Project Description

1 State of the art and preliminary work

I. State of the art

The project aims at examining the distribution of relative pronouns and relative complementisers introducing relative clauses in West Germanic languages. Apart from examining the syntactic features involved in the particular options, as well as the relation of the relative clause to the head noun in the matrix clause, the aim is to identify the status of the West Germanic patterns against an empirically adequate typological background

Relative clauses constitute a well-studied area of syntax both in generative grammar and in typological research, yet there are many questions related to certain asymmetries (and their potential interrelatedness) that deserve more systematic investigation, both as far as empirical data and theoretical conclusions are concerned. Such asymmetries include ones that are directly related to the morphological properties of functional elements, as well as ones that follow from the syntactic function associated with the gap. The phenomena to be investigated have relevance also for other constructions and have more general implications concerning the syntax-morphology interface. The investigation of West Germanic languages in this respect is of primary importance: regarding generative grammar, several properties of relative clauses were first discussed based on Standard English data. Yet the Standard English pattern is by no means universal cross-linguistically and even dialects of West Germanic (including English dialects) exhibit data that are problematic for existing assumptions.

A. Relative operators versus complementisers

In (Standard) English, relative clauses can be introduced by relative pronouns, (1a), or by the complementiser that, (1b).

(1) a. I know the man who painted your car blue.
b. I know the man that painted your car blue.

The use of relative pronouns, which is attested in other West Germanic languages as well (Romaine 1984), is a typically European strategy and is rare cross-linguistically (Van Gelderen 2009, citing Comrie 2002). In generative grammar, it is a standard assumption that the relative pronoun moves to the left periphery of the clause. In Standard English, an overt relative pronoun does not co-occur with an overt relative complementiser; this is traditionally attributed to there being a "Doubly Filled COMP Filter" (going back to Chomsky & Lasnik 1977). Yet it is evident that the availability of such constructions depends on the dialect, as shown in (2) below (% refers to the construction being grammatical in certain dialects/idiolects but not in others).

(2) % This is the city in which that I live.

In more recent analyses which do not assume a "COMP" position, that is taken to be the head of the complementiser phrase (CP), while the relative pronoun (or the entire phrase containing the relative pronoun) moves to the specifier of that projection. This makes the two kinds of elements categorically and syntactically distinct, yet it has to be mentioned that relative pronouns may be reanalysed as complementisers in language change (as was the case for that in Old English, the process termed as the "relative cycle" by Van Gelderen 2009; see also Romaine 1984), making the syntactic category distinction less rigid.

Regarding the choice between complementisers and relative pronouns, Van Gelderen (2009) mentions that while wh-pronouns are promoted by prescriptive rules, there is a clear preference (with a 4:1 ratio) for that in spoken English (see also Romaine 1982; Montgomery & Bailey 1991; Van Gelderen 2004; Tagliamonte et al. 2005). In line with this, the study of Herrmann (2005) shows that the use of the relative pronouns who and which is not very frequent in the regional dialects of Britain.

Standard German allows only the relative pronoun strategy but studies on South German dialects show that the complementiser strategy involving wo (and in certain Bavarian dialects was) is the default pattern (see, for instance, Brandner & Bräuning 2013 on Bodensee Alemannic; Salzmann 2017 on Zurich German; Fleischer 2004a, 2016 on Hessian; and Weiß 2013 on Bavarian). Combinations of the form (2) are attested in these dialects as well. Standard Yiddish has relative clauses introduced by the uniform particle vos, as well as relative clauses introduced by pronouns (Fleischer 2004b), but combinations are not reported.

In Standard Dutch, d-pronouns are used in relative clauses; in colloquial Dutch, wh-pronouns are also possible (Boef 2013). As reported by Boef (2013: 141) on the basis of the SAND1 data, some dialects like Waasland Dutch may employ the complementiser dat ‘that’ in addition to the relative pronoun (rendering doubling patterns similar to (2) above):

(3) Dat is de man die dat het gedaan heeft.
  that is the man who that it done has
  ‘That is the man who has done it.’ (Boef 2010)

However, in the same dialects, no dat-relatives occur without an overt relative pronoun (unlike the English pattern in (1b) above), which suggests that dat in these cases is merely the regular finite complementiser (not typing the clause as relative) and not a genuine relative complementiser (typing the clause as relative). The same conclusion was drawn by Bennis & Haegeman (1984) for West Flemish. Boef (2013: 141) also reports that other dialects (such as the ones in Vlaams-Brabant) have dat-relatives, but these dialects do not have doubling patterns (see also Haegeman 1992). Frisian is similar to Dutch in essentially applying relative pronouns: these are regularly declinable relative pronouns but an indeclinable relative pronoun as a complement to prepositions is also possible (Hoekstra 2015). There is altogether little literature on Afrikaans relative clauses: wat is generalised as an invariable particle in most constructions (except for adverbials and complements of prepositions, see Meyer 2016). According to Den Besten (2012), it is not a pronoun in these constructions anymore, making it similar to the other relative complementisers mentioned above.

In addition to doubling patterns like (2) in relative clauses with nominal heads (ordinary relative clauses), doubling patterns are reported to exist in free ("headless") relative clauses in German and Dutch dialects. Consider:

(4) a. wem dass des zvei is, kann aa wenger zoin
    who.DAT that that.N too.much is can.3SG also less pay.INF
    ‘Whoever finds it too much can pay less as well.’ (Bavarian; Weiß 2013)
  b. Wie dat er nou trouwt zijn stommerike.
    who that there now marries are stupid
    ‘Whoever gets married nowadays is stupid.’
    (South Brabant; Zwart 2000, citing Vanacker 1948)

These combinations involve a wh-pronoun and the regular finite complementiser; the same kind of combination in free relatives is not reported for English (while English has doubling both in embedded questions and in ordinary relatives).

B. Subject/object asymmetries

Subject/object asymmetries are well-known to exist in relative clauses (and in various other clause types involving movement). A crucial generalisation made by Keenan & Comrie (1977) concerns the "Noun Phrase Accessibility Hierarchy", according to which nominal expressions can be accessible to different degrees; subjects are the most accessible, followed by other roles according to the following scheme: subject > direct object > indirect object > oblique object (complement of preposition) > genitive (possessor phrase) > object of comparison. The study of Keenan & Comrie (1977) concerned the distribution of resumptive pronouns in relative clauses: resumptive pronouns tend to occur with less accessible noun phrases (if a resumptive pronoun is required for a certain type of noun phrase, it is also required for all the less accessible noun phrase types as well). Accordingly, resumptive pronouns with subjects are cross-linguistically very rare (Comrie & Kuteva 2005). Various studies (e.g. Wanner & Maratsos 1978) also provide evidence for subject relative clauses being processed more easily than object relative clauses.

