very self..." Previous scholars (Rivero 1986, Fontana 1993, i.a.) have argued that separation from the verb, “interpolation,” is evidence for se as a DP in OS. Further support comes from auxiliary selection following McGinnis (2004). In OS, unaccusatives select the be auxiliary (9) while reflexives select have (10), suggesting transitive syntax in the latter. Thus, se is still a DP argument in OS. (9) Minaya Alvar Fáñez essora es llegado. (10) quando el se   ha  echado en tierra... Minaya Alvar Fáñez then is  arrived when he Reflse has thrown on ground “Minaya Alvar Fáñez then arrived." "...when he has cast himself to the ground..." In MidS, interpolation is lost. Auxiliary selection patterns as in OS (Aranovich 2003), so se is still a DP that moves as a D-head/determiner clitic. In ModS, se is inflection (Maddox 2016). The subject agreement cycle: Subject agreement affixes are grammaticalized from subject pronouns (Givón 1976, Lambrecht 1981, Jelinek 1984, Van Gelderen 2011). Spanish subject affixes from Latin and German subject affixes from Old High German are from Proto-Indo-European pronouns (Bopp 1857, Shields 1992). The subject cycle has three stages. At stage (a), the pronoun is a DP that moves to Spec,T to contribute interpretable φ-features. At stage (b), it is reanalyzed as a D-head and feature loss begins. At stage (c), it is reanalyzed as uninterpretable φ-features on T, triggering another element to merge. This “renewal” restarts the cycle. Spanish is at stage (c): there is subject agreement on the verb and the pronouns are full DPs; i.e., they can be coordinated (11), modified and be separated from the verb, as in (12). Crucially, the affix can be doubled by an overt pronoun or pro (13); i.e., renewal. Thus, having pro licensed by agreement in the sense of Rizzi (1982) is a result of the subject agreement cycle. (11) Tú y yo somos amigos. (12) Yo mismo no quiero ir. (13) Tú / (pro) com-es you and I are friends I myself not want-1.S to-go you pro eat-2S “You and I are friends.” “I myself do not want to go.” “You eat.” Since German is a partial null subject language (D’Alessandro 2014), it has gone through this cycle as well. German has pro, but this is only one of the ingredients needed to make passive se. Cyclic interaction: We adopt MacDonald’s (to appear) structures for AntiCse and Passse: (14) AntiCse: [VoiceP Voicese [VP DP ] ] (15) Passse: [VoiceP pro Voicese [VP DP ] ] Se as Voice head (due to the reflexive-cycle) and pro in Spec,Voice (due to the subject agreement cycle) in Passse in (15) is a type of cyclic interaction. Pro is in Spec,Voice in Passse because of renewal. It is absent in AntiCse. In both, se heads Voice. The difference reduces to distinct configurations of se (Voice) and the presence or absence of pro. Two predictions fall out of this: A) languages that have not grammaticalized the reflexive as a Voice head may have AntiCse but will not develop Passse due to the reflexive occupying Spec,Voice; B) a language that lacks subject agreement affixes of the kind that license pro per Rizzi (1982) will not develop Passse even if se heads Voice since there is no pro to merge in Spec,Voice; i.e., no renewal. Both predictions are consistent with the data. German sich is a full DP and not a Voice head (Schäfer 2008). It merges in Spec,Voice and thus pro cannot merge there, as in (16): (16) German: [VoiceP sich Voice [VP DP ] (17) Latin: [VoiceP sē Voice [VP DP ] In Latin, a null subject language, sē was a full DP, but it only had Reflse (see 7 and 8) and AntiCse as in (18) below (Geniušienė 1987). While Latin did have pro, it could not merge in Spec with anticausatives since sē was there (17). Passse developed in Late Latin/early Romance (Cennamo 1999, Adams 2013), when se begins to be reanalyzed as a head. This explains the presence of Passse in early Spanish texts, despite interpolation showing that, for some speakers, se was still a full DP. In Latin, interpolation with sē was frequent while in OS, it was rare (Maddox 2016), suggesting reanalysis as a head was still in progress but almost complete in OS. (18) dum calor sē frangat. (19) Les pieds (*pro) se bougent sur la piste de danse. while heat AntiCse breaks the feet pro Passse move on the floor of dance "While the heat breaks..." “One moves one’s feet on the dance floor.” French appears to contradict our second prediction since it has Reflse, AntiCse, and Passse, as in (19) above. However, Passse existed in Old French (Cennamo 1993), which was a null subject language (Roberts 1993, Vance 1997), so pro could merge in Spec,Voice in Passse. Thus, French is actually consistent with our prediction: it developed Passse at a time when it had subject pro and se as Voice head. Passse survived into Modern French as a historical remnant of Old French. References: Bahtchevanova, M. & E. van Gelderen. 2016. The interaction between the French subject and object cycles. In E. van Gelderen (ed.), Cyclical Change Continued. Amsterdam: John Benjamins. Geniušienė, E. 1987. Typology of Reflexives. Berlin: Mouton De Gruyter. Schäfer, Florian. 2008. The Syntax of (Anti-)Causatives. Amsterdam: Benjamins. Shields, K. 1992. A History of Indo-European Verb Morphology. Amsterdam: John Benjamins. Diachronic data: (7), Terence, Phormio, 2.4; (8), Caesar, de Bello Gallico, 20.5; (9) Cid, l. 2449; (10), Ferrer Sayol, Libro de Pallado BNM 10211, para. 115; (18), Cicero, De oratore 1.265.

ABSTRACT.  MEDIEVAL  ROMANCE AND ITS PLACE IN THE  VERB  SECOND  TYPOLOGY

BACKGROUND: Claims that the Romance languages went through a period in which they showed a multitude

of ‘Germanic-like’ properties are widespread (Adams 1987; Vance 1988, 1995, 1997; Roberts 1993; Mathieu

2006, 2009; Benincà 2013; Salvesen 2013; Poletto 2014) and have at their heart a proposal that some or all

of the Romance languages went through a Verb Second stage (cf. in particular Benincà 1983-4 et seq.).

Despite the fact that a number of scholars have highlighted significant differences between the instantiation

of the V2 property in Medieval Romance and (Modern) Germanic (Sitaridou 2012) and differences between

the Medieval Romance languages (Wolfe 2016, 2016b), there is to date no account to formally capture these

differences and integrate the data into a pan-Romance and Germanic typology of the V2 property.

METHODOLOGY AND ASSUMPTIONS: The present paper outlines a first attempt at such an account, by drawing

on a new corpus of fourteen Medieval Romance texts representing seven different varieties from the

10th-14th centuries (sample size between 1000-2000 clauses per text). We assume an articulated set of leftperipheral

projections in line with Rizzi (1997, 2004 et seq) and draw on this intuition to suggest that the

features responsible for the V2 property, namely a Probe triggering finite verb movement and an Edge

Feature requiring merger of a phrasal constituent in a CP-Specifier can have different loci between V2

languages. This facilitates a novel account of intra-Romance V2 variation both synchronically and

diachronically, along with that observed between the Romance and Germanic language families.

VERB MOVEMENT: We present a range of qualitative and quantitative arguments that suggest, contrary to the

situation in the modern (non-Raeto-) Romance languages that finite verb movement reaches a target higher

than the T-layer in the Medieval languages, as in most early and modern Germanic varieties (cf. Vikner 1995;

Holmberg 2012; Walkden 2014). The key arguments for this parallelism between Germanic and Medieval

Romance come from  (i) the attestation of Germanic inversion constructions, which differ from Modern

Romance ‘free inversion’, in that the postverbal subject is unambiguously in Spec-TP (1a),  (ii) Matrix clauses

with direct object fronting with no clitic resumption, which are shown not to be marginal in frequency but to

account for 12-31% of V2 clauses (1b),  (iii) the position of the finite verb which precedes high temporal

deictic adverbials lexicalising the leftmost projections of the T-layer (Cinque 1999; Ledgeway in press) (1c)

and ( iv) quantitatively significant matrix/embedded asymmetries, with embedded clauses showing absence

of Germanic-inversion and large reductions in XPNon-Subject—V-S orders, yielding a dominant SVO order in all

varieties bar Old Sardinian where embedded VSO dominates (1d):

(1) a. Ja vos= avoit il si longuement servi

already you=have.3SG.PST he so long serve.PTCP

‘He has already served you such a long time’ (French, Quête 119)

b.

c.Lo abbate  vene ora ’n derieto

the abbot come.3SG.PST now in behind

‘The abbot now came back’ (Sicilian, DSG 15, 15)

c.(E) co  dis=el plusor fiade

and this say.3SG.PST=he often

‘And he said this often’ (Venetian, Lio Mazor, 19)

d. c’ avia ego binkidu per rasone

because-have.1SG.PST I win.PTCP for reason

‘…because I had justly won’ (Sardinian, S.Maria di Bonarcado 2)

STRUCTURE OF THE LEFT PERIPHERY: Although we argue that V-to-C movement and a locus of ‘EPP-effects’ in

the C-domain are properties common to all Medieval Romance languages considered, the structure of the

left periphery shows notable diachronic and synchronic variation (as may well be the case in Early and

Modern Germanic (Petrova & Solf 2008; Walkden 2014; Holmberg in press). In particular, we argue that in all

Early (11th-12th century) Medieval Romance languages and 13th century Sicilian, Occitan and Neapolitan

the locus of the V Probe and Edge Feature responsible for V2 is Fin, the lowest head in an articulated left

periphery. In common with the Early Germanic varieties discussed by Axel (2004, 2005, 2007) and Walkden

(2014, 2015) this Fin-V2 syntax manifests itself most clearly in the ability for multiple left-peripheral

constituents to precede the finite verb, yielding V>4 orders across the corpora (2a). By contrast, in

approximately 1200 a reanalysis towards an innovative stricter V2 syntax takes place in western Spanish

varieties, Venetian and French where verb movement targets Force and the preverbal field is impoverished

to only consist of the Hanging Topic-Frame-Force fields. V>4* orders are no longer licensed in the Force-V2

texts, with a restricted set of left-peripheral speaker-oriented adverbials, Hanging Topics and Frame-Setting

clauses acting as the sole triggers for V3 (2b). We argue that this situation receives clear parallels in the

Later Old and Middle High German data discussed by Axel (2004) and the contemporary West Flemish data

in Haegeman & Greco (in press), where Frame-Setters readily permit V3 orders whilst other left-peripheral

constituents do not:

(2) a. Illi, per amor del Senhor, lur= lavava los pes

she for love of-the Lord them=wash.3SG.PST the feet

‘Through her love of the Lord, she washed their feet’ (Occitan, Douceline 45)

b. Et luego que llego a la puerta el diablo  abrioge=la

And soon that arrive.3SG.PST at the door the devil open.3SG.PSt=it

‘And as soon as he arrived at the door, the devil opened it’ (Spanish, Lucanor 204)

Our proposal is that Modern German and Dutch occupy a distinct sub-type of the Force-V2 typology where

Hanging Topics can trigger V3, whilst Frame-Setters cannot. We analyse this as result of partial syncretism

in the C-system between the functional heads Frame and Force (Force V2 B). These data together yield the

tentative V2 typology in (3), where the presence of a V-movement trigger and an Edge Feature in the Cdomain

are the commonalities, whilst the precise makeup of the left-peripheral sequence of heads yields the

variation between systems:

(3) a.  Fin-V2 > [HT [Frame [Force [Topic [Focus [Fin  V2-XP [Fin⁰  V] [T… v…V…]]]]]]

b.  Force-V2 A > [HT [Frame [Force  V2-XP [Force⁰  V] [Topic [Focus [Fin V2-XP [Fin⁰ V] [T… v…V…]]]]]]

c.  Force-V2 B > [HT [Frame/Force  V2-XP [Frame/Force⁰  V] [Topic [Focus [Fin V2-XP [Fin⁰ V] [T… v…V…]]]]]

LICENSING OF NULL ELEMENTS: Although the implications of this approach with variation in the functional

structure between V2 systems have been explored for overt left-peripheral constituents (cf. Poletto 2002 and

Walkden 2015), we suggest the typology can fruitfully account for the distribution of null elements. Drawing

on the wide literature on V1 in Germanic and extending this to Romance, we propose that V1 is licensed  (i)

in the presence of a null discourse operator in Spec-ForceP in Narrative Inversion contexts,  (ii) in contexts of

discourse continuity or rhematicity with a null pronoun in Spec-TopP and  (iii) in contexts where a null topic

associated with a discourse participant occupies Spec-FrameP, typically in spoken but not written language.

We propose that whilst  (i-iii) are all attested in Romance and Germanic Fin-V2 systems, option (ii) which

leads to widespread verb-initial structures is not found in Force-V2 systems with heavy restrictions on V>4.

This is linked to the licensing condition for a special variant of pro, proTop, whose licensing we propose is

contingent upon TopP forming part of the verbal prefield. We suggest this yields both heavy restrictions on

the conditions permitting V1 clauses in Force-V2 systems and the fact that Null Subjects and Topics in Later

Old French, Later Western Old Spanish and Later Old High German (Axel 2005) are only licensed postverbally.

PATHWAYS OF DIACHRONIC CHANGE: Taken together our data shed new light in the pathways of change

attested in both Germanic and Romance which show a remarkable degree of similarity. In particular, we

suggest that the putative Germanic ‘incipient’ VSO stage hypothesised by Hinterhölzl & Petrova (2010) finds

a cognate in the syntax of 10th century Sardinian. This we argue is a ‘relic’ of a precursor to the Fin-V2

syntax found across Early Romance and Germanic, which in certain varieties is reanalysed as a Force-V2

system. We analyse the SOV > VSO > Fin-V2 > Force V2 pathway as being analogically conditioned by the

spread of a featural [+EF, +V] makeup of a growing class of functional heads ‘upwards’ within the clausal

spine and suggest that the changes concerning the functional structure in the left-periphery yield a cascade

of similar changes in the null argument system and syntax-pragmatics interface of both Germanic and

Romance.

ABSTRACT. The purpose of this paper is be to explore the ramifications of contact on interpretations of language history. Nigerian Arabic (NA henceforth) has constructions which, if compared to other varieties of Arabic, typologically adheres to processes of grammaticalization. Looked at in isolation, NA as a self-contained entity, the examples in (4-6) below can be read as the result of internal developments. Triangulating between other varieties of Arabic and languages of the Lake Chad basin, however, it is clear that (4-6) were added via contact.

The issue NA has a set of demonstratives largely identical in form and function with demonstratives in other Arabic regions, marking the categories SG/PL, M/F and near/far. The near demonstratives, for instance, are identical with those of some dialects of Upper Egypt, the area from which ancestral NA migrated in the late fourteenth century.

(1) NA EA (Upper Egypt, Bʕeeri, near Luxor) near far near far SG da/di (M/F) ďaak/ďiik da/di dukkaati/dukkiiti PL dool/deel ďoolak/ďeelak dool/deel dukkuṃṃa/dikkinna

As in all varieties of Arabic, these demonstratives are used adnominally (2) and pronominally (3, all sentential citations are from recorded, natural speech).

(2) al-hadiid-e di bi-diss-uu-ha DEF=iron-F this.F 3-stick-MPL-it.F This iron they stick in … (TV76b) (3) di al-arabiyye kuure ‘this.F DEF-Arabic formerly ‘This was Arabic formerly’ (TV76b = text from which citation taken)

At the same time, unlike virtually all varieties of Arabic outside of the Sudanic region, the demonstrative occurs in a series of further functions:

4. Marking the closure of relative, conditional and temporal clauses [kan ma gaal-a da] ma ba-arf-a if not said-it DEM.M not I-know-him ‘If he didn’t say it, I wouldn’t recognize him’ (IM17a) 5. Marking personal pronouns ana di ma ind-í gurs I DEM.F not at-my money ‘Me (F) I don’t have money’ (GR103) 6. Marking adverbs hineen da kula kulu arab bas here DEM.M as well all Arabs only ‘Here also there are only Arabs’ (IM82)

Note that in Egyptian Arabic, and indeed in all varieties outside of the Sudanic area, demonstratives in positions such as (4-6) are unthinkable.

At first blush it might appear that the extensions in (4-6) are a “natural” result of grammaticalization processes. (4), for instance, has a number of characteristics of the grammaticalization of demonstratives documented in Diessel (1999: 19-20). The demonstrative in (4), for instance, is specific to the dependent clause context and if not obligatory, is at least highly frequent in this context, it only occurs in the singular, it tends to be M da to the exclusion of F di (though di also occurs) and it has a boundary marking, not deictic function. It turns out, however, that all of (4 - 6) are attributes of demonstratives or demonstrative-like elements which are found throughout the languages of NE Nigeria and beyond (Diessel 1999: 26). For instance, Glavda, a very small Central Chadic (Biu-Mandara) language spoken in the Ngoshe area about 20 miles south of the most southern border of NA in Nigeria, marks relative and conditional clauses (almost obligatorily), personal pronouns and adverbs (the latter two optionally, as with NA (5, 6)) with a discourse specificness identifier (variously –n, -íin, -áan, termed DSI here).

D makarant-íin [an-a tág ŋ-áyə-n] kəna ba school-DSI I-PER say SbjP-I-DSI DM DM (that is) the school which I talked about’. It stands to reason that the NA demonstrative besides maintaining its inherited deictic functions has borrowed the function in (4) via contact and hence in this respect aligns itself with other languages of the Lake Chad region, rather than with other varieties of Arabic. This is not the only domain where contact has led to a significant change in structure (Owens 2014, 2016 on idiomaticity).

In this presentation I will first describe the special behavior of the NA demonstrative in contexts such as (4 - 6), drawing on an extensive, online corpus of NA. Since the corpus is semi-automatized for concordance searches, it allows a complete overview of nearly 300,000 words of oral text. Of the 2,801 tokens of ana ‘I’, for instance, 247 or nearly 9% of the total are marked by demonstratives (either da ‘this.M or di this.F, as in (5)). I will briefly outline the comparable behavior of the discourse specificness marker in Glavda. The choice of this language is dictated by availability of material (Owens 2010) and by a randomness factor. Glavda, so far as we know, would not have had extensive enough direct contact with NA to have had a structural impact on it. However, the fact that the Glavda specificness marker and the NA demonstratives show a high degree of functional overlap confirms that one is dealing with a widespread areal (Sprachbund) phenomenon to which NA belongs. As shown elsewhere (Owens 1996), the moving force in the formation of this Sprachbund were the dominant Kanuri of the Kanem-Bornu empire, under whose suzerainty the Nigerian Arabs have lived since their arrival in the area.

The NA material shows that grammaticalization effects can be extensively borrowed. From an historical perspective it warns that what in a given language might be assumed to be a case of grammaticalization is in fact one of borrowing. Further ramifications of this perspective will be elaborated upon.

ABSTRACT. Having spent the last millennium in close contact with several Indo-European languages, Maltese, the modern descendant of Siculo-Arabic (Brincat 2011), possesses various pluralization strategies. In this paper I explore the historical development of two such strategies of Semitic-origin: the suffixes -iet and -(i)jiet (e.g. papa ‘pope’, pl. papiet; omm ‘mother’, pl. ommijiet). Specialists agree that both suffixes originated from the Arabic plural suffix -āt (Borg 1976; Mifsud 1995); however, no research has explained the development of -(i)jiet, nor connected its development to that of -iet. I argue that the development of -(i)jiet was driven by the influx of i-final words which resulted from contact with Italian: Maltese speakers affixed -iet to such words, triggering a glide-epenthesis that occurs elsewhere in Maltese (e.g. Mifsud 1996: 34) and in other varieties of Arabic (e.g. Erwin 1963; Cowell 1964; Owens 1984; Harrell 2004). With a large number of plurals now ending in ijiet, speakers reanalyzed this sequence as a unique plural suffix and began applying it to new non¬-i-final words as well. Since only Maltese experienced this influx of i-final words, it was only in Maltese that speakers reanalyzed this sequence as a separate suffix. Additionally, I argue against an explanation of the development of -(i)jiet that does not rely on epenthesis. With regard to the suffix -iet, I argue that two properties unique to it – the obligatory omission of stem-final vowels upon pluralization, and the near-universal tendency to pluralize only a-final words – emerged from a separate reanalysis of the pluralization of collective nouns that reflects a general weakening of the Semitic element in Maltese under Indo-European contact.

To test these hypotheses I first surveyed Aquilina’s (1987-90) Maltese-English-Maltese dictionary for all words pluralized by -iet and -(i)jiet (1,449 and 2,387 words, respectively). Three patterns emerged: (1) -iet exclusively pluralizes words having an a-final (1,386 words, or 95.7%) or consonant-final (63 words, or 4.3%) singular, and never singulars ending in any other vowel; (2) among words which Aquilina identifies as having entered Maltese through English, and which I thus assume represent the most recent loanwords in Maltese (Borg & Azzopardi-Alexander 1997: xii-xiv; Brincat 2011), -iet only pluralizes those for which the singular ends in a; (3) -(i)jiet can pluralize words ending in any segment, but tends to pluralize i-final words (1,122 words, or 47%), the majority of which Aquilina identifies as having entered the language through Italian.

Extensive contact with Italian introduced a large number of i-final loanwords into Maltese. I hypothesize that speakers historically pluralized such words with -iet, bringing stem-final and suffix-initial vowels into contact and so triggering the epenthesis of j. Glide-epenthesis is attested elsewhere in Maltese between stem-final high vowels and various suffix-initial vowels (e.g. mieli ‘wealthy’ + -a ‘FEM’ → mielija, + -in ‘PL’ → mielijin; gru ‘crane’ + -a ‘SG’ → gruwa, + -iet ‘PL’ → gruwiet) (Mifsud 1996:34), as well as in Maghrebi varieties of Arabic closely related to Maltese (e.g. Eastern Libyan Arabic (Owens 1984), Moroccan Arabic (Harrell 2004)) and in more disparate varieties (e.g. Iraqi Arabic (Erwin 1963), Syrian Arabic (Cowell 1964)). With many plurals suddenly ending in ijiet, Maltese speakers reanalyzed this sequence as a plural suffix and began applying it to new words: as -ijiet on consonant- and some vowel-final words (deleting the stem-final vowel in the latter case), and as -jiet elsewhere (preserving the vowel). Although they likewise separate stem-final vowels and the reflex of -āt with an epenthetic glide, other Arabic varieties did not experience a comparable influx of i-final words and so a similar reanalysis never occurred. Thus, although a general, non-Maltese-specific epenthesis process was responsible for the glide of -(i)jiet, I attribute the reanalysis of the resulting ijiet sequence as a unique suffix and its subsequent extension to new forms to the extensive contact with Italian which only Maltese experienced.

One may wonder whether the glide of -(i)jiet instead originated from the suffixation-triggered reemergence of a root-final glide, or whether such a process could have played any role in the development of the suffix. To answer this, I surveyed Aquilina (1987-90) for all roots having j or w as the final radical. When a stem having such a root and ending in a vowel instead of the glide takes a suffix, suffixation triggers both the restoration of the glide and the loss of the vowel, creating a consonant-glide cluster (e.g. xidi ‘gadfly’ + -a ‘SG’ → xidja, + -iet ‘PL’ → xidjiet; root: x-d-j). This is inconsistent with the behavior of -(i)jiet: otherwise, -jiet marks only vowel-final stems. Thus, I regard such plurals as bearing the -iet suffix and reject any connection between this process and the development of -(i)jiet as both unnecessary and implausible.

