Number-sensitive reflexive pronouns in Danish: Optionality, microvariation, and cyclic change

Danish, in common with the other Scandinavian languages but in contrast with e.g. English and German, has a possessive reflexive pronoun sin (with the inflectional variants sit and sine). Sin must have a subject antecedent within a particular binding domain, and this subject antecedent must be singular (or non-plural). This is an unusual demand in a cross-linguistic perspective: In the other Scandinavian languages and in earlier stages of Danish (before 1000 AD, at least), sin places no number restrictions on its antecedent. It is also an unusual demand in a linguistic theory-perspective: Various researchers (Burzio 1991 and Reuland
2011, among others) suggest that the defining trait of reflexive pronouns is precisely that they lack morphological content (number, gender, sometimes even person). With the Danish sin, we have a reflexive pronoun that can be specified as a 3rd person singular pronoun, which certainly seems like some amount of morphological content. However, this may not be a lasting trait. In modern
Danish, it is not unusual to hear or read sin used with plural antecedents, and I suggest that sin may be in the process of losing its singular feature.

Danish having number-restricted reflexives is not a new feature of the language. It is attested in the earliest non-runic sources of Danish, going back to the 13th century. In the same period of time, Danish object reflexive sig/sig selv seems to have changed from patterning with sin in mainly allowing singular antecedents to the present situation where it patterns with the other Scandinavian
languages in allowing both singular and plural antecedents. This is a relatively recent development, however, and it was finalized as recently as the 20th century in the spoken language (cf. Pedersen 2017). I hypothesize that sin may be going the same way as sig and changing (back) to allow both singular and plural antecedents. The change in sin lags behind because there has been less external
pressure on speakers to acquire a system with a number-neutral sin.

I investigate the properties of sin (and sig) in modern and historical Danish in this thesis. I view the topic through a lens of generative grammar with a particular focus on microvariation. My thesis is a contribution to a field of research into reflexive pronouns within generative grammar which has become increasingly eclectic with the inclusion of more languages, more niggling details, and new theory-internal perspectives. My thesis is not aimed at arguing e.g. for or against an agreement-based or movement-based approach to binding. I offer an explorative, empirically based perspective on diachronic and synchronic microvariation within a single language. I hope to offer some insight into an interesting case of cyclic change and linguistic optionality.

In the first chapter of the thesis, I introduce the theories and technical machinery behind the approach to linguistics that I adopt in this thesis. I situate my thesis within a framework of generative grammar, and within a framework of variation and change with a special focus on microvariation. I sketch the development in the research on reflexives from the relatively unified Government and Binding approach of the 1980s to the current, rather less unified range of approaches.

In the second chapter of the thesis, I narrow the focus and look specifically at Danish and the Danish reflexive system. I present an analysis of the reflexive systems in standard Danish (an analysis which first saw the light of day in Vikner 1985), and I outline some of the ways that the Jutlandic dialects of Danish differ from the standard. The ways that reflexive use in the Jutlandic dialects differ

from standard Danish are somewhat stigmatised and I hypothesize that this could lead to speakers hyper-correcting and using sin with plural antecedents more frequently. Finally, I present the use of sin in standard Danish that I call number restricted or number-sensitive.

The third chapter of the thesis is a diachronic study of the use of sin and sig in Danish over the last millennium. There is textual evidence from runic stones from Denmark that sin was used with both plural and singular antecedents before 1000 AD. There is no direct evidence that this was also the case for sig. This use is, however, a direct continuation from the Common Germanic stage where both sig and sin were used with antecedents of all numbers. By the 13th century, where the earliest Danish manuscripts are from, the use of sin and sig had changed so that both forms are primarily used with singular antecedents and their non-reflexive counterparts (deres and dem) are used with plural antecedents. Texts in the following 6-700 years shows different developmental trajectories for sig and sin. Sig becomes the predominant form in the written language with both singular and plural antecedents after the Reformation (with a great deal of variation). This is probably due to influence from German, and locally bound dem remains frequent in the spoken language until the early 20th century. Locally bound dem must be considered a very marginal form in the modern language outside of the contexts where there is structurally conditioned optionality between sig and dem. Sin stays limited to mainly singular antecedents, although examples of plural antecedent sin can be found sporadically all through the period.

The fourth chapter of the thesis is a corpus study. I investigate the occurrence and distribution of plural antecedent sin in KorpusDK, which is a written corpus with texts from 1983 to 2002. I found 1218 examples of plural antecedent sin in the corpus out of 188,585 instances of sin with any kinds of antecedents. Overall, plural antecedent sin occurs relatively more often with inanimate antecedents, with full noun antecedents (rather than pronoun antecedents), with coordinated antecedents with singular coordinands, and in complex clauses (particularly a type of clause where a partitive is modified by a relative clause). The data does

not support suggestions from the literature that plural antecedent sin is better or more likely to occur within distributive contexts.

The fifth chapter is an acceptability judgment experiment with a sample of 550 young Danish speakers form various Danish gymnasiums (upper secondary schools). I find in the study that plural antecedent sin is more acceptable if it is contained with a plural nominal, if the antecedent is inanimate, and if the antecedent is a full noun. I find a little statistically significant evidence of sociolinguistically conditioned variation but the most striking result here is that the effects of e.g. region or gender are not very striking. Participants from all over the country rated plural antecedent sin as quite acceptable, just like participants from all over the country rated plural antecedent sin as quite unacceptable. I believe that this could show a linguistic state of ongoing change where some speakers have plural antecedent sin as part of their grammars, and some speakers do not. Whether or not a speaker accepts plural antecedent sin seems more dependent on other factors in that speaker’s own grammar than on the sociolinguistic factors that I coded the data for. Specifically, those speakers who rate other non-standard uses of reflexives (such as using locally bound singular possessive pronouns) higher also tend to rate plural antecedent sin higher.

In the final chapter of the thesis I sum up the results from the previous chapters, suggest areas of improvement and topics for follow-up studies, and speculate on the factors that could have an impact on this possible ongoing change. I suggest that some of these factors are the change in number-restriction on sig, hyper-correction driven by other stigmatised features of the Danish reflexive system, semantic agreement, and – very speculatively – the great deal of optionality or variation in number agreement on Danish adjectives, which could in principle also work as evidence for acquiring a reflexive system with less number-sensitivity on sin.