Dialogue & Discourse 15(1) 45-76 doi: 10.5210/dad.2024.102 ©2024 Derya Çokal and Klaus von Heusinger This is an open-access article distributed under the terms of a Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/). 45 German Demonstrative Pronouns in Contrast Derya Çokal DCOKAL@UNI-KOELN.DE University of Cologne Klaus von Heusinger KLAUS.VONHEUSINGER@UNI-KOELN.DE University of Cologne Editor: Manfred Stede Submitted 12/2023; Accepted 07/2024; Published online 08/2024 Abstract German has two demonstrative pronouns: the der, die, das paradigm and the dieser, diese, dies(es) paradigm. Previous studies mainly compared the anaphoric use of der with the personal pronoun er and observed that der refers to less prominent antecedents. However, there are only very few studies that have investigated the differences between these two demonstrative pronouns. We hypothesize that they differ in signaling topic persistence and in accessing contrastive antecedents. We tested these hypotheses in short texts that manipulated the contrast of the antecedent by inducing the expression ‘in contrast to’ vs. ‘together with’ (e.g., the cellist in contrast to the flautist vs. the cellist together with the flautist). Results from our eye-tracking reading Experiment (Experiment 1), in which participants’ eye- movements were monitored while reading sentences, show that (i) readers preferred dieser when referring to the topic of a sentence, and (ii) dieser caused less processing difficulties than der in both contrast and no-contrast contexts. Our sentence completion Experiment (Experiment 2) also confirmed that der and dieser are both used for anaphoric reference to a topical antecedent. Collectively, our experiments provide evidence that dieser functions as inducing topic persistence. These results suggest that there is a need for further experimental investigation into the semantic factors and informational structures influencing the usage of demonstrative pronouns in German. Keywords: demonstratives, anaphora, online reading, contrast, prominence. 1 Introduction Discourse management encompasses the organization and flow of information in a conversation or written text, and so plays a crucial role in effective communication. Effective discourse management relies on establishing and maintaining clear referential structures to avoid ambiguity. Referential structures deal with how referential expressions refer to entities in discourse. We can refer to entities/individuals in discourse by personal or demonstrative pronouns. In particular, German includes various types of demonstratives that can be utilized anaphorically. The most prevalent ones are the demonstratives from the der paradigm (i.e., der, die, das – also known as ‘d- pronouns’) and those from the dieser paradigm (i.e., dieser, diese, dieses – ‘dem-pronouns’). For example, in (1) a speaker can say: (1a) Ich habe einen Perkussionisten und einen Gitarristen auf der Bühne gesehen. I saw a percussionist and a guitarist on the stage. http://creativecommons.org/licenses/by/3.0/ ÇOKAL AND VON HEUSINGER 46 (1b) Der+ Perkussionist / er / der / dieser/ hat sehr fleißig gearbeitet. The percussionist / he / that one / this one/ worked very diligently. When referring to the percussionist or guitarist introduced in (1a), one can use a definite noun phrase (der Perkussionist), a personal pronoun (er), or one of the two d-pronouns (e.g., der vs. dieser)1. With respect to (1b), scholars have been trying to determine which anaphoric form is selected and what factors play a role in the selection process. The most studied factors are structural elements such as grammatical role (subject vs. object vs. other arguments), linear position of antecedents (first vs. last), topichood (topic vs. non-topic), and thematic role (agent vs. patient) – (See the section on structural factors below.). In addition, der has been extensively investigated and specifically compared to the personal pronoun er using acceptability judgment tasks, sentence completion experiments, forced-choice tasks (e.g., Bader & Portele, 2019; Bouma & Hopp, 2007; Schumacher et al., 2016), corpus data (Bosch et al., 2003), self-paced reading, (Bosch & Umbach, 2007), visual world paradigm studies (Ellert, 2013; Schumacher et al., 2017; Wilson, 2009), ERP studies (Schumacher et al., 2015; Repp & Schumacher, 2023), and an eye-tracking reading study (Patterson & Schumacher, 2020). A common result across these studies is that while der refers to less prominent antecedents, er refers to more prominent antecedents (e.g., potential antecedents of der: object in Kaiser, 2011, or recent/last-mentioned entities, anti-topic bias in Bosch & Hinterwimmer, 2016; Bosch & Umbach, 2007; Wilson, 2009; and potential antecedents of er: topic, subject, first-mentioned entity in Bosch et al., 2003; Bouma & Hopp, 2007). Furthermore, while the antecedent choice of er is flexible, the antecedent of der is not (Bader & Portele, 2019; Kaiser, 2011; Schumacher et al., 2015; 2016; 2017). It is worth noting that the comparison of the demonstrative pronouns der and dieser has received far less attention with online processing methods (e.g., eye-tracking reading or ERP experiments). Dieser has been primarily investigated using offline studies: (i.e., Abraham 2002 (Centering Theory); Ahrenholz, 2007; Weinert 2007 (corpus studies); Fuchs & Schumacher, 2020 (sentence completion task); Patil et al. 2020; 2023 (forced-choice and acceptability judgement tasks); Patterson & Schumacher, 2021 (acceptability judgement tasks); Patterson & Schumacher, 2023 (story completion tasks); Zifonun et al., 1997; Wöllstein et al., 2022 (in grammar); as well as Diessel, 1999; Himmelmann, 1997 (more typological research)). When discussed in sentences with two arguments (e.g. subject vs. object), the pronominal use of der and dieser is assumed to signal a topic shift. According to Zifonun et al. (1997), Abraham (2002); and Bosch et al. (2007), der and dieser refer to a non-topic antecedent that introduced a referent that is taken up by the demonstrative as a topic in the current sentence, typically in the first position of the sentence. However, there is no explicit research on the contrastive or topic persistent functions of demonstrative pronouns in German. To fill this gap, we focus on the pronominal use of der and dieser, rather than on their determiner uses (i.e., prenominal use). We specifically focus on the masculine singular forms der/dieser, since these forms are fully specified for number and case, while the feminine forms die/diese are underspecified for case and number, which is not within the scope of the current study. Moreover, to the best of our knowledge, there are still no studies that have examined the online reading processing of der and dieser in contrast and no-contrast contexts. By using both online and offline methods (i.e., eye-tracking reading and sentence completion experiments), the current study fills this gap and makes several novel contributions. Firstly, we employ eye-movement recording during reading to examine the processing of der and dieser in contrast and no-contrast contexts. Secondly, since previous studies have explored the thematic roles of verbs (i.e., agent vs. patient) 1 Note that the use of der as in der Perkussionist can be either the definite article der, die, das or the demonstrative determiner der, die, das. Both paradigms partly overlap, but show different forms for the genitive singular, genitive plural and dative plural (see Wöllstein et al., 2022: 744, §1297). GERMAN DEMONSTRATIVE PRONOUNS 47 in the antecedent selection of demonstratives and personal pronouns, we contribute new results with respect to der and dieser. Our study explores antecedent preferences of der and dieser when all parameters (e.g., thematic roles, recency, and animacy) are kept the same but the contrastiveness of the subject and topic of the previous sentence are controlled. Such investigation provides a deeper understanding of the discourse and referential functions of demonstratives and how prominence – boosted by contrast – affects the selection of antecedents of der and dieser. 2 Review of Literature Before delving into our experimental results, we give a brief review of previous studies that investigated the status of noun phrases in anaphora processing and production, including concepts such as the first-mentioned entity and the recency effect. Following this, we discuss the outcomes of studies focusing specifically on the d-pronouns der and dieser. 