Some words, especially those of Semitic-origin, have collective (a type of syntactically-singular noun denoting an uncountable genus, species, or material), singular, and plural forms; the singular and plural derive from the collective via the affixation of -a and -iet, respectively (e.g. collective lawż ‘almond’ + -a ‘SG’ → lawża, + -iet ‘PL’ → lawżiet) (Mifsud 1996). Reflecting a general weakening of the Semitic features of Maltese under prolonged contact with Indo-European languages, Mifsud (1995, 1996) finds that Maltese speakers have tended to abandon grammatical features atypical of Indo-European languages (e.g. collectivity) and to treat collectives as plural nouns. I propose that Maltese speakers reanalyzed -iet here as marking the a-final singular (rather than the collective form), deleting the vowel in the process, and so associated -iet as a pluralizer of ¬a-final singulars. Hence, -iet tends to mark new words having an a-final singular, and deletes the vowel even when it is not a suffix (e.g. papiet ‘popes’ < papa, *pap) in modern Maltese.

To summarize, I have argued that extensive contact with and borrowing from Italian ultimately drove the development of the Maltese plural suffix -(i)jiet. Additionally, I have argued that the suffix -iet acquired two unique properties due to a separate reanalysis that occurred as part of a general weakening of the Semitic features of the language under Indo-European influence.

References Aquilina, Joseph. 1987-90. Maltese-English-Maltese dictionary. Malta: Midsea Books. Borg, Albert J. 1976. The imaala in Maltese. Israel Oriental Studies VI. 191-223. Borg, Albert J. & Marie Azzopardi-Alexander. 1997. Maltese. London: Routledge. Brincat, Joseph. 2011. Maltese and other languages: a linguistic history of Malta. Sta Venera, Malta: Midsea Books. Cowell, Mark W. 1964. A reference grammar of Syrian Arabic, based on the dialect of Damascus. Washington, D.C.: Georgetown University Press. Erwin, Wallace M. 1963. A short reference grammar of Iraqi Arabic. Washington, D.C.: Georgetown University Press. Harrell, Richard S. 2004. A short reference grammar of Moroccan Arabic. Washington, D.C.: Georgetown University Press. Mifsud, Manwel. 1995. The productivity of Arabic in Maltese. Proceedings of the 2nd International Conference of AIDA, 151-60. Cambridge: Cambridge University Press. Mifsud, Manwel. 1996. The collective in Maltese. Rivista di Linguistica 8. 29-52. Owens, Jonathan. 1984. A short reference grammar of Eastern Libyan Arabic. Wiesbaden, Germany: Harrassowitz.

ABSTRACT. A Quantitative Investigation of Noun Pluralization in Cypriot Maronite and Maltese Arabic

Cypriot Maronite Arabic (CMA) is an endangered variety with speakers all bilingual in Cypriot Greek. Maltese is a well-known example of a mixed morphological system which has adopted significant amounts of productive Romance morphology as well as vocabulary. In this study, I investigate the effects of language contact and attrition/endangerment on their noun plural formation systems. Like other varieties of Arabic, these two include both suffixal noun plurals and non-concatenative “broken” plurals. However, inflectional morphology, particularly for nouns, is known to be vulnerable in heritage language and language attrition, in general (Dorian 1978, Preston 1982) and for Arabic in particular (Albirini & Benmamoun 2014, Benmamoun et al. 2014). This leads to the following research questions for these two varieties: To what extent is the broken plural system still active? If less active, is the change categorical/complete, or in a gradient transitional stage? If less active, what pluralization patterns are replacing broken ones? More frequent broken one(s)? Etymologically Arabic suffix(es)? Or Greek/Romance suffix(es)? If still active, is the distribution of attested broken patterns the same as in other Arabic varieties?

Cypriot Maronite Arabic The set of 251 native-origin pluralizable nouns in Borg’s (2004) CMA dictionary is thus distributed:

Suffixation Broken plural Other Total N 92 134 25 251 % 37 53 10 100 Table 1. Distribution of CMA plural types

In CMA, broken plurals exceed previous estimates for Arabic (20-40%, though 59% in Khouloughli (1992)). Within the category “other,” the largest group is nouns with Greek-suffixed forms. Within the category of broken forms, plurals are distributed as follows:

CaCaa?iC/CawaaCiC CiCaaC/aCCaaC CuCuuC Other Total CMA N 69 42 19 5 135 CMA % 51 31 14 4 100 MSA % 30 42 12 18 102 (rounding) Table 2. Distribution of CMA and MSA (Khouloughli 1992) broken plural types.

Trisyllabic plurals have expanded in CMA relative to MSA, accounting for over half of broken plurals. However, the top two broken plural categories remain the same, although reversed, and both use CuCuuC plurals as the next most common option. Notably, MSA makes more use of additional patterns, while these have all but disappeared from CMA. Item analysis reveals that 28% (26/92) of CMA suffixed plurals take broken plurals in other varieties of Arabic. The change is unidirectional – there are no CMA broken plurals which etymologically take sound plurals, and only rare cases in which nouns switch between broken plural categories.

Maltese Previous research has claimed that the broken plural accounts for under 10% of the Maltese lexicon, versus various estimates of ~20% to ~40% for Modern Standard Arabic. However, I find that broken plurals are far more robust for the core lexicon exemplified in two different pedagogical grammars of Maltese, accounting for approximately 40% of plural forms in both. At least in this subset of the lexicon, broken plurals as a class comprise the largest percentage of plural forms, even though the Romance plural /-i/ suffix is the second largest class and outnumbers any individual broken plural pattern. Within the broken plurals, the CCVVCVC pattern is most common in both sources (52.5%) followed by the CCVC pattern (27.5%), then a variety of other patterns. This tallies with the distribution among the broken plurals attested in the Maltese Language Resource Server online corpus (40% CCVVCVC, 27% CVC). However, it differs from the distribution in Modern Standard Arabic, as will be explained.

Conclusion Broken plurals remain a robust component of CMA grammar. However, their variety is now restricted, somewhat vulnerable to regularization, and no longer productively extended to new forms. In Maltese too, broken plurals remain a robust part of the Maltese lexicon, though also apparently no longer productive. Lack of productivity in tandem with centuries of intense contact has not led to the shrinking of this morphological class to a handful of exceptional forms. I speculate that this is possibly due to their high salience as grammatical structures, as found for L2 speakers (Walter 2011).

ABSTRACT. Introduction. Consider the sentences in (1):

(1) a. Juan está bailando.

          John dance.PRS.PROG.3.SG

          ‘John is dancing’

      b. Juan baila

          John dance.PRS.IMPF.3.SG

          ‘John dances/is dancing’

From a synchronic perspective, the sentences in (1) show that Spanish has two markers to express an event-in-progress meaning: the unambiguous periphrastic marker in (1a) and the ‘simple present’ marker in (1b), which is ambiguous without a context because it can also be used to express generic and habitual meanings. These two markers have been claimed to be alternative options to convey events-in-progress and hence to be in free variation [1, 2]. However, other descriptions of the Spanish Progressive do not treat them as free alternants, and present the intuition that the progressive form actualizes an event [3] or enhances its duration [4], in contrast to the use of the simple present form.

These markers are also diachronically related. It has been shown that linguistic markers that are linked to specific meanings evolve in consistent ways across times and cross-linguistically in what are known as grammaticalization paths [5]. One frequently observed change is the Progressive to Imperfective shift [5, 6, 7], where a progressive marker arises to express an event-in-progress meaning whence it encroaches upon the more general imperfective domain. In this change, a new marker gets recruited to express a specific meaning, becomes categorical over time and thus obligatory in some contexts, and slowly generalizes to all contexts to express a broader imperfective meaning. In this vein, if Spanish presents two markers that alternate freely to refer to events-in-progress, it can be claimed that the language is in a recruitment-to-categoricalization stage. Nevertheless, these stages comprise sub-stages that can differ in more minimal ways.

To better characterize the relation between these two markers and the sub-stages within the diachronic path, we collected experimental data to test the hypotheses of whether the markers are in free variation or if the use of the variants is contextually determined. If so, we intended to understand what characteristic of the context facilitated/inhibited the use of one or the other marker. We considered that three different varieties of Spanish –Iberian Spanish (IS), Rioplatense Spanish (RS) and Mexican Altiplano Spanish (MS)– could show variation as a manifestation of different stages of the diachronic path. The ultimate goal of this study was to explain the reasons behind the choices of the speakers when using one or the other marker, given that those choices will lead to following stages in the shift.

Our hypothesis was that the imperfective marker is already a dispreferred form to express an event-in-progress meaning in Spanish, and that the progressive marker has specialized its meaning and has taken over most instances of event-in-progress utterances. We claim that it is the context   –in particular, the richness of the cues to aspectual and temporal properties of the event provided by it, and the degree to which the speaker can assume that the hearer has access to them– the one that drives the choice of the speakers between using an unambiguous marker or a more general one that relies on contextual information to disambiguate the intended reading. 

Methods. We assessed the degree of acceptability of each of the markers (PROG/IMPF/and Pretérito as a baseline condition) to express an event-in-progress in three dialects of Spanish (IS, RS, MS). 114 subjects (39 IS, 38 RS, 37 MS) rated the acceptability of 180 context-sentence pairs in a questionnaire task on a Likert scale from 1 to 5. The independent variables were the amount of temporal and aspectual information provided in the context (Rich/Poor), the aspectual marker (PROG/IMPF/PRET) and the grammatical person (1^st/2^nd/3^rd). 

Results. A linear mixed effects model analysis showed an effect of Context in the choice of Aspectual Marker by Dialect, and an effect of Grammatical Person in the choice of Aspectual Marker. The three dialects showed a clear preference for the PROG-marker to express events-in-progress regardless of contextual information (IS = 4.77, RS = 4.67; MS = 4.49) in comparison to the IMPF-marker (IS = 3.93, RS = 3.67, MS = 3.54). However, as is shown in the graph, RS and IS speakers showed that their choice of IMPF-marker was significantly modulated by contextual information (IS: Rich = 4.18; Poor = 3.69, p < .0001 -- RS: Rich = 3.90; Poor = 3.43, p < .0001). MS speakers, in contrast, were not sensitive to contextual information, and dispreferred the marker across contexts (Rich = 3.57, Poor = 3.51, p =.673). The Grammatical Person effect showed that speakers give significantly higher ratings to the IMPF-marker to express events-in-progress when the sentence is in the second person in comparison to the first person in all dialects (IS: 2^nd person = 4.16, 1^st person = 3.67, p < .0001 -- RS: 2^nd person = 3.86, 1^st person = 3.38, p < .0001 – MS: 2^nd person = 3.79, 1^st person = 3.26, p < .0001).

Discussion. We showed that when conveying information about an event in progress, Rioplatense and Iberian Spanish speakers show sensitivity to the temporal and aspectual properties of the situation provided by the context and the extent to which hearers are aware of them. These speakers are able to use the ambiguous, more general imperfective marker when they can rely on contextual information to disambiguate the meaning at play. They use the unambiguous progressive marker otherwise. In contrast, for Mexican Altiplano Spanish speakers, enrichment of the context does not result in an increased acceptability in their use of the imperfective marker for events-in-progress.

We conclude that all varieties are in a recruitment-to-categoricalization stage. The variation however is not free. The use of the progressive marker is not yet completely obligatory to express an event-in-progress meaning. While Rioplatense and Iberian Spanish still allow for the use of the imperfective marker when the contextual information is sufficiently rich, Mexican Altiplano Spanish appears as one step further in the categoricalization process, given its systematic dispreference of the use of the imperfective marker to express event-in-progress meanings. We thus understand progressive-imperfective variation in Spanish as the manifestation of sub-stages within a grammaticalization path determined by speaker sensitivity to independently motivated properties of the context. 

References. 1. Comrie, 1976; 2. Bertinetto, 2000; 3. Fernández de Castro, 1999; 4. Roca Pons, 1958; 5. Bybee et al. 1994; 6. Dahl 1985; 7. Deo, 2015.

ABSTRACT. According to a widely-cited statement by Bolinger, ‘the moment a verb is given an infinitival complement, that verb starts down the road of auxiliariness’ (1980: 297). However, to the best of our knowledge, there have been few attempts to explain how verbs acquire infinitival complements in the first place. The goal of this paper is to provide such an explanation for a single class of lexical verbs that are known to grammaticalize into tense-aspect auxiliaries, namely, verbs meaning FINISH, which is understood here as verbs that denote the act of bringing a process to completion. In this talk, we conduct a logistic regression analysis over of all 1962 tokens of FINISH constructions (involving the verb acabar) in a corpus of Spanish historiographical texts from the mid-13th to the end of the 18th century. Our main findings are two. First, we identify a hitherto undescribed process of language change in which previously inferred meanings are later obligatorily expressed in an overt fashion, a process which we call here overtification. Second, we account for the construction's subsequent acquisition of a temporal meaning as the result of a context-induced reanalysis, motivated by Eckardt's principle of 'Avoid Pragmatic Overload.'

ABSTRACT. Historical Sociolinguistics has afforded a new interest in tracing heterogeneity and the vernaculars in the history of languages, reconstructing the sociolinguistic contexts and direction of language changes developed longitudinally as well as socially-based variation patterns in remote speech communities –legitimating the historical validity of some ‘sociolinguistic universals’: the curvilinear hypothesis, overt-covert prestige patterns, life-span-change, generational change, age-grading, or ‘changes from above’ and ‘changes from below. But this treatment of language variation and change holistically, longitudinally, macroscopically, uni-dimensionally and focused on the speech community as a macro-cosmos so far can sometimes be complemented with other views atomistically, cross-sectionally, microscopically, multidimensionally and privileging the community of practice as a micro-cosmos. This conveys a shift from the sociolinguistic study of collectivity and inter-speaker variation to that of individuality, intra-speaker variation and authenticity. The aim of this paper is to show results and conclusions on the microscopic investigation of the linguistic and extralinguistic mechanisms and motivations for style-shifting within the micro-cosmos of late Medieval England applying and thus testing the validity of current theoretical models of intra-speaker variation. The study is carried out through the analysis of the behaviour of variable (TH) in members of the Paston family from the archival source of The Paston Letters. The results show that letters also shed light onto the motivation(s) for variability in individuals and their stylistic choices for the construction of identity, in addition to tracing language change. This would therefore provide us with the possibility of reconstructing the sociolinguistic values, in terms of attitudes, ideologies and identities, for social interaction and communication in those centuries when the supralocalisation processes imposed some ‘norm-enforcing’ practices, and of accounting for the social meaning of inter- and intra- speaker variation in the sociolinguistic behaviour of speakers as a resource for identity construction, representation, and even social positioning.

ABSTRACT. The Old Indo-Aryan language (OIA) displayed a complex system of nominal inflection: three numbers, three genders, and eight cases were distinguished morphologically. Moreover, several declension classes were distinguished based on the terminations of noun stems. This system, however, underwent a gradual decline as OIA began to develop into Middle Indo-Aryan (MIA). By the time the modern Indo-Aryan languages began to emerge, the OIA case system had largely broken down, and had been replaced by an alternative strategy of marking arguments’ roles with postpositions. This research examines the major developments in the collapse of the OIA case system from a usage-based perspective, according to which the emergence and continual reshaping of language structure is primarily the result of language use in discourse.

Specifically, this research focuses on the loss of OIA dual declensions, nominative-accusative and genitive-dative syncretism, the leveling and complete loss of distinct OIA declension classes, and the rise of postpositions as a replacement for inflectional case marking. These changes were chosen in particular because they represent the major (i.e. most widespread) changes in the development of nominal inflection from OIA up until the emergence of the modern Indo-Aryan languages—most, if not all, varieties of MIA represented in the textual record underwent these changes. Drawing data from grammatical descriptions as well as from historical texts, this research traces the development of these changes through successive stages of Indo-Aryan. It is demonstrated that the facts of the collapse of the OIA case system are compatible with a usage-based view of language change, and it is argued furthermore that these facts would be difficult to fully account for under the view that language use itself cannot be a primary mechanism for the actuation of (morphosyntactic) changes in language.

Take, as an illustrative example, the change in which the OIA suffix -tas was extended to mark the ablative form of nouns from all declension classes in varieties of early MIA. This suffix was used in OIA to form adverbs, usually with clear ablative senses (e.g. sarva- ‘all’, sarva-tas ‘from all directions’). This suffix could combine with a wide range of nominals (Burrow 2001: 167)—in other words, it had a high type frequency. Usage-based theory makes the claim that type frequency is the main factor in determining the degree of productivity of a construction (Bybee & Beckner 2009: 842). This is because the more diverse the contexts in which a speaker encounters a particular construction, the more salient that construction becomes in the speaker’s mind as a possible strategy for conveying its meaning/function in contexts never before encountered (cf. -ness vs. -ity in English). The -tas suffix was extended to mark the ablative case for all MIA nouns because it was very productive, and thus easily accessed as a strategy for conveying ablative-like meanings.

Another similar example involves the extension of a-stem noun declensions to MIA nouns which previously fell into distinct inflection classes in OIA. For example, the OIA noun manas- ‘mind’, which was an s-stem noun and so appeared in the genitive singular as manas-as, adopted a-stem declensions in MIA, resulting in the genitive singular form mana-ssa in Pāli, which exhibits the MIA form of the OIA a-stem suffix -sya, cf. deva- ‘god’ and deva-sya ‘of god’. The a-stem nouns represented over half the nouns of OIA, and so a-stem declensions were the highest in type frequency when compared to declensions of other classes. Just as in the previous example, the high type frequency of a-stem declensions indicates that they were productive—they were easily accessed as means for expressing cases on nouns and were easily extensible to new contexts (i.e. nouns of other classes).

A non-usage-based view might explain these changes as simple cases of analogical leveling without delving any further into the issue, given that analogy is a well-established process by which languages change. An explanation of this kind, however, would neglect to account for the role of productivity in influencing the direction of analogy, and would in turn fail to acknowledge that productivity, as a property of constructions, necessarily emerges out of language use—a speaker’s repeated experience with a particular construction in discourse affects the construction’s representation in the mind of that speaker, in terms of, for example, its degrees of salience, schematicity, entrenchment etc. (Barðdal 2008, Bybee 2010, Langacker 2008).

This research adds to the growing body of literature on usage-based theory and thus contributes to a better understanding of inflectional systems in at least two meaningful ways. First, a consequence of the view that grammar is dynamic and emergent from language use is that inflectional systems cannot be fully understood without reference to how they developed (and continue to develop) diachronically. Inflectional systems, like other grammatical structures, comprise sets of emerging generalizations. This means that inflectional systems should not be considered abstract, a priori rule systems, but rather real-time, temporal, and social phenomena whose structure is epiphenomenal, a result of spreading systematicity from concrete instances of language use (Hopper 2014). The emergent grammar view lends support to theories of grammar such as Radical Construction Grammar (Croft 2001), which emphasize local rather than global regularities, and which do not relegate “irregularities” to the periphery of grammatical description.

The second way in which this research contributes to a better understanding of inflectional systems is by elucidating some of the possible mechanisms and pathways by which inflection is lost over time. In particular, this research provides additional support for several of the pathways of nominal inflection loss identified in Barðdal & Kulikov (2009) such as analogical leveling, semantic and/or functional overlap of argument structures, etc. However, like Barðdal (2009), this research concludes that these pathways are best explained in many cases as resulting from language use. Thus, language use is a primary mechanism of inflection loss (and of language change in general). Additionally, by demonstrating how a typologically commonplace development like the collapse of a case system can and should be accounted for in terms of usage-based principles, this research also provides additional support for the claim that crosslinguistic similarities and universals are often functionally-motivated.

References

Barðdal, Jóhanna (2009). The Development of Case in Germanic. In Jóhanna Barðdal and Shobhana Chelliah (eds.) The Role of Semantic, Pragmatic and Discourse Factors in the Development of Case. Amsterdam: John Benjamins.

Barðdal, Jóhanna, and Leonid Kulikov (2009). Case in Decline. In Andrej Malchukov and Andrew Spencer (eds.) The Oxford Handbook of Case. Oxford: Oxford University Press.

Burrow, Thomas (2001 [1995]). The Sanskrit Language. Delhi: Motilal Banarsidass.

Bybee, Joan (2010). Language, Usage, and Cognition. Cambridge: Cambridge University Press.

Bybee, Joan, and Clay Beckner (2009). Usage-Based Theory. In Bernd Heine and Heiko Narrog (eds.) The Oxford Handbook of Linguistic Analysis. Oxford: Oxford University Press.

Croft, William (2001). Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press.

Hopper, Paul (2014 [2002]). Emergent Grammar. In Michael Tomasello (ed.) The New Psychology of Language: Cognitive and Functional Approaches to Language Structure, Volume 1. New York: Psychology Press.

Langacker, Ronald (2008). Cognitive Grammar: A Basic Introduction. Oxford: Oxford University Press.

ABSTRACT. The loss of inflectional categories of the verb is a well attested phenomenon in the history of Indo-European. For example, Middle Persian has lost the future tense that was marked on the verb in Old Iranian, German has given up the morphological distinction between active and middle voice, and Hindi has removed the dual from its conjugation. Little is known, however, about the general propensity for indvididual morphosyntactic categories to be lost, maintained, or gained, and no attempt has ever been made to estimate in a systematic way how likely these processes were in the history of the family.

In Indo-European, inflectional categories of verbs are typically expressed cumulatively, e.g. the Latin verb ending -ō [prs.1sg.act.ind] expresses tense, person, role, diathesis, and mood at once. This raises the question of how likely it is that individual categories develop separate marking (e.g. tense vs. aspect in Slavic), and if they do develop separate marking, how likely these are lost again. To find out we sampled six separately marked categories from the AUTOTYP verb inflection database Bickel and Nichols 2005, namely aspect, mood, diathesis, tense, reflexives, and evidentiality.

In order to determine the propensity of these categories for loss in the history of the Indo-European languages in our sample, we performed Bayesian phylogenetic analyses. For each category we modeled its absence or presence as discrete states in a Continuous Time Markov Chain over the posterior samples of trees from Chang et al. 2015. We fitted two models, sc. one with equal rates (i.e. not biased towards one of the states) and one with unequal rates (i.e. biased either towards presence or absence), and assessed the likelihood of the models with Bayes Factors, where (log) BFs > 2 are commonly interpreted as positive, BF > 5 as strong evidence for the model that assumes unequal rates.

Results suggest that separate marking of mood, tense and evidentials is dispreferred (BFs > 2), i.e. unlikely to arise and likely to be lost over time (see the Markov Chains in Fig. 1). By contrast, aspect, diatheses and reflexives are equally likely to be gained, lost, or maintained (BFs < 2) as separate markers, all with similar rates. To complement these results, we performed the same analysis on the cumulative expressions of tense, aspect, mood, and evidentiality, encoded as present in a language if any of the combinations TAM or TA or TEvid or TM is available, else absent. We found that when co-expressed, these categories are likely to be preserved or gained (BF > 2).