Regarding the distribution of relative markers in regional dialects of Britain, Herrmann (2005) shows that the Noun Phrase Accessibility Hierarchy is relevant in the spread of the relative complementisers that and as: subjects are more accessible than objects, which predicts not only that subject relative clauses are more frequent but also that the complementiser strategy is more frequent in subject relative clauses than in object relative clauses. In the standard language, the occurrence of that is restricted to subjects and (direct/indirect) objects. With oblique objects, it is possible only if the preposition is stranded but not when the preposition is omitted altogether; this is, however, possible in non-standard varieties, as in (5) below:

(5) I haven’t been to a party yet that I haven't got home the same night.
(Van Gelderen 2009, citing Miller 1993)

The differences in the frequency of the various functions can be enormous; as pointed out by Fleischer (2004b), relative clauses with indirect objects are very rare in corpora and hence the investigation of dialectal variation must also include dialect grammars (which also make use of directly elicited data). Fleischer (2004a, 2004b) examined the distribution of relative markers in dialects of German (including Yiddish) and concluded that differences occur in line with the Accessibility Hierarchy: subjects and direct objects tend to pattern together with respect to the possible/preferred strategies, while obliques tend to have different strategies; indirect objects either pattern with subjects and direct objects or with obliques.

Apart from the Accessibility Hierarchy, which affects the differences among all the functions located below the subject as well, the specificity of the subject as a function (Comrie & Kuteva 2013) is also relevant for subject/object asymmetries: accordingly, one can distinguish between relativisation on subjects and relativisation on "obliques" (obliques in this case encompassing all categories but the subject). Relevant asymmetries are attested in Germanic languages as well: for instance, Danish relative clauses are generally introduced by som but subject relatives can also be introduced by der (Vikner 1991).

The special status of subjects is also reflected in generative grammar, where arguments of a lexical verb can be either external or internal: subjects are either external (e.g. with transitive verbs) or internal (e.g. with unaccusative verbs), while all other argument types are internal. Subjects agree with the verb in Germanic languages; in English, subjects also have a canonical position in the specifier of the TP (tense phrase) projection. In other words, the special status of subjects is given in the syntax independently of relative clauses

II. Preliminary work

The previous project of the principal investigator (DFG BA 5201/1-1) examined the left periphery of various finite embedded clauses, including relative clauses, with special attention paid to the possible combinations of clause-typing elements. The findings constitute the theoretical basis of the present proposal, in that the basic syntactic structure of the left periphery and especially that of multiple elements in the clausal left periphery (CP-domain) is to be modelled in a minimalist, non-cartographic approach (Bacskai-Atkari to appear c). The project also gathered data from five Germanic languages (Dutch, Danish, Icelandic, Norwegian, Swedish) via a questionnaire examining the left periphery of various embedded clauses, including relative clauses (Bacskai-Atkari & Baudisch 2018).

A. Relative operators versus complementisers

Bacskai-Atkari (2021b) investigated the syntactic possibilities of relative pronouns and complementisers in dialects of German; it was found that the co-occurrences are severely limited by constraints that can be best explained by a minimal structure rather than by a pre-given cartographic template. In this way, doubling patterns in relative clauses are similar to doubling effects in interrogative clauses (Bacskai-Atkari 2018a) as they can be analysed with a single CP (unlike, for instance, comparative clauses, Bacskai-Atkari 2018a, 2018f, 2020b, 2021a). The complementiser appears overtly in West Germanic embedded interrogatives because a finite C position is regularly lexicalised by overt elements, resulting in V2 orders in German main clauses and in T-to-C movement in English main clause questions (Bacskai-Atkari 2018b, 2018e, 2020d). This produces doubling patterns in constituent questions since the wh-element must be overt, but doubling is less likely to appear in polar questions where the disjunctive operator and the disjunctive complementiser overlap in their functions (Bacskai-Atkari 2019, 2020c, to appear a). The lexicalisation requirement on [fin] is not entirely specific to (West) Germanic: similar effects can even be detected in languages in which functional domains occur lower than the CP-domain (Bacskai-Atkari 2018c, to appear b). While relative pronouns and relative complementisers are sufficiently distinct syntactically regarding whether they can lexicalise [fin], the grammaticalisation of relative pronouns into complementisers is also attested (as in the case of English that), referred to as the "relative cycle"; this phenomenon can be detected in other clause types as well (Bacskai-Atkari 2014, 2016, 2018d) and can be therefore considered as a fairly robust syntactic phenomenon that is not restricted to the particular clause type. Its availability is, however, strongly restricted by the morphological properties of the respective pronouns, which in turn are only partially predictable from the general system of morphological case and gender in the language (Bacskai-Atkari 2020e).

B. Subject/object asymmetries

Bacskai-Atkari (2020a, 2020e) presents the first results of an ongoing corpus study on the King James Bible and its modernised version (New King James Bible). This study involves the examination of the relative marker in the original version for all relative clauses that contain “who” or “whom” in the new version. This design allows for detecting the various options (including zero relatives) in the original by relying on a simple search in the new version. Relative clauses involving a human referent are preferably introduced by who/whom in the standard language. The New King James Bible is fairly conservative: in other words, deviations the other way round (that is, the newer version using less standard options than what is found in the original) are not likely to occur (as also verified by the investigation on a smaller, representative sample including the hits for “which” and “that” in the new version). The new version is also very consistent in using who for subjects only but never for objects. There are 3569 subject relative clauses involving who and 398 object relative clauses involving whom in the new version; the considerable differences in the frequency of subject and object relative clauses are in line with the Accessibility Hierarchy, according to which subject relative clauses are more accessible and more frequent than object relative clauses. Similarly, the frequency of that-relatives is significantly higher in subject relatives (around 75% in the representative sample mentioned above) than in object relatives (around 28%). The examples in (6) illustrate the difference between the two versions and between subject and object relatives:

(6) a. And Abraham called the name of his son that was born unto him, whom Sarah bare to him, Isaac.
    (King James Bible; Genesis 21:3)
  b. And Abraham called the name of his son who was born to him – whom Sarah bore to him – Isaac.
    (New King James version; Genesis 21:3)

As can be seen, the original version uses that with the subject relative but whom with the object relative; the preferences are so strong that they may occur in the same sentence, overriding potential parallelism effects (which can be detected among coordinated subject relative clauses). The frequency of which-relatives with human referents is roughly the same (around 20%) in subject and in object relative clauses, suggesting that the subject/object asymmetry makes reference to the category of the relative marker but not to the chosen pronoun.

1.1 Project-related publications

1.1.1 Articles published by outlets with scientific quality assurance, book publications

Bacskai-Atkari, Julia. 2014. Cyclical change in Hungarian comparatives. Diachronica 31.4. 465–505.

Bacskai-Atkari, Julia. 2016. On the diachronic development of a Hungarian declarative complementiser. Transactions of the Philological Society 114.1. 95–116.

Bacskai-Atkari, Julia. 2018. Clause typing in main clauses and V1 conditionals in Germanic. In: Mailin Antomo and Sonja Müller (eds.) Linguistische Berichte (LB) Sonderhefte 25: Non-canonical verb positioning in main clauses. Hamburg: Helmut Buske Verlag. 179–200.

Bacskai-Atkari, Julia. 2018. The relative cycle in Hungarian declaratives. In: Ana Maria Martins and Adriana Cardoso (eds.) Word order change. Oxford: Oxford University Press. 68–87.