2.1 Structural factors in anaphora processing Researchers have determined the factors that play a role in anaphora processing, including structural factors (e.g., first-mention advantage, subjecthood, parallelism, last-mentioned entity: Arnold et al., 2000; Gordon, et al., 1993; Stevenson et al., 1994; 1995; 2000) and semantic factors (e.g., implicit causality, animacy: Järvikivi et al., 2017; Pyykkönen & Järvikivi, 2010; van den Hoven & Ferstl, 2018). It has long been recognized that a recency effect results in ‘more recently introduced’ entities being more likely antecedents of referential expressions (Hobbs, 1979). However, choosing the most recent mention is not necessarily the most effective strategy in anaphora resolution. Previous studies also show a first-mention advantage, indicating the first- mentioned noun phrase is the preferred antecedent of an ambiguous pronoun (e.g., Arnold et al., 2000; Gordon, et al., 1993). In addition, while the results of Kaiser and Trueswell’s (2004) eye- tracking study that examined the temporal progression of the first-mention effect initially revealed an inclination towards recency, this preference later shifted to the first-mentioned character. In a subsequent eye-tracking reading study, Fukumura and van Gompel (2015) demonstrated that the influence of mention order on pronoun comprehension could depend on the type of referential expression. For instance, the personal pronoun er in German has been shown to refer to the subject (e.g., Bouma & Hopp, 2007; Schumacher et al., 2016), whereas der refers to the object (Schumacher et al., 2015; 2016). Another hypothesis, supported by numerous researchers, suggests that an entity’s prominence is influenced by its syntactic function rather than its sentence position. According to the subject preference account, prominence derives from grammatical role, with entities in a subject position being more salient than those in other grammatical roles (Arnold et al., 2000; Crawley et al., 1990; Gordon et al., 1993; Grosz et al., 1995).2 While providing an extensive characterization of topicality or subject matter is not within the scope of this paper, it is important to acknowledge that numerous researchers have pointed out that subject position often hosts topical entities (e.g., Chafe, 1976; Gundel, 1988; Lambrecht, 1994; Reinhart, 1981). The topic is what the sentence is about (Reinhart, 1981). A relationship between topicality and givenness/pronominalization has been proposed (e.g., Ariel, 1990; Beaver, 2004). This implies a link between being a suitable antecedent for a subsequent pronoun and being topical. For instance, in the constructions ‘Peter in contrast to Paul came early’ or ‘Peter together with Paul came early’, Peter is in the subject and topic position whereas the second argument Paul is in a prepositional phrase. Paul holds a lower position in terms of grammatical role hierarchies compared to Peter. Additionally, the use of contrastive construction further contributes to a higher prominence level for Peter than Paul. Peter rather than Paul is likely to be a more prominent antecedent for referential expressions. As a result, one could deduce that 2 We use the term saliency and prominence in the same way (cf, von Heusinger & Schumacher 2019 for further discussion). ÇOKAL AND VON HEUSINGER 48 topicality functions as the overarching influence that shapes the prominence of a referent and steers pronoun interpretation by narrowing possible antecedent options.3 Therefore, to the best of our knowledge there is still a lack of understanding of how contrastive structure (in contrast to) and topic persistence affect the online processing and production of der and dieser. Before discussing the existing theories and accounts concerning the pronominal use of der and dieser (see section 2.3), we will briefly overview eye-tracking studies for demonstratives. 2.2 Eye-tracking reading studies on demonstratives To our knowledge, there have been very few eye-tracking reading studies that have examined the online processing of demonstrative pronouns and compared them with the processing of singular pronouns (e.g., it) or personal pronouns (e.g., he) (cf. review article on demonstratives Peeters et al., 2021). In this section, we specifically focus on eye-tracking reading studies on demonstratives, excluding visual world paradigm studies due to differences in eye-movement measures and data analysis (e.g., visual world paradigm studies: Brown-Schmidt et al., 2005; Ellert, 2013; Wilson, 2009). The eye-tracking reading studies have examined whether different referential expressions encode the distinct cognitive status of the intended referent in the mind of the addressee (Çokal et al., 2014; 2018; Patterson & Schumacher, 2020). For instance, to examine whether singular pronoun (it) and demonstrative (this) signal different procedural instructions, Çokal et al. (2018) ran eye-tracking reading experiments. They used three eye-movement measures (i.e., regression path times, second pass reading times, and total time) and four regions of interest (i.e., context, pronoun, disambiguation, and spillover). The interaction between referential expression type (it vs. this) and antecedent types (concrete entity vs. proposition) was observed in late measures such as second pass reading times and total time in the context region. Similarly, the same interaction pattern was also evident in the regression path times (i.e., early measures) for the disambiguation region. Their results were robust across eye-movement measures (early and late measures) and areas of interests (i.e., context and disambiguation). In another study, Çokal et al., (2014) explored whether two English demonstratives (i.e., this and that) direct readers’ attention to different segments of a written discourse (i.e., adjacent frontier vs. distant frontier).4 To address this, they ran two eye-tracking reading experiments, with five regions of interest and three eye-movement measures (i.e., early measures: first pass reading and regression path times; late measure: second pass reading times). Again, robust findings (i.e., this and that more readily access the adjacent frontier) were observed across early and late eye-movement measures and areas of interests (i.e., anaphora and disambiguation regions). While the previous two studies focused on English demonstratives and singular pronoun, there is another eye-tracking reading study (Patterson & Schumacher, 2020) that examined the processing of German personal pronoun (er) and demonstratives (der). Patterson and Schumacher (2020) predicted that in both accusative and dative items, it would be easier to process er when it refers to NP1 (proto-agent) compared to NP2 (proto-patient). Conversely, it should be easier to process der when it refers to NP2 compared to NP1. In Experiment 1, they had two regions of interest (i.e., pronoun and spill-over regions) and four eye-movement measures (i.e., early 3 It is also worth mentioning that antecedents joined by with showed no plural and a singular entity preference in a sentence completion experiment (Hielscher & Musseler, 1990). Similarly, Moxey et. al (2004) observed that the context ‘X with Y’ allows individuals to be focused upon entities separately, and thus resulted in few plural continuations in the with condition. 4 The adjacent frontier-only hypothesis uses Polanyi’s (1986) left-frontier/right-frontier distinction. Polanyi uses “right frontier” to refer to the clause or group of contiguous clauses immediately adjacent to the referential expression, which can thus be argued to be salient; the “left frontier” refers to a clause or clauses separated from the deictic expression by intervening clauses or units (i.e., by the right frontier) and can thus be argued to be less salient. GERMAN DEMONSTRATIVE PRONOUNS 49 measures: first-pass times and right-bound reading times and late measures: re-reading times and total times). In Experiment 2, they had three regions of interest (i.e., pronoun, buffer, and repeated NP regions) but used the same eye-movement measures as in Experiment 1. In both experiments, der took longer to read than er in all eye-movement measures in the pronoun region. Additionally, irrespective of anaphora type, in both first-pass times and right-bound reading times in the repeated NP region, NP2 conditions – specifically with the personal pronoun – had longer reading times/fixation times than NP1 conditions. The precise timing for when prominence (i.e., NP1 reference preference) is used to resolve ambiguity was not determined. Overall, these studies employed meticulous experimental designs and analysis methods, incorporating a minimum of three eye-movement measures and examining between two to five regions of interest depending on their hypotheses. The studies examined the effects in both early and late eye-movement measures because each measure signals different linguistic and cognitive processing strategies (cf. Rayner & Liversedge, 2012 for linguistic and cognitive influences on eye movements in reading). One common challenge in anaphora resolution in eye-tracking reading experiments is discerning a clear and robust interaction pattern across all eye-movement measures and areas of interest. In addition, capturing the precise region where disambiguation occurs can be quite challenging (cf. Patterson & Schumacher, 2020). However, this does not diminish the insightful information provided by eye-movement studies. In fact, such studies elucidate how linguistic elements interact with discourse structure and cognitive processes in language comprehension, highlighting the complexity of cognitive mechanisms involved. 