We conclude that for Indo-European languages, category loss, gain, and maintenance are tightly tied to cumulation. Some categories favor cumulation and have little survival chances on their own, while other categories seem to evolve independently, subject only to chance and contact.

ABSTRACT. Proto-Indo-European (PIE) noun cases are divided into ”strong” and ”weak”, with the ”weak” cases being accented later in the word in all but one declension. Strong cases are accented on or near the root syllable. As cases are lost in Indo-European, the ”weak” cases with their stem or ending accents are the first to merge or vanish.

”Weak” cases are specifically the instrumental, locative, ablative, dative and genitive, separated by their accent from the ”strong” cases nominative, accusative and vocative. Weak cases are the ”oblique” cases in other languages, while strong cases are mostly the same as ”core” cases. Merger of oblique cases with other oblique cases is specifically an Indo-European phenomenon, according to a worldwide survey by Baerman, Brown, and Corbett (2005). Other families tend to merge core cases with other core cases, or with oblique cases, instead. I give an account here of IE case loss that accounts for the fact that the oblique-oblique merger is so specific to IE.

Case loss in Indo-European begins with the merger of the ”weak” cases, but seems to be triggered by the shift from PIE’s pitch accent to the stress accent of its descendants. The weak cases begin vanishing in each family around the time of that family’s shift to stress accent, or when stress accent triggers changes to vowel systems. In each family, it is the instrumental, locative and ablative cases that merge first, all of which have accent (now stress accent) furthest from the root.

Examples include Proto-Germanic, which according to Lehmann (1961) has its ”upper boundary” at the shift from free pitch accent to fixed stress accent, triggering stress-sensitive sound changes including Verner’s Law. Lehmann and Slocum (2007) mention that the Proto-Germanic language begins weak-weak case mergers at this time, with the locative merging with the dative, the ablative with the genitive, dative or instrumental. Lehmann’s conclusion of Proto-Germanic is also stress-sensitive: he chooses the loss of /e, a/ when final and weakly stressed, a form of vowel reduction. By this point the Germanic dialects begin to split, and the instrumental case begins to disappear.

Other case-losing Indo-European families have a similar pattern. Either the shift to stress accent or the subsequent reduction of unstressed vowels seems to coincide with the merger of the locative, ablative, and instrumental with other ”weak” cases in all these languages. In Italic, the locative, ablative and instrumental merge in pre-Latin days, shortly after stress accent and vowel reduction are attested (Allen, 1973; Herman, 1967). Stress accent and vowel reduction are attested in Celtic before Old Irish (Schrijver, 1995); by the time of Old Irish we see the mergers of locative, ablative and instrumental cases. Similar timing is observed in Prakrit as a typical Middle Indo-Aryan example (Edgerton 1939; Schwarzschild 1956), which merges all of Sanskrit's weak cases after shifting to stress accent.

Latvian and Lithuanian never shift entirely from pitch to stress accent, retaining elements of both. Neither suffers significant vowel reduction (Bond 1994; Veikova and Savickiene 2014), and both retain all eight PIE cases. Armenian did develop a stress accent, but a word-final one, always stressing the case endings, and it retains 7 of the 8 PIE noun cases, including the locative, ablative and instrumental.

I argue that the typologically unusual pattern of case loss in IE languages is related to the weak case endings being accented in most paradigms. The shift from pitch to stress accent triggered changes that rhythmically treated the weak cases differently from the strong ones, which accounted for the weak cases being consistently the first to merge.

References

Allen, W. S. (1973). Accent and rhythm. Cambridge: Cambridge University Press. Baerman, M., Brown, D., & Corbett, G. (2005). The syntax - morphology interface: A Study of Syncretism(1st ed.) (No. 2000). Cambridge: Cambridge University Press. doi: 10.1017/CBO9780511486234 Bond, D. (1994). A note on the quality of Latvian vowels. Journal of Baltic Studies, 25(1), 3-14. Retrieved from http://www.jstor.org/stable/43211868 Edgerton, F. (1939). Endingless noun case-forms in Prakrit. Journal of the American Oriental Society ,59 (3),369-371. Retrieved from http://www.jstor.org/stable/594693 Herman, J. (1967). Vulgar Latin. University Park, Pennsylvania: Pennsylvania State University Press. Lehmann, W. (1961). A Definition of Proto-Germanic: A Study in the Chronological Delimitation of Languages. Language,37 (1), 67–74. doi: 10.2307/1938717 Lehmann, W. P., & Slocum, J. (2007). A Grammar of Proto-Germanic (J. Slocum, Ed.). Texas. doi:10.1515/9783110197280 Veikova, M., & Savickiene, I. (2014). The Acquisition of the First Case Oppositions by a Lithuanian and a Russian Child. In A. Usoniene, N. Nau, & I. Dabasinskiene (Eds.), Multiple Perspectives in linguistic research on Baltic languages (pp. 165–188). Cambridge Scholars Publishing. Schrijver, P. (1995). Studies in British Celtic historical phonology (Vol. 5). Rodopi. Schwarzschild, L. A.(1956). Notes on the declension of feminine nouns in Middle Indo-Aryan. Journal of the Royal Asiatic Society of Great Britain and Ireland (3/4), 181-190. Retrieved from http://www.jstor.org/stable/25222840

ABSTRACT. The birth of new syntactic features in world Englishes under multilingual conditions has fostered active debate over the relative role played by universals (e.g. Kortmann and Szmrecsanyi 2004) and language transfer (e.g. Bao 2015) in shaping these new systems. The tendency in such analysis has been to evaluate formal properties of the abstract linguistic systems involved—the L1/substrate, the L2/superstrate, and the contact variety. In this talk, I investigate possible sources of observed variation by focusing on a question less commonly asked in the study of world Englishes: Why do only certain syntactic structures stabilise over time out of a wider pool of variation? In answering this question, I move beyond a pure consideration of linguistic systems and argue for the relevance of sequential acquisitional states of individual language learning in understanding historic contact situations.

The study compares two postcolonial varieties of English: Indian English (IndE) and Singapore English (SgE). The IndE data come from a cline of 24 non-dominant and dominant bilingual speakers of English; the SgE data are from secondary sources (primarily Ho and Platt 1993) and the corpus ICE-Singapore. The different substrates of these two contact varieties—primarily Indo-Aryan vs. Sinitic for the data examined—offer an interesting test case for the question of universals and substrate-driven change in world Englishes. The analysis first provides a brief overview of several syntactic features: past tense marking, use of progressive –ing, omission of copula and auxiliary be, omission of definite and indefinite articles, subject-verb agreement, use of modals, and use of the past perfect construction. Quantitative analysis of conditioned variation shows that substrate systems play a leading role in restructuring in all cases, occasionally giving rise to ‘third systems’, as L1-L2 contrasts at times lead learners to infer a distinct form-meaning pairing to either of the contributing systems. There is little support for the presence of ‘pure’ universal effects, even where apparent similarities in IndE and SgE arise.

However, a meta-analysis of variation across the dominant and non-dominant IndE speakers shows that only a subset of these variable IndE syntactic forms are deeply embedded across the whole population (e.g. use of progressive –ing, omission of articles). Others behave more like learner features that remain restricted to individuals with little use of English. It is these more widespread features that have diverged sufficiently over time and across the population that they are widely recognised as characteristic of the variety. In this light, substrates offer a less satisfactory account, as they over-predict long-term outcomes. To address this, I turn to a sociohistorical hallmark of World Englishes: diminishing input from the original target variety over time. Two theories from Second Language Acquisition are invoked to model sensitivity to input in learning—the Subset Principle (Wexler & Manzini 1987; White 1989) and the Interface Hypothesis (Hulk and Müller 2000; Sorace and Filiaci 2006). Although developed in acquisition research, I suggest that these kinds of models are important for understanding contact varieties in which exposure to target variety input has declined dramatically over time. The data are re-examined not just in relation to substrate contrasts, but also in terms of how much input is required for the acquisition of each feature. Relative input-demand enhances our ability to account for which features that exhibit more deeply established outcomes. The Subset Principle predicts that learner grammars that generate structures (e.g. sentences) that form a subset of the target grammar will more easily expand to converge on the target grammar than those that generate a superset. All but two of the stable IndE features are superset features that would have required significantly more input to converge on the historical target. However, the Subset Principle does not account for article omission in IndE, which is a subset grammar compared to British English. The analysis is therefore revised and generalized to cover any acquisitional state requiring rich input (e.g. complexity of the syntax-pragmatics interface, in the case of the English article system). This expanded model, which resembles the Interface Hypothesis, is able encompass the full set of stable IndE syntactic features.

In the final discussion, I develop a four-way typology whereby contact features can be assessed as varying across these two dimensions — high/low contrast between substrates and superstrates, and high/low richness of input required for acquisition. The set of stable indigenized features in IndE appear to combine high substrate-superstrate contrast and high need for rich, sustained, and consistent input. An extension of the typology to include SgE and further Englishes, however, begins to reveal that input alone cannot be the basis for stable contact outcomes. In the expanded typology, substrates remain an extremely strong force in determining long-term stable outcomes, possibly even stronger overall than input. Most importantly, the typology is able to model an interplay between these two forces, reframing emergent contact features as part of dynamic input-sensitive phases of L2 development, not simply static semiotic systems in contact.

References

Kortmann, Bernd, and Benedikt Szmrecsanyi. 2004. Global synopsis: morphological and syntactic variation in English. In Bernd Kortmann, Kate Burridge, Raj Mesthrie and Edgar Schneider, eds. A Handbook of Varieties of English, Vol. 2: Morphology and Syntax. Berlin, New York: Mouton de Gruyter, 1122-82.
Zhiming Bao. 2015. The Making of Vernacular Singapore English: System, Transfer and Filter. Cambridge: Cambridge University Press.
Sorace, Antonella. and Francesca Filiaci. 2006. Anaphora resolution in near-native speakers of Italian. Second Language Research 22(3): 339-368
Hulk, Aafke, and Natascha Müller. 2000. Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition 3 (3): 227-244.
Wexler, Kenneth, and  Rita Manzini. 1987. Parameters and learnability in binding theory. In Parameter Setting, ed. by Thomas Roeper and Edwin Williams. Dordrecht: Reidel, 41-76.
White, Lydia. 1989. Universal Grammar and Second Language Acquisition. Amsterdam: John Benjamins.
Ho, Mian-Lian & John T. Platt. 1993. Dynamics of a Contact Continuum: Singaporean English. Oxford: Oxford University Press.

ABSTRACT. Jarvis and Pavlenko (2008:3) describe cross-linguistic influence as a “phenomenon that takes place in the minds of individuals, and which is subject to the effects of various cognitive, linguistic, social and situational factors”. Such a combination of psycholinguistic, sociolinguistic and sociocultural factors must have also shaped past contact-induced change, which raises the question explored in this talk: in what ways can the contexts that underpin prehistoric contact-induced change be reconstructed?

Often, the outcomes of past contact-induced change are easily identified, but their contribution to our understanding of linguistic history relies on reconstruction of the undocumented sociolinguistic and sociohistorical contexts of the changes. In this talk, I argue that more careful consideration of models of bi-/multilingual cross-linguistic influence, ethnographies of language use, and information on past population structures, can lead to more robust reconstructions of the factors that likely influenced instances of prehistoric contact-induced change. This in turn allows us to begin to bridge the gap between individual-level mechanisms of cross-linguistic influence and community-level outcomes of contact-induced change, and so gain a deeper understanding of the likely roles that different types of bi­/multilingual speakers play as agents of diachronic language change.

The evidence of past contact-induced change among the Austronesian and Papuan languages of island Melanesia is prominent in the language contact literature (e.g. Thurston 1987; Ross 2008, 2013). In southern Bougainville of Papua New Guinea, for example, the Austronesian languages have undergone changes in word order at both the clausal and phrasal level (see Evans and Palmer 2011) and both the Austronesian and Papuan languages display semantic restructuring of some kinship terms; changes which can be attributed to language contact. Speaker-based models of cross-linguistic influence and bi-/multilingual acquisition (e.g. Van Coetsem 2000; Jarvis and Pavlenko 2008; Meisel 2011) suggest the reconstruction of a psycholinguistic process of imposition, such that non-dominant speakers of the recipient language have imposed onto it structural features from their dominant language (cf. Van Coetsem 2000). Such a reconstruction, based on proposed correlations between speaker behaviours and particular outcomes of contact-induced language change, is valuable for understanding the history of contact-induced change in southern Bougainville, but also raises further questions about the agents of change. Who were these non-dominant speakers of the recipient languages? Were they adult second-language learners (cf. Siegel 2016), or speakers, bilingual from childhood, whose linguistic repertoire was influenced by proposed patterns of community-level language use (cf. Ross 2013)?

I propose that resolving such questions must rely on the reconstruction of the broader sociolinguistic and sociohistorical context of contact-induced change. That is, on the reconstruction of some of the other factors which affect cross-linguistic influence (cf. Jarvis and Pavlenko 2008). For southern Bougainville, contemporary patterns of language use, oral histories of clan origins, marriage preferences, trading partners, and migration, as well as information on past population structures all provide a starting point for reconstructing past patterns of both individual- and community-level language use. For example, oral histories of trade and migration suggest that community-level linguistic dominance was likely a Papuan language in the north of the region, but an Austronesian one in the south. Population genetic studies indicate that there has been considerable intermarriage between Austronesian and Papuan speakers in the history of contemporary ethnolinguistic groups (see Duggan et al. 2014). This sociohistorical context points to childhood bilinguals, dominant in a Papuan language, as the likely drivers of contact-induced change in the Austronesian language Torau. However, it also indicates the need to consider the range of bi-/multilingual speakers that would have underpinned these community-level linguistic patterns, and so driven other patterns of contact-induced change. Complementing the linguistic reconstruction of contact-induced change in southern Bougainville in this way results in a more nuanced reconstruction of linguistic history that begins to incorporate the dynamic nature of bi­/multilingual speakers and the complex layering of contact-induced change across the region.

Thus, it is argued in this talk that linguistic reconstructions which encompass the “cognitive, linguistic, social and situational factors” of contact-induced change can begin to make connections between understandings of bi-/multilingual acquisition and cross-linguistic influence, and diachronic contact-induced change.

References Duggan, Ana T., Bethwyn Evans, François R. Friedlaender, Jonathan S. Friedlaender, George Koki, D. Andrew Merriwether, Manfred Kayser & Mark Stoneking. 2014. Maternal history of Oceania from complete mtDNA genomes: contrasting ancient diversity with recent homogenization due to the Austronesian expansion. The American Journal of Human Genetics 94:1-13. Evans, Bethwyn & Bill Palmer. 2011. Contact-induced change in southern Bougainville. Oceanic Linguistics 50(2):484-523. Jarvis, Scott & Aneta Pavlenko. 2008. Crosslinguistic influence in language and cognition. New York/London: Routledge. Meisel, Jürgen M. 2011. Bilingual language acquisition and theories of diachronic change: bilingualism as cause and effect of grammatical change. Bilingualism: Language and Cognition 14(2):121-45. Ross, Malcolm. 2008. A history of metatypy in the Bel languages. Journal of Language Contact 2:149-64. —. 2013. Diagnosing contact processes from their outcomes: the importance of life stages. Journal of Language Contact 6(1):5-47. Siegel, Jeff. 2016. Contact-induced grammatical change in Melanesia: who were the agents of change? Australian Journal of Linguistics 36(3):406-28. Thurston, William R. 1987. Processes of change in the languages of north-western New Britain. Canberra: Pacific Linguistics. Van Coetsem, Frans. 2000. A general and unified theory of the transmission process in language contact. Heidelberg: Universitätsverlag C. Winter.

ABSTRACT. This presentation explores the combined operation of L2 acquisition and dialect contact via accommodation as driving factors in determining the selection of linguistic features in settings involving both forms of adult learning. Although both processes have usually been treated as separate sources of grammatical restructuring in contact situations, we will argue that they may operate jointly in settings where dialectal accommodation and adult language learning favor the same linguistic outcomes. We will exemplify this joint effect by focusing on two changes in the history of Castilian/Spanish: the restructuring of the 3rd person object clitic system in medieval southern Iberian Castilian (MSIC) and the merging of the medieval sibilants in early colonial Spanish (ECS). We will argue that, in both situations, the ecological conditions of contact in a hybrid demographic environment caused language speciation to proceed not just by accommodation among speakers of L1 Iberian varieties, but also via L2 acquisition.

Although the absence of a clear linguistic criterion to distinguish between “languages” and “dialects” is a well-established axiom among linguists, in practice the fields of language contact (Thomason and Kaufman 1988, van Coetsem 2000, Winford 2007) and dialect contact (Trudgill 1986, Kerswill 2002) have traditionally had little interface, both theoretically and in the types of phenomena they strive to understand. Exceptions are few and far in between (e.g., ‘code contact’, Johanson 2002). While the existence of a cline between adult language and dialect acquisition is usually acknowledged, the wider field of contact linguistics continues to grapple with the hermeneutic validity of this divide (Siegel 2010, Erker 2016). A survey of research in these two areas shows that certain general principles seem to apply in both kinds of situations. For instance, for both ‘dialect’ and ‘language’ contact, speakers are sensitive to the levels of demographic presence of language elements; they tend to favor structurally simpler or more systemically unmarked alternatives; and they may map characteristics of their native system onto the target system (Chambers 1992, Silva-Corvalán 1994, Deumer 2002). From an evolutionary perspective (Croft 2000, 2006; Mufwene 2008), this overlap between ‘dialect contact’ and ‘language contact’ effects may be conceptualized as a function of selection of features from the speakers’ feature pool, which may include elements of heterogeneous provenance, acquired by the individual at various life stages.

Most relevant to the topic of this panel is the fact that these general principles of language processing and language acquisition are best exemplified by historical situations that involve the simultaneous contact among linguistically close varieties (i.e., ‘dialects’) and less close varieties (i.e., ‘languages’), where both the progression and the outcome of contact may reflect the simultaneous workings of dialect and L2 acquisition strategies by adults (e.g., Spanish and English in New York City, Otheguy and Zentella 2012; English and German in Indiana and Wisconsin, Nützel and Salmons 2011). The possibility of interaction among both types of influence is particularly relevant in cases of demographic resettlement of populations speaking a set of closely-related varieties into territories where markedly divergent codes are used, and where post-resettlement local demographic interaction patterns lead to mutual accommodation among speakers of the transplanted language, as well as to acquisition by speakers of the local or other languages (i.e., not tabula rasa situations, cfr. Trudgill 2004).

This presentation explores the interface between dialect and language contact in two historical situations of language expansion in the history of Castilian/Spanish that match the above ecological scenario: the southward expansion of Castilian in the Iberian Peninsula in the medieval period, and the formation of early colonial Spanish in the Americas. Both of these situations involved contact among varieties of Indo-European Romance stock as well as contact between the latter and non-Indo-European languages (Iberian Arabic in the first case, African and indigenous American languages in the second). In the two cases, the outcome of contact reveals patterns of restructuring of the phonological and morphological Castilian/Spanish input system(s) involving the selection of simpler options among the alternatives present in the contact feature pool, which may be attributed to both dialect contact among L1 speakers and L2 language acquisition by adults. To illustrate both situations, data from the result of contact in the third-person clitic system in MSIC and the sibilant phonemic inventory in ECS will be presented. In the case of the clitic system in MSIC, speakers maintained the straightforward case-determined system (e.g., le for 3p sing. dative, lo and la for 3p. sing. masc./fem. accusative) and rejected the outcomes most commonly used among speakers in northern Iberia: the semantically-determined and the hybrid or interdialectal systems based on the absence/presence of a more complex set of semantic features (count, human, animate). In the case of ESC sibilants, simplification involved the rearrangement of a system of four segments (/s̺, z̺, s̪, z̪/) into one (/s̪/), an outcome that was already available in the Iberian feature pool, but only as one of several merged alternative systems. Although both phenomena have been explained in the past as the result of dialect contact via accommodation (among others, Tuten 2003 for MSIC and Parodi 2001 for ESC), we will argue that the selected options in both cases also presented advantages as the target of L2 acquisition. In a sociolinguistic environment where both forms of learning were demographically prominent, the historical explanation of either situation as primarily the result of dialect contact obscures the actual dynamics of contact within the community and the role that different forms of learning must have had in language speciation. Similarly, this evidence underscores the need to consider other language acquisition effects in contact situations beyond those strictly involving the transfer of grammatical properties from system A to system B, i.e. interference (Thomason and Kaufman 1988) or imposition (van Coetsem 2000).

Chambers, JK. 1992. Dialect acquisition. Language 68: 673-705. van Coetsem, Frans. 2000. A General and Unified Theory of the Transmission Process in Language Contact. Heidelberg: Winter. Croft, William. 2000. Explaining Language Change: An Evolutionary Approach. New York: Longman. Croft, William. 2006. The relevance of an evolutionary model to historical linguistics. In Nedergård Thomsen, Ole (ed.), Different Models of Linguistic Change, 91-132. Amsterdam: John Benjamins. Deumert, Ana. 2003. Markedness and salience in language contact and second-language acquisition: Evidence from a non-canonical contact language. Language Sciences 25:561-613. Erker, Daniel. 2016. Social salience as a determinant of contact-induced change: The view from Spanish in the United States. Paper presented at the 2016 Hispanic Linguistics Symposium, October 9, 2016. Johanson, Lars. 2002. Contact-induced change in code-copying framework. In Jones, Mary and Edith Esch (eds.), Language Change: The Interplay of Internal, External and Extra-Linguistic Factors, 285-313. Berlin: Mouton de Gruyter. Kerswill, Paul. 2002. Koineization and accommodation. In Chambers, Trudgill, and Natalie Schilling-Estes (eds.), The Handbook of Language Variation and Change, 669-702. Malden, MA: Blackwell. Mufwene, Salikoko S. 2008. Language Evolution: Contact, Evolution and Change. New York: Continuum. Nützel, Daniel, and Joseph Salmons. 2011. Language contact and new dialect formation: Evidence from German in North America. Language and Linguistics Compass 5: 705–717. Otheguy, Ricardo & Ana Celia Zentella. 2012. Spanish in New York: Language Contact, Dialect Leveling and Structural Continuity. Oxford: Oxford University Press. Parodi, Claudia. 2001. Contacto de dialectos y lenguas en el Nuevo Mundo: La vernacularización del español de América. International Journal of the Sociology of Language 149: 33–53. Siegel, Jeff. 2010. Second Dialect Acquisition. Cambridge: Cambridge University Press. Silva-Corvalán, Carmen. 1994. Language Contact and Change: Spanish in Los Angeles. Oxford: Oxford University Press. Thomason, Sarah & Terrence Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell. Trudgill, Peter. 2004. New-dialect Formation: The Inevitability of Colonial Englishes. Cambridge: Cambridge University Press. Tuten, Donald. 2003. Koineization in Medieval Spanish. Berlin: Mouton de Gruyter. Winford, Donald. 2007. Some issues in the study of language contact. Journal of Language Contact – THEMA 1: 22-40.