Bacskai-Atkari, Julia. 2018. Doubly Filled COMP in Czech and Slovenian interrogatives. In: Denisa Lenertová et al. (eds.) Advances in formal Slavic linguistics 2016. Berlin: Language Science Press.

Bacskai-Atkari, Julia. 2018. Deletion phenomena in comparative constructions: English comparatives in a cross-linguistic perspective. Berlin: Language Science Press.

Bacskai-Atkari, Julia. 2020. German V2 and Doubly Filled COMP in West Germanic. The Journal of Comparative Germanic Linguistics 23(2). 125–160.

Bacskai-Atkari, Julia. 2020. English relative clauses in a cross-Germanic perspective. Nordlyd 44. 93–115.

Bacskai-Atkari, Julia. to appear. Discourse-driven asymmetries between embedded interrogatives and relative clauses in West Germanic. In: Nicholas Catasso, Marco Coniglio & Chiara De Bastiani (eds.) Language change at the interfaces: Intrasentential and intersentential phenomena. Amsterdam: John Benjamins.

1.1.2 Other publications

Bacskai-Atkari, Julia. 2020. Changes affecting relative clauses in Late Modern English. In: Merja Kytö & Erik Smitterberg (eds.) Late Modern English: Novel encounters. Amsterdam: John Benjamins. 91–115.

2 Objectives and work programme

2.1 Anticipated total duration of the project

3 years (funding requested for 3 years)

2.2 Objectives

The main objective of the project is to gain new insights into the syntactic and morphological factors underlying relativisation, as attested in finite relative clauses characteristic of European languages. The project examines the distribution of relative markers in West Germanic in particular and aims to account for the observed variation in a principled way. West Germanic languages are an ideal testing ground regarding the syntactic and morphological properties of relative clauses since these languages allow various relative markers and their combinations, which is cross-linguistically by no means a common pattern. Not all languages display finite relative clauses (see, for instance, Ackermann and Nikolaeva 2013); further, the use of wh-pronouns as relative pronouns is essentially restricted to European languages (Van Gelderen 2009, citing Comrie 2002). The traditional, regular pattern in West Germanic involves the use of demonstrative-based pronouns (Brandner & Bräuning 2013, as in German and Dutch), potentially along with the wh-pronoun strategy. As mentioned before, the pronoun strategy co-exists with the complementiser strategy.

The project has two fundamental assumptions. First, it is assumed that the origin of the various relative markers (wh-based or demonstrative-based) has a bearing on which elements can occur in which constructions and how pronouns and complementisers can be combined. This hypothesis needs to be investigated as this question has not been examined in the literature in detail. Second, it is assumed that the choice between the pronoun strategy and the complementiser strategy is affected by the function associated with the gap in the relative clause (subject, direct object etc.). This assumption follows from the Noun Phrase Accessibility Hierarchy (Keenan & Comrie 1977; see section 1, I.B). The more specific hypothesis regarding this is that while complementisers are assumed to spread from the unmarked subject function (since they do not overtly identify the gap), relative pronouns may spread either from the unmarked subject function (due to frequency effects) or from the lower, marked functions (as they identify the gap overtly); either way, however, in a system maintaining both the complementiser and the pronoun strategy, complementisers will be eventually associated with higher functions, while pronouns will rather be associated with lower functions. This hypothesis needs to be tested as the literature so far offers only partial insights in this respect: for Old High German, Coniglio (2019) argues that the d-pronouns spread from the subject function; for Middle English, Romaine (1982) and Gisborne & Truswell (2017) suggest exactly the opposite. The question still remains, however, how the distribution changes over time and to what extent the specific properties of Germanic and general typological implications apply (Romaine 1984). For Frisian, Hoekstra (2015) shows that the indeclinable relative pronoun that was attested for the subject and object functions in Old Frisian (just like in Old Dutch and Old English) is restricted to complements of prepositions in Modern Frisian, while subject and object relatives use a declinable relative pronoun. This suggests that the presence of a preposition interacts with the morphosyntactic properties of the pronoun itself and inflection may not be required to identify a gap once an overt pronoun and an overt preposition are both present.

These hypotheses can be ideally tested in West Germanic since the languages (or their relevant varieties) allow for multiple strategies: in this way, any differences can be attributed to the formal properties of these elements rather than to other differences between the languages that are governed by independent factors.

I. Relative operators versus complementisers

One major result of previous investigations (see section 1 part II) was that West Germanic languages preferably have an overt element in a finite C: this means that in most syntactic patterns, the absence of an overt element in C leads to ungrammaticality, and altogether very few patterns involve a lexically specified null complementiser. In some cases, however, a certain degree of optionality arises; for these cases, it is assumed that (i) all things being equal, the pattern with an overt C element will be more frequent, and (ii) patterns in which the overt C element is missing may be governed by independent factors that override (i).

This constellation results in a general preference for relative complementisers (especially in non-standard varieties and in informal contexts): all things being equal, an overt relative complementiser is used rather than an overt pronoun, and this preference is overridden only in cases where there are additional reasons to lexicalise the gap.

Since the relative operator is not necessarily overt if the complementiser is already overt, doubling effects (traditionally referred to as "Doubly Filled COMP", see section 1) do not automatically arise (as they do with embedded constituent questions); in fact, doubling is expected to be constrained by economy. While the findings so far predict a preference for the complementiser strategy, certain asymmetries still have to be accounted for. In particular, free relatives are syntactically more similar to embedded interrogatives since they contain wh-pronouns across West Germanic (that is, even in languages that have d-pronouns in ordinary relatives). The operator cannot be silent due to information structural constraints (it is the constituent that corresponds to the focus in the answer in question-answer sequences and as such it identifies information that is not recoverable from the preceding discourse, Bacskai-Atkari to appear a), so in dialects that by default have overt elements in C, doubling arises. In other words, there is an important theoretical distinction between ordinary relatives and free relatives, which leads to the following hypothesis:

Hypothesis 1: Doubling is more frequent in free relatives than in ordinary relatives.

Hypothesis 1 needs to be investigated as it is an empirical expectation based on the theoretical analysis pursued so far. In most theoretically informed analyses in generative grammar (see section 1), relative clauses with nominal heads were considered with respect to the possibility of doubling or the prohibition thereof: combinations of the form "wh-pronoun + that" are known to exist in English non-standard dialects in ordinary relatives but not in free relatives. The difference between the two clause types has not yet been studied in this respect. The purpose of the project here is to examine whether the empirical data support Hypothesis 1. Regarding this question, the morphological complexity of the pronoun and (if applicable) its associated phrase must also be considered: Bayer & Brandner (2008) show that in embedded questions in Alemannic and Bavarian, morphologically simplex single wh-elements (such as wer ‘who’) are less likely to occur in doubling patterns than morphologically complex phrases (such as mit was für eine Farbe ‘with what colour’).

Regarding the distinction between complementisers and relative pronouns, it is evident that there is some flexibility in the CP-domain: relative pronouns may grammaticalise into relative complementisers (see section 1). This is attested in the historical development of English that. The appearance of the given element as a complementiser in relative clauses is thus not unexpected since it is already there in the relative clause anyway, albeit in a partly different role (and with partly different feature properties). However, not all relative complementisers end up in relative clauses this way. In German, as shown by Brandner & Bräuning (2013), such a reanalysis of the ordinary wh-pronoun wo ‘where’ into a regular relative complementiser is unlikely. Brandner & Bräuning (2013) suggest that the appearance of wo instead of the earlier so in relative clauses is a matter of change from the d-series of pronouns (subsuming elements like so and als) to the w-series. This is rather an instance of complementiser change than grammaticalisation. Regarding such phenomena, the following hypothesis emerges:

Hypothesis 2: Complementiser change is subject to paradigm effects.

Paradigm effects are known to be responsible for certain changes in phonology: the shape of a word may influence the shape of another word if both are members of the same paradigm (see the paradigm uniformity of Steriade 2000). Crucially, a paradigm in the sense of Steriade (2000) is not restricted to inflectional paradigms but applies to derivational morphology as well. The relevant phenomena are related to the phonology-morphology interface.

Hypothesis 2 supposes that similar effects can arise on the morphology-syntax interface. Hypothesis 2 stems from the assumption made by Brandner & Bräuning (2013) that the change from so to wo in relative clauses happened at the same time as als changed to wie in equative clauses. However, while Brandner & Bräuning (2013) claim that the two changes are in fact the same, in the sense that both so/als and wo/wie are variants of the same element (the establishment of wo as relative and wie as equative taking place later and not uniformly across dialects), Hypothesis 2 supposes that no common syntax and semantics is necessary in order to derive the observed effect. For Brandner & Bräuning (2013), the common underlying base between the two constructions is a coordination-like component: this is, however, problematic, especially as it does not take into account potential differences with respect to the presence or absence of a matrix equative element. One possibility is to suppose that analogy between the clause types is responsible for such a change (much like in the spirit of König 2015). This would presuppose that there is an observable time gap between the two changes. Such timing differences are indeed attested between non-degree equatives and degree equatives, as well as comparatives (Jäger 2010), but it is not clear whether and how the changes in ordinary relative clauses relate to this: corpus studies are necessary in this respect. However, Hypothesis 2 makes no direct reference to a time gap and is thus compatible with the assumption of Brandner & Bräuning (2013) that the two changes took place simultaneously. According to Hypothesis 2, analogical changes affecting the complementiser system or a subsystem thereof are not necessarily triggered by a common semantic base but they arise precisely because the individual members are taken to be part of a syntactic paradigm, whereby changes on one member affect the system as a whole. In other words, if relative and equative complementisers belong to a shared paradigm, it is expected that any change affecting either complementiser may trigger changes on another complementiser, such that the resulting new paradigm is at least as uniform regarding its morphosyntactic properties as the previous one.

Hypothesis 2 needs to be investigated as the empirical data regarding complementiser change should be examined (existing diachronic studies on the emergence of complementisers concentrate on grammaticalisation, not on the kind of complementiser change described here). Moreover, if paradigm effects can be detected, the question arises how and to what extent complementisers constitute a paradigm, that is, which complementisers can have an immediate effect on which complementisers.

II. Subject/object asymmetries

As discussed in section 2.2, subject/object asymmetries in relative clauses are well-known in the literature; the previous investigations mentioned in section 1 also indicate that such asymmetries are relevant. On the one hand, there are two possible sources of these asymmetries: either subject relative clauses are markedly different from all other kinds of relative clauses (referred to as the "special status of subjects" for the sake of simplicity) or the entire Accessibility Hierarchy is relevant, in which case the asymmetry between subjects and direct objects is not markedly distinct from the asymmetry between, for instance, direct objects and indirect objects. On the other hand, the particular methodology applied in a given study may also affect the results (see section 2.3, "Methods"). This suggests that experimental and corpus results may indeed deliver different conclusions. Specifically, the following hypothesis arises:

Hypothesis 3: The special status of subjects is decisive in corpora.

Hypothesis 3 builds on the observation that subject relative clauses are the easiest to process and the most frequent to occur: this makes them the unmarked type. It is supposed that this sets them significantly apart from all other relative clause types in their actual occurrences in corpora. This is because in actual language use, speakers have various ways to convey the same message and therefore (i) they can choose to formulate the sentence in such a way that the subject is relativised and not a lower function and (ii) they can choose to convey the message by not using a relative clause when a lower function would be relativised. Keenan & Hawkins (1987) provide experimental evidence for informants changing a lower relativised function into a higher one (e.g. when asked to repeat an indirect object relative, they change it into a subject relative). This preference is expected to be higher in natural language use, i.e. when no pre-given lower relativisation is present. Hypothesis 3 has to be investigated because there are no explicit and comprehensive results so far regarding this question. Hypothesis 3 is complemented by another hypothesis concerning grammaticality judgements:

Hypothesis 4: The Accessibility Hierarchy is decisive in grammaticality judgements.

Hypothesis 4 essentially says that as far as the grammar and acceptability are concerned, the special status of subjects diminishes and a more balanced picture emerges, where the functions below the subject show more clear distinctions than they do in corpora. This follows from the assumption mentioned above that the high frequency of subject relatives in corpora is at least in part due to this strategy being preferred over various other alternatives, which does not imply that the other relativisation strategies would be ungrammatical. Hypothesis 4 has to be investigated since the relevant grammaticality judgement experiments (see section 2.3) have to be carried out first. Ultimately, testing Hypothesis 3 and Hypothesis 4 would contribute to a better understanding of which factors are responsible for subject/object asymmetries and to what extent.

2.3 Work programme incl. proposed research methods

applicant: Dr. Dr. Julia Bacskai-Atkari


The project aims at examining the syntax of relative clauses in West Germanic from a generative perspective, applying the minimalist framework (as proposed by Chomsky 2001, 2004, 2008, among others) in the analysis of syntactic structure. As language variation is a central issue of the project, the findings of linguistic typology will also be considered, since these findings help to understand the cross-linguistic status of the West Germanic patterns. In this respect, the project aims at contributing to a dialogue between formal grammar and linguistic typology, the necessity of which was pointed out already by Polinsky (2010). The various methods of gaining data (especially regarding the traditional opposition between text corpora and grammaticality judgements) are treated as complementary options that can be helpful in approaching complex theoretical questions from various perspectives (see Haspelmath 2009).

Since relative clauses are a well-studied area of linguistics (see section 1), part of the project is devoted to the analysis of already known patterns. This is naturally important because it constitutes the basis of further investigations. In addition, however, the project aims at gaining novel empirical data via corpus studies, questionnaires, and grammaticality judgement experiments. Regarding this, it must be kept in mind that the individual West Germanic languages (and their varieties) under scrutiny differ considerably in terms of how accessible the relevant data are and to what extent they have been discussed in the literature. The proposed project takes this into consideration regarding the choice of the methods.

Part 1: Relative operators versus complementisers

A. General considerations

This part of the project aims at examining asymmetries regarding the distribution of pronouns and complementisers. There are two kinds of asymmetries to be examined here: first, the one related to the exact clause type (headed relative, free relative, embedded question); second, the one related to the origin of the relative marker (demonstrative-based or wh-based). As described in section 2.2. above, the expectation is that doubling patterns are more likely to arise in free relatives and that combinations are further restricted by the origin of the given elements. Regarding clause types, the hypothesis is that West Germanic languages primarily seek to lexicalise the C position overtly and operators appear in addition if their presence is motivated by other factors, which can be information structural in nature (e.g. focussed elements cannot be left silent) but may be related to processing as well. The project aims at gaining robust empirical data that allow for a direct comparison between clause types. Further, regarding possible combinations, so far only asymmetric patterns (d-pronoun + wh-complementiser, wh-pronoun + d-complementiser) seem to be confirmed reliably as involving the doubling of a relative pronoun and a relative complementiser, while symmetric combinations seem to involve a mere finite complementiser (d-pronoun + d-complementiser in certain Dutch dialects, cf. Boef 2013) or do not seem to be possible at all (wh-pronoun + wh-complementiser).

B. Languages to be studied

One major focus will be on English, which allows both the complementiser strategy and the pronoun strategy in its standard variety as well. A good point of comparison is South German, since the relevant dialects not only use the complementiser strategy by default, but they are also known to allow doubling patterns even in free relatives, which is not confirmed for English. Since the aim of the project is not to provide a large-scale dialect survey but rather to gain reliable empirical data via case studies in order to test the theory, the project will concentrate on Alemannic in this respect, as this variety has been extensively documented and speakers can be easily recruited via the University of Konstanz. Equally, English regional dialects will not be investigated further (for comparative purposes, the study of e.g. Herrmann 2005 can be used as a reliable source) but the focus will be more on standard usage, with the aim of carrying out new empirical research on Standard English (see "Methods" below”). The results are to be compared with languages that do not have a choice between the pronoun strategy and the complementiser strategy, such as Standard German and Dutch. Afrikaans will also be considered in detail via a questionnaire study to gain the primary data. The results will be compared to novel findings concerning Mennonite Low German and Yiddish.

Part 2: Subject/object asymmetries

A. General considerations

This part of the project examines subject/object asymmetries. The main research question here concerns the choice between relative pronouns and relative complementisers. As described above in section 2.2, it is expected that corpus results and experimental results differ in this respect, as the proportion of subject relative clauses is expected to be high in actual language use and this configuration is supposed to show the most unmarked pattern(s). In the experimental settings, however, the limits of variability are tested. In varieties with a choice between the pronoun strategy and the complementiser strategy, such as Standard English, it is expected that the complementiser strategy is favoured higher on the Accessibility Hierarchy. Further, the project aims at examining how these results compare to the ratings in varieties that have no such choice. Finally, it is to be examined whether a variety like Standard German, which in principle allows for both wh-pronouns and d-pronouns, may show similar asymmetries in terms of the distribution of these patterns as the pronoun/complementiser asymmetry in varieties like Standard English.

B. Languages to be studied

In line with what was said above, the major focus will be on English (standard variety and historical data), as well as on German (standard variety, historical data, and Alemannic). The results will again be directly compared to Dutch. Afrikaans will also be considered here but as the primary source for the Afrikaans case study is centred on the questionnaire, it is not expected that a direct comparison with English, German and Dutch will follow as gaining the essential quantitative data for Afrikaans is not a major focus of this project.



Especially as far as historical data are concerned, the project uses parsed corpora to identify which patterns were used in the given periods and what their frequency is. Regarding English, the following corpora will be used: Penn Parsed Corpora of Historical English, Michigan Corpus of Middle English Prose and Verse, ARCHER, British National Corpus, Freiburg Corpus of English Dialects, Corpus of Contemporary American English, Corpus of Historical American English, Michigan Corpus of Academic Spoken English. Regarding German, the following corpora will be used: DDD Referenzkorpus Altdeutsch, Referenzkorpus Mittelhochdeutsch, Corpus of Historical Low German, GerManC, Bonner Frühneuhochdeutschkorpus, Datenbank für Gesprochenes Deutsch.

In the previous project of the principal investigator, questionnaires exploring the left periphery of embedded clauses were completed for Dutch (as well as for Scandinavian languages). The data gained for relative clauses can be used as a basis for the investigation of Dutch in the present project. However, gaining comparable data for Afrikaans would be necessary, since this language is understudied. It is planned that a questionnaire on relative clauses and equative clauses (consisting of about 90 questions) should be completed (building on the previous questionnaire, see Bacskai-Atkari & Baudisch 2018, but tailored to the aims of the present project). This involves the translation of sentences from English to Afrikaans, as well as answering questions about particular elements in the clause.


Altogether 5 experiments are planned to be carried out in the project (mostly grammaticality judgement experiments; see Schütze 2016 on the methodology). As described in section 2.2, certain relative clauses can be rare in corpora: in particular, indirect object relative clauses are very infrequent. Most studies relating to the Accessibility Hierarchy are based on corpora or on other databases (for instance, dialect speech) and do not consider direct elicitation. This also means that the relative frequency of various relative markers is likely to be confounded with the frequency of the given function (subject, object etc.), making it difficult to assess whether subject/object asymmetries are primarily related to the Accessibility Hierarchy (that is, whether direct and indirect objects show similar asymmetries) or rather to the special status of subjects. The project aims at bridging this empirical gap by testing the availability of pronoun and complementiser strategies in different kinds of relative clauses (the gap/operator corresponding to subject, direct object, indirect object, prepositional complement). The results of the experiments will then be compared to the corpus data and the data from dialect syntax databases; the differences between the elicited data and corpus data may cast light upon the differences between what the grammar allows (grammaticality) and what counts as a preferred option (accessibility/parsing).

Experiment 1 is a grammaticality judgement experiment on English, with the primary aim of examining the differences between syntactic functions (subject, direct object, indirect object, prepositional complement) and how they correlate with the choice between the pronoun strategy and the complementiser strategy (that). Grammaticality judgements are to be expressed on a scale (1–5). Apart from syntactic roles and relative markers, it will also be taken into account whether the referent is human or non-human: in Standard English, this involves a distinction between who(m) for human referents and which for non-human referents, but it should be examined whether the acceptability (and actual distribution) of that-relatives may also show differences in this respect. As most corpus and dialectal data discussed in the relevant literature refer predominantly to British English (and to English English in particular), Experiment 1 should be carried out on British English as well, to be compared with a control group of US speakers. This comparison would make it possible to detect potential differences which may be regional in nature, but which may be also due to differences in norm-oriented behaviour.

Experiment 2 is a production experiment on English. This experiment also aims at detecting differences in the choice of relative markers with respect to the syntactic function involved (and the human vs. non-human distinction regarding the referent). The experimental design involves gapped sentences with a missing relative marker; the sentences are to be parts of a single text or of smaller texts, and distractor items should include other cases where there is some optionality (for instance, between if and whether in embedded polar questions). Ideally, Experiment 2 should run in parallel with Experiment 1: the two experiments should have a similar (but not the same) set of experimental subjects to achieve comparable data (again, the focus is on British speakers and the control group of US speakers). The purpose of Experiment 2 is twofold. On the one hand, by comparing the results of Experiment 1 and Experiment 2, the differences between grammaticality judgements and language production in an experimental setting can be detected. This provides information about the possible discrepancy between grammaticality and preference regarding the choice of the relative marker. On the other hand, this design allows for creating contexts that otherwise rarely occur in actual corpora: indirect object relative clauses are known to be infrequent, and the low number of occurrences makes it hard to arrive at conclusive results. Naturally, the results of Experiment 2 are to be compared with actual corpora as well, since completing a gapped text experiment constitutes a linguistic behaviour different from ordinary instances of writing (and, of course, speech).

Experiment 3 is on German and intends to examine the distribution and acceptability of wo-relatives compared to so-relatives. The experiment consists of a production part (similar to Experiment 2) and a grammaticality judgement part (Experiment 1). The aim is to achieve comparable results both with the English experiments and with existing databases, especially the SynAlm (Syntax of Alemannic dialects) database that also used both grammaticality judgement and production tasks on the exact same speakers. The investigation is necessary in order to achieve controlled results for all types of relative clauses (according to syntactic function) and to test so-relatives. Brandner & Bräuning (2013) mention that while so-relatives were not tested directly, some speakers produced them spontaneously. This indicates that this relative clause type is still possible. The hypothesis of Brandner & Bräuning (2013) is that so-relatives and wo-relatives are essentially the same: the difference lies in whether the pronoun is from the d-series (so, etymologically related to ordinary d-pronouns like das ‘that’) or from the w-series (wo). However, the doubling patterns attested with wo (with a d-pronoun) are not known to occur with so (historically or in present-day dialects); this may be a restriction on so-relatives and if so, the distribution of the two elements is sufficiently distinct. Since so-relatives are reported so far only from the Bodensee Alemannic area, the experiment should be carried out there as well (via the University of Konstanz).

Experiment 4 is also on German and focuses on the differences between the distribution of d-pronouns versus w(h)-pronouns in the standard variety. The results also serve as a basis for comparison with Experiment 3. As mentioned before, Standard German allows both wh-pronouns and d-pronouns. The expectation is that asymmetries will be found in terms of the distribution of the less marked pattern (d-pronouns, cf. Brandner & Bräuning 2013), namely that the less marked pattern is more likely to appear lower in the Accessibility Hierarchy than the more marked one. Given that such an asymmetry cannot be drawn back to a syntactic difference like the difference between relative pronouns and complementisers, any attested difference in ratings is primarily due to processing reasons. The differences will be compared to the differences observed between the pronoun strategy and the complementiser strategy in English (Experiment 1 and also Experiment 2): this comparison may reveal something more of the nature of the English asymmetries, i.e. to what extent they can be attributed to processing differences and whether syntactic principles in addition play a significant role.

Experiment 5 is on Dutch. As described by Boef (2013), among others, many Dutch dialects are similar to Standard Dutch in allowing only the pronoun strategy, what is more, only with d-pronouns. Experiment 5 should be run on such speakers and it intends to measure the acceptability of relative clauses in terms of the Accessibility Hierarchy. This provides an optimal testing ground for observing differences purely due to the Hierarchy since in this case no interference from competing strategies is expected to occur. The results will be compared with the acceptability judgement experiments for English (Experiment 1) and German (Experiment 4). This again may provide additional clues as to the importance of the Hierarchy per se in the acceptability of various relative clauses, since some functions may be dispreferred regardless of the employed strategy.

Databases and collaboration

There are various databases to complement the methods mentioned above. The previous project of the principal investigator started a database on relative clauses in the King James Bible and its modernised version. Since the purposes of that project were different from the present proposal, the database so far, with the exception of a small sample, includes only the hits for who and whom in the modern version and their counterparts in the original version. However, the counterparts of which and that should also be considered and the database should be extended accordingly. Taking which into account may cast light upon the differences between relative clauses with human referents and relative clauses with non-human referents, and the distribution of the pronouns who(m)/which between these two categories: the data so far indicate that which was used with human referents even more extensively than in present-day dialects. Taking that into account increases the hits for relative clauses with human referents as well, since this pattern is also possible in modern Standard English (and actually attested in the modernised version); in this way, a more accurate picture regarding the frequencies of the individual types can be drawn. The results arising from this database will be compared with other corpus data, including the two versions of the Wycliffe Bible from the Middle English period.

Concerning other databases and ongoing data collection by other projects, collaboration with other researchers is indispensable. Regarding German dialects, the SyHD atlas on Hessian dialects will be consulted; in addition, two researchers who were involved in SyHD (Prof. Dr. Helmut Weiß and Prof. Dr. Jürg Fleischer) agreed to be collaborative partners on the present proposed project. Likewise, regarding Alemannic, the SynAlm database will be used (collaboration is agreed on with PD Dr. Ellen Brandner, the principal investigator of that project). Data from Mennonite Low German relative clauses were collected by PD Dr. Göz Kaufmann, who has also agreed to collaborate on the present project (apart from Mennonite Low German, this involves the ongoing studies on Pomeranian). Yiddish data will be accessible via the SEYD (Syntax of Eastern Yiddish Dialects) project, led by Dr. Lea Schäfer (a collaborator of the present project). Data on Dutch dialects was collected by the SAND project; collaboration is agreed on with Prof. Dr. Jeroen van Craenenbroeck. As mentioned earlier, data on Afrikaans is scarce: apart from the data collection intended to be carried out by the present project, collaboration is agreed on with Dr. Theresa Biberauer.


The results of the project are to be made accessible in various publications. At least 3 articles should be placed in international journals; the following journals are planned to be targeted: The Journal of Comparative Germanic Linguistics, Glossa: A Journal for General Linguistics, Journal of Linguistics. In addition to publications, the results are planned to be discussed and presented at international conferences and workshops. The events that are planned to be attended are listed in section Finally, the results of the questionnaire on Afrikaans are to be made accessible online via the university library in Konstanz (similarly to how it was done by Bacskai-Atkari & Baudisch 2018 at the University of Potsdam).


The proposed project is to be carried out over a period of 3 years (36 months). Theoretical work will be continuous; the time schedule given in the table below is primarily defined by the collection and evaluation of empirical data, as these constitute the basis for further theoretical investigations, and by the specific topic addressed (as given above).