2.3 Existing work on pronominal use of two demonstratives in German In addition to several studies that have focused on English demonstratives (Çokal et al., 2014; 2018; Fossard et al., 2012; Maes et al., 2022a; 2022b), there has been behavioral research on demonstrative determiners (der and dieser +NP) in written discourse in German (Bader et al., 2022; Fuchs & Schumacher, 2020; Patil et al., 2020). It has been claimed that dieser is conventionally limited to selecting a noun phrase at the end of a clause as its antecedent, since demonstratives generally avoid the most prominent discourse referents as antecedents (Hinterwimmer & Bosch, 2018; Patterson & Schumacher 2021; Zifonun et al., 1997). Specifically, Patil et al. (2020) focused on transitive or ditransitive sentences with distinct grammatical and thematic roles and used a forced-choice task with a drop-down menu to select der, dieser, and er. They demonstrated that dieser can refer to NPs at the beginning of sentences. However, this occurs particularly when the object comes before the subject in the preceding context sentence. Consequently, Patil et al. (2020) proposed that dieser exhibits a preference for an object reference. Similarly, in a sentence completion experiment, Fuchs and Schumacher (2020) investigated whether er, der, and dieser bring arguments with different grammatical or thematic roles in a preceding discourse into focus. To test this, they provided a sentence with two arguments (i.e., proto-agents and proto-patient) for two verb types: (1) active accusative verbs (subject/ proto-agent and direct object/proto-patient) and (2) dative experiencer (i.e., indirect object/proto-agent and subject/proto-patient) as in (2a) and (2b). (2a) example for an active accusative verb Jeden Morgen hat der Pfleger den Heimbewohner gekämmt. Dabei hat er/der/dieser oft... Every morning, thenom (male) nurse combed theacc (male) resident. During this process, he/he.DEM /he.DEM often... (2b) example for a dative experiencer verb Im Hafen ist dem Segler der Urlauber aufgefallen. Wenig später hat er/der/dieser dann... At the harbour, thedat (male) sailor noticed thenom (male) tourist. Shortly afterwards, he/he.DEM/he.DEM then.... ÇOKAL AND VON HEUSINGER 50 They observed that in 65% of cases, the personal pronoun er refers to the first-mentioned entity, while both demonstrative pronouns tend to refer to the second-mentioned entity (73% for der and 74% for dieser). It should be noted that in both types of sentences, the first-mentioned entity exhibits a high level of agentivity involving either an agent or experiencer as opposed to a proto- patient. In addition, in Fuchs and Schumacher’s (2020) study, the second mentioned entity came immediately before anaphoric expressions. This suggests that the proximity of the second entity (i.e., recency effect) to the demonstrative pronouns could have an impact. However, the preference of a second-mentioned referent for demonstrative pronouns may not always be observed. For instance, in an online electroencephalogram (EEG) experiment, Repp and Schumacher (2023) investigated the processing of the d-pronoun der and personal pronouns in naturalistic discourse contexts (specifically during story listening). Their analysis of the Tschick corpus highlighted the unexpected characteristics of the antecedents of der (such as their occurrence as subjects and agents), suggesting the prominent role of the perspectival center in their corpus. Consequently, like personal pronoun (e.g., er) der can also refer to subjects/agent antecedents. Repp and Schumacher (2023) pointed out that their findings on der dramatically deviate from the observations in previous studies (e.g., object antecedent preference in Bosch et al., 2003; 2007; proto-patient antecedents in Schumacher et al., 2016). Moreover, in the EEG experiment, where participants listened to audiobooks, Repp and Schumacher (2023) also found that a biphasic N400- Late Positivity pattern at posterior electrodes emerged in response to d-pronouns compared to personal pronouns. Due to its relative unexpectedness in the context of stories when compared to a personal pronoun, they interpreted that the observed N400 associated with der was indicative of higher processing costs. On the other hand, in a sentence completion experiment, with a subject-experiencer verb (e.g., ‘Bernhard fears the real-estate agent because dieser/der has a bad reputation’), Bader et al. (2022) extended earlier studies, finding that both dieser and der referred to the object referent. In contrast, with an object-experiencer verb (e.g., ‘The real-estate agent frightens Bernhard because dieser/der makes a fraudulent impression’), a majority of references with dieser and der were to the subject referent. Additionally, the two demonstratives predominantly referred to the subject in the case of object-experiencer verbs (Hinterwimmer & Bosch, 2018; Patterson & Schumacher, 2021; Zifonun et al., 1997). Given the research indicating that dieser is conventionally associated with selecting a noun phrase positioned at the end of a clause – especially when the object precedes the subject in the preceding context sentence (Patil et al., 2020) – dieser was observed to refer to the subject in the case of object-experiencer verbs (Bader et al., 2022). To sum up, previous studies have shown that demonstrative pronouns in German are most often associated with a recently mentioned entity/second-mentioned entity as an antecedent compared to the personal pronoun er. However, a recent EEG study has shown that subjects’ preference for the second-mentioned entity is not always observed for der (Repp & Schumacher, 2023), and thematic roles are also significant in determining references of der/dieser to the object/subject (the subject/first NP references in Bader et al., 2022). These findings provide a nuanced understanding of how these two demonstratives are used in different situations, such as verb semantics. 2.4 Topic persistence with demonstrative pronouns and demonstrative determiners Extending this summary of recent research on demonstrative pronouns to demonstrative determiners in complex demonstratives (e.g., that book), complex demonstratives are distinguished from definite noun phrases (e.g., the book) in their anaphoric use by the following properties: First, they express contrast to other referents of the same type. They also express an anti- uniqueness/contrast condition, evidenced by the ungrammaticality of this sun, this pope in a neutral context (Ahn, 2019; Bisle-Müller, 1991; Bosch & Umbach, 2007; King, 2001; Wolter, 2006). Second, demonstrative determiners are topic shifters (i.e., they refer to a non-topical antecedent and promote it to the topic of the current clause (Bosch & Umbach, 2007; Diessel, 1999; van GERMAN DEMONSTRATIVE PRONOUNS 51 Kampen, 2007; Zifonun et al., 1997). Third, they signal topic persistence, which is understood as a primitive form of Givón’s (1983) concept of topic continuity. Topic persistence means the cataphoric or forward-looking function (Fuchs & Schumacher, 2020) of demonstratives is best investigated for indefinite demonstratives (Deichsel & von Heusinger, 2011, Gernsbacher & Shroyer, 1989, Prince, 1981). These patterns are observed more frequently with complex demonstratives with the pronominal dieser than with definite noun phrases with the prenominal der, which is in most cases understood as being the definite article (e.g., the book). From the observations on demonstrative determiners in complex demonstratives, we assume that some functions are similar for demonstrative determiners and demonstrative pronouns – perhaps in a different distribution. Therefore, we assume that the pronominal dieser is used for topic persistence and thus signals that in the subsequent discourse the referent is used as topic. 2.5 Contrast function with demonstratives The semantics of German demonstratives also involves a contrast function. Contrast function, as discussed by Diessel (1999), refers to the capability of demonstratives to distinguish a specific object or person from a group of similar entities (cf. further discussion, Ahrenholz, 2007). This is described as “…pointing out one member of a group” (Anderson & Keenan, 1985, p. 289). Bisle- Müller (1991) also emphasized that the contrast and delimitation functions of demonstrative noun phrases with dieser imply a distinction from other referents, suggesting a sense of “only this one matters, forget about the others.” This distinction goes beyond just spatial functions of demonstratives and signifies a fundamental differentiation. Bisle-Müller’s (1991) perspective on the meaning of dieser is that it serves to address differentiation from other possible referents. All these accounts describe the demonstrative determiner dieser (dieser NP), since a complex demonstrative der NP cannot easily be distinguished from a definite noun phrase der NP. Other researchers have posited that the demonstrative pronoun der is used if there is a contrast in alternation with the non-accented personal pronoun er (Bosch, 1988; Zifonun et al., 1997). Additionally, Bader et al. (2022) propose that both dieser and der demonstratives are capable of indicating contrast. While these studies focused on German pronouns, van Kampen (2007) also examined how referent preferences can also be modulated by contrast in Dutch, German and Swedish. Van Kampen (2007) suggests that demonstratives can only shift a topic referent if that referent holds a contrastive position in the previous sentence. While van Kampen (2007) investigated antecedents in the focus position, one could also test accented antecedents or antecedents that are contrastively marked by explicit expressions. However, it remains uncertain how sensitive der and dieser are to topic contrast context. Consequently, we investigated their usage in both contrast and non-contrast contexts (See current study & design sections below.). 