ABSTRACT. Since the time of Whitney (1879: §761b) and Delbrück (1888: 267-268) certain Vedic Sanskrit verbs in -ya-, inflected in the middle voice, have received special attention for exhibiting variation between root and suffix accentuation, with some Vedic branches preferring one accentuation, others the other. See e.g. (1). Both scholars agree on classifying these verbs as non-passive, belonging to the 4th (or div) class of verbs; but by discussing the verbs in a paragraph following his coverage of “aorist passives” in -i, and before his discussion of passives, Delbrück recognizes a special affinity to passives.

(1) a. śvātréṇa yát pitrór múc-ya-se pári swelling.ins.sg.n when father.loc.du.m release-ya-3sg.mid on (RV 1.31.4c) ‘when you (Agni) get free through swelling on your parents (the kindling sticks)’ b. imā́ uptā́ mṛtyupāśā́ this.nom.pl.m cast.nom.pl.m death-fetter.nom.pl.m yā́n ākrámya ná mucyáse which.acc.pl.m enter.cvb neg release-ya-3sg.mid (AV 8.8.16ab) ‘These fetters of death are cast, having entered which you do not become free.’

Structures labeled passive, by contrast, only have accent on the suffix. At the same time, as Delbrück notes (1888: 268), while the function of these structures is overwhelmingly passive, structures from certain roots (‘dhar und mar, und etwa noch darç dhū parc lup vac vañc …’) have “neutral” or “medial” function (see also his detailed discussion on pp. 269-271). Delbrück finds similar variation in the so-called aorist passive in -i, with the added observation that ‘sehr häufig neben diesen Aoristen mediopassivische Praes. auf yate stehen, z. B.: kshī́yate neben ákshāyi …’ (265-267)

In a series of recent publications (2001 ≈ 2012, 2011, 2013), Kulikov argues that there is a fundamental difference between the structural type in (1) and passives. The former is to be classified as anticausative and non-passive, and its root accentuation must be considered original, as shown by the fact that this is what is found in the Rig Veda, and suffix accentuation occurs in the later Atharva Veda.

I examine Kulikov’s evidence and arguments, confront his analysis with that of Pāṇini, and show that the latter provides a better account.

As regards the issue of root accent and accent variation, the difference between Rig Veda and Atharva Veda may be a matter of variation between different Vedic schools, rather than of chronology. Pāṇini considers root accent optional and limited to a subclass of verbs, which makes suffix accent the default. Root accent and accent variation therefore are not reliable criteria for distinguishing anticausatives from passives.

There is, however, syntactic evidence that differentiates the two categories. In passives, converbs are controlled by the underlying subject (kartṛ in Pāṇini’s terms), whether overtly present or implied (2). In anticausatives, by contrast, it is the surface subject that controls the converb, as in (3). Further, as (3b) shows, the converb-control test makes it possible to identify anticausatives in texts that do not provide accent information (and which Kulikov did not include in his survey of anticausatives).

(2) na vā a̱hiṅkṛtya sā̱ma gīyate (ŚB 1.4.1.1) neg ptcle neg.making.hiṅ.cvb sāman.nom.sg.n chant.pass.3sg ‘For the sāman is not chanted without making (the sound) hiṅ.’

(3) a. imā́ uptā́ mṛtyupāśā́ this.nom.pl.m cast.nom.pl.m death-fetter.nom.pl.m yā́n ākrámya ná mucyáse which.acc.pl.m enter.cvb neg release-ya-3sg.mid (AV 8.8.16ab) ‘These fetters of death are cast, having entered which you do not become free.’ b. tās sīmānam eva+ ūrdhvā udīrya + that.nom.pl.f parting.acc.sg emph above.nom.pl.f rise.cvb asṛjyanta (JB 3.104) release.ya.impf.3pl ‘They, rising up above (his) hair parting, came into existence.’

Finally, it is important to keep in mind Delbrück’s observation that structures with suffix accent and classified as passives may exhibit “neutral” or “medial” functions and that the same holds for the “passive aorist” (even Kulikov notes that the distinction between passive and non-passive anticausative function is often difficult to discern). To this must be added that for some verbs, such as the one in (3b), the anticausative value seems to be a later development, and passive the original function.

Under the circumstances, it is legitimate to speculate that anticausatives developed out of earlier passives and that root accent, where found, is an innovation, probably modeled on the root accent of other, solidly non-passive ya-presents and serving to (optionally, in some Vedic branches) differentiate anticausatives from passives.

References

Delbrück, Bert[h]old. 1888. Altindische Syntax. Halle: Waisenhaus. Repr. 1968, Darmstadt: Wissenschaftliche Buchgesellschaft. Kulikov, Leonid. 2001. The Vedic ya-presents. Universiteit Leiden Proefschrift. Kulikov, Leonid. 2011. Drifting between passive and anticausative: True and alleged accent shifts in the history of Vedic -ya-presents. Вопросы языкового родства/Journal of Language Relationship 6: 185-215. (Also containing discussion by other scholars and a reply by Kulikov.) Kulikov, Leonid. 2012. The Vedic –ya-presents: Passives and intransitivity in Old Indo-Aryan. Amsterdam: Rodopi Kulikov, Leonid. 2013. Language vs. grammatical tradition in Ancient India: How real was Pāṇinian Sanskrit? Evidence from the history of late Sanskrit passives and pseudo-passives. Folia Linguistica Historica 34: 59-92. Whitney, William Dwight. 1879. Sanskrit grammar including both the Classical Language, and the Older Dialects, of Veda and Brahmana. Leipzig: Breitkopf & Härtel.

ABSTRACT. Most of the older Indo-European language branches show a relatively consistent nominative-accusative alignment which remains remarkably stable through time. However, Indo-Aryan and Anatolian, which represent two archaic and yet typologically diverse branches of Indo-European, both develop split ergative systems in the course of their attested history. In Indo-Aryan, notably Vedic the change primarily affects a verbal construction originally representing a predicated verbal adjective with resultative stative meaning, developing into a finite ergative construction with past perfective semantics (cf. Dahl 2016, Dahl and Stroński 2016 for discussion). In the Anatolian language Hittite, on the other hand, the development primarily affects the nominal system, where an originally derivational suffix develops into an ergative case marker restricted to a subclass of nouns in later stages of Hittite (cf. e.g. Goedegebuure 2015). The two development paths instantiated in Vedic and Hittite result in two basically distinct types of alignment splits, one based on different Tense/Aspect/Mood (TAM) properties and another based on different noun classes with more or less consistently distinct lexical semantic properties. Various scholars have observed that split case marking may be conditioned by lexical semantic factors or by grammatical factors like TAM distinctions and that these two types of split tend not to co-occur (cf. e.g. Trask 1979, Payne 1980, Khokhlova 1995, 2001 and Filimonova 2005). It is likely that this tendency may be explained as a consequence of the fact that TAM-based alignment splits and NP-based splits result from distinct grammaticalization processes arising in different areas of grammar. This paper provides a contrastive analysis of diachronic data from Vedic and Hittite and explores to what extent the different paths leading to split ergativity in these two language branches bear typologically relevant resemblances or whether they in fact represent typologically complementary processes. Specifically, recent work by Dahl (2016 cf. also Condoravdi and Deo 2015) suggests that the Vedic construction follows a grammaticalization path from predicated resultative adjective via passive to inverse and that it develops into an ergative construction in Early Middle Indo-Aryan, a development chain which finds parallels in other languages (cf. e.g. Gildea 1997). Apart from this, the Indo-Aryan alignment system maintains a general nominative-accusative character, which still characterizes many present-day Indo-Aryan languages. In Hittite, on the other hand, we encounter a fundamentally different situation. There are two noun classes in this language, common gender nouns and neuter nouns. Both common and neuter nouns are used as the subject of intransitive (i.e. one-place) verbs, but unlike common gender nouns, bare neuter nouns never occur in subject position with transitive (i.e. two- or three-place) verbs. Another significant difference between the two noun classes is that common gender nouns have separate forms for nominative and accusative case, (cf. e.g. antuḫšaš ‘man (nom.)’ vs antuḫšan ‘man (acc.)’), a distinction which is not found with neuter nouns. Furthermore, bare neuter nouns in subject position primarily occur with unaccusative intransitive verbs and rarely, if ever with unergative intransitive verbs, a distributional fact which may be indicative of semantic alignment. It should be noted, however, that neuter nouns could be accommodated to the common gender class by the addition of a suffix -ant- which in Old Hittite has a clear derivational character (cf. Goedegebuure 2015). In later stages of Hittite this marker seemingly represents an inflectional ergative ending characteristic of neuter nouns. It is reasonable to claim that the development in Hittite represents a change from an alignment system based on a lexically determined split between semantic and accusative alignment to a system based on an analogous split between ergative and accusative alignment. Apart from providing a contrastive analysis of the two development paths instantiated in Hittite and Vedic, this paper examines to what extent the resulting split ergative patterns have different syntactic properties. That is, it aims to clarify whether the ergative- and absolutive-marked arguments show analogous behavior in Anatolian and Indo-Aryan or not, and to what extent their behavior deviates from corresponding nominative- and accusative-marked arguments in other constructions. References: Dahl, Eystein. 2016. ‘The origin and development of the Old Indo-Aryan predicated –tá construction.’ In Eystein Dahl and Krzysztof Stroński (Eds.) Indo-Aryan Ergativity in Typological and Diachronic Perspective. Amsterdam: John Benjamins, 63-110. Dahl, Eystein and Krzysztof Stroński. 2016. ‘Ergativity in Indo-Aryan and beyond.‘ In Eystein Dahl and Krzysztof Stroński (Eds.) Indo-Aryan Ergativity in Typological and Diachronic Perspective. Amsterdam: John Benjamins, 1-38. Filimonova, Elena. 2005. The noun phrase hierarchy and relational marking: Problems and counterevidence. Linguistic Typology 9 (2005), 77–113. Gildea, Spike. 1997. Evolution of grammatical relations in Cariban: How functional motivation precedes syntactic change. Grammatical Relations: A Functionalist Perspective, Typological Studies in Language, v. 35, ed. by T. Givón, 155-198. Amsterdam: John Benjamins Press. Goedegebuure, Petra. 2015. ‘The rise of split-ergativity (or rather split-accusativity) in Hittite.’ Handout from the Workshop Language, Variation and Change 2015, June 8 - Chicago. Hoffner, Harry A. and H. Craig Melchert. 2008. A Grammar of the Hittite Language. Part 1. Reference Grammar. Winona Lake, IN.: Eisenbrauns. Khokhlova Ludmila V., 1995. The Development of Patient-Oriented Constructions in Late Western NIA Languages. Osmania Papers in Linguistics 21, 15–51. Khokhlova Ludmila V., 2001. Ergativity Attrition in the History of Western New Indo –Aryan languages (Punjabi, Gujarati and Rajastahani). The Yearbook of South Asian Languages and Linguistics, 159–184. Payne, John R. 1980. The Decay of Ergativity in Pamir Languages. Lingua 51, 147–186. Trask Larry R. 1979. On the Origin of Ergativity. In Ergativity. Towards a Theory of Grammatcal Relations. Frans Plank (ed.). 385–404. London: Academic Press.

ABSTRACT. In this talk I will discuss possible paths through which split intransitive systems may arise, and their relationship with  the emergence  of head-marking  (Nichols  1986a).  The study  will be conducted  on Latin  and  the transition  to Romance (on which see also Vincent 1997; 1998; Ledgeway 2011; 2012), but other languages might reveal analogous paths of development.

In particular, I will consider the diachronic relationship between two clusters of changes occurring in Late Latin in the encoding of transitivity, namely the active-inactive and neutral realignment of grammatical relations, testified by the extended accusative (Plank 1985) — the use of the accusative in subject function, initially confined to unaccusative structures and sensitive to semantic features such as animacy and control — and the temporary loss of the grammatical dimension of voice, whereby voice forms become interchangeable, no longer matching their canonical function(s).

Three parameters appear to play a role in the gradual spread of the accusative into the functional domains of the nominative  for non-object  core arguments:  semantic (the inactive nature of the arguments), syntactic (the degree of syntactic cohesion between the argument and its verb/predicate), pragmatic (the grammaticalization of a constituent originally  denoting  the  topic  of  the  clause), interacting, in the course  of  time, with  the  reorganization  of  voice distinctions.

I will argue that the two phenomena investigated reflect a more general change, the rise of  head-marking patterns in a predominantly dependent-marking  language such as Latin, which determines a deep restructuring in the encoding of the argument structure of the clause in the transition to Romance (Cennamo 2009; 2011).

More specifically, evidence will be given for the shift, in Late Latin, from dependent-marked split intransitivity systems  (signaled  on the S argument  and initially  reflecting  its inactive  nature),  to head-marked  split intransitivity systems (marked on the verb and applying to clusters of verbs/patterns, irrespective of purely semantic factors such as control) (cf. Nichols 1986b: 144).

The Late Latin data, therefore, offer  interesting insights into possible routes through which a dependent-marking language  may  develop  head-marking  coding  patterns, as a result  of changes  in the alignment  of core  arguments in various transitivity domains, an  issue  that  ties  in  with  recent  discussion  on  the  emergence  of  semantically  aligned systems (Mithun 2008 and contributions in Donohue & Wichmann 2008).

References

Cennamo, M. (2009) ”Argument structure and alignment variations and changes in Late Latin”. In J. Bardðal & Sh. Chelliah (eds), The Semantics and Pragmatics of Case, Amsterdam: Benjamins, 307–346.

--(2011) “Impersonal  constructions  and  accusative subjects  in Late Latin. In  A. Malchukov & A.  Siewierska  (eds), Impersonal Constructions. A Cross-linguistic Perspective. Amsterdam: Benjamins: 169-188. Donohue, M. & S. Wichman (eds) The Typology of Semantic Alignment, Oxford: Oxford University Press.

Ledgeway, A. (2011) “Morphosyntactic typology and change”. In M. Maiden and J.  Ch. Smith (eds), The Cambridge History of the Romance Languages, Cambridge: Cambridge  University Press, 382–471.

--(2012). From Latin to Romance. Morphosyntactic Typology and Change. Oxford: Oxford University Press.

Mithun, M. (2008), “The emergence of agentive systems in core argument marking”, in M. Donohue & S. Wichmann (2008) (eds), 297–333.

Nichols, J. (1986a) “Head-marking and dependent-marking  grammar”. Language 62: 56–119.

--. 1986b. On form and content in typology. In W. P. Lehmann (ed),  Language Typology 1985. Amsterdam: Benjamins, 141–162.

Plank, F. (1985) "The  extended accusative/restricted nominative in perspective", in F. Plank (ed) Relational  Typology, Berlin:  Mouton de Gruyter, 269–310.

Vincent, N. (1997) "The emergence of the D-system in Romance", in A. Van Kemenade & N. Vincent (eds) Parameters ofMorphosyntactic Change, Cambridge: Cambridge University Press,149–151.

--. (1998) “Tra grammatica e grammaticalizzazione: articoli e clitici nelle lingue (italo)- romanze. In P. Ramat & E. Roma (eds), Sintassi Storica: Atti del XXX Congresso Internazionale della Società di Linguistica Italiana, Rome: Bulzoni, 411–440.

ABSTRACT. Stylistic fronting, the fronting of an element to a position preceding the finite verb, in subordinate clauses is assumed to be found in both Old Norwegian (Faarlund, 2004:251) and Old Spanish (Fontana, 1993:75ff) :

1.En ef boren værðr þa er uppnæmt And if carried is then is confiscated 'And if it is brought along, it will be confiscated.' (Magnus Lagabøtes landslov)

2.De asia e de affrica oydo auedes ya en otros of asia and of africa heard have already in other

libros quamannas son e quales. books how-big are and how 'You have already heard from other books of how big Asia and Africa are and what they’re like.' (Estoria de Espanna)

These examples show the key properties of Stylistic fronting: a subject gap is required, only heads can be fronted and it follows the accessibility hierarchy (Maling, 1980, 1990).

In our talk, we show that patterns similar to Stylistic fronting in both Old Norwegian and Old Spanish do not comply with the key properties of Stylistic fronting. Using data from the syntactically annotated PROIEL and MENOTEC-corpus, we present data from both Old Norwegian and Old Spanish that show that phrases, not only heads may be fronted and that a subject gap is not necessary. In the following examples a phrase has been fronted:

3.En ef a skipi byr hoggvi eigi meira en þeim And if on ship prepare hew not more than those

vinzt samdægres nema suo sem aðr var sagt accomplish within-same-day except so comp previously was said 'And if they hew more on the ship than they can accomplish within the same day, except if this was previously agreed' (Magnus Lagabøtes landslov)

4.ca por aventura estos vuestros consejeros vos lo dizen for for fortune these your advisors you it say

porque saben que desque en tal fecho vos ovieren metido because know comp from in such deed you they-had put 'For maybe these advisors of yours say it to you because they know that from the moment they have put you in such a situation'… (El conde Lucanor)

Data from Old Norwegian and Old Spanish also show that an element can be fronted even when a subject is present. In the following examples fronting occurs in the presence of a lexical subject:

5.Sem Milun þetta frá. þa var hann mioc feginn. COMP Milun this knew then was he very happy. 'When Milun came to know this, he became very happy' (Strengleikar)

6.et dize en este lugar la glosa que aquel aruol çino. and say in this place the word COMP that tree cino

era al que los latinos llamamos lentisco. was to-the that the latins call lentisque 'And here says the bible that that tree was a cino, that which we Romans call a lentisque.'

We discuss whether stylistic fronting is a valid theoretical concept in the languages under scrutiny, since data from both Old Norwegian and Old Spanish show that Stylistic Fronting-like movement does not comply with the restrictions traditionally given for Stylistic Fronting. We argue that what has previously been called Stylistic Fronting is a more generalized movement option in these languages.

Bibliography: Faarlund, J. T. (2004). The syntax of Old Norse: with a survey of the inflectional morphology and a complete bibliography. Oxford: Oxford University Press. Fontana, J. M. (1993). Phrase Structure and the Syntax of Clitics in the History of Spanish. (Phd), Univ. of Pennsylvania Philadelphia. Maling, J. (1980). Inversion in embedded clauses in Modern Icelandic. Íslenskt mál og almenn málfræði, 2, 175-193. Maling, J. (1990). Inversion in {Embedded} {Clauses} in {Modern} {Icelandic}. In J. Maling & A. Zaenen (Eds.), Syntax and semantics 24: Modern Icelandic syntax (Vol. 24, pp. 71-91). San Diego: Academic Press.

ABSTRACT. In this paper I will compare the use of resumption structures in Old French and Old Swedish and show how changes in the semantic properties of the structure may explain why it was lost in French, while it has been maintained in Swedish. The use of resumptive structures may be linked to the Verb Second property.

1. Background Resumption is a structure in which a semantically bleached element comes in second position, typically between an adverbial element and the finite verb. The structure is attested in all stages of the Germanic V2 languages, and it is also found in certain Old Romance varieties, first and foremost in the Gallo-Italic ones (see among others Vance, 1997; van Reenen and Schøsler, 2000; Ferraresi and Goldbach, 2003; Ledgeway, 2008; Poletto, 2014; Wolfe, 2015).

1. Enn er Fáfnir screið yfir gröfna þá lagði Sigurðr hann með sverði til hiarta. yet when Fafnir walked over ditch ÞÁ laid Sigurd him with sword to heart `As Fafnir walked across the ditch, Sigurd poked him in the heart with the sword.' (Old Norse) 2. Et quant il sont issuz de Kamaalot si chevauchent tant qu' il sont en la forest venu and when they were left of Camelot SI rode so.much that they were in the forest come `And when they had left Camelot, they rode until they arrived in the forest.' (Old French)

2. An implicational hierarchy In principle, resumptive structures may occur after initial adverbial clauses, initial PPs and initial adverbials. In some languages, resumption may also occur after an initial thematic element containing a relative clause or even a bare thematic DP (Sollid & Eide 2007).

Languages differ with respect to what kind of elements permit a resumptive structure, and these structures constitute an implicational hierarchy (Salvesen, ms). This hierarchy is represented in (3). The cline must be understood in the following way: A language that accepts resumption after an initial adverb, will also accept resumption after an initial PP and an initial adverbial clause. I will show that this hierarchy may be further subdivided.

3. Adverbial clause > PP > Adv

3. Expansion and loss Examining data from Old Swedish and Old French, we find that Old French accepts resumption in all kinds of syntactic configurations. In Old Swedish, the possibilities are more restricted, and it is not until Late New Swedish that the language has the same possibilities of construction that we find in Old French. In addition, we find that the frequency in which resumption is used is remarkably higher in Old French than in Old Swedish.

This implies that both from a syntactic and a merely statistic point of view, resumption seems to be very well rooted in Old French. However, resumption disappears from the French language in the 16th century, while the use of the resumptive has expanded in Swedish, making it a prominent feature of the (spoken) modern language (Egerland and Falk, 2010; Nordström, 2010).

It is by examining the semantics of the resumptive structure that we may find an explanation as to why this rather surprising change could take place. Based on extensive corpus searches and a total of about 10,000 clauses, I will show that there are important semantic changes linked to the use of the resumptive.

In 10th century French, the resumptive may occur after different kinds of initial adverbial elements: temporal, locative, causative, and instrumental. In the 13th century the fronted adverbial in a resumptive structure is almost exclusively temporal (4).

4. Puis si avint aprés que une chités que li empereres avoit conquise revela encontre lui then SI happened after that one city that the emperor had conquered revolted against him ‘Then it happened that a city that the emperor had conquered revolted against him.’ (Old French)

In Old Swedish, like in the other Old Germanic languages, the most common environment for resumption is after an initial conditional clause. Over time, we find that more and more types of initial constituents may be followed by a resumptive. Resumption in Late Old Swedish is found after locative (5), concessive and temporal adverbial elements.

5. oc j the landeno som är ey mykith frosth tha lates winträn swa sta baar affskwrna and in the lands that are not much frost THA let winters so stand naked cut and in the countries where they do not get much frost, they let them stand uncovered after having been cut off.’ (Late Old Swedish)

4. Conclusion In the Swedish, the resumptive structure has been robust over time, while it disappeared from French at the same time V2 word order was lost. The robustness of the Swedish structure may be related to its capacity to gain ground and follow different kinds of semantic elements. The resumptive in Old French took on a specific temporal reading, which made it less versatile and thus it lost ground.

