In this doctoral thesis, Christiane Müller takes on one of the major issues that has shaped much of modern formal linguistic research, namely long-distance dependencies. These dependencies have been important in particular because of their proclaimed universality (starting with Chomsky Reference Chomsky1964, Ross Reference Ross1967) and because they illuminate the locality of syntactic computations more generally (Belletti Reference Belletti2019). They also carry significant implications for learnability, since there is very little if any negative data that will help the child figure out that certain dependencies are licit, and others are not. Universality could solve this problem as the constraints then did not have to be learned to begin with. However, the picture is far more complicated. Rizzi (Reference Rizzi1982) was among the first to argue that there is variation, but ever since his work, this variation has been thought to be limited and systematic (Phillips Reference Phillips, Sprouse and Hornstein2013). Starting in the early 1980s, Mainland Scandinavian languages have been reported to be unusually permissive when it comes to extracting from domains that are typically considered to be islands (Hagström Reference Hagström, Svensson, Wieselgren and Hansson1976, Engdahl & Ejerhed Reference Engdahl and Ejerhed1982). Quite tellingly, reference grammars for all the Mainland Scandinavian languages report relative clause extraction to be acceptable given that certain conditions are fulfilled (Faarlund, Lie & Vannebo Reference Faarlund, Lie and Vannebo1997, Teleman, Hellberg & Andersson Reference Teleman, Hellberg and Andersson1999, Hansen & Heltoft Reference Hansen and Heltoft2011). However, since then it has always been unclear how general and systematic the permissibility is. In recent years, extraction from domains typically considered to be islands has again become very popular. Currently, islands are being investigated intensely across the Scandinavian languages: Christensen, Kizach & Nyvad (Reference Christensen, Kizach and Nyvad2013a, b) and Nyvad, Christensen & Vikner (Reference Nyvad, Christensen and Vikner2017) study Danish, Lindahl (Reference Lindahl2017), Tutunjian et al. (Reference Tutunjian, Heinat, Klingvall and Wiklund2017), and Wiklund et al. (Reference Wiklund, Heinat, Klingvall, Tutunjian, Escobar, Torrens and Parodi2017) study Swedish, and Bondevik (Reference Bondevik2018) and Kush, Lohndal & Sprouse (Reference Kush, Lohndal and Sprouse2018, Reference Kush, Lohndal and Sprouse2019) study Norwegian. Müller’s dissertation fits into this current trend nicely as she investigates adjuncts in Swedish in great detail, a topic that has not been investigated much in Swedish in recent years.
Müller’s dissertation studies extraction from adjunct clauses in Swedish in comparison with English. A couple of examples of adjunct islands are provided in (1) from English.
(1)
Occasional examples have been presented in the literature intended to show that Swedish does not obey this constraint, as seen in (2).
(2)
Müller’s objective is to more rigorously investigate the extent to which Swedish allows violations of the adjunct clause constraint. She has written a dissertation which altogether comprises seven chapters.
Chapter 1 is a brief introduction which outlines the research questions and the aims of the dissertation. The latter are as follows (p. 11):
To identify and investigate factors that have an impact on the acceptability of adjunct island extraction in Swedish.
To examine the role of two factors (coherence and finiteness) more closely in Swedish and English, with the goal to investigate the possibility of cross-linguistic variation with regard to their impact on the acceptability of sentences involving adjunct island extraction.
To investigate how coherence and finiteness affect the online processing of adjunct clause extraction sentences in English, with the goal to look for online support in favor of the permeability of such structures.
The remaining six chapters seek to realize these aims.
Chapter 2 provides important background information concerning the topic of island constraints, zooming in on adjunct islands specifically. The chapter narrows down the scope of the thesis by presenting the various types of adjunct clauses that will be studied. In addition, the chapter also contains a rich and systematic overview of different syntactic and non-syntactic analyses in the literature that have been invoked to account for the data, especially in the Scandinavian languages.
The focus on the Scandinavian languages continues in Chapter 3, which offers a detailed overview of the Mainland Scandinavian situation. Müller reviews a range of different analyses that have been proposed to account for the unexpected island violations in Mainland Scandinavian languages. She looks at the relevant previous research and identifies several knowledge gaps. In particular, we have little knowledge about the conditions that affect the felicity of adjunct clause extraction in languages like Swedish, and even Scandinavian more generally.
The next three chapters are case studies of adjunct extraction in Swedish. Chapter 4 contains an acceptability judgment study whose goals are to identify which factors determine whether or not adjunct extraction is acceptable. The factors that Müller considers are: (i) the degree of semantic coherence between the matrix and the adjunct clause (building on the work by Truswell Reference Truswell2011 for English), (ii) the degree of syntactic integration of the adjunct clause (this includes both its internal and external syntax, see Haegeman Reference Haegeman, Coene, De Cuyper and D’hulst2004, Reference Haegeman2012), and (iii) the grammatical function of the extracted element (adjuncts vs. arguments). The main finding from the chapter is that all these factors are relevant for both Swedish and English. However, there is a residue of instances of licit extraction from a subset of finite adjunct clauses reported to be unacceptable in English, suggesting that finiteness plays a potential key-role in account for this difference. A caveat is in order, though: The methodology employed in this chapter is not very rigorous, with few items per condition. Müller acknowledges the limitations, in part through labeling it a ‘semi-formal’ study. Nevertheless, the findings are suggestive of certain patterns worth exploring further.