3 The current study Examining demonstratives in such a contrast context will provide new results with respect to how der and dieser are processed in context and how they are integrated into a discourse model. Overall, there is a lack of comprehensive characterization for dieser, and no systematic distinction has been established between der and dieser in terms of their interpretive preferences. Therefore, the current study aims to answer the un-investigated questions: (1) Do we observe any distinction between these two demonstrative pronouns uses? and (2) How does contrast affect the function of der and dieser to find an antecedent and their integration into the discourse model? Above, we argued that previous studies leave open questions of how contrast influences real-time processing of der and dieser. We predict that the functions are not equally distributed over the two demonstrative pronouns: (1) dieser signals topic persistence (2) der signals that the antecedent argument is contrastive. In order to test these two hypotheses, we constructed short texts that allow us to (i) neutralize the parameter of distinct grammatical and thematic roles; (ii) manipulate the contrast of one argument; (iii) neutralize immediate recency, and (iv) see whether demonstrative pronouns can ÇOKAL AND VON HEUSINGER 52 access topics and whether they also continue the previous topic (i.e., topic persistence). We, therefore, designed short texts of the following type: (3) S1: I listened to a cellist and a flautist at a Munich classical music festival. S2: The cellist, in contrast to the flautist /together with the flautist, played a few wrong notes. S3: A few minutes before the opening concert of the event, dieser/der changed the strings skillfully and tuned his instrument. The first sentence (S1) introduces two indefinite noun phrases in direct object position. In the second sentence (S2), the first indefinite noun phrase introduced in S1 is taken up as a definite noun phrase in subject/topic position. The second noun phrase is taken up as a prepositional phrase, either expressing a contrast (NP1 in contrast to NP2) or as a comitative phrase (NP1 together with NP2). On the one hand, the syntactic role of NP2 is clearly degraded in comparison to the NP1, which functions as the subject/topic. On the other hand, NP2 has the same thematic role as it refers to an alternative or comitative expression to NP1. This design enables us to investigate the subtle difference between der and dieser keeping the thematic role stable. The third sentence (S3) starts with the demonstrative pronoun in the subject (and topic) position and then predicates a property that is only coherent with NP1. In other words, we have an ambiguous pronoun that is disambiguated after the verb phrase. To examine subtle differences between the functions of der and dieser, in Experiment 1 (reported below), we used eye-movement recording during reading to examine how der/dieser are processed in German, using contrast (e.g., NP1 in contrast to NP2) and no-contrast contexts (e.g., NP1 together with NP2). Additionally, we ran a sentence-completion experiment (Experiment 2 reported below) to examine participants’ antecedent preferences regarding information structure without time pressure (i.e., der/dieser references to NP1 or NP2 in contrast and no-contrast contexts). 4 Experiment 1 Experiment 1 was an eye-tracking study using a 2 × 2 within-subject, within-item design, crossing contrast type (contrast and no-contrast) with anaphora type (dieser and der) as seen in Example (4) below. (4a) Dieser in contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist im Gegensatz zum Flötisten hat ein paar schiefe Töne gespielt. Einige Minuten vor dem Eröffnungskonzert der Veranstaltung hat dieser die Saiten routiniert ausgewechselt und sein Instrument gestimmt. (4b) Der in contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist im Gegensatz zum Flötisten hat ein paar schiefe Töne gespielt. Einige Minuten vor dem Eröffnungskonzert der Veranstaltung hat der die Saiten routiniert ausgewechselt und sein Instrument gestimmt. (Translations of 4a/4b: At a classical music festival in Munich, I listened to a cellist and a flautist at a Munich classical music festival. The cellist, in contrast to the flautist, played a few wrong notes. A few minutes before the opening concert of the event, dieser/der changed the strings skillfully and tuned his instrument.) GERMAN DEMONSTRATIVE PRONOUNS 53 (4c) Dieser in no-contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist zusammen mit dem Flötisten hat ein paar schiefe Töne gespielt. Einige Minuten vor dem Eröffnungskonzert der Veranstaltung hat dieser die Saiten routiniert ausgewechselt und sein Instrument gestimmt. (4d) Der in no-contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist zusammen mit dem Flötisten hat ein paar schiefe Töne gespielt. Einige Minuten vor dem Eröffnungskonzert der Veranstaltung hat der die Saiten routiniert ausgewechselt und sein Instrument gestimmt. (Translations of 4c/4d: At a classical music festival in Munich, I listened to a cellist and a flautist. The cellist, together with the flautist, played a few wrong notes. A few minutes before the opening concert of the event, dieser/der changed the strings skillfully and tuned his instrument.) 4.1 Predictions Our first prediction is that dieser is a marker for topic persistence (Deichsel & von Heusinger 2011, Gernsbacher & Shroyer, 1989, Prince, 1981). If this principle does play a role, then dieser in (4a/4c) would likely lead to less processing difficulties, resulting in lower odd ratios of regressions-out as well as shorter total times in the disambiguation5 and/or spill-over regions. Both conditions suggest that we talk more about “the cellist” and therefore allow dieser to be processed easily. Our second prediction is that der has a congruent feature of contrast and therefore it can be anaphorically linked to a contrasted item in the previous context. If that is the case, in (4b) der would be preferred to der in (4d) because in (4b) a contrast with respect to a competitive referent is made (i.e., the cellist in contrast to the flautist). Consequently, we expected to observe processing difficulty (i.e., high proportions of regressions-out and longer total times) in (4d) immediately after der, (the disambiguation region), in comparison to der in (4b). Given this experimental design, the presence or absence of topic contrast in the previous context would lead to an interaction (all other things being equal) between contrast type and anaphora type, due to the prediction of a difference between (4b) and (4d) and the absence of a contrast effect in (4a) and (4c). It should be noted that this prediction assumes that der and dieser have a difference in meaning and function. 4.2 Method 4.2.1 Participants Fifty-two paid, native German-speakers aged 22-24 (46 Females and 6 Males) from the University of Cologne participated in the experiment. They were not exposed to another language before they were three years old. All were unaware of the study’s purpose. The experiment involving human participants was reviewed and approved by the Ethics Board of the German Linguistic Society. The participants provided written informed consent to participate in this study. 4.2.2 Apparatus We used an Eyelink 1000 eye-tracker (SR Research Ltd, Ottawa, Canada) in tower-mounted mode, with a chin rest to stabilize each participant’s head. 5 In the disambiguation region, we disambiguated the antecedents of der and dieser by employing a referential expression closely related to NP1 (strings related to cellist). ÇOKAL AND VON HEUSINGER 54 4.2.3 Materials Forty items were created based on Example 4 above. Each item appeared in four conditions, which crossed contrast type (contrast vs. no-contrast) with anaphora type (der vs. dieser). The explicit topic contrast was manipulated using im Gegensatz zu (‘in contrast to’) in the contrast condition and zusammen mit (‘together with’) in the no-contrast condition. Anaphora type was manipulated by including either der or dieser. We disambiguated the antecedents of der and dieser by using referential expressions/noun phrases (e.g., strings) coming after them, thereby disambiguating the demonstrative pronouns (e.g., der/dieser changed the strings – referring to a cellist in example 4). Throughout the experiment, der and dieser were disambiguated to NP1, i.e. the subject and topic of the antecedent sentence (i.e. cellist in the example above. While both referents (i.e., cellist and flautist) were congruent in gender and number with the pronoun, reference to the subject/topic was ensured via the plausibility of the critical sentence (e.g., in example 4 it is only plausible that a cellist would check the strings, not the flautist because a flute does not have strings). References to the subject/topic was done to avoid the recency/last-mentioned entity as a confounding factor for demonstrative reference (Bosch & Hinterwimmer, 2016; Bosch & Umbach, 2007; Kaiser, 2011; Wilson, 2009; Zifonun et al., 1997). In addition, to reduce the recency effect (i.e., the second NP as an alternative antecedent), temporal