Selected references Verner Egerland and Cecilia Falk. Si och så. Mellan narrativitet och grammatik. In Gunilla Byrman, Anna Gustafsson, and Henrik Rahm, editors, Svensson och svenskan. Med sinnen känsliga för språk. Festskrift till Jan Svensson den 24 januari 2010. Lunds universitets pub- likationer, Lund, 2010. Gisella Ferraresi and Maria Goldbach. Particles and sentence structure: a historical perspec- tive. In Uwe Junghanns and Luka Szucsich, editors, Syntactic structures and morphological information. Mouton de Gruyter, Berlin, 2003. Adam Ledgeway. Satisfying V2 in early Romance: Merge vs. Move. J. Linguistics, 44:437– 470, 2008. Jackie Nordström. The swedish så-construction, a new point of departure. Working Papers in Scandinavian Syntax, 85:37–63, 2010. Cecilia Poletto. Word Order in Old Italian. Oxford University Press, 2014. Pieter van Reenen and Lene Schøsler. The Pragmatic Functions of the Old French Particles AINZ, APRES, DONC, LORS, OR, PUIS, and SI. In Susan C. Herring, Pieter van Reenen, and Lene Schøsler, editors, Textual Parameters in Older Languages. John Benjamins, Amsterdam / Philadelphia, 2000. Chrisitne Meklenborg Salvesen. Resumptive particles and Verb Second. Ms. Hilde Sollid and Kristin M. Eide. On verb second and the så-construction in two Mainland Scandinavian contact situations. Nordlyd, 34(3):7–28, 2007.
 Barbara S. Vance. Syntactic Change in Medieval French: Verb-Second and Null Subjects. Kluwer, Dordrecht, 1997. Sam Wolfe. Microvariation in Medieval Romance Syntax, A Comparative Study. PhD thesis, University of Cambridge, Cambridge, 2015.

ABSTRACT. Markers and models in linguistic change

The present paper is comparative in three respects. First, as is fitting for this workshop, it seeks to

compare — and contrast — the development of items within the history of Romance and Germanic.

In the context of this empirical material we then go on to consider the divide that has sometimes been

claimed to exist between functional and formal approaches to language and language change (see for

example Newmeyer 1998 and much subsequent discussion) and to argue that such a dichotomy is not

merited. We will argue instead for an approach that is both formalist and functionalist. To that end, we

will explore alternative formal approaches, comparing in particular two non-derivational frameworks,

LFG and HPSG, with the derivational model of Minimalism.

We focus in particular on three sets of data drawn from the domain of adpositions, more

specifically in all the cases we consider here prepositions. Adpositions and the constructions they enter

and develop into are a prototypical example of what we can call ‘Gabelentz’s Law’, namely that: “Alle

Afformativen waren ursprünglich selbständige Wörten [all grammatical formatives were originally

independent words]” (von der Gabelentz 1891: 250-1). The theory-neutral cover term for these items

that we will adopt is ‘marker’, defined as “a word that is ‘functional’ or ‘grammatical’ as opposed to

substantive, in the sense that its semantic content is purely logical in nature (perhaps even vacuous)”

(Sag & Pollard 1994: 45). Whether these items are to be considered heads and, if so, what constitutes

the definition and empirical coverage of such a concept is a theoretical issue to which we return below

when we investigate the various approaches that have been developed in order to model data of this

kind.

The first case-study concerns locative-directional prepositions — henceforth ‘to’ words — and

their development into case and/or infinitival markers. The sources are two PIE items *ad ‘at, near’ and

*do ‘to, toward’. Items such as French à and Italian a come via Latin ad to cover the semantic domain of

both proto-items and then develop in three ways: as indirect object markers in all the Romance

languages; as markers of animate direct objects in Spanish and various southern Italian dialects and as

complementizers also across the whole family as in French il m’a invité à venir ‘he invited me to come’. In

Germanic on the other hand we find a split. The Scandinavian languages retain the reflex of *ad as an

infinitival marker as in Swedish hon försökte att dansa ‘she tried to dance’ and have recourse to a different

item altogether for the directional and indirect object function: Sw till < Proto-Germanic *tila ‘goal’ (cf

Germ Ziel). German zu by contrast continues both the directional and indirect object marking and

develops the infinitival marking function while the locative meaning is taken up by the item in. English

retains both prepositions but aligns with German in the use of to as an infinitival marker. The

theoretical issue is then whether we have here different functional heads: P, K and C/I or whether we

allow the category P to be functionally polivalent. Minimalist analyses in general go down the former

route albeit with some discussion over the status of so-called ‘prepositional complementizers’ (Kayne

1999); work within LFG has in general accepted the P vs C/I distinction but not the introduction of a

case head K; while HPSG has in the main opted to retain a single head P (cf Borsley 2001, Abeillé et al

2006).

Our second empirical domain is what we may call ‘of’ words. Here again there are interesting

similarities and differences between Romance and Germanic. In most Romance languages (Romanian

is the exception) we find a preposition derived from Latin de ‘(down) from, about’ serving to mark the

argument of a non-verbal head as in French le président de France ‘the president of France’. Within

Germanic, English of and German von play a similar role whereas in the Scandinavian languages a

number of different prepositions fulfil this function. A major difference between the two families,

however, is that within Romance this item has also developed a complementizing function, as in Italian

ha deciso di partire ‘he decided to leave’, which leads to a three-way contrast between infinitives

introduced by a/à, those introduced by de/di, and those with no overt complementizing item. The

details of which element occurs in which subset of contexts vary across the family but the fact of the

split within the complementation system is a pan-Romance feature and provides a notable contrast to

the developments within Germanic. The individual changes are consistent with cross-linguistically

attested patterns of grammaticalization but once again raise the issue of how they are to be represented

within formal systems. In particular, if the changes are modelled as shifts from the category P to C in

the verbal domain and K in the nominal domain, this amounts to treating the phonological identity of

these items as the accidental residue of a common historical source. An alternative approach involves

revising the concept of head itself and distinguishing between ‘strong’ and ‘weak’ heads as in the

HPSG-based account advanced by Abeillé et al (2006). Our paper will review the arguments for and

against both approaches but either way it is clear that formal accounts are possible — and, we would

argue, on general methodological grounds desirable — without forsaking the possibility of adducing

functional explanations for the changes of the kinds proposed within the literature on

grammaticalization (see for example Haspelmath 1989).

In the two case studies above the focus has been items that were already prepositional and

considers how they have moved on from there to a new range of functions. In the terms of one

common definition of what is covered by grammaticalization, this involves the shift ‘from a less to a

more grammatical status’ (Kuryłowicz 1969: 65). Our final dataset concerns an earlier stage in which a

grammatical item evolves from a lexical one as in what we may call ‘house’ words. The case of the

French preposition chez ‘at, with’ which derives from Latin casa ‘hut’ and is cognate with Spanish and

Italian casa ‘house is familiar from the discussion of its development by Longobardi (2001) and more

recently Harrison & Ashby (2003), Sornicola (2011), Plank (2015). Less widely discussed are the

apparently parallel developments in the Scandinavian languages, e.g. Danish som det hedder hos Byron ‘as

Byron has it’ or Swedish sitt hos mig ‘sit next to me’. That an item such as this may go on to the next

stage is evidenced by the fact that in modern Faroese hos serves a function akin to that of English of and

French de described above. At the same time, unlike in French, where the grammaticalization of casa

goes together with the loss of the original lexical item, in the Scandinavian languages hos co-exists with

its etymological source hus ‘house’. Crucial to Longobardi’s analysis of chez is the availability of the

theoretical option of raising to an empty D head. We will seek to show however that the Scandinavian

data are not compatible with such an account and that once again an alternative formal solution is

available within the context of what in other respects appears to be a functionally motivated change.

References

Abeillé, Anne, Olivier Bonami, Danièle Godard & Jesse Tseng (2006) The syntax of French à and de: an

HPSG analysis. In Patrick Saint-Dizier (ed.) Dimensions of the Syntax and Semantics of Prepositions.

Dordrecht: Springer, 147-162.

Borsley, Robert (2001) What do ‘prepositional complementizers’ do? Probus 13. 155–171.

Gabelentz, Georg von der (1891) Die Sprachwissenschaft: Ihre Aufgaben, Methoden und bisherigen Ergebnisse.

Leipzig: Weigel Nachf.

Harrison, Annette & William J. Ashby (2003) Remodelling the house: the grammaticalization of Latin

casa to French chez. Forum for Modern Language Study 39.4: 386-399.

Haspelmath, Martin (1989) From purposive to infinitive — a universal path of grammaticization. Folia

Linguistica Historica 10: 287–310.

Kayne, Richard (1999) Prepositional complementizers as attractors. Probus 11. 39–73.

Kuryłowicz, Jerzy (1965) The evolution of grammatical categories. Diogenes 13: 55-71.

Longobardi, Giuseppe (2001) Formal syntax, diachronic Minimalism, and etymology: the history of

French chez. Linguistic Inquiry 32: 275-302.

Newmeyer, Frederick J. (1998) Language Form and Language Function. Cambridge, MA: MIT Press.

Plank, Frans (2015) Time for change. In Carlotta Viti (ed.) Perspectives on Historical Syntax. Amsterdam:

John Benjamins, 61-92.

Sag, Ivan & Carl Pollard (1994) Head-Driven Phrase Structure Grammar. Chicago: Chicago University

Press.

Sornicola, Rosanna (2011) Il lessema latino CASA e i suoi continuatori galloromanzi. Un problema di

storia culturale. In A. Overbeck, W. Schweickard & H. Völker (eds), Lexicon, Varietät, Philologie,

Romanistische Studien Günter Holtus zum 65. Geburtstag, Berlin, De Gruyter, 611-634.

ABSTRACT. Building on the work of Thomason and Kaufman (1988), Ratcliffe (2005 ) proposes that the sociolinguistic parameters of contact can be defined on two dimensions: direction of influence—shift (L1>> L2) or borrowing (L2>L1)—and degree of prior stable bilingualism. The first parameter is binary and the second is contiguous. Ratcliffe further proposes that these parameters determine the structural result of contact. In this communication I observe structural result by comparing the dialects with classical Arabic and core dialects in terms of four broad linguistic categories: basic vocabulary as defined by the hundred word Swadesh list, consonant inventory, verbal inflectional morphology, and key morphosyntactic features including basic word orders and the form of negatives and interrogatives. Based on observations from a variety of languages, the ypothesis predicts that shift in a situation of minimal prior bilingualism should result in loss of complex phonology and inflectional morphology, the typical creole phenomenon (McWhorter 1988, Romaine 1988, Lefebvre 1998, Mufwene 2001, etc,)., and the oposite extreme, borrowing in a situation of maximal bilingualism hould result in morphosyntactic convergence with minimal borrowiing of lexicon or inflectional morphology (Ross 1996, Aikhenvald 2006). The data from Nubi and Buxari confirm these predictions for the respective contact situations. The data from Maltese and Kormatiki, however, are more problematic and require further analysis.

degree L1>L2 L2>L1 of prior bilingualism LOW creole adstrate

HIGH substrate morphosyntactic convergence

ABSTRACT. In Matras’ (2009: 175) words, ‘there appears to be a near-consensus view that the borrowing of verbs is made more cumbersome in some languages due to the widespread tendency of verbs to be morphologically more complex’. Nevertheless verbs do get borrowed, and various cross-linguistically common strategies have been identified for how they are integrated (or not) into the structure of the recipient language (see, e.g. Wichmann and Wohlgemuth 2008, Wohlgemuth 2009, and Matras 2009: 176). Semitic languages present special challenges in this area, with their unusual non-concatenative morphology, the ‘root-and-pattern’ system. A verbal form is typically formed from an abstract tri-consonantal or quadri-consonantal root: √C1-C2-C3, or √C1-C2-C3-C4). The stem is formed on a pattern or template (e.g. C1aC2aC3a or mC1uC2əC3-) on the basis of the derivational class (broadly associated with, e.g., transitivity, causativity, passivity etc.) and the tense-aspect-mood category required. This is then inflected for person by means of affixes. Thus, a typical verbal lexeme in a Semitic language is defined by (1) its consonantal root and (2) its derivational class (Hebrew/Aramaic binyan/Arabic wazn). Any loan-verb – if fully integrated into the system – must conform to this system, i.e. the loan-verb should be allocated a root and a derivational class. It is these requirements that make the Semitic languages a special case for the study of verb loans. It might be expected that loan-verbs between Semitic languages would be easier to adopt: a loan-verb from another Semitic language already has a root and derivation class. But it is not quite so simple. Semitic languages do not always have the same rules as to what constitutes a possible root. They also do not have the same inventories of derivational classes, (although there is overlap). Even those which are historically cognate may vary considerably in form and in function. The North-Eastern Neo-Aramaic (NENA) dialects spoken in N. Iraq have borrowed many verb lexemes from Arabic, both vernacular and standard, as a result of intense contact, especially since the 20th century. Such loan-verbs come ready with a consonantal root and derivational class. The question arises: does the Arabic root also constitute a root in the Neo-Aramaic dialect, and if not, will it be adapted (and how?) to conform to the existing rules of the recipient language? For instance, in NENA there are no ‘geminated roots’, i.e. roots where C2=C3, in the first (the simplest) derivational class. In Arabic these are very common, e.g. dqq i ‘to knock’ and hzz i ‘to shake’. These are in fact borrowed into NENA as middle-y verbs (√C1-y-C3), which are already common in the language: NENA dyq I ‘to knock’ and NENA hyz I ‘to shake’. The other main question is: On what basis is a derivational class selected for a loan-verb? There are some cognate derivations shared by Arabic and NENA. Naturally these do not have identical forms, although there are some similarities in form and function. Although speakers cannot be expected to be expert in Comparative Semitics, they might well allocate loan-verbs to derivations that are somehow similar in form and/or function: and these are often cognate. For those Arabic loanverbs whose derivational class has no cognate in NENA, there are two choices: (1) select a non-cognate derivation or (2) transfer some derivational morphology along with the verb lexeme: the latter strategy can result ultimately in a borrowing of a derivational class (see Coghill 2015 for an investigation of various stages of this process in several Neo-Aramaic languages). This paper will address the tendencies and underlying motivations of strategies used to adopt and adapt loanverbs from Arabic into North-Eastern Aramaic, in particular the adaptation of the root and the selection of a derivational class. It will also compare the results with some other cases of intra-Semitic verb borrowing.

References Coghill, Eleanor. 2015. ‘Borrowing of verbal derivational morphology between Semitic languages: The case of Arabic verb derivations in Neo-Aramaic’, in F. Gardani, P. Arkadiev, and N. Amiridze (eds), Borrowed Morphology (Language Contact and Bilingualism 8). Berlin: De Gruyter Mouton, 83–107. Matras, Yaron. 2009. Language Contact. Cambridge: Cambridge University Press. Wichmann, S. & Wohlgemuth, J. 2008. ‘Loan verbs in a typological perspective’, in Stolz, Bakker and Paloma (eds.) Aspects of language contact. Berlin: Mouton de Gruyter. Wohlgemuth, Jan. 2009. A Typology of Verbal Borrowings. Berlin: Mouton de Gruyter.

ABSTRACT. The Islamic conquests of the late antique Near East and North Africa brought speakers of Arabic into contact with speakers of a number of other languages, both closely related, most notably Aramaic, others only distantly related or not at all. Arabic dialectologists have taken great interest in identifying features in the modern dialects transferred from the substrate and adstrate languages spoken by the populations who eventually adopted Arabic. When the donor language is not related to Arabic, identifying these features is often relatively clear. However, in cases where speakers of another Semitic language adopted Arabic, identifying transferred features is quite challenging. Aramaic was the lingua franca of much of the Levant and Mesopotamia, and speakers of Arabic and Aramaic were in frequent contact for centuries. It is thus not surprising that scholars of Syrian and Mesopotamian dialects have identified the origin of a number of features of these dialects as Aramaic. Most studies dedicated to this topic have concentrated on the lexicon (e.g., Hopkins 1995; Contini 1999; Müller-Kessler 2003). Much less work has explicitly focused on phonological, morphological and syntactic features (Diem 1971; Behnstedt 1991; Arnold and Behnstedt 1993; Retsö 2000); nevertheless, a number of features have become widely (though not universally) accepted as examples of transfer from Aramaic into Arabic via second-language acquisition by Aramaic speakers after the Islamic conquests; for example, the shift ā > ō (e.g., lisān > lisōn; fallāḥ > fallōḥ) in a number of Lebanese and Anatolian dialects; 3rd and 2nd plural pronouns (e.g., hinnōn/hinnēn; kon/ken) in a number of Syrian dialects, and others. In this paper, we reexamine some of these proposed features. We will argue that many of these features are better explained as the result of common linguistic processes. Others are well attested in a number of Semitic languages, and may therefore constitute retentions in Arabic. Overall, we will argue that the influence of Aramaic on the contemporary Arabic dialects of the Levant and Mesopotamia is indeed quite limited. Further, we will argue that we should not expect a great deal of Aramaic influence on these dialects for several reasons. First, Arabic was more widespread in the pre-Islamic period than is typically appreciated (Al-Jallad forthcoming). Second, there has been a great deal of convergence in modern dialects. Finally, given the long history of Aramaic/Arabic contact, bilingualism was probably extremely common, and thus rapid imperfect acquisition of Arabic by Aramaic speakers in the early Islamic period, which would have ostensibly led to the transfer of many Aramaic features, can now be considered very unlikely. The results should have implications to our understanding of the social structure of the region and the difference between speakers in urban versus rural areas.

ABSTRACT. Internet language is often criticized, parodied, and (less) often celebrated for its use of novel word forms, abbreviations, and (un)grammatical practices (Thurlow, 2006). ‘Grammar’ in the internet context is understood to include syntax, morphology, orthography, and typography (Herring, 2012). But where does this “Netspeak,” as it has been dubbed by Crystal (2001), originate? What motivates individual internet users to generate nonstandard forms in the first place, and in what contexts do they do so? Which forms are subsequently taken up by others?

Insight into these questions can be gained from an examination of what I call Special Internet Language Varieties (SILVs) and their contexts of use. SILVs are sociolects associated with particular internet subcultures that self-identify, in large part, in terms of their creative nonstandard linguistic practices. SILVs have been identified in several languages; English examples include Leet, LOLspeak (from LOLcats), and Doge. The latter two are multimodal, incorporating both text and images. Although SILVs flourish in specialized contexts, and most internet users are unaware of them, some elements of SILVs are adopted more widely and become part of general “Netspeak” in their respective languages, and some elements spread to offline usage, typically on clothing, novelty items, and in advertising contexts (Author et al., 2012).

In this talk, I present an analysis of four SILVs based in languages with contrasting grammatical structures and writing systems: Leet (English), Fakatsa (Hebrew), Padonki (Russian), and Martian Language (Chinese). Data were collected several years ago from the online contexts in which each variety most naturally occurs and analyzed by a multilingual team of researchers. The characteristic features of each SILV were identified and described at the levels of typography, orthography, word formation, lexis, syntax, pragmatics, and rhetoric. A speech act analysis was also conducted of the pragmatic functions of SILV usage in context.

While we found many language- and context-specific differences among the four SILVs, our findings also reveal a surprising number of similarities, even though there is little evidence of any direct contact between the the individual varieties. This leads us to posit several general principles to account for the similarities, which we argued could have also motivated individual internet users to create the deviant forms in the first place. At the same time, the social context – especially the gender demographics of the group and the values around which each subculture originally formed – create a context in which individual innovations are judged appropriate or not relative to subcultural norms.

In this talk, these findings are presented, along with evidence of the spread of SILV features outside the subculture associated with each into general internet “slang” in the respective languages and thence into offline contexts. This spread itself raises interesting questions; for example, what determines which few features, out of a large set of creative and variable practices, will spread? In concluding, I propose explanations for this spread, and consider the broader implications of the characteristics, origins, and spread of SILVs for theories of language change.

References

Author et al. (2012). Crystal, D. (2001). Language and the Internet. Cambridge: Cambridge University Press. Herring, S. C. (2012). Grammar and electronic communication. In C. Chapelle (Ed.), Encyclopedia of applied linguistics. Hoboken, NJ: Wiley-Blackwell. Preprint: http://ella.slis.indiana.edu/~herring/e-grammar.pdf Thurlow, C. (2006). From statistical panic to moral panic. Journal of Computer-Mediated Communication, 11, 667–701.

ABSTRACT. The mechanisms of propagation in language change is driven by the selection motives of the speakers of a language community. In recent studies, the utterance selection model was applied for a formal analysis of language change dynamics driven by speaker selection (cf. Blythe & Croft 2012). Blythe & Croft used the utterance selection model to show that empirically frequent S-curved trajectories of language change can only be captured if replicator selection drives the speakers' selection motives. In our study, we reimplemented the utterance selection model as applied by Blythe & Croft, but tested four differnt types of replicator selection. Our results reveal that the typical S-shape trajectory is only systematically reproduced by one of the four types. Furthermore, the study points to the fact that Blythe & Croft's replicator selection type models functional selection instead of - as they argued - social selection.

ABSTRACT. Introduction: A sample typology of diachronic developments was compared with patterns of categorical errors from experimentally elicited misperception and natural errors in perception and production. All of these correlations are highly significant These results demonstrate parallels between patterns in synchronic data and sound changes. They do not necessarily indicate that such errors are a primary source of sound change, but it is possible that some changes result from frequently recurring errors in perception and production. Gradual changes and the spread of changes may also be facilitated by the same confusability that causes these patterns of errors. Even if such errors are not a source of change, the correlations demonstrate that synchronic data can be used as a line of evidence in reconstructing sound changes.

Production: Natural patterns of variation in production can demonstrate biases resulting from motor planning, gestural mechanics, and aerodynamics (Garrett and Johnson 2013), as well as social factors (Anttila 2007). Existing variability in production makes it possible for speakers to gradually shift their productions, or for listeners to reanalyze the target. Categorical mispronunciations provide a different source of evidence for aspects of pronunciation that may influence sound changes. Collections of such errors include e.g. van den Broecke and Goldstein’s (1980) collection from English, Dutch, and German. Such errors have been interpreted as evidence for phonological categorization and motor planning. Sound changes may reflect similar biases (Garrett and Johnson 2013).

Perception: Perception is also a potential source of change. Acoustic similarity facilitates misperception and subsequently change, if listeners frequently enough fail to account for particular contextual influences on phonetic realization (Ohala 1993, a.o.). Though experimental conditions do not provide an exact parallel for historical change, misperception patterns have been demonstrated to parallel some common sound changes, e.g. k > tʃ (Chang et al. 2001). Patterns of perception errors in natural speech provide evidence for phonological processing and lexical retrieval as well as the salience of different acoustic cues (Browman 1980). Tang (2015) provides a tabulated collection of single-phoneme replacement errors.

Typological correlations: I collected a small typology of unconditioned and conditioned consonant developments (989 observations) in Indo-European, Semitic, Sino-Tibetan, Uto-Aztecan, Uralic, Mayan, Austronesian, and Otomanguean. These were selected to represent a range of unrelated but relatively well-described language families, using inputs of segments that are present in most of these proto-languages (p t k b d g s z ʃ m n l r j w h); the outputs also include tʃ dʒ ʒ v f, because these are common outcomes in the synchronic perception and production data, which is primarily from English. This typology of diachronic developments is positively correlated with the matrix of perceptual confusions collected by Miller and Nicely (1955) in English nonce words, pooled across all listening conditions (white noise of different intensities, some with frequency-based filtering). The correlation is highly significant: r(174) = 0.93, p < 0.001, though largely due to cells for unchanged segments and accurately identified segments. Omitting these cells to avoid the influence of stability as well as the relative stability of different segments, the correlation is smaller but still significant: r(163) = 0.17, p = 0.027. Correlations with confusions from other misperception experiments are similar. Perception errors in natural speech from Tang’s (2015) compilation across English dialects provide the strongest parallel for diachronic developments. The correlation with my diachronic matrix is highly significant: r(382) = 0.94, p < 0.001. Omitting cells of unchanged segments and correctly identified segments from the two respective matrices, the correlation is still highly significant: r(366) = 0.49, p < 0.001. Patterns in production errors from van den Broecke and Goldstein’s (1980) collection, pooled across the collections from English, German, and Dutch, are also significantly correlated with the diachronic developments: r(270) = 0.33, p < 0.001. This collection doesn’t provide comparisons with the number of correctly produced segments, so this data is based only on the relative frequency of different errors compared with diachronic changes.

Conclusions: Patterns in misperception and production errors provide a parallel for phonetically rapid diachronic changes. These errors are generally limited to sounds that already exist within the system, so they provide the strongest parallel for diachronic mergers or shifts within an inventory. These connections shed some light on possible factors facilitating sound change. Experimental methods can complement some of the limitations of traditional comparative reconstruction, particularly to demonstrate the plausibility of developments without established diachronic parallels. Differences in correlations under different conditions and for different segments may further help identify the mechanisms of particular diachronic changes.

References Anttila, Arto. 2007. Variation and optionality. In Paul de Lacy (ed.), The Cambridge handbook of phonology, 519–536. Cambridge: Cambridge Univ. Press.

Browman, Catherine. 1980. Perceptual processing: Evidence from slips of the ear. In Victoria Fromkin (ed.), Errors in linguistic performance: Slips of the tongue, ear, pen, and hand, 213–230. New York: Academic Press.

Chang, Steve, Madelain Plauché, & John Ohala. 2001. Markedness and consonant confusion asymmetries. In Elizabeth Hume & Keith Johnson (eds.), The Role of speech perception in phonology, 79–101. San Diego: Academic Press.

Garrett, Andrew, & Keith Johnson. 2013. Phonetic bias in sound change. In Alan Yu (ed.), Origins of sound change: Approaches to phonologization, 51–97. Oxford: Oxford Univ. Press.

Miller, George, & Patricia Nicely. 1955. An analysis of perceptual confusions among some English consonants. Journal of the Acoustical Society of America 27(2): 338–352.

Ohala, John. 1993. The phonetics of sound change. In Charles Jones (ed.), Historical linguistics: Problems and perspectives, 237–278. New York: Longman Publishing.

Tang, Kevin. 2015. Naturalistic Speech Perception. London: University College London dissertation.

van den Broecke, Marcel, & Louis Goldstein. 1980. Consonant features in speech errors. In Victoria Fromkin (ed.), Errors in linguistic performance: Slips of the tongue, ear, pen, and hand, 47–65. New York: Academic Press.

ABSTRACT. Languages of the Cariban family are generally described as moderately polysynthetic (Gildea 2012):  verbs generally take multiple TAM suffixes, a set of person prefixes/proclitics organized hierarchically, and verbal number suffixes/enclitics (p.461-462);  nouns take person prefixes/proclitics, nominal number suffixes/enclitics, and case-markers/postpositional enclitics (p.465);  postpositions take person prefixes/proclitics and the adverbial number suffix/enclitic (p.465);  adverbs take only the adverbial number suffix  no other parts of speech take inflectional morphology. In contrast, initial results from a documentation project show that Yawarana (yar), spoken by fewer than 30 elders in central Venezuela, has an almost entirely isolating profile:  verbs take various TAM suffixes (all homophonous with derivational suffixes) but have lost all etymological person-marking prefixes/proclitics;  nouns have also lost etymological person prefixes/proclitics and are in the process of losing case-marking;  etymological verbal number suffixes are lost, with etymological nominal and adverbial number suffixes now occurring on all parts of speech. This paper describes the main historical changes that have conspired to so radically reduce the inflectional complexity of Yawarana and discusses the likely role played by language contact. The first type of change, the one that led to most of the loss of inflectional complexity, is the replacement of the Proto-Cariban verbal inflectional system with analytical constructions (parallel to modern English auxiliary constructions): constructions with subordinate verb forms (deverbal nouns and deverbal adverbs) in biclausal and copular constructions got used to convey TAM distinctions, after which the prior system dropped out of use (Gildea 1998; 2012:465-473). This change has the effect of eliminating nearly all etymological categories of verbal inflectional morphology (only the imperative suffix survives) and of introducing nominal and adverbial person- and number-marking into the domain of verbal predication. In terms of alignment, this change introduces ergative case-marking (formerly oblique) and absolutive verbal person prefixes/proclitics and number suffixes/enclitics (formerly marking nominal possession). A parallel change is complete also in Cariban languages Kuikuro and Makushi, and is well advanced in Panare, Pemón, and Akawaio. The second type of change is loss of inflections within specific constructions that continue to be used. In Yawarana, the Proto-Carib set of nominal/postpositional person prefixes/proclitics is almost entirely lost, leaving most heads unmarked for person. There are incipient first and second person proclitics wï= ‘1’ and më= ‘2’, both transparently derived from free pronouns and both occurring equally freely on nouns, verbs, and postpositions. Third person is less consistent: nouns continue to use Proto-Venezuelan *i- before consonants and *it- before vowels; postpositions use t(ë)- (source unclear); intransitive verbs and transitive verbs with 1A/2A have no prefix, whereas transitive verbs with 3A3P take the prefix ta- (source unclear). There is no trace of a third person reflexive distinction (Proto-Cariban *t-) nor of a reflex of the inclusive Proto-Cariban person prefix *k- (although the distinction is pronominally preserved). Changes of this second type that are still in progress include the incipient loss of the ergative suffix (simply absent in many clauses, cf. ex 3-4) and the loss of a part-of-speech distinction in the formerly nominal and adverbial number markers, which now occur regularly marking any of the three parts of speech (Table 1). Applying these examples to the research questions of the workshop, we can say: (1) the features lost are case-marking, person-marking, and TAM-marking; (2) both prefixes and suffixes are lost; and (3) given the continued robustness of cognate morphology in related languages, complexity is Loss of inflection in Yawarana (Cariban) – 2/2 not a plausible factor in the loss. Regarding (4), change in syntax is the motivation for the first kind of loss, but not for the second, although the incipient loss of ergative case may be correlated with the innovative order SVO. We are especially intrigued by (5) the possible role of contact in these changes. The first type of change is attested in 5-7 other Cariban languages, all except Panare characterized by extensive multilingualism and contact with other cultures; however, more limited reanalyses of these types are attested in other Cariban languages, the difference appears to be in quantity rather than process. In contrast, the incipient loss of ergative case is attested only in Yawarana and sister language Mapoyo, both in an advanced stage of obsolescence. This loss could be a consequence of ongoing language shift, a particularly extreme form of language contact. Examples illustrating intransitive (ex. 1) and transitive constructions (exs. 2 and 3) (1) a. wïrë sepan-se. b. ¿më=kopi-che ka më=pëkëpene? 1SG get_married-PST 2SG=bathe-PST QP 2SG=alone ‘I got married’ {HistAnFo.49} ‘did you bathe by yourself? {ELICITATION:GrMe} (2) wïrë pojtë-po-ri wïrë makë-ri entë=ma 1SG want-DES-IPFV 1SG mother-POS here=INTS ‘I want my mother here’ {ConvAmGu.100} (3) a. tajne yakëtë-se t-apë-ri, ta wïrë=ya 3PL cut-PST 3-arm-POS say.IPFV 1SG=ERG ‘they cut their arms, I say’ {{CtoWaruMaFl.51} b. u=koyoweti ta-yakëtë-se, u=yarë-se ta t-ïwïj=yaka 1=navel 3/3-cut-PST 1=bring-PST then 3-house.POS=into ‘[She] cut my umbilical [cord], then brought me to her place’ {Hist2MaPe.10} c. wï=yakëtë-sapë 1SG=cut-PERF ‘I was cut’ {CtoWaruMaFl.68} Singular Plural Pronouns aniki ‘who’ aniki-santomo(-jne) Nouns chiyakomo ‘friend’; më=chiyakomo ‘your=friend’ chiyakon-tomo(-jne) putiya ‘bottle’; i-putiya-ri ‘his-bottle.POS’ iputiyari-jne Adverb asakë ‘two’ asakë-jne Post-positions wïy=akërë ‘with me’; t-akërë ‘with him’ ? më=pëkëpene ‘by yourself’; të=pëkëpene ‘by himself’ ? Verbs wïnïkï-rï ‘sleep-IPFV’ or ‘sleep-NZR’ wïnïkï-kontomo(-jne) sekujnë-se ‘comb_self-PST’ or ‘comb_self-PTCP’ sekujnë(-pëj)-se-jne epe-se ‘flee-PST’ or ‘flee-PTCP’ epe-se-kontomo wereta-sapë ‘sit_down-PERF’ or ‘sit_down-NZR’ wereta-saj-kontomo waimu-po-jra ‘speak-DES-NEG’ waimu-po-jra-kontomo Table 1. Yawarana number, person and TAM inflection by part of speech References Gildea, Spike. 1998. On Reconstructing Grammar: Comparative Cariban Morphosyntax. New York: Oxford University Press. Gildea, Spike. 2012. “Linguistic Studies in the Cariban Family.” In The Indigenous Languages of South America: A Comprehensive Guide, ed. by Lyle Campbell and Verónica Grondona, 441–94. Berlin: Mouton de Gruyter.

ABSTRACT. This paper describes ongoing inflectional change in Copala Triqui, an endangered language of Mexico. Older speakers use a complex system of tonal and segmental changes for aspectual inflection. Younger speakers have reduced and simplified the segmental inflection, and place a heavier load on tonal change to indicate aspect.

ABSTRACT. In this talk, we first demonstrate the object marking system in Cuwabo (Bantu, Mozambique), addressing its morphological restriction on classes 1/2 nouns, the underlying semantic aspects of these classes related to animacy, and the syntactic requirement of obligatory object marking. Adopting a functional-comparative perspective, we then relate the absence of object marking for the remaining classes to a number of other morphosyntactic developments in the language. The talk shows, on the one hand, the development of a range of formal means to express discourse-pragmatic relations, thus illustrating different grammaticalisation paths, and on the other hand, contributes to understanding the function of typical Bantu object marking systems through an analysis of its loss. More widely, the talk illustrates the interplay of loss of inflectional morphology and the development of alternative morphosyntactic means to address the functions associated with the lost morphology.

ABSTRACT. Bilingual mixed languages typically emerge in situations of community bilingualism when minority speakers are under great pressure to adopt a dominating language but nevertheless resist to total linguistic assimilation (Thomason 2003). The preset paper argues (i) that the language under investigation, Amish Shwitzer, is a so far unreported example of a grammar-lexicon mixed language, (ii) that Shwitzer is exceptional among grammar-lexicon mixed languages insofar as its ancestral languages are genetically closely related to one another, and (iii) that the sociolinguistic circumstances under which the mixed language emerged are untypical, too, insofar as Shwitzer speakers borrowed many structural features from another language, Pennsylvania Dutch, that has never been a dominant language in their community. Therefore, the development of Shwitzer is a spectacular case of contact-induced language change where non-dominant (i.e., Pennsylvania Dutch) speakers played a crucial role.

The present paper consists of three parts. Part 1 gives a short overview of the history and the present situation of the language, based on a fieldtrip in 2016. Shwitzer is a language spoken by approximately 3000 speakers in Adams County, Indiana (USA) (Meyers & Nolt 2005; Humpa 1996; Fleischer & Louden 2010). Shwitzer speakers are descendants of Swiss Anabaptists who immigrated in the mid-19th century. Whereas the 19th century immigrants spoke Bernese Swiss German, Shwitzer has diverged from its European ancestor in interesting ways. Today, there are two clearly distinct varieties of Shwitzer, a Mennonite and an Amish variety. The Mennonite variety is a typical example of a moribund American German(ic) heritage language. Amish Shwitzer, however, is not only fully vital and acquired by children as their first language, but it has also converged with Pennsylvania Dutch to a remarkable degree.

In part 2, I will give an outline of the linguistic structure of Amish Shwitzer. The language maintained many specifically Bernese characteristics, mainly in lexicon and phonology. However, its morphosyntactic structure is almost fully pennsylvanized (case inflection, tense and aspect, nonfinite complementation, function words, etc.). I conclude that the structure of Amish Shwitzer must be analyzed as a specific type of mixed language, namely a stabilized grammar-lexicon mixed language (Meakins, to appear), reminiscent of e.g. Ma’á/Mbugu (Humpa 1996), with grammatical structure and morphemes from one language (Pennsylvania Dutch) and vocabulary from another (Bernese Swiss German). To our knowledge, Amish Shwitzer is the first reported grammar-lexicon mixed language whose ancestral languages are close genetic relatives (cf. Matras & Bakker 2003 for an overview).

Part 3 seeks for a sociolinguistic explanation of the emergence of the mixed nature of Amish Shwitzer. The explanation for the maintenance of many Bernese traits is rather obvious, since this particular group of Amish is a minority within the greater Amish community and retained a separate Swiss-Amish identity not only in language, but also in dress or custom (Meyers & Nolt 2005). However, it is the great degree of pennsylvanization that calls for an explanation. It is a challenging observation that, within the Swiss-Amish community of Adams County, Pennsylvania Dutch has never played the role of a dominant language as we would expect it when comparing the Amish Shwitzer situation to the circumstances under which other mixed languages emerged. Also, bilingualism in both ancestral languages (again a typical prerequisite for the emergence of mixed languages, cf. Bakker 1997, Thomason 2001, Velupillai 2015) does not seem to have been widespread within the community. Instead, spoken Pennsylvania Dutch must be rather classified as a non-dominant variety among the Adams County Amish. Shwitzer speakers even joke about the Pennsylvania Dutch speaking Amish: si visse nid vie racht shwatze (‘they do not know how to speak right’). I will discuss three factors that eventually made possible the shift towards Pennsylvania Dutch grammatical structure. First, Pennsylvania Dutch speaking Amish from other parts of North America have married into the Swiss-Amish community, especially after the 1960s (Humpa 1996, Meyers & Nolt 2005). Children born to Pennsylvania Dutch speaking mothers in mixed dialect families were probably the agents of the shift. However, this does not yet explain why their newly emerging variety was so successful and had a chance to spread throughout all of the Swiss-Amish community. Second, there is an interesting asymmetry in the intelligibility of Shwitzer vs. Pennsylvania Dutch. Speakers report that Pennsylvania Dutch is easily understandable for Shwitzer speakers but not vice-versa. Therefore, there has never been a need for outsider Amish entering the Shwitzer-speaking community to fully adapt to the language of Adams County. However, this does not explain why Pennsylvania Dutch is understandable for Shwitzer speakers. Third, Pennsylvania Dutch happens to be more similar to written German in a number of salient phonological features. This is a linguistic accident and has little to do with the sociolinguistic situation. All Amish, Pennsylvania Dutch or Shwitzer speaking, have some passive knowledge of written German (hochtitsch ‘High German’ or bibutitsch ‘Bible German’). Despite the otherwise great linguistic differences between written German and Pennsylvania Dutch, the two are perceived as the same language by the Shwitzer speaking Amish.

I conclude that bilingual mixed languages may emerge, too, even when they descend from closely related varieties. Furthermore, (asymmetrical) intelligibility of the ancestral languages may give rise to new types of sociolinguistic circumstances for the emergence of mixed languages where concepts such as ‘dominant language’ or ‘widespread bilingualism’ are not fully applicable.

References

Bakker, Peter. 1997. "A Language of our Own": The Genesis of Michif - the Mixed Cree-French language of the Canadian Métis. New York, Oxford : Oxford University Press. Fleischer, Jürg & Louden, Mark. 2010. Das Amish Swiss German im nordöstlichen Indiana: eine alemannisch-pfälzische Mischmundart? In: Christen, Helen et al. (eds.): Alemannische Dialektologie: Wege in die Zukunft. Stuttgart: Steiner. 231-245. Humpa, Gregory J. (1996): Retention and loss of Bernese Alemannic traits in an Indiana Amish dialect: a comparative-historical study. Unpublished PhD thesis Purdue University. Matras, Yaron & Bakker, Peter. 2003. The study of mixed languages. In: Matras, Yaron & Bakker, Peter (eds.). 2003. The Mixed Language Debate: Theoretical and Empirical Advances. Berlin: Mouton de Gruyter. pp. 1-20. Meakins, Felicity. To appear. Mixed languages. In: Aronoff, Mark (ed.): Oxford Research Encyclopedia of Linguistics. Oxford: Oxford University Press. Meyers, Thomas J. & Nolt, Steven M. 2005. An Amish Patchwork. Indiana’s Old Orders in the Modern World. Bloomington: Indiana University Press. Thomason, Sarah G. (2001): Language Contact. An Introduction. Washington D.C.: Georgetown UP. Thomason, Sarah G. 2003. Social factors and linguistic processes in the emergence of stable mixed languages. In: Matras, Yaron & Bakker, Peter (eds.). 2003. The Mixed Language Debate: Theoretical and Empirical Advances. Berlin: Mouton de Gruyter. pp. 21-39. Velupillai, Viveka. 2015. Pidgins, Creoles and Mixed Languages. An Introduction. Amsterdam/Philadelphia: Benjamins.

ABSTRACT. This paper addresses a cluster of research questions, including:

1. What kinds of evidence are needed to trace the introduction of non-dominant speaker features into a new language? 2. What kinds of features do non-dominant speakers ‘successfully’ introduce? 3. How do these innovations find their way into monolingual speech?

Whereas related studies often focus, in historical linguistics, on the distant past (‘substrates’) or, in heritage language linguistics, present-day contact settings, our data come from the recent, well-documented past, previously German-speaking communities in eastern Wisconsin. Contact emerged in the mid-nineteenth century, and some bilingual speakers remain even today. To the first question, the setting we’re working with gives us a ‘sweet spot’ for contact effects. It’s shallow enough in time that data from letters and even audio recordings exist but deep enough to spot what effects linger several generations after most speakers have shifted to English. Recent work (Wilkerson et al. 2014) has traced evidence on the demographic setting of language shift in these communities, showing that a key generation of learners—born after about 1920—would have had tremendous amounts of input in English from non-native speakers. In van Coetsem’s terms (2000), these speakers imposed features of German onto English. To the second question, we compare the effects of German found in the English of eastern Wisconsin with findings on the challenges that adult native speakers of German face learning English as described in the second language acquisition literature. Aspects of English shown to pose difficulty for proficient German speakers of English include loss of distinction between count and non-count nouns, variation in tense and aspect, and devoicing of obstruents, among others (cf. Biersack 2003, Erling 2002). Writings by early English L1 German-Americans show a very similar range of non-native-like structures found among adult L2 learners. Bagwell et al. (forthcoming) found German-sourced features in family letters, showing variation in tense and aspect, overgeneralization of German modal particles, and German-like word order as well as spellings, which is highly suggestive of German pronunciations. To the third question, contemporary speakers no longer show many of these features, but produce instead syntactic and phonological structures typical of broader American English (Sewell & Salmons 2014). (Wilkerson et al. (2014:6) show some features and their anticipated trajectories over time. We are about to survey Wisconsin speakers to test these patterns, with results expected by the time of the workshop. We conclude that shallow-time depth historical work provides a critical source for understanding the effects of non-dominant speakers in shaping community speech. Such speakers introduce a full range of L2 effects into the community and with sufficient input from L2 speakers, these appear in L1 speech. Longer term success of such features appear subject to patterns familiar from koine formation (Kerswill 2002), though we await data to confirm this.

References Bagwell, Angela, Samantha Litty and Mike Olson. Forthcoming. Wisconsin immigrant letters: German influence and imposition on Wisconsin English. In Raymond Hickey (ed.), Keeping in Touch. Familiar Letters across the English-speaking World. Amsterdam: John Benjamins.

Biersack, Sonja. 2003. Systematische Aussprachefehler deutscher Muttersprachler im Englischen. Eine phonetisch-phonologische Bestandsaufnahme. Forschungsberichte des Instituts für Phonetik und Sprachliche Kommunikation der Universität München (FIPKM) 39: 37-130.

van Coetsem, Frans. 2000. A general and unified theory of the transmission process in language contact. Heidelberg: Winter.

Erling, Elizabeth J. 2002. ‘I learn English since ten years’: The global English debate and the German university classroom. English Today 18: 8-13.

Kerswill, Paul. 2002. Koineization and Accommodation. In Chambers, Trudgill, & Schilling-Estes (eds.), The Handbook of Language Variation and Change, 669-702. Malden, MA: Blackwell.

Sewell, Alyson & Salmons, Joseph. 2014. How far-reaching are the effects of contact? Parasitic gapping in Wisconsin German and English. Questioning Language Contact: Limits of Contact, Contact at its Limits, ed. Robert Nicolai, 217-251. Leiden: Brill.

Wilkerson, Miranda, Mark Livengood & Joseph Salmons. 2014. The Socio-Historical Context of Imposition in Substrate Effects. Journal of English Linguistics 42.1-23.

ABSTRACT. Experiential verbs are known for their tendency to display a range of construction types both within and across languages. As suggested in many studies, the common variation associated with these verbs finds a partial explanation in the dual nature of the experiencer argument, the prominent human participant which seems to be the obvious candidate to appear as the topical clausal subject, but which happens to carry features of affectedness more typical of objects or obliques (e.g. Bossong 1998). How experiencers are treated syntactically thus becomes a matter of specific lexicalization processes, although languages often possess valency-affecting constructions which allow speakers to modify a verb’s basic alignment pattern (cf. It frightens me > I am frightened of it; I fear it > It is fearful to me: Talmy 1985:99). Spanish, with which the present work is concerned, illustrates the above. In this language, the largest sub-class of emotional verbs (alegrar ‘to make happy’, espantar ‘to frighten’, preocupar ‘to worry’, etc.) has the stimulus argument as subject and the experiencer as direct object (1a), but in usage two alternative constructions show up. The “middle” (Maldonado 1999) variant brings the experiencer into subject position and downgrades the stimulus to oblique status (1b) or eliminates it (1c); the other option, akin to a “dative-subject” type of construction (Melis/Flores 2013, and references therein), features a clause initial experiencer marked with dative case and a postverbal nominative stimulus controlling verbal agreement (1d):

(1) a. La insistencia (NOM) espantó un poco a Felipe (ACC) ‘The insistence frightened Philip a little’ (2004, CORPES XXI)

b. Vio su reflejo en el cristal de una sombrerería y se (MIDDLE) espantó (3SG) de su aspecto (OBL) ‘He saw his reflection in the window of a hat store and was frightened by his appearance’ (2005, CORPES XXI)

c. Eran gritos tan endemoniadamente desgarradores que Jennifer (NOM) se (MIDDLE) espantó (3SG) ‘They were such terribly heartbreaking screams that Jennifer was frightened’ (2005, CORPES XXI)

d. y le (DAT-3SG) espantan (3PL) un poco esos aullidos (NOM PL) ‘and he is a little frightened of those howls’ (2001, CORPES XXI)

From a diachronic perspective, as will be shown, the issue of major interest lies in that the dative alignment pattern, known for its historical decline in other European languages (Haspelmath 2001), was expanded in Spanish, at the cost of the old, well-established, and in some sense less “deviant” middle voice construction. The surprising diachronic shift raises a number of questions, which this corpus-based study seeks to resolve, paying attention to the formal mechanisms that enabled the change and to the semantic criteria that may have operated as motivating forces.