Period Issues to be investigated
Year 1, m. 1–4 ● designing the questionnaire on Afrikaans
● designing Experiment 1 & Experiment 2 (English)
● pretesting Experiment 1 & Experiment 2 (English)
Year 1, m. 5–8 ● finding informants and distributing the questionnaire on Afrikaans
● running Experiment 1 & Experiment 2 (English)
● designing Experiment 3 & Experiment 4 (German)
Year 1, m. 9–12 ● glossing and evaluating the questionnaire on Afrikaans
● pretesting Experiment 3 & Experiment 4 (German)
● designing Experiment 5 (Dutch)
Year 2, m. 1–4 ● publishing the questionnaire on Afrikaans (webserver)
● running Experiment 3 & Experiment 4 (German)
● pretesting Experiment 5 (Dutch)
Year 2, m. 5–8 ● running Experiment 5 (Dutch)
● examining operators and complementisers in dialect syntax databases
● examining subject/object asymmetries in other (dialect) databases
Year 2, m. 9–12 ● evaluating the results of Experiment 1 & Experiment 2 (English)
● evaluating the results of Experiment 3 & Experiment 4 (German)
● evaluating the results of Experiment 5 (Dutch)
Year 3, m. 1–4 ● extending the database on the King James Bible
● examining German historical corpus data
● writing up an article on complementisers and operators
Year 3, m. 5–8 ● adding the parallel loci from the Wycliffe Bible to the database
● examining English historical corpus data
● writing up an article on subject/object asymmetries
Year 3, m. 9–12 ● examining contemporary English corpus data
● examining contemporary German corpus data
● examining cross-linguistic data from other languages

3 Bibliography

Ackerman, Farrell & Irina Nikolaeva. 2013. Descriptive typology and linguistic theory: A study in the morphosyntax of relative clauses. Stanford, CA: CSLI Publications.

Bacskai-Atkari, Julia. 2014. Cyclical change in Hungarian comparatives. Diachronica 31.4. 465–505.

Bacskai-Atkari, Julia. 2016. On the diachronic development of a Hungarian declarative complementiser. Transactions of the Philological Society 114.1. 95–116.

Bacskai-Atkari, Julia. 2018a. Deutsche Dialekte und ein anti-kartografischer Ansatz zur CP-Domäne. In: Augustin Speyer & Philipp Rauth (eds.) Syntax aus Saarbrücker Sicht 2: Beiträge der SaRDiS-Tagung zur Dialektsyntax, 9–29. Stuttgart: Steiner.

Bacskai-Atkari, Julia. 2018b. Clause typing in main clauses and V1 conditionals in Germanic. In: Mailin Antomo & Sonja Müller (eds.) Linguistische Berichte (LB) Sonderhefte 25: Non-canonical verb positioning in main clauses, 179–200. Hamburg: Helmut Buske Verlag.

Bacskai-Atkari, Julia. 2018c. Marking finiteness and low peripheries. Huba Bartos et al. (eds.) Boundaries crossed, at the interfaces of morphosyntax, phonology, pragmatics and semantics, 183–198. Dordrecht: Springer.

Bacskai-Atkari, Julia. 2018d. The relative cycle in Hungarian declaratives. In: Ana Maria Martins & Adriana Cardoso (eds.) Word order change, 68–87. Oxford: Oxford University Press.

Bacskai-Atkari, Julia. 2018e. Doubly Filled COMP in Czech and Slovenian interrogatives. In: Denisa Lenertová et al. (eds.) Advances in formal Slavic linguistics 2016. Berlin: Language Science Press.

Bacskai-Atkari, Julia. 2018f. Deletion phenomena in comparative constructions: English comparatives in a cross-linguistic perspective. Berlin: Language Science Press.

Bacskai-Atkari, Julia. 2019. Syntactic features and clause typing in Middle English polar questions. In: Merja Stenroos et al. (eds.) Current explorations in Middle English, 137–153. Berlin: Peter Lang.

Bacskai-Atkari, Julia. 2020a. Changes affecting relative clauses in Late Modern English. In: Merja Kytö & Erik Smitterberg (eds.) Late Modern English: Novel encounters, 91–115. Amsterdam: John Benjamins.

Bacskai-Atkari, Julia. 2020b. Non-degree equatives and reanalysis: A case study of doubling patterns in German and Hungarian. In: Veronika Hegedűs & Irene Vogel (eds.) Approaches to Hungarian 16: Papers from the 2017 Budapest conference, 5–23. Amsterdam: John Benjamins.

Bacskai-Atkari, Julia. 2020c. Doubly Filled COMP und polare Fragen im Westgermanischen. In: Augustin Speyer & Julia Hertel (eds.) Syntax aus Saarbrücker Sicht 3: Beiträge der SaRDiS-Tagung zur Dialektsyntax, 213–232. Stuttgart: Steiner.

Bacskai-Atkari, Julia. 2020d. German V2 and Doubly Filled COMP in West Germanic. The Journal of Comparative Germanic Linguistics 23(2). 125–160.

Bacskai-Atkari, Julia. 2020e. English relative clauses in a cross-Germanic perspective. Nordlyd 44. 93–115.

Bacskai-Atkari, Julia. 2021a. Degree semantics, polarity, and the grammaticalisation of comparative operators into complementisers. In: Jóhannes Gísli Jónsson & Thórhallur Eythórsson (eds.) Syntactic features and the limits of syntactic change, 15–39. Oxford: Oxford University Press.

Bacskai-Atkari, Julia. 2021b. Verdoppelungen in Relativsätzen und der Status von Relativpartikeln in Dialekten des Deutschen. In: Augustin Speyer & Julia Hertel (eds.) Syntax aus Saarbrücker Sicht 4. Beiträge der SaRDiS-Tagung zur Dialektsyntax, 11–30. Stuttgart: Steiner.

Bacskai-Atkari, Julia. to appear a. Discourse-driven asymmetries between embedded interrogatives and relative clauses in West Germanic. In: Nicholas Catasso, Marco Coniglio & Chiara De Bastiani (eds.) Language change at the interfaces: Intrasentential and intersentential phenomena. Amsterdam: John Benjamins.

Bacskai-Atkari, Julia. to appear b. Information structure, functional left peripheries, and the history of a Hungarian interrogative marker. In: Barbara Egedi & Veronika Hegedűs (eds.) Functional heads across time: Syntactic reanalysis and change. Oxford: Oxford University Press.

Bacskai-Atkari, Julia. to appear c. Complementisers, word order and a non-cartographic approach to the CP-domain. In: Ángel J. Gallego and Dennis Ott (eds.) Cartography and explanatory adequacy. Oxford: Oxford University Press.

Bacskai-Atkari, Julia & Lisa Baudisch. 2018. Clause typing in Germanic: A questionnaire and its results. Potsdam: University of Potsdam.

Bayer, Josef & Ellen Brandner. 2008. On wh-head-movement and the Doubly-Filled-Comp Filter. In Charles B. Chang & Hannah J. Haynie (eds.), Proceedings of the 26th West Coast Conference on Formal Linguistics, 87–95. Somerville, MA: Cascadilla Proceedings Project.

Bennis, Hans & Liliane Haegeman. 1984. On the status of agreement and relative clauses in West-Flemish. In Wim de Geest (ed.), Sentential complementation: Proceedings of the International Conference Held at UFSAl, Brussels June, 1983, 33–53. Dordrecht: Foris Publications.

Boef, Eefje. 2010. The structure of embedded CP-domains: Evidence from Dutch dialects (and new questions for research on language in aphasia). Paper presented at: Aphasia and Syntactic Movement, Aarhus University, Denmark, September 10, 2010.