Chapter 5 investigates the hypothesis that adjunct islands in English and Swedish are similar when it comes to the coherence factor, but they differ when it comes to the role of finiteness in determining whether or not extraction is possible. Two acceptability judgment studies are devised to test this hypothesis, considering extraction from after-adjunct clauses in Swedish and English, respectively. An example from English is given in (3) (Truswell Reference Truswell2007: 166, see also Manzini Reference Manzini1992).
(3)
Müller finds that finiteness causes a decline in acceptability in English but not in Swedish, even though the acceptability never got ratings above mid-point on a seven-point scale. Nevertheless, this demonstrates that there is constrained cross-linguistic variation in extractability between the two languages. She argues that these results support the idea that filler-gap integration also is attempted in some island structures, which leads Müller to develop what she calls the permeability hypothesis. This hypothesis holds that ‘integrative processes related to dependency formation are to some degree active in islands, at least in adjunct islands of the type investigated here’ (p. 159).
Chapter 6 sets out to test the permeability hypothesis based on data from English. Müller designed an online experiment using self-paced reading aimed at testing the role of coherence and finiteness in sentences with extraction from after-adjunct clauses. The main finding is that the reading times are faster for coherent adjuncts compared to non-coherent adjuncts. Furthermore, she finds a finiteness-related slowdown at regions associated with gap integration in coherent adjuncts. This shows that some kind of filler–gap dependency is attempted in the adjuncts investigated, providing support for the hypothesis that integrative processes take place inside island domains.
The last chapter, Chapter 7, provides a summary of the main research questions and findings of the dissertation. Müller also discusses certain further theoretical implications. Among others, she argues that certain adjunct clauses in English and Swedish should be considered weak islands. As such, she contributes to the growing inventory of weak islands:
Up until the late 1980s nothing much beyond wh-islands had been thought to be weak (selective) islands. Beginning with Relativized Minimality, however, and ever-growing range of WIs has been recognized. Thus, theories of WIs have mushroomed, each coming with a significant set of new data and important new connections to other domains. (Szabolcsi & Lohndal Reference Szabolcsi, Lohndal, Everaert and van Riemsdijk2017:3)
Müller also attempts to provide a more general typology of languages depending on whether or not they allow extraction from adjuncts (p. 199). She distinguishes between languages that obey the Strong Adjunct Condition and languages that adhere to the Weak Adjunct Condition. In the former case, adjuncts are always strong islands (such as Dutch, French and Greek, according to Müller), whereas in the latter case adjuncts can either be strong or weak. Those that are strong are the Peripheral Adverbial Clauses whereas the weak ones are Central Adverbial Clauses, drawing heavily on Haegeman’s (Reference Haegeman, Coene, De Cuyper and D’hulst2004, Reference Haegeman2010, Reference Haegeman2012) typology. Given this typology, Müller argues that it may be possible to maintain the universality of the Adjunct Condition (Stepanov Reference Stepanov2007). However, new findings regarding extraction from sentential adjuncts in Norwegian cast additional doubt on the universality of adjuncts (Bondevik Reference Bondevik2018, Kush, Lohndal & Sprouse Reference Kush, Lohndal and Sprouse2019). These findings suggest that adjuncts are not syntactic islands in Norwegian, although there may be potentially universal semantic factors that could account for some of the data (see Kush, Lohndal & Sprouse Reference Kush, Lohndal and Sprouse2019 for additional discussion).
In general, the overview of the relevant literature is comprehensive, and Müller clearly demonstrates substantial critical and analytic skills in integrating her own research questions into this literature. Her own analyses are creative and mature. In particular, it is striking how close the relationship between the empirical generalizations and the proposed theoretical analyses are. This is a great virtue of the present dissertation where the theoretical analyses are modeled on substantive empirical tests and not governed by a particular pre-chosen model. Rather, Müller engages critically with the previous research in this area, and she lets her empirical generalizations together with sound analytical principles guide the theoretical analysis.
Let us return to the question posed at the outset, namely how general and systematic island violations in Mainland Scandinavian languages are. Are we closer to being able to answer that question now? In some ways we are, due to the resurgence of island studies. In other ways, we may not be, since the various studies use a different methodology and look at different phenomena. As such, it is difficult to provide an overarching picture given that most of the studies are not comparable. In turn, this also means that we do not have a solid answer to the question of why Mainland Scandinavian languages are exceptional, that is, what the locus of variability is. The hope would be that future work would address this by doing rigorous comparative studies across the Mainland Scandinavian languages.
That said, this does not in any way detract from the merits of Permeable Islands: A contrastive study of Swedish and English Adjunct Clause Extraction. This dissertation advances our understanding of Swedish and illuminates core differences between Swedish and English. Furthermore, it provides scholars working on the other Mainland Scandinavian languages with homework in testing whether similar (or different) patterns can be obtained. Müller argues that ‘[E]xtraction possibilities might vary for different types of extraction dependencies such as topicalization, question formation, and relativization or cleft formation’ (p. 107). Her work combined with the other recent work show that we need a more refined view of islands, since movement operations do not seem to cluster the way much previous research has suggested. As such, Mainland Scandinavian languages will likely continue to set the agenda when it comes to studying islands for years to come.