Additionally, the paper will discuss the resulting emergence of a kind of split-S phenomenon, illustrated in (2), where a middle coding contrasts with the dative marking in a “transimpersonal” (Malchukov 2008) type of construction:

(2) a. Me (MIDDLE) alegro (1SG) ‘I am glad.’ (2001, CORPES XXI)

b. Me (DAT-1SG) alegra (3SG) ‘I am glad.’ (2003, CORPES XXI)

REFERENCES Bossong, Georg. 1998. “Le marquage de l’expérient dans les langues d’Europe”. In J. Feuillet (ed.), Actance et valence dans les langues de l’Europe, 259-294. Berlin: Mouton de Gruyter. Haspelmath, Martin. 2001. “Non-canonical marking of core arguments in European languages”. In A. Y. Aikhenvald, R. M., W. Dixon, and M. Onishi (eds.), Non-canonical marking of subjects and objects, 53-83. Amsterdam: John Benjamins. Malchukov, Andrej. 2008. “Split intransitives, experiencer objects, and ‘transimpersonal’ constructions: (re-)establishing the connection”. In M. Donohue and S. Wichmann (eds.), The typology of semantic alignment, 76-100. Oxford: OUP. Maldonado, Ricardo. 1999. A media voz: problemas conceptuales del clítico se en español. México: UNAM. Melis, Chantal, and Marcela Flores. 2013. “On the historical expansion of non-canonically marked ‘subjects’ in Spanish”. In I. Seržant and L. Kulikov (eds.), The diachronic typology of non-canonical subjects, 163-184. Amsterdam: John Benjamins. Talmy, Leonard. 1985. “Lexicalization patterns: semantic structure in lexical forms”. In T. Shopen (ed.), Language typology and syntactic description, vol. 3, 57-149. Cambridge: CUP.

ABSTRACT. Non-finite constructions constitute an important component of the syntactic system of Indo-Aryan (IA) languages. On the one hand, some non-finite verb forms - such as converbs - have already received proper attention in the diachronic and typological literature, particularly with regard to Old Indo-Aryan (OIA), Middle Indo-Aryan (MIA) and New Indo-Aryan (NIA) (cf. Tikkanen 1987; Peterson 1998: Subbarao 2012 among others). Other forms, like participles, have been usually analysed in the wide context of reorganization of finite verbal system which led to alignment change (for recent discussion see Dahl and Stroński 2016). On the other hand, adverbial participles or infinitives have so far been understudied (cf. Sigorski 2005), particularly within Early NIA – the period in the development of IA languages which witnessed several important morphosyntactic evolutions and which still requires in-depth studies, particularly due to lack of well edited corpora. The aim of this presentation is to partly fill this gap by highlighting major trends in the development of constructions based on various non-finite verb forms in Early NIA. It is the first study of this kind based on multilingual annotated corpora comprising four dialectal groups – early Rajasthani, Awadhi, Braj and Pahari, each comprising 10000 words of prose, verse and inscriptional data.

We will focus on the category of 'converb' in early NIA. This category has already been analysed from typological and diachronic perspectives (see e.g. Haspelmath and König 1995) and as such forms the point of departure of our study. In the present paper we are studying the main argument marking in converbal chain constructions and its interplay with the animacy hierarchy. The results of our research clearly show that: - differential object marking primarily shows a semantic motivation, i.e. the marking is based on animacy and definiteness similar to the conditions in Modern IA (Aissen 2003) at quite early stages. However, its final establishment occurred in different periods in the various dialectal groups; - differential subject marking primarily shows a grammatical motivation, i.e. it is dependent on the transitivity of the main verb, but secondary factors such as animacy and definiteness cannot be excluded; - O marking in non-finite verb construction preceded the O marking in finite constructions. We hypothesize that this order of appearance is due to the indifference of converbs to tense/aspect; - the "same subject constraint" was basically preserved but there are examples of its violation which may receive a functional explanation (cf. Tikkanen 1995).

Converbal chains will be compared with so called absolute constructions (i.e. verbal constructions which do not share a subject argument with the main clause) which are either based on adjective or adverbial participles. From MIA we know that both types of construction were in complementary distribution. In Early NIA, however, there appears to be a conflation of the converbal chain construction and the absolute construction with regard to the Subject Identity Constraint. Following Subbarao (2012), we offer a possible explanation for this phenomenon, notably that the functional differentation of the two constructions had chiefly syntactic motivation - converbal chains were predominantly subject oriented.

The facts observed on non-finite constructions are of major importance for a deeper understanding of the interplay between syntax and semantics in NIA languages, for instance, with regard to the gradual decay vs. the reinforcement of ergative alignment.We add data from other dialectal groups in order to shed new light on the intricacies of alignment patterns in their earliest forms in NIA.

References Aissen, Judith. 2003. Differential object marking: Iconicity vs. Economy. Natural Language & Linguistic Theory 21:435–448. Dahl Eystein and Stroński Krzysztof. 2016. Ergativity in Indo-Aryan and beyond. In: Eystein Dahl and Krzysztof Stronski (eds.) Indo-Aryan Ergativity in Typological and Diachronic Perspective. Haspelmath, Martin and Ekehard König (eds.). Converbs in Cross-linguistic perspective. Berlin: Mouton de Gruyter. Khokhlova, Ludmila V. 1992. Trends in the Development of Ergativity in New Indo-Aryan. Osmania Papers in Linguistics 18, 71–89. Khokhlova, Ludmila V. 1995. The Development of Patient-Oriented Constructions in Late Western NIA Languages. Osmania Papers in Linguistics 21, 15–51. Khokhlova, Ludmila V., 2006. Sintaktičeskaja evolucija zapadnych novoindijskich jazykov v 15–20 vv. In: Dybo (et al. eds.) Aspekty komparativistiki. Moskva: Rosijskij Gosudarstvennyj Gumanitarnyj Universitet (Orientalia et Classica: Trudy Instituta Vostočnych Kultur i Antičnosti: Vypusk VIII). 151–186. Peterson John, 1998. Grammatical Relations in Pali and the Emergence of Ergativity in Indo-Aryan. München: Lincom Europa. Sigorskij Aleksandr, 2005. Deepričastija v jazyke xindi. Sbornik naučnych trudov/MGIMO(U) MID Rossii, 21 (36), 54-63. Stroński, Krzysztof. 2014. On Syntax and Semantics of the Past Perfect Participle and Gerundive in Early NIA – Evidence from Eastern Pahari. Folia Linguistica Historica 35, 275-305 Subbarao, Kurumuri V. 2012. South Asian languages: a syntactic typology. Cambridge: Cambridge University Press. Tikkanen Bertil 1987. The Sanskrit gerund: a synchronic, diachronic and typological analysis. Studia Orientalia, 62, Helsinki. Tikkanen, Bertil. 1995. Burushaski converbs in their South and Central Asian areal context. In M. Haspelmath & E. König (eds.). Converrbs in Cross-linguistic perspective. Berlin: Mouton de Gruyter. 487-528. Verbeke Saartje 2013. Ergativity and alignment in New Indo-Aryan languages. Berlin-New York: Mouton de Gruyter.

ABSTRACT. 1. Introduction. This talk offers a new diachronic split-CP account of the actuation of Locative Inversion (LI) in Early Modern English producing non-canonical XVS, e.g. here comes Johnny. I argue that LI stems from the reanalysis of remnants of the Middle English verb-second (V2) constraint as A-movement in which formal [±loc(ative)] features were innovated on Fin°(itness) in the C-layer. This reanalysis in English is perhaps unsurprising when one considers that English is a so-called ‘strong satellite-framed language’ (Acedo-Matellán 2016) in which result states and locations are expressed not by verbs but by other independent elements. Moreover, this paper looks to make tentative comparison between the outcomes of the loss of V2 in Old Spanish and Old French and modern XVS inversion mirroring but crucially differing from English LI. Present Day English (PDE) is SVO without V2; lexical verbs cannot move to T, a requirement for V2 languages where finite verbs then move to C (Holmberg 2015). The loss of historical V2 in English can be attributed to a parameter cascade (Biberauer & Roberts 2008) in which the loss of one parameter led to knock-on reanalyses others. A causational cline can be assumed: new restrictions on fronting > the loss of V-to-T-to-C movement, i.e. generalized V2; > the loss of V-to-T movement > blocked residual microparametric V2 of lexical verbs, i.e. no V2. This helps explain the sparsity of XVS structures in PDE, but does not hold for Romance varieties which have retained V-to-T. In sum, PDE generally prevents XVS (1a) orders in clauses containing only a lexical verb. SV(O)X (1b) is the most canonical order, but topicalisation can produce XSV(O) (1c). (1) a. *the letter wrote John on Thursday b. [CP [TP John … [vP wrote … [DP the letter [PP on Thursday]]]] c. [CP [PP on Thursday][TP John [vP wrote [DP the letter…]]] PDE is, however, a residual V2 language (Rizzi 1996:64) containing ‘…construction-specific manifestations of T-to-C movement […] which does not generalize the V2 order to main declarative clauses’. Such cases are limited to microparametric raising of modals/auxiliaries from T-to-C in WH-questions (2) and Negative Inversion, eg. never did I say such a thing. (2) [CP who[C have [TP you T [have [vP [v met [VP [V meet [PRN who] 2. Locative Inversion and V2. Locative Inversion (LI) in Present Day English (3a) is an anomalous XVS construction. In LI a spatially deictic complement may raise above an unaccusative verb leaving the subject in a late position. This is only possible with a lone lexical verb and a full DP subject; pronominal subjects are ungrammatical in this distribution (3b) and must raise above the verb (3c). (3) a. Down the hill tumbled Jack. b. * down the hill tumbled he c. Down the hill he tumbled LI superficially resembles V2 constructions including lexical verb V-to-C movement. However, Rizzi & Shlonsky (2006) show otherwise. Assuming a Split-CP ([ForceP [TopP [FocP [TopP* [FinP…) (Rizzi 1997) and roughly adopting the cartographic account of Rizzi & Shlonsky (2006), I assume A-movement of the locative complement bearing both locative and topicalisation features [i:LOC] [i:TOP] through FinP to check [u:LOC] on Fin° then followed by raising to TopP to check [u:TOP] on Top°. Crucially, Fin is nominal (Rizzi & Shlonsky 2006) meaning that verbal constituents are blocked from passing through it. This combination can satisfy the EPP on T as the PP passes through Spec-T leaving a low in-situ subject in Spec-VP (4).

(4) [ForceP[TopP[PPdown the hill[Top{u:TOP}[FinP PP[Fin {u:LOC}][TP PP [T[vP PP[vtumbled[VP[DP Jack] [V [PP {i:LOC, i:TOP}]]]]… Yet in Middle English this surface order was straightforward V2 since all lexical verbs moved to C (Fin) (5). Thus Fin in V2 systems is a ‘verbal Fin’ allowing T-to-C, indicating the innovation of ‘nominal Fin’ in English. Moreover, Middle English V2 allows the fronting of many elements; adjuncts, PP arguments, quantified objects etc.; fronting was dependent on the syntax-discourse interface, but there is no diachronic evidence to suppose [±LOC] features for V2 English. In (5) the fronting of the PP is simply due to its topic feature. (5) [ForceP[TopP[PP of þese seuene heuedes [Top {u:TOP}[FinP PP[Fin comen{+V}][TP PP [Tcomen[vP PP[vcomen[VP[DP alle manere of synnes] [Vcomen [PP {i:TOP}]]]]… ‘from these seven heads come all manner of sins’ (Middle English) An interesting point of comparison regarding Romance is found in the findings of Corr (2016), who shows how wide-focus in Ibero-Romancevarieties can be analysed as LI, but topicalisation is not involved as in English. Moreover, unlike English, some of these varieties contain LI with temporally deictic elements. Drawing on Siggurðsson’s (2010) articulated analysis of FinP containing both temporal and locative heads relating to the discourse, we can conclude that English only lexicalised the latter in movement contexts, while other languages might do both. I argue that frequent historical V2 [frame-setter Temp-adverb [ PN Subject [VFIN] orders can be considered evidence for why only spatial deictic elements, i.e. locatives developed such inversion in English. 3. Contribution. I offer a novel diachronic account of the emergence of the feature constellation present in English LI from historical V2 to its current composition. I show that this reanalysis appears to be a radical acquisitional response of last resort when children were faced with uninterpretable target values as V2 collapsed. For LI, the V2 parameter disintegrated gradually leaving remnants which became opaque to the acquirer. The child was faced with two options: dump the structure or find any possible analysis allowed by the system. They did the latter positing [±LOC] on the perceived deictic properties of certain utterances, a type of Exaptation. Unaccusatives were especially productive for this reanalysis due to their late subjects and correlation with verbs of motion, facilitating XVS.The findings show that spatial deixis acted as a mediator for opaque PLD because children were able to ‘go beyond the input’ (Biberauer 2016). This shows the diachronic effects of what Ramchand & Svenonius (2014:185) call a ‘cognitive proclivity to perceive experience in terms of events, situations, and propositions (with analogous ontologies for other extended projections)’. In this case children drew on the strategy that in English satellite elements consistently mark such relationships (Acedo-Matellán 2016), facilitating reanalysis. I conclude that, in the right circumstances, deixis is a Chomsky(an) 3rd (2005) factor capable of driving language change. Selected References: Acedo-Matellán, V. 2016. The morphosyntax of Transitions. OUP. ➤ Biberauer, T. 2016. Going beyond the input (and UG): an emergentist generative perspective on syntactic variation, stability and change. Presented at KNAW: Language Variation in Action, Amsterdam. ➤ Corr, A. 2016. Wide-focus subject-verb inversion in Ibero-Romance: a locative account. Glossa 1(1):11.1-33. ➤Holmberg, Anders. 2015. Verb second. In T Kiss & A Alexiadou (eds.), Syntax: Theory and Analysis,242–283. de Gruyter. ➤ Ramchand, G & P Svenonius. 2014. Deriving the functional hierarchy, Language Sciences 46b,152-174. ➤ Rizzi, L. 1996. Residual verb second and the Wh-Criterion. In A Belletti & L Rizzi (eds.) Parameters and Functional Heads: Essays in Comparative Syntax, 63-90. OUP. ➤ Rizzi, L & U Shlonksy. 2006. Satisfying the subject criterion by a non-subject: English locative inversion and heavy NP shift. In M Frascarelli (ed.), Phases of interpretation, 341-361. De Gruyter

ABSTRACT. In this study, I apply principles of language contact theory to contact between related Arabic dialects in order to propose a new account for the language’s so-called “pluriform developments.” These innovations, whereby “a general trend … has occurred in all Arabic dialects, and an individual translation of this trend in each area” (Versteegh, 2001:108), have long occupied students of historical Arabic dialectology. In such cases, developments are encountered in typologically analogous yet etymologically diverse realizations across the modern Arabophone world. Their diversity of form has generally been seen as precluding common inheritance as a source for their similarity, and mutual influence via dialect contact has likewise been commonly written off due to a belief that “typically dialect contact leads to the borrowing of another dialect’s markers [i.e., surface forms], not to the borrowing of a structure which is then filled independently” (Versteegh, 2001:108). Though alternative proposals have been circulated, none has found widespread acceptance.

I propose that contact between Arabic dialects does in fact underlie the occurrence of these forms, via the mechanism of contact-induced grammaticalization (Heine & Kuteva, 2003; 2005). Previous analyses of the Arabic phenomenon have generally defined linguistic contact processes as consisting of the transfer of linguistic “matter,” and lag behind current understandings in the broader field which recognize the possibility of convergence in “pattern” alone (cf. Matras, 2009). When attention is instead paid not the forms of the innovations in question but to the grammaticalization pathways underlying their evolution, the apparently chaotic dialect data is transformed: numerous distinct etymologies consolidate into a considerably smaller number of cross-linguistically attested grammaticalization paths, which in turn cluster in a series of contiguous geographic blocs. The pluriform development of novel future tense markers provides an example. Descendents of *ɣādī, *māʃī, *sājir, and *rājiħ, all meaning ‘going’, represent grammaticalizations of de-allative future tense markers clustered in the north and west of the Arab world, while those of *biddu, *jibbī, * jiʃā, and *jidūr ‘want’ attest de-volitive futures in the south and east. While the sources and development of such forms have been treated individually (e.g., Stewart, 1998), their geographic distribution and potential systematic linkage to one another have not previously been discussed on a comprehensive scale capable of assessing widespread mutual influence.

When such factors are considered, the result strongly resembles the outcome of what Heine and Kuteva (2003, 2005) label contact-induced replica grammaticalization, in which a grammaticalization process active in one language is replicated by speakers of another with which it is in contact, where it is then independently advanced using native etymological material. In my study, I compared data for four morphsyntactic operators typically identified in the discussion of pluriform developments (future tense markers, genitive exponents, nonverbal predicate negators, and the temporal adverb ‘now’), drawn from a geographically comprehensive sample of 76 published dialect descriptions. In each case, I submitted the data to a three-part heuristic, developed with reference to Heine and Kuteva’s original theorization and informed by the critiques/refinements of Ross (2007) and Law (2014). First, etymologies were identified for all attested forms of a given feature represented in the sample, and these were evaluated as viable examples of grammaticalization processes with reference the cross-linguistic and theoretical literature. Second, grammaticalization processes so identified were assigned, when possible, to more general grammaticalization pathways comprising the multiple products of synonymous but distinct evolutions plausibly linked by processes of replication (e.g., de-volitive futures, ‘now’ from multimorphemic constructions meaning ‘this time’, etc.), in a manner similar to the concept of “gram families” as proposed by Dahl (2001). In the third step, the geographic incidence of each grammaticalization pathway’s diverse instantiations was mapped and assessed as consistent or inconsistent with a history of areal diffusion via contact.

My results, presented in the form of isogloss maps, conclude that the conditions of this heuristic are met by the majority of forms of all four features examined; these thus represent the products of geographically contiguous, analogous grammaticalization paths whose decidedly non-random distribution is most likely the result of replication stemming from dialect contact. Contact-induced grammaticalization is thereby advanced as a viable explanatory mechanism which accounts for the pluriform development data in a manner previous proposals have failed to match. This finding also provides a valuable empirical corroboration of the contact-induced grammaticalization paradigm in a scenario of contact between related varieties: previous authors have largely avoided analyses of such contexts for the sake of methodological clarity, but at the same time assert that they are critical for a complete understanding of the phenomenon (Heine & Kuteva, 2005; Dahl, 2001). This study thus contributes to general conversations in language contact theory while simultaneously increasing recognition of contact as a continual force in the rich and varied history of the Arabic language.

References

Dahl, Ö. (2001). Principles of areal typology. In M. Haspelmath, E. König, W. Oesterreicher and W. Raible (Eds.), Language typology and language universals: An international handbook (pp. 1456-1470). Berlin, New York: Walter de Gruyter.

Heine, B. and T. Kuteva. (2003). On contact-induced grammaticalization. Studies in Language 27(3), pp. 529-572.

Heine, B. and T. Kuteva. (2005). Language contact and grammatical change. New York: Cambridge University Press.

Law, D. (2014). Language contact, inherited similarity and social difference: The story of linguistic interaction in the Maya Lowlands. Amsterdam, Philadelphia: John Benjamins.

Matras, Y. (2009). Language contact. Cambridge: Cambridge University Press.

Ross, Malcom D. (2007). Calquing and metatypy. Journal of Language Contact 1(1), pp. 116-143.

Stewart, D. (1998). Clitic reduction in the formation of modal prefixes in the post-classical Arabic dialects and Classical Arabic sa-/sawfa. Arabica 45, pp. 104-128.

Versteegh, K. (2001). The Arabic Language. Edinburgh: Edinburgh University Press.

ABSTRACT. In the history of Arabic varieties, a prominent phonological process that has been observed as early as the 8th century is one of palatalisation and subsequent affrication of dorsal stops. In many contemporary Arabic dialects this process has reached saturation such that it is nowadays showing signs of an apparent reversal. All the while, a parallel process of palatalisation of alveolar stops is occurring in a number of dialects, and in some cases, affricates are shifting towards simple fricatives. This paper analyses data from a number of such dialects from the Levant and the Arabian Peninsula and discusses them from a socio-historical linguistic perspective. We argue that while processes of palatalisation are phonetically motivated, their reversal, where applicable, appears to stem from social factors revolving around contact among neighbouring dialects and regional koinés. In Labovian terms, we are currently witnessing a change from above in linguistic variables that had previously undergone changes from below (the level of speakers’ consciousness).

Historically, the two main consonants that are candidates for palatalisation are the velar and uvular stops /k/ and /q/, respectively. The latter is somewhat more complex, as in many dialects /q/ has shifted to [ɡ], whereas proto-Arabic */ɡ/ has by-and-large shifted to some form of coronal affricate or fricative (or, in extreme cases, a glide). Conversely, the phoneme labelled /q/ is often realised as a glottal stop in urban varieties of the Levant and North Africa, while some non-Urban Levantine dialects realise it as [k] or [kˤ].

Following are a number of shifts that have occurred in a number of speech communities in the Middle East:

Rural Palestinian Arabic (e.g., Umm al-Fahm): q> kˤ k>ʧ [in the environment of high front vowels] Bethlehem Palestinian Arabic: q>k k>ʧ [unconditionally] This constitutes a minimal chain shift.

Najdi Arabic—Shagra, Gasim (central Saudi Arabia): k>ʦ *q>ɡ>ʣ

Elsewhere in Najd: k>ʧ *q>ɡ>ʤ

Bani Hasan (central Jordan): *q>ɡ>ʤ [very restricted lexically, in a handful of cases where /q/ precedes a front vowel]

Sūf (northern Jordan): *q>ɡ k>ʧ [conditionally]

In a number of other dialects (e.g., Cairo, Damascus, some Moroccan dialects), we witness the palatalisation of coronals rather than dorsals, resulting in such variable shifts as: t>tʲ~ʧ; d>dʲ~ʤ This shift occasionally extends to the ‘emphatic’ /dˤ/ and /tˤ/, where palatalisation leads to de-emphaticisation as well.

At the same time, several dialects seem to undergo a seemingly reverse process of de-palatalisation or de-affrication, such as the following:

Rural Palestinian Arabic (e.g., Umm al-Fahm): ʧ>k Various Jordanian dialects – Horan and central Jordan (e.g, Amman, Sūf, Sult): ʧ>k Najdi speakers in the Hijazi city of Jeddah (western Saudi Arabia): ʦ>k; ʣ>ɡ Elsewhere in Saudi Arabic (various provinces): ʦ~ʧ>k; ʣ~ʤ>ɡ

Affrication may be construed as a type of lenition, which originated in palatalisation, either as a result of phonetic conditioning, such as an adjacent high front vowel, or as an unconditioned shift. In this respect, we can view it as a natural process in the spirit of the Principle of Least Effort I (“effort reduction in speech is restricted by the need to satisfy one’s addressees’ need to understand”), occasionally encroaching on the addressees’ comprehension as in the second iteration of this principle by Labov (2001).

The reversal of affrication, however, is much less intuitive, and does not appear to follow any sort of phonetic motivation. In all cases where this reversal occurs, there are strong social forces that lead to the unaffricated outcome. Affricates are typically stigmatised insofar as they indicate that a speaker hails from a certain geographical region or is part of a social group other than the dominant, urban, educated group in their greater environs. Replacing an affricate with the stop from which it had originally shifted is therefore just that—an abrupt replacement, not a phonetic approximation of a specific target.

When considering a sound change such as k>ʧ, it must be assumed that several intermediate variants had existed (and indeed some dialects have variants such as [kʲ]. We may thus formulate a hypothesis that k>ʧ actually represents a lengthy historical process of the type: k>kʲ>tʲ>ʧ (kʲ & tʲ being mere hypothetical, yet phonetically logical examples). The sudden reversal of, e.g., [ʧ] to [k], seems to have skipped these intermediate variants. Social factors, in particular the pressure to sound less local and more like the surrounding (usually urban) koiné, break the phonetic constraints otherwise in place. Unlike the original process of affrication, which was a change from below, the reversal thereof is a change from above, the target stop being readily available in other, more socially dominant dialects to which speakers have access as a result of close contact.

The data analysed in this paper, which draw upon detailed studies of a number of dialects (e.g., Al-Wer et al. 2015 and Al-Hawamdeh 2016 for Jordan, Alessa 2008 for Jeddah, Haeri 1994 for Cairo) offer new insights into the reversal of a historical process, for which the descriptive labels “de-palatalisation” or “de-affrication” may be suitable. We argue that rather than the phonetic processes these labels apply, what we witness is a more complex scenario, whereby dominant dialects coming in contact with more localised varieties exert social pressure such that the latter conform to the former despite the lack of linguistic internal motivation to do so. Rather than arguing that a historical phonological process is reversed, per se, we postulate that it is ongoing a new change, the results of which coincide with a very old phonological reality. The role of contact, not only between Arabic and neighbouring or substrate languages, but also among Arabic varieties, is crucial to the understanding of this and similar processes.

ABSTRACT. The earliest available account of the Arabic dialect spoken in Gaza City (henceforth GCA) dates to the period of the First World War (Bergsträßer 1915). Later texts published by Salonen (1979, 1980) painted a complex, and potentially unreliable (see de Jong 2000) view of the city’s dialect. The result is that despite being an important urban center in the Arabic speaking world, scholars know little about Gaza City from a linguistic perspective. This paper draws on data collected during fieldwork in Gaza City in 2013 (15 speakers) and with Gaza City refugees in Jordan in 2015 (8 speakers). This paper investigates how contact between different dialects of Arabic have shaped one aspect of GCA, the realization of the Arabic feminine gender marker /a/: a word-final vocalic morpheme that shows considerable variation across Arabic dialects.

Arabic dialectologists have generally considered GCA to be a dialect that lacks raising of this morpheme, realizing it as [a]. This apparent lack of raising positions GCA as unique given that raising of /a/ to [e], except after back consonants, is a prominent feature of urban Palestinian dialects (Al-Wer 2007, Shahin 2007). However, acoustic analysis of the data collected in 2013 and 2015 conducted in Praat (Boersma & Weenink 2016) suggests that the realization of /a/ in GCA may have been higher than the [a] that has been treated as a given by Arabic scholars, but that dialect contact in both Gaza City and among refugees in Jordan has resulted in substantial change to their dialect of Arabic over successive generations.

In examining the speech of elderly speakers the vowel is raised notably higher than [a], closer to [æ] or [ɛ]. However, it is not consistently realized as high as the [e] typical of other urban Palestinian varieties. This raising occurs across both those elderly indigenous Gaza City residents who remained in Gaza City and their elderly counterparts in Jordan. Additionally, in both communities there is an identical processes of lowering and backing for this vowel across age generations, moving its realization in the direction of [a]. Within Gaza City this reflects that dialect contact happening in the Gaza Strip between the indigenous communities and the large numbers of Palestinian refugees that have entered the territory after the creation of Israel in 1948 has had an effect. Previous work (Author X 2015) has also suggested that Palestinian refugees in Gaza City are undergoing a similar process of change, suggesting the potential leveling of /a/ in Gaza City (Al-Wer 2003, Trudgill 1986). However, for Gaza City refugees in Jordan this trajectory of change challenges the expected dialect contact narrative (see Trudgill 1986, 2004), which would suggest convergence towards the local Jordanian Arabic realization of this vowel, [ɛ] (Al-Wer et al 2015, Herin 2014).

The results of this study suggest two things. First, despite a history within Arabic linguistics of treating GCA as an anomaly with respect to its purported realization of /a/ as [a], GCA may have realized the vowel closer to what one expects to find for /a/ in other urban Palestinian dialects. Second, a long history of refugee migration and dialect contact in Gaza City has had a discernable effect on the realization of /a/, resulting in a process of change that moves the realization of this vowel closer [a]. Even outside of Gaza City proper, Palestinian refugees living in Jordan, but whose roots are in Gaza City, show a similar process of change.

These findings suggest that across the Palestinian community, a long history of forced migration and dialect contact is resulting in the leveling (Trudgill 1986, 2004) of certain features of the dialect, even across wide geographic areas. This suggests that the complex sociocultural history of the Palestinian community, one marked by decades of political and social upheaval, is contributing to linguistic change at a number of levels within Palestinian Arabic varieties (see Author X 2015; Horesh 2014, 2015 for additional cases of change). Beyond this, the results obtained from Gaza City refugees and its challenge to the dominant dialect contact narrative suggests that in moving forward with work on Palestinian Arabic, scholars should more closely consider spatial constructs like the refugee camp and their role in structuring daily life as a potential influence on the directionality of language change.

References

Al-Wer, Enam. 2003. New dialect formation: The focusing of –kum in Amman. In D. Britain & J. Chesire (Eds.), Social Dialectology: In honour of Peter Trudgill. Amsterdam: John Benjamins. 59-67. Al-Wer, Enam. 2007. The formation of the dialect of Amman: From chaos to order. In Catherine Miller, Enam Al-Wer, Dominique Caubet and Janet Watson (eds.) Arabic in the City. New York: Routledge. 55–76. Al-Wer, Enam, Uri Horesh, Bruno Herin, & Maria Fanis. 2015. How Arabic Regional Features Become Sectarian Features: Jordan as a Case Study. Zeitschrift für Arabische Linguistik 62: 68-87. Author X. 2015. Bergsträßer, Gotthelf. 1915. Sprachatlas von Syrien und Palästina. Leipzig, Germany: Hinrichs. Boersma, Paul & Weenink, David (2016). Praat: doing phonetics by computer [Computer program]. Version 6.0.18, retrieved 23 May 2016 from http://www.praat.org/ De Jong, Rudolf E. 2000. A Grammar of the Bedouin Dialects of the Northern Sinai Littoral: Bridging the Linguistic Gap between the Eastern and Western Arab World. Leiden, The Netherlands: Brill. Horesh, Uri. 2000. Toward a phonemic and phonetic assessment of Jaffa Arabic: Is it a typical urban Palestinian dialect? In Manwel Mifsud (ed.) Proceedings of the Third International Conference of AIDA: Association Internationale de Dialectologie Arabe. Paris, France: AIDA / Malta: Salesian Press. 15–20. Horesh, Uri. 2014. Phonological outcomes of language contact in the Palestinian Arabic dialect of Jaffa. Ph.D. Dissertation, Colchester, U.K.: University of Essex. Horesh, Uri. 2015. Structural change in Urban Palestinian Arabic induced by contact with Modern Hebrew. In: A. Butts (Ed.), Semitic Languages in Contact. Leiden: Brill. 198- 233. Rosenhouse, Judith. 2007. Jerusalem Arabic. In: Kees Versteegh (ed.) Encyclopedia of Arabic Language and Linguistics. Leiden, The Netherlands: Brill. 481–493.

Salonen, Erkki. 1979. Zum arabischen Dialekt von Gaza, Teil I. Helsinki, Finland: Finnish Oriental Society.

Salonen, Erkki. 1980. Zum arabischen Dialekt von Gaza, Teil II. Helsinki, Finland: Suomalainen Tiedeakatemia.

Shahin, Kimary. 2007. Palestinian Arabic. In Kees Versteegh (ed.) Encyclopedia of Arabic Language and Linguistics. Leiden, The Netherlands: Brill. 526–538. Trudgill, Peter. 1986. Dialects in Contact. New York: Basil Blackwell

Trudgill, Peter. 2004. New-dialect Formation: The Inevitability of Colonial Englishes. Oxford: Oxford University Press.

ABSTRACT. This paper presents a usage-based, Construction Grammar perspective on linguistic change and argues that this, in combination with findings from social network/communities of practice analysis, allows us to model the details of language change on the atomic micro-level of individual speakers and hearers, and to connect these findings in a systematic way to the macro-level of the speech community.

Usage-based Construction Grammar (Cxg) assumes that language(s) comprise conventional and prototypically non-compositional from-meaning pairings at levels of linguistic structure. These are organized in a so-called mental constructicon. Linguistic knowledge is knowledge of constructions and the constructicon. The acquisition and use of constructions in the individual speaker are strictly usage-based, i.e. they are dynamic across a life time in actual acts of language use, under the influence of frequency effects, rich contextual input, and saliency (cf. Tomasello 2006; Bybee 2010; Zima & Bergs).

Modelling language change in such a framework involves a number of different mechanisms, many of which are still not fully understood. On the one hand, we find traditional factors (hearer-based) such as imperfect replication (cf. Croft 2000). But CxG also allows for other mechanisms, such as coercion and snow-cloning, which are mostly speaker-based and usually involve deliberate manipulation of linguistic material at the atomic, micro-level.

Coercion is the systematic creation and resolution of mismatch (cf. Lauwers & Willems 2011; Bergs 2017). Constructions and constructs can be matched (‘unified’) to form larger units. This matching is subject to certain constraints, such as morphosyntactic congruence. Yet, speakers can create new and innovative utterances such as (1) and (2).

(1) she tried to eat her way out of her clothes, (COCA, Autumn 2012, Vol. 25 Issue 3, p466-493. 28p. Gettysburg Review) (2) Vicari bribed himself off Shosha. (COCA, 2008, Stealing Adriana)

Example (1) shows the monotransitive verb eat as complemented by an abstract NP her way and the AdvP out of her clothes. The result is a novel, non-compositional utterance intended to mean something like ‘She deliberately ate in such a way that her clothes would not fit anymore’. Similarly, in (2) the verb bribe usually needs a simple nominal object (the recipient of the bribe), but here it is combined both with a reflexive and an AdvP of direction. Again, the resulting meaning is non-compositional ‘Vicari bribed other people in such a way that he could leave Shosha’. Note that coercion is not only part and parcel of spectacular, note-worthy utterances such as (1) and (2), but that it also can be seen in more mundane contexts, in the form of metaphor and metonymy, as in (3) and (4).

(3) He drank the bottle of Vinho Verde. (COCA, 2005, The Folly of Being Comforted) (4) Two empty hours were a sinus in which infections bred. (Jonathan Franzen, The Corrections, 2001, p.1 )

Example (3) is a case of coercion as drink usually requires some liquid that can be drunk as complement. (3), in contrast, is a textbook case of metonymy, where container stands for contained. (4) has hours (which cannot be empty) that cannot be a sinus. This is a fairly conventional example of a metaphor, which, like (3) can be elegantly modeled as coercion (cf. Gonzálvez-García 2011). Coercion can thus be characterized as a very widespread and powerful type of speaker innovation.

A second type of speaker innovation involves the deliberate violation of linguistic rules and constraints not unlike coercion, and yet sufficiently different to warrant an independent category. Examples (5) - (8) illustrate this second category.

(5) But motivation alone does not assure success: Because circumstances. (COCA 1996, International gifted women: Developing a critical human resource.) (6) Fill us in briefly, if you would. (COCA 215, PBS, Traugott 2017) (7) A fun thing to talk about. (COHA, 1972, Harpers, Wayne Rindels) (8) It was not safe to drive after dark, let alone walk. (COCA 2014, Mother Jones, Nick) Baumann

In (5) the conjunction because is complemented by a bare NP, in (6) we see a case of insubordination (Traugott 2017), in (7) thing is modified by the noun/adjective fun (Denison 2013) and let alone in (8) is the textbook example for a traditionally ‘extragrammatical’ idiom (Fillmore et al. 1988). These illustrate the deliberate manipulation of grammatical rules; the motivations are similar to those for coercion (see below). Yet, this second type seems to be rarer compared to the almost ubiquitous process of coercion. The impact of this process on the macro-system of the speech community can be equally massive, though.

A third type of speaker innovations involves snow-cloning, i.e. the exploitation of linguistic cliché templates, such as San Antonio is the Paris of Texas (X is the N of Y; Pullum 2004). Speakers not only repeat the original cliché (create more tokens), but deliberately add new tokens until a more abstract type emerges, which then in turn allows for the creation of even more new tokens, including those that break the original rules of the type itself and thus may ultimately lead to new, even competing types in the constructicon. This third type of innovation is also comparatively rare, but it is particularly helpful in illustrating the actual mechanism of diffusion in the community and in discussing the motivations for change on both the micro- and macro-level from a socio-psychological point of view.

This paper argues that all three mechanisms illustrate micro-level, atomic speaker innovations that can trigger linguistic change in the speech community. The innovative act itself is a manipulation of existing constructions and constructs performed by the speaker. Important socio-psychological motivations for such manipulative acts are discussed by Keller (e.g., “Talk in such a way that you are noticed”, Keller 1994) and Haspelmath (“extravagance”, Haspelmath 1999).

The novel viewpoint is that CxG is a framework that allows for the elegant modelling of acts of speaker innovation. As a usage-based model, it also helps us to trace the propagation of such innovations. In contrast to traditional generative models, usage-based CxG regards the constructicon as being dynamic, as in constant flux. New constructions come in, old ones lose some of their activation potential. This not only happens in first language acquisition, but across a lifetime. The development of new constructions is driven by frequency effects, salience, and socio-psychological factors. It is couched in a rich contextual context. In other words, the success or failure of an innovation can now be identified in a social network or a community of practice. Construction Grammar in combination with those micro-scale sociolinguistic tools of analysis helps us to trace the innovation and spread of constructions, from their innovation as atoms on the speaker/micro-level to the community/macro-level.

References

Bergs, Alexander. 2017/in press. “Under pressure: Norms, Rules, and Coercion in Linguistic Anaylses and Literary Readings”. In: Michael Burke & Emily Troscianko (eds.), Cognitive Literary Science. Oxford: Oxford University Press, 279-302. Bybee, Joan. 2010. Language, Usage, and Cognition. Cambridge: Cambridge University Press. Croft, William. 2000. Explaining language change: an evolutionary approach. Harlow, Essex: Longman. Denison, David. 2013. „Parts of speech: Solid citizens or slippery customers?” Journal of the British Academy, 1, 151–185. http://www.britac.ac.uk/sites/default/files/JBA-001-151-Denison.pdf (accessed 26 Nov 2016) Fillmore, Charles J., Paul Kay & Mary Catherine O’Connor. 1988. “Regularity and idiomaticity in grammatical constructions: The case of Let alone.” Language 64/3: 501-538. Gonzálvez-García, Francisco. 2011. “Metaphor and metonymy do not render coercion superfluous: Evidence from the subjective-transitive construction”. Linguistics 49.6: 1305–1358 Haspelmath, Martin. 1999. “Why is grammaticalization irreversible?”. Linguistics 37–6, 1043–1068 Keller, Rudi. 1994. On Language Change: The Invisible Hand in Language. London: Routledge. Lauwers, Peter and Dominique Willems. 2011 (eds.). New Refelections on Coercion. Special Issue of Linguistics 49. Pullum, Geoffrey. 2004. „Snowclones: Lexicographical Datin to the Second.” Language Log http://itre.cis.upenn.edu/~myl/languagelog/archives/000350.html (accessed 26 Nov 2016) Tomasello, Michael. 2006. “Construction Grammar for Kids”. Constructions, Special Issue ed. by Doris Schönefeld https://www.constructions.uni-osnabrueck.de/wp-content/uploads/2014/06/2006-SI-Tomasello-26-84-1-PB.pdf (accessed 26 Nov 2016) Traugott, Elisabeth. 2017/to appear. “Insubordination in the light of the Uniformitarian Principle”. Special issue of English Language and Linguistics “Cognitive Approaches to the History of English”, ed. by Alexander Bergs and Thomas Hoffmann. Zima, Elisabeth & Alexander Bergs. 2016/in production. “Multimodal Construction Grammar”. Linguistics Vanguard, Special Issue “Towards a Multimodal Construction Grammar”, ed. by Elisabeth Zima. https://www.degruyter.com/view/j/lingvan (accessed 26 Nov 2016)

ABSTRACT. It is now generally recognized that language contact is a major force shaping grammar. As Drinka points out, ‘Language contact, rather than representing some sort of deflection from the norm or peripheral chance occurrence, renders to us some of our most essential insights as to how language change occurs’ (2012: 244). It is also now understood that ‘internal’ and ‘external’ processes of language change are usually intertwined; in the words of Heine & Kuteva, ‘language-internal change and language contact are not mutually exclusive; rather, they may work in conspiracy with each other’ (2010: 87). The mounting body of work on contact-induced grammatical change is providing new opportunities for investigation of principles underlying these interactions, at the outset and through subsequent lines of development.

            Copied grammatical patterns are rarely perfect replicas of their models. So what tends to be transferred in the early stages, and to what extent are subsequent directions of development constrained? Johanson (2008, 2013) points out that semantic, combinational, and frequential properties of grammatical markers and patterns may be copied together or separately, without their substance. Drinka, Johanson, Heine & Kuteva, and others concur that copies are often less frequent and less grammaticalized than their models, but not the reverse. Heine & Kuteva propose that ‘contact-induced innovations tend to be confined to an increase in frequency of use and extension of existing pieces of discourse structure to new contexts, some of which are associated with novel meanings . . . Some form of consistency is attained when, as a result of language contact, clusters of discourse pieces turn into new use patterns.’ (2010: 89)           

            We hypothesize contact effects in similarities across language boundaries when we can rule out common inheritance. We see product of contact, but in the absence of a rich philological record, not necessarily the process behind it. If the similarities extend across multiple languages in an area, however, what has been called by Heine & Kuteva a grammaticalization area, we may gain some insight into the varieties of ways that a particular grammatical feature or structure may enter a language, and the kinds of factors that may affect its subsequent development.

            A set of clause-combining constructions sometimes referred to as “switch reference” (Jacobsen 1967, 1983) offer a rich opportunity for the investigation of these questions. In these constructions, dependent clauses carry markers specifying their relation to the matrix clause. The terminology originated in the idea that the markers distinguish coreference or disjoint reference between the subject of the dependent clause and the matrix, though closer examination of spontaneous speech has shown that in many of these systems, subject reference is just one factor in marking events as tightly integrated or more loosely associated. What is interesting is that such systems show strong areality: they are particularly common in certain areas of North America, Australia, and Papua New Guinea, for example. In his survey of switch-reference (SR) in North America, for example, McKenzie (2015: 423) reports:

Concerning the distribution of SR languages in North America, they are almost all in the West, in a mostly contiguous geographic area. Also, many language families only have one or two members that have SR; these members are in that contiguous area. These facts support previous observations that areal diffusion has played a role in its cross-linguistic distribution.

            Among the languages of the West cited by McKenzie are those of the Pomoan family, indigenous to an area north of San Francisco. The Pomoan languages all contain highly-developed systems of clause-combining, which distinguish dependent clauses portrayed as elements, with their matrix clause, of a single event (same) from those portrayed as separate but related events (different). Like many similar systems elsewhere in the world, they make further simultaneous/sequential and realis/irrealis distinctions. Markers indicating event cohesion are verb suffixes, while those indicating a looser connection are clause enclitics. The system can be reconstructed for their common parent, Proto-Pomoan, whose time depth has been estimated to be comparable to that of Germanic. The markers occur in a secondary development, the formation of sentence-initial adverbs based on a verb ‘be/do’ followed by a suffix or enclitic, which indicate discourse relations to a previous sentence.

            The Pomoan languages are surrounded on three sides by five unrelated languages from three, unrelated families: Yuki and Wappo (long separated from each other geographically but remotely related in the Yukian family), Nisenan and Patwin (Wintun family), and Western Miwok (Utian family). The languages share a number of clearly contact-stimulated grammatical innovations, but show little lexical borrowing. None of them has a well-developed clause-combining system like those of their Pomoan neighbors, but very early effects of contact with the Pomoan systems can be seen in each of them. These effects differ across the languages in interesting ways, even across related languages. As predicted, the most salient effect of contact is the frequency of specification of relations among sentences. Each language has exploited a different native resource to mark the relations, but all have apparently taken as their model apparently more analytic markers, the sentence-initial adverbs, as their models, rather than the more deeply embedded suffixes and clitics. They vary in the number of distinctions made in the system, but none distinguish as many features as their Pomoan models. They vary in the degree to which their markers have become grammaticalized, but none are as advanced as their models.

            Taken together, the incipient systems show that an initial step in contact-stimulated grammatical change can indeed be heightened frequency, but that beyond that, there is not a single pathway along which particular grammatical systems must develop.

References

Drinka, Bridget 2012. The Balkan perfects: Grammaticalization and contact. in Bjorn Wiemer, Bernhard Wälchli, and Björn Hansen (eds).). Grammatical replication and borrowability in language contact. The Hague: Mouton. 511-558.
Heine, Bernd & Tania Kuteva. 2010. Contact and grammaticalization. In Raymond Hickey (ed.)  The Handbook of Language Contact, 86-105. Oxford: Blackwell.
Jacobsen, William. 1967. Switch-reference in Hokan-Coahuiltecan. In Dell Hymes & William Bittle (eds.), Studies in Southwestern ethnolinguistics, 238-63. The Hague: Mouton.
Jacobsen, William. 1983. Typological and genetic notes on switch-reference systems in North American Indian languages. In John Haiman & Pamela Munro (eds.), Switch reference and universal grammar, 151-83. Amsterdam: John Benjamins.
Johanson, Lars. 2008. Remodeling grammar: Copying, conventionalization, grammaticalization. In Peter Siemund & Noemi Kintana (eds), Language contact and contact languages, 61-79. Hamburg Studies on Multilingualism. Amsterdam: John Benjamins
Johanson, Lars. 2013. Isomorphic processes: Grammaticalization and copying of grammatical elements. In Martine Robbeets & Hubert Cuyckens (eds.), Shared grammaticalization, with special focus on the Transeurasian languages, 101-111. Amsterdam: John Benjamins.
Kuteva, Tania & Bernd Heine. 2012. An integrative model of grammaticalization. In Björn Wiemer, Bernhard Wälchli and Björn Hansen (eds.) 2011, Grammatical replication and grammatical borrowing in language contact. Berlin: Mouton de Gruyter. Pp. 159-198.