Boef, Eefje. 2013. Doubling in relative clauses: Aspects of morphosyntactic microvariation in Dutch. Utrecht: LOT.

Brandner, Ellen & Iris Bräuning. 2013. The particle wo in Alemannic: Only a complementizer? Linguistische Berichte 234. 131–169.

Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz (ed.), Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press.

Chomsky, Noam. 2004. Beyond explanatory adequacy. In Adriana Belletti (ed.), The cartography of syntactic structures 3: Structures and beyond, 104–131. Oxford: Oxford University Press.

Chomsky, Noam. 2008. On phases. In Robert Freidin et al. (eds.), Foundational issues in linguistic theory: Essays in honor of Jean-Roger Vergnaud, 133–166. Cambridge, MA: MIT Press.

Chomsky, Noam & Howard Lasnik. 1977. Filters and control. Linguistic Inquiry 8(4). 425–504.

Comrie, Bernard. 2002. Rethinking relative clause types. In Paolo Ramat & Thomas Stolz (eds.), Mediterranean languages, 87–98. Bochum: Universitätsverlag Brockmeyer.

Comrie, Bernard & Tania Kuteva. 2005. Relativization strategies. In Martin Haspelmath et al. (eds.), The world atlas of language structures, 494–501. Oxford: Oxford University Press.

Comrie, Bernard & Tania Kuteva. 2013. Relativization strategies. In Matthew S. Dryer & Martin Haspelmath (eds.), The world atlas of language structures online, Leipzig: Max Planck Institute for Evolutionary Anthropology. Available at (last accessed: 15 August 2018).

Coniglio, Marco. 2019. Relative clause marking in historical German. Linguistische Berichte 258. 139–177.

Den Besten, Hans. 2012. Afrikaans relative wat and West-Germanic relativization systems. In Tom van der Wouden (ed.) Roots of Afrikaans: Selected writings of Hans Den Besten, 41–59. Amsterdam: John Benjamins.

Fleischer, Jürg. 2004a. A typology of relative clauses in German dialects. In Bernd Kortmann (ed.), Trends in linguistics: Dialectology meets typology, 211–243. Berlin: Mouton de Gruyter.

Fleischer, Jürg. 2004b. Zur Typologie der Relativsätze in den Dialekten des Deutschen. In Franz Patocka & Peter Wiesinger (eds.), Morphologie und Syntax deutscher Dialekte und historische Dialektologie des Deutschen, 60–83. Vienna: Edition Praesens.

Fleischer, Jürg. 2016. Relativsatz-Einleitung. In SyHD-atlas. Available at (last accessed: 16 May 2019).

Gelderen, Elly van. 2004. Grammaticalization as economy. Amsterdam: John Benjamins.

Gelderen, Elly van. 2009. Renewal in the left periphery: Economy and the complementiser layer. Transactions of the Philological Society 107(2). 131–195.

Gisborne, Nikolas & Robert Truswell. 2017. Where do relative specifiers come from? In Eric Mathieu & Robert Truswell (eds.), Micro-change and macro-change in diachronic syntax, 25–42. Oxford: Oxford University Press.

Haegeman, Liliane. 1992. Theory and description in generative syntax: A case study in West Flemish. Cambridge: Cambridge University Press.

Haspelmath, Martin. 2009. Welche Fragen können wir mit herkömmlichen Daten beantworten? Zeitschrift für Sprachwissenschaft 28. 157–162.

Herrmann, Tanja. 2005. Relative clauses in English dialects of the British Isles. In Bernd Kortmann (ed.), A comparative grammar of British English dialects: Agreement, gender, relative clauses, 21–124. Berlin: Mouton de Gruyter.

Hoekstra, Eric. 2015. A historical note on the (in)declinable relative pronoun and its syntactic functions. It Beaken 77 (3–4). 165–174.

Jäger, Agnes. 2010. Der Komparativzyklus und die Position der Vergleichspartikeln. Linguistische Berichte 224. 467–493.

Keenan, Edward L. & Bernard Comrie. 1977. Noun phrase accessibility and universal grammar. Linguistic Inquiry 8(1). 63–99.

Keenan, Edward L. & Sarah Hawkins. 1987. The psychological validity of the Accessibility Hierarchy. In Edward E. Keenan (ed.), Universal grammar: 15 essays, 60–85. London: Croom Helm.

König, Ekkehard. 2015. Manner deixis as source of grammatical markers in Indo-European languages. In Carlotta Viti (ed.), Perspectives on historical syntax, 33–60. Amsterdam: John Benjamins.

Meyer, Jean-Jacques. 2016. A nominal shell analysis of restrictive relative clause constructions in Afrikaans. Stellenbosch University MA thesis.

Miller, Jim. 1993. The grammar of Scottish English. In James Milroy & Lesley Milroy (eds.), Real English, 99–138. London: Longman.

Montgomery, Michael & Guy Bailey. 1991. In which: A new form in written English. American Speech 66. 147–163.

Polinsky, Maria. 2010. Linguistic typology and formal grammar. In Jae Jun Song (ed.), The Oxford handbook of typology, 650–665. Oxford: Oxford University Press.

Romaine, Suzanne. 1982. Socio-historical linguistics. Cambridge: Cambridge University Press.

Tagliamonte, Sali, Jennifer Smith & Helen Lawrence. 2005. No taming the vernacular! Insights from the relatives in northern Britain. Language Variation and Change 17. 75–112.

Romaine, Suzanne. 1984. Towards a typology of relative clause formation strategies in Germanic. In Jacek Fisiak (ed.) Historical syntax, 437–470. Berlin: De Gruyter.

Salzmann, Martin. 2017. Reconstruction and resumption in indirect A'-dependencies: On the syntax of prolepsis and relativization in (Swiss) German and beyond. Berlin: Mouton de Gruyter.

Schütze, Carson T. 2016. The empirical base of linguistics: Grammaticality judgments and linguistic methodology. Berlin: Language Science Press.

Steriade, Donca. 2000. Paradigm uniformity and the phonetics/phonology boundary. In Michael Broe & Janet Pierrehumbert (eds.), Papers in laboratory phonology V: Acquisition and the lexicon, 313–334. Cambridge: Cambridge University Press.

Vanacker, Valeer Frits. 1948. Syntaxis van het Aalsters dialect. Tongeren: Michiels.

Vikner, Sten. 1991. Relative der and other C° elements in Danish. Lingua 84. 109–136.

Wanner, Eric & Michael Maratsos. 1978. An ATN approach to comprehension. In Morris Halle, Joan Bresnan & George Miller (eds.), Linguistic theory and psychological reality, 119–161. Cambridge, MA: MIT Press.

Weiß, Helmut. 2013. Satztyp und Dialekt. In Jörg Meibauer et al. (eds.), Satztypen des Deutschen, 764–785. Berlin: Walter de Gruyter.

Zwart, Jan-Wouter. 2000. A head raising analysis of relative clauses in Dutch. In Artemis Alexiadou (ed.), The syntax of relative clauses, 349–385. Amsterdam: John Benjamins.