phrases were also used after the critical sentence (e.g., Einige Minuten vor dem Eröffnungskonzert der Veranstaltung ‘A few minutes before the opening concert of the event’) to put a distance between der/dieser and their possible antecedent options. Except for our manipulations (the cellist & the flautist), there were no alternative antecedents for der or dieser in the previous context. In the spillover region, the length of adverb (routiniert ‘skillfully’) in each item was matching across four conditions. The forty stimuli were distributed into four lists, following a Latin Square procedure. In each list, every item appeared in only one condition, and each condition appeared an equal number of times. Each list was assigned to 10 participants. Additionally, there were 73 fillers with plural pronouns but not personal or demonstrative pronouns in the singular. Since our critical items consisted of first NP references with der and dieser, the plural pronouns in our filler items referred to the second-mentioned NP or its referent, resulting in ambiguity/underspecification (See the filler item below.). In addition, there were three practice items, all of which were similar in length to the experimental sentences. Here is an example of a filler item: (5) Es war in einem Operationssaal eines Krankenhauses in Köln Lindenthal. Die Ärzte standen am Operationstisch. Die Anästhesisten standen daneben. Es war ein typischer arbeitsreicher und produktiver Tag. Sie checkten sorgfältig ihre Aufgaben und erledigten die Arbeit. (It was in an operating room of a hospital in Cologne Lindenthal. The doctors were standing at the operating table. The anesthesiologists stood by. It was a typical busy and productive day. They carefully checked their task and got the job done.) 4.2.4 Pre-testing items We pre-tested our stimuli using Qualtrics with 120 native speakers of German. We provided the initial text and asked them to choose “who would do the action”. A sample pre-test item is as follows: (6) Es gab einen Cellisten und einen Flötisten. Der hat die Saiten routiniert ausgewechselt. Wer hat die Saiten ausgewechselt? (Who changed the strings?) (a) Cellist (b) Flötist GERMAN DEMONSTRATIVE PRONOUNS 55 Our results showed that 75% of referents were selected as NP1, whereas 25% of cases were selected as NP2. Using Qualtrics, we also ran an acceptability judgment task to check that the items and conditions did not differ in their acceptability. The participants did not take part in the previous pre-testing session (n = 46). They rated 40 experimental stimuli and 40 filler items from 1 to 5. We ran a linear mixed-effects model including random intercept for participants and items (intercept: β = 1.028, SE = .144, t = 71.084). While the interaction between anaphora type and contrast type was not significant (β = 0.038, SE = .054, t = .712), there were main effects of anaphora and contrast types, anaphora: β = 0.064, SE = 0.27, t = 2.369; contrast: β = -0.092, SE = 0.27, t = -3.337. Overall, the use of dieser was significantly preferred to der, der with explicit contrast context with M = 3.22, SE = 1.40; dieser with explicit contrast M = 3.36, SE = 1.39; der with no-contrast context M = 2.91, SE = 1.38; Dieser with no-contrast M = 3.33, SE = 1.40. In addition, to a significant degree participants had less of a preference for the use of der with a no-contrast context to der with a contrast context: β = -0.325, SE = 0.111, t = -2.969. The main effect of contrast indicates that participants less preferred contrast context than no-contrast context. Overall, our results indicate that sentences are not considered ungrammatical, receiving an average score of 3 on a scale of 1-5. This was expected given that the demonstrative dieser is often preferred in formal and /or written discourse, while the demonstrative der is preferred in informal or oral discourse (Patil et al., 2023, for further discussion). Moreover, participants might have learned that the use of der is impolite for human referents. Both assumptions would predict a certain preference of dieser over der in our written test items. 4.2.5 Procedures We presented 116 texts in Times New Roman 18 font, in a fixed random order, with no experimental items adjacent to each other. The texts were presented on six or seven written lines, with each line containing between 76 and 85 characters. The critical regions including der/dieser and their referents (i.e., die Saiten/the strings) always appeared near the middle of a line and were not the last sentence on the screen. To familiarize participants with the experimental procedure, the experiment began with three fillers. While viewing was binocular, only the right eye was tracked. The items appeared on a 19” monitor, approximately 70 cm away from the participants’ eyes. In order for the experimenter to check the calibration of each participant, the participant fixated on a black square before each item, indicating the position of the first character of the text. The black square was automatically replaced with text once a stable fixation was detected. After reading each item, the participant pressed a button to indicate the end of the sentence. For 37% of the items, a comprehension question then appeared, which the participant answered by pressing a button on the left or right side of the button box. The comprehension questions never probed an anaphora and the antecedent (‘cellist’; or ‘flautist’). The eye-tracking Experiment took 50 minutes, and included informing participants about consent forms, instructions, and three short breaks. 4.2.6 Data analysis Texts were divided into three regions (see Table 1). Below, we report data for the following regions: anaphora, disambiguation, and spillover. It should be noted there is a length difference in the anaphora region between der (i.e., 3 characters) and dieser (i.e., 6 characters) rendering this main effect in the anaphora region uninterpretable. Additionally, due to the skipping rate of functional words in previous studies (Rayner, 2011), we have typically not found early effects of syntactic or referential processing difficulty in the anaphora region. We predicted that we would observe an interaction between two factors (i.e., anaphora type and type of contrast) in the disambiguation and spill-over regions, which immediately follow the anaphora region. These regions were matched for length in all conditions. ÇOKAL AND VON HEUSINGER 56 Anaphora Disambiguation Spillover der/dieser der/dieser die Saiten the strings Routiniert skillfully Table 1. Regions in Experiment 1 Fixations of less than 80, or more than 1200 ms, were excluded from analysis. The percentages of data points for total time in each region that were excluded are as follows: 21% in the pronoun region, 8% in the disambiguation region, and 5% in the spillover region. 21% of the excluded data points in the pronoun region involve instances where the pronoun was skipped. All participants correctly answered at least 90% of comprehension questions. von der Malsburg and Angele’s (2017) study showed that conducting multiple comparisons (i.e., rate of false positives when dependent measures in eye movements with multiple regions are tested) increased the probability of incorrectly rejecting the null hypothesis (Type 1 error). Since reporting a large number of eye- movement measures might result in false positives due to family-wise errors (von der Malsburg & Angele, 2017), we selected only two eye-movement measures: regression out and total time. We selected regressions-out, which is the proportion of trials where readers looked back from the region to an earlier piece of the text between the time when the region was first entered from the left to the time when the region was first exited to the right. This measure indicates that lexical semantic information is processed during regressions-out and used in recovery from processing difficulty from the previous text where the current word/text does not meet the expectation of readers (cf. further discussion on regressions in reading Sturt & Kwon, 2018). For completeness, we also report total time (i.e., the sum of all fixations in the region) as a general measure of processing, even though this does not provide information about initial processing. In cases where the region received no fixations (total time), the trial was treated as missing data and excluded from analysis. In order to keep false positives in check (Type I error), we conducted the Bonferroni correction (von der Malsburg & Angele, 2017). There were two measures on three regions, then there were 6 analyses. To remove familywise error, we divided our critical value by 6 (i.e., p < 0.05/6). After Bonferroni adjustment, our significance threshold value was p < .008. Then, since we assumed |t| had to be greater than 2 for p < .05, we determined the new |t| value 2.65 (based on Z distribution). A given co-efficient was judged to be significant at α = .008 if the absolute t-value exceeded 2.65. We used the following packages: lme4 to run logistic mixed effects regressions models for the analysis of regressions-out. We used Sjplot to calculate odds ratios, random effects; library (emmeans) to calculate standard errors and confidence Intervals; ggplot2 to graph estimate proportions of each condition. The analysis was based on whether or not participants had regressions-out in the region. Therefore, trials without regressions-out were retained in the analysis. Trials that had no regressions-out were included and coded as 0. We used logistic-mixed effect regression (GLMER). We contrast-coded our fixed effects (i.e., anaphora type and contrast type) and centered the coding around zero (i.e., using -0.5 and 0.5). Random slope parameters corresponding to the two experimental factors and their interactions were included in the maximal model for both participants and items (Barr, Levy, Scheepers, & Tily, 2013; Bates et al., 2015). Below we report odd ratios for the analysis of regressions-out. To aid convergence, and to avoid spurious over- estimates of correlations, random correlation parameters were excluded from the model. In anaphora and disambiguation regions for regressions-out, the model failed to converge, and random slope parameters with the least variance were removed until convergence was achieved. To decide which factors needed to be removed first to achieve convergence, we assessed the variability at GERMAN DEMONSTRATIVE PRONOUNS 57 each level in our data. It should be noted that when there is little variability between the levels of a particular random effect, it may not be necessary to include it in our model. The following models converged for the regressions-out: • Anaphora region: (anaphora type * contrast type +1||subject) + (anaphora type +1|item) + anaphora type * contrast type • Disambiguation region: (anaphora type +1|subject) + (1|item) + anaphora type * contrast type • Spillover region: (1|subject) + (1|item) + anaphora type * contrast type For each region and total time, linear mixed effects regression (LMER) models using lme4 R packages were constructed, incorporating all fixed effects and interactions in a single step. An additional package (plyr) was used to compute mean values. Factor labels were transformed into numerical values and centered prior to analysis. We performed analyses on log-transformed reading times, and all analyses reported below incorporated crossed random intercepts for participants and items. The model did not converge in the anaphora and disambiguation regions for total time. To achieve convergence, random slope parameters with the lowest variance were eliminated. Converged models for each region are as follow: • Anaphora region: (1|subject) + (1|item) + anaphora type * contrast type • Disambiguation region: (contrast type +1|subject) + (1|item) + anaphora type * contrast type • Spillover region: (1|subject) + (1|item) + anaphora type * contrast type After running the above models, for any significant interactions and main effects, we examined the effect of exposure to test items. Specifically, we explored the impact of reading first NP references with der and dieser during the experiment and whether participants adapted the referential patterns in the experiment (see Çokal & Ferreira, 2015 for normalization of pronoun errors during an eye-tracking reading experiment; Johnson & Arnold, 2023 for participants’ adaptation to experimental manipulation/referential patterns within an experiment). To account for this, we added ‘trial order’ as a covariant to our models. Considering the possible effects of nesting, we included participants and items as random intercepts. Random slope was specified for the variable ‘trial order’ within the grouping variable ‘subject’. This random slope allowed the effect of ‘order’ on the outcome variable to vary across different subjects. Each subject may have their own unique relationship between the order of items and their responses. Including random slopes allows the model to capture this variability in individual subject’s responses to the variable ‘order’. While regressions-out are reported with odds ratios, standard errors, and p-values, the results for total times also include coefficients, standard errors, and t-values for each fixed effect and interaction (see Baayen et al., 2008 for absolute t-value in linear-mixed effect models). Data, scripts, and stimuli for all experiments are available at https://osf.io/ubxkr/?view_only=01bc62fb4fd0460f9e7831b0eedfbe4e. 4.3 Results and discussion In this section, we present our results region by region. 4.3.1 Anaphora region Regressions-out did not show main effects of anaphora and contrast types (see Tables 2 & 3, Figure 1). Total time showed a main effect of anaphora type, with longer reading times for dieser than der but this effect was not significant since the absolute t-value did not exceed 2.65 (Dieser: M = 297, ÇOKAL AND VON HEUSINGER 58 SE = 11, Der: M = 272, SE = 14). This effect is likely due to the length differences between dieser and der, as discussed above. The interaction between anaphora and context in total times was not significant (see Tables 2 & 3 below). Regions/parameters Regressions-out Total Time Anaphora Odds Ratio SE p-value β SE t Intercept 0.17 (0.12-0.23) 0.03 .001* 5.70 0.039 142 Contrast 0.97 (0.70-1.35) 0.16 .873 -0.023 0.023 -1.019 Anaphora 0.77 (0.54- 1.11) 0.14 .167 -0.052 0.023 -2.249 Contrast* Anaphora 0.101 (0.53-1.92) 0.33 .976 0.011 0.046 .244 Disambiguation Intercept 0.22 (0.15-0.30) 0.04 .001* 6.065 0.067 89.791 Contrast 0.78 (0.62-0.99) 0.09 .043 -0.027 0.026 -1.025 Anaphora 0.32 (0.25-0.41) 0.04 .001* -0.054 0.021 -2.548 Contrast *Anaphora 1.13 (0.171-182) 0.27 .606 0.026 0.042 .617 Spillover Intercept 0.12 0.01 .001* 5.755 0.041 138 Contrast 1.49 (1.12-1.97) 0.21 .006* 0.022 0.027 .818 Anaphora 0.81 (0.61-1.07) 0.12 .136 0.001 0.019 .050 Contrast* Anaphora 1.37 (0.78-2.40) 0.39 .279 -0.056 0.039 -1.441 Table 2. Results of mixed-effects analysis for regressions-out and total time for Experiment 1. Regressions-out are reported with odds ratios, standard errors, and p-values, the results for total times also include coefficients, standard errors, and t-values for each fixed effect and interaction. After Bonferroni adjustment, statistically significant affects are indicated with asterisk (*). The effects are considered significant when the absolute t-value is greater than 2.65. GERMAN DEMONSTRATIVE PRONOUNS 59 Pronoun Disambiguation Spillover Regressions-out M (SE) M (SE) M (SE) Der, contrast .209 (.034) .288 (.033) .144 (.020) Dieser, contrast .177 (.022) .143 (.017) .132 (.018) Der, no-contrast .208 (.035) .336 (.033) .113 (.014) Dieser, no-contrast .179 (.028) .158 (.024) .080 (.013) Total time Der, contrast 252 (17) 531 (29) 355 (11) Dieser, contrast 293 (15) 511 (25) 340 (12) Der, no-contrast 292 (21) 555 (29) 350 (13) Dieser, no-contrast 302 (16) 509 (24) 356 (15) Table 3. Means (and standard errors) for regressions-out and total time for Experiment 1 4.3.2 Disambiguation region Regressions-out showed a main effect of anaphora but not a main effect of contrast types or a two- way interaction between anaphora type and contrast context (see Tables 2 & 3, Figure 1) 6. The odds ratios with der were higher than those with dieser, which indicates that irrespective of contrast type (i.e., contrast or no-contrast), conditions with der led to more processing difficulties compared to dieser (see Figure 1 & Table 2). In addition, irrespective of anaphora types, no-contrast context led to more regressions-out than the contrast context. However, after Bonferroni correction, contrast-effect was not significant. Our results show that references to a topic with dieser led to less processing difficulties than for der, irrespective of contrast type pairwise comparison of dieser and der with contrast context (β = -1.028, SE = .173, z = -5.921, pairwise comparisons of dieser and der with no-contrast context: β = -1.284, SE = .237, z = -5.408). 6 We ran right-bound reading time (re-reading time as a late measure) and observed the same patterns in regressions-out. (See Appendix A). While anaphora and spillover regions did not show any main effects or a significant interaction between the variables, in the disambiguation region, we observed a main effect of anaphora β = -.0046, SE = .019, t = - 2.326; but no main effect of contrast and an interaction between the two variables (contrast: β = -.018, SE = .019, t = - .944; .anaphora * contrast: contrast: β = -.001, SE = .039, t = -.015). compared to dieser, der, in the disambiguation region, led to longer right bound re-reading times for der irrespective of contrast type contexts. Readers’ initial and late processing strategies for der and dieser did not change. ÇOKAL AND VON HEUSINGER 60 Figure 1. Estimated proportion of regressions-out across regions for Experiment 1. Error bars show the 95% confidence intervals. GERMAN DEMONSTRATIVE PRONOUNS 61 Figure 2. Estimated total times (ms) across regions for Experiment 1. Error bars show the 95% confidence intervals. In the disambiguation region, total time did not show a significant interaction between the two variables (see Tables 2 & 3 and Figure 2). There was a main effect of anaphora, with longer total times with der than dieser but this effect was not significant after the Bonferroni and t-value adjustments. There were no significant main effect of trial order and an interaction between the two variables and trial order (regression-out, β = .011, SE = .006, z = 1.794; total-time: β = .002, SE = .002, z = -.856). 4.3.3 Spillover region Regressions-out showed a main effect of contrast context but no main effect of anaphora type or an interaction between the factors (see Tables 2 & 3, Figure 1). Compared to the no-contrast context, the contrast context led to more regressions-out, showing that contrast contexts were difficult to process irrespective of anaphora type in the spillover region. Regressions-out and total time did not show a two-way interaction (i.e., anaphora * context) or main effects of anaphora type (see Tables 2 & 3, Figure 2). In addition, trial order as a co-variant did not affect the results of regressions-out and total-time (regression-out, β = .001, SE = .007, z = .046; total-time: β = .001, SE = .002, z = .381). ÇOKAL AND VON HEUSINGER 62 4.4 Overall summary In the disambiguation region, the main effect of anaphora in regressions-out indicates that overall, readers had less processing difficulties with dieser. In the initial processing, regardless of contrast type participants preferred dieser over der. The less processing difficulty observed for dieser might be attributed to its topic persistence function in both contrast and no-contrast contexts. This finding appears to support our initial prediction that dieser serves as a marker for topic persistence. Our second prediction was that in the disambiguation region, der in the contrast context would lead to less processing difficulty than der in the no-contrast context, due to the presence of a contrastive antecedent. This assumption is not supported by the findings in the disambiguation region of regressions-out. We observed a pattern where der with the contrast context resulted in lower regressions-out than der with the no-contrast context, although this difference was not statistically significant. Regardless of anaphora type, processing a contrast context resulted in higher regressions-out compared to a no-contrast context in the spillover region. Initially, due to the presence of two alternatives, readers in the disambiguation region might find no-contrast contexts demanding. While these alternatives are bundled with the use of such phrases as ‘together with’, in the disambiguation region they need to ‘unbundle’ them. After accomplishing this, in the spillover region readers may find no-contrast contexts less taxing than contrast contexts, since contrastive information in online processing, emphasizing a particular element or concept in a sentence or discourse, creates a mental ‘placeholder’ for related alternatives (Repp & Spalek, 2021). This placeholder can be connected to likely candidates or alternatives presented after the initial focus. Our results indicate that these alternatives can be generated or considered as the discourse unfolds. Our findings suggest that the effect of contrast contexts in online discourse processing needs to be further tested. In previous studies with offline methods, where participants read two-sentence stories on Amazon Mechanical Turk and answered comprehension questions about the reference of referential expressions, readers adapted referential patterns within the study after exposure to certain structures in the stimuli (Johnson & Arnold, 2023). Similarly, in an online eye-tracking reading study, native speakers of English who encountered pronoun errors (i.e., gender mismatching) normalized the input and resolved the pronoun ambiguity using available relevant information (Çokal & Ferreira, 2015). However, their sensitivity to errors was weaker when the antecedent of a pronoun error was embedded in the density of information and declined after certain exposure to errors (Çokal & Ferreira, 2015). In the current study, we did not observe any trial order effect (i.e. the possibility of learning effects) on our results. It should be noted that since the effect of trial order on the processing of der and dieser was not our main goal in the current study, further studies need to be conducted on this topic. While Experiment 1 focused on participants’ online processing of der and dieser in contrast and no-contrast contexts, in Experiment 2 we explored participants’ offline antecedent preferences for these anaphoric expressions. In this way, we were able to examine which individuals (NP1) and (NP2) were selected as antecedents using these expressions. Experiment 2 also provided further information on whether participants preferred topic references (NP1) with der and dieser. 5 Experiment 2 Experiment 2 tested participants’ antecedent preferences using a sentence completion method approved by the Ethics of Research Committee at University of Cologne. 5.1 Method In this section, we will describe our participants, materials, and data analysis. GERMAN DEMONSTRATIVE PRONOUNS 63 5.1.1 Participants Experiment participants (n =32) were native German-speakers aged 21-23 (24 Females and 8 Males) from Cologne University, and they were compensated for their participation. All were unaware of the study’s purpose, and none had participated in Experiment 1. To recruit participants, we posted fliers to the Faculty of Philosophy at University of Cologne. Our main selection criteria were being a native speaker of German and not being bilingual (defined as not growing up in an environment where more than one language was spoken). 5.1.2 Materials We used the same sentences as in Experiment 1, but we made a slight change to avoid the demonstrative determiner uses of der and dieser in their completions (e.g., der Cellist, der Flötist). While in Experiment 1, the auxiliary hat ‘has’ came before der and dieser (i.e., hat der/dieser ...), in Experiment 2, hat (i.e., der/dieser hat) followed der and dieser (please see below). This forced a demonstrative pronoun usage and disallowed a demonstrative determiner usage. Each participant was provided with an initial context and asked to provide a completion answer for the sentence fragment ending with der hat or dieser hat in a manner consistent with the previous text. In each condition, participants could refer to either the first NP (NP1) (e.g., the cellist) or the second NP (NP2) (e.g., the flautist). One experimental item with its conditions is given below: (7a) /(7b) Dieser/Der in the contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist im Gegensatz zum Flötisten hat ein paar schiefe Töne gespielt. Es waren einige Minuten vor dem Eröffnungskonzert. Dieser / Der hat ... (I listened to a cellist and a flautist at a Munich classical music festival. The cellist, in contrast to the flautist, played a few wrong notes. It was a few minutes before the opening concert of the event. Dieser / Der has …) (7c) /(7d) Dieser/Der in the no-contrast context Bei einem Münchener Klassikfestival habe ich einem Cellisten und einem Flötisten zugehört. Der Cellist zusammen mit dem Flötisten hat ein paar schiefe Töne gespielt. Es waren einige Minuten vor dem Eröffnungskonzert. Dieser / Der hat ... (At a classical music festival in Munich, I listened to a cellist and a flautist. The cellist, together with the flautist, played a few wrong notes. It was a few minutes before the opening concert of the event. Dieser / Der has …) There were 40 experimental and 60 filler sentences. There were four experimental conditions as per Experiment 1. Four lists were constructed, using Latin Square counterbalancing. In each list, each sentence appeared in only one condition, with an equal number of items from each condition. Sentences were presented in a Word document in fixed random order. All filler sentences had referential expressions (See a sample filler with a personal pronoun below.). (8) Es war in einem Operationssaal eines Krankenhauses in Köln Lindenthal. Die Ärzte standen am Operationstisch. Die Anästhesisten standen daneben. Es war ein typischer arbeitsreicher und produktiver Tag. Sie hatten … (It was in an operating room of a hospital in Cologne Lindenthal. The doctors were standing at the operating table. The anesthesiologists stood by. It was a typical busy and productive day. They had …) 5.1.3 Procedure Each participant made an appointment for a sentence completion experiment session. We conducted a two-block data collection process, which has been previously used in studies of plural and singular anaphors (Çokal et al., 2023; Koh & Clifton, 2002). In the first-block, participants ÇOKAL AND VON HEUSINGER 64 finished all sentence completions, and we saved their completions to a different folder. In the second-block, immediately after saving their document, participants were asked to review their completions and underline to what the referential expressions (e.g., der, dieser) referred. Since filler sentences included other referential expressions (e.g., sie ‘she’, ‘they’, das ‘that’, er ‘he’), participants also underlined their referents. No feedback was given to participants. The whole experiment took 1 hour 15 minutes. 5.1.4 Data analysis We used the following continuation codings and samples for der and dieser. The data analysis below was based on participants’ underlined interpretations. If der or dieser referred to the protagonist introduced first, then its referent was coded as the first NP (NP1), as in (9) and (10) below: (9) In meiner Stammkneipe habe ich mit einem Koch und einem Barmixer geredet. Der Koch im Gegensatz zu dem Barmixer war durchgehend sehr freundlich. Es gab ein unterhaltsames Gespräch. Der hat den anderen viele Tipps und Ratschläge zum Kochen gegeben. (In my regular pub, I had a conversation with a chef and a bartender. The chef, unlike the bartender, was consistently very friendly. It was an entertaining conversation. Der gave the bartender many tips and pieces of advice about cooking.) (10) Auf einem Jazzfestival in dem berühmtesten Kölner Konzerthaus habe ich einen Sänger und einen Saxophonisten gemeinsam auftreten sehen. Der Sänger zusammen mit dem Saxophonisten hat stehende Ovationen bekommen. Es herrschte eine angenehme Stimmung in der schönen Kölner Philharmonie. Dieser hat den Saxophonisten umarmt, weil er so glücklich über die tolle Performance war. (At a jazz festival in the most famous concert hall in Cologne, I saw a singer and a saxophonist perform together. The singer, along with the saxophonist, received a standing ovation. There was a pleasant atmosphere in the beautiful Cologne Philharmonic Hall. Dieser hugged the saxophonist because he was so delighted with the fantastic performance.) If der or dieser referred to the protagonist introduced as the second noun, then its referent was coded as the second NP (NP2), as in (11) below: (11) In meiner Stammkneipe habe ich mit einem Koch und einem Barmixer geredet. Der Koch zusammen mit dem Barmixer war durchgehend sehr freundlich. Es gab ein unterhaltsames Gespräch. Dieser hat schon mit 16 seine ersten Cocktails gemixt. (In my regular pub, I talked with a chef and a bartender. The chef, along with the bartender, was consistently very friendly. There was an entertaining conversation. Dieser started mixing cocktails at the age of 16.) Ungrammatical or incoherent sentence completions were excluded from the logistic mixed effects regression analysis. For instance, while der/dieser is used for masculine nouns in nominative singular, they can also be used for feminine nouns in the genitive or dative singular form (see Appendix C for other cases). 5.2 Results We ran logistic mixed effects regressions, taking pronouns (der vs. dieser) and type of contrast (contrast vs. no-contrast) as the fixed effects, and including crossed random intercepts and slopes for participants and items. When participants had ungrammatical or incoherent sentence completions, these cases were coded as ‘other cases’ using 2 and excluded from the main analysis (as illustrated above). The percentage of other cases was 3.59 (n = 23 cases out of 1280. In the logistic mixed-effects regression, we coded references to the first protagonist (NP1/ der Cellist) as 1 and to the second protagonist (NP2/ der Flötist) as 0. We contrast-coded our fixed effects (i.e., GERMAN DEMONSTRATIVE PRONOUNS 65 anaphora type and contrast type). However, since the full model did not converge, we reduced the fixed random effect for items and ran the following model instead: Response ~ (1 | subject) + (1|item) + contrast type * anaphora type The intercept was significant (β = 2.27, SE = .258, z = 8.18) and the analysis revealed main effects of anaphora and contrast type (Anaphora: β = -0.366, SE = .180, z = -2.039, p = .041; OR = 0.67, SE = .12, p = .024; Contrast type: β = -0.406, SE = .180, z = -2.249; p = .024; OR = 0.69, SE = .14, p = .041). However, the interaction between these two factors was not significant (β = 0.157, SE = .355, z = 0.441; OR = 1.17, SE = .42, p = .659; See Appendix B, Figure 4). Most references were resolved to NP1. Additionally, there were slightly fewer NP1 references in contrast than no- contrast conditions: Der-contrast: OR = 0.905, SE = .025; Dieser-contrast: OR = 0.873, SE = .032; Der-no-contrast: OR = 0.937, SE = .018; Dieser-no-contrast: OR = 0.902, SE = 0.026 (See Appendix B and Figure 4 below). Figure 4. Estimated proportions of NP1 references out of all NP references in topic contrast and no-contrast contexts with der and dieser. Note: Error bars show the 95% confidence intervals. Overall, our results suggest that participants preferred NP1 references with der and dieser but this preference slightly changed in the contrast context. 6 General Discussion In this paper, we investigated whether a contrast context (i.e., NP1 in contrast to NP2) or no-contrast context (i.e., NP1 together with NP2) affects online processing of der and dieser regarding integration to a discourse model and antecedent preference. In addition, we also explored whether der and dieser are used to refer to NP1 or NP2 in these contexts. We predicted that the functions of der and dieser are not equally distributed. Our first assumption was that dieser signals topic persistence. Our eye-tracking reading experiment (Experiment 1) showed that in both contrast and no-contrast contexts the processing of dieser was easier than that of der, with lower regressions-out in the disambiguation region. This suggests that dieser led to less processing difficulty compared to der when referring to the first NP in both contrast and no-contrast contexts. In a sentence completion task (Experiment 2), participants mostly 0.00 0.25 0.50 0.75 1.00 Der contrast Der.no contrast Dieser.contrast Dieser.no contrast Pr op or tio n of N P1 re fe re nc es contrast no contrast ÇOKAL AND VON HEUSINGER 66 preferred continuations in which dieser and der refer to NP1 rather than NP2 in both contrast and no-contrast contexts (e.g., the cellist (NP1) in contrast to/together with the flautist (NP2). Der/Dieser hat). These results support our first prediction on the topic persistence function of dieser. A related question is, ‘Why do we not observe such online processing difficulty with dieser?’ One reason may be that dieser serves a ‘singling out’ function, making it a more suitable anaphor for ‘unbundling’ compared to der (Ahrenholz, 2007; Anderson & Keenan, 1985: Bisle- Müller, 1991). Our findings from the eye-tracking reading Experiment show longer reading times for der compared to dieser (i.e., regressions-out in the disambiguation region). In a sentence completion task (Experiment 2), participants mostly preferred continuations in which der and dieser refer to NP1 rather than NP2 in both contrast and no-contrast contexts (e.g., the cellist (NP1) in contrast to/together with the flautist (NP2). Der/Dieser hat). Our sentence completion experiment results suggest that these reading difficulties might not be attributed to topic references with der and dieser (i.e., ‘the cellist’ but not ‘the flautist’ and ‘the cellist’ together with ‘the flautist’ played a few wrong notes – ‘Der / Dieser fixed the strings’). One possible reason behind longer reading times for der might be that readers’ prior expectations in the disambiguation region lean towards the demonstrative determiner use of der (der + cellist) instead of a pronominal use. To investigate this further, we conducted a brief corpus analysis to examine the distribution of pronouns der and dieser as well as determiners der and dieser. To accomplish this, we retrieved instances from the Cosmas Tagged T2 corpus (https://www2.ids-mannheim.de/cosmas2/projekt/referenz/archive.html). Our results indicate that the percentage of determiner use of der is 95%, while the pronominal use is 5% (cases: Der-pronoun (n) = 13.352; Der-determiner (n) = 253.801). In contrast, determiner use of dieser accounts for 50% of cases, with demonstrative pronoun use also making up 50% (cases: Dieser-pronoun (n) = 9.402; Der-determiner (n = 9.525)). What is interesting is that, even though the assumption is less usage of der in written discourse (see Ahrenholz, 2007; Patil et al., 2020; Weinert, 2007), our small-scale corpus analysis indicates that people more frequently use der than dieser in written (newspaper) discourse. While further research is needed, it is not the focus of the current study. The combination of our results and the corpus analysis suggests readers tend to prefer the d-pronoun use of der (i.e., determiner), while the d-pronoun use of dieser does not significantly deviate from their expectations. Our second assumption was that der would be preferred in the contrast context rather than the no-contrast context because it requires a topic contrast. Experiment 1 (an eye-tracking reading study) did not support this prediction. We observed fewer regressions-out in the contrast context with der, compared to the no-contrast context with der or a similar pattern for der in the same region of total times. However, the interaction between anaphora and contrast type was not statistically significant. This finding appears to be inconsistent with the theoretical claims made by Bosch (1988) and Zifonun et al. (1997) for German, as well as van Kampen (2007) for Dutch and Swedish, who suggested that der could signal a contrastive antecedent. There might be two reasons for the inconsistency. The first reason is methodological differences. The previous studies’ claims were based on observations. Therefore, there were no controlled empirical data on der in the contrast context. The second reason is the limitation of the reading study, as the manipulation of the accented antecedent (i.e., the use of prosody) was not possible with our current experimental setting 6.1 Implications of topic reference/NP1 in the current study Our sentence completion experiment results suggest that a topic reference is preferred not only with dieser but also with der. These findings can be aligned with those of other studies that have explored demonstrative determiners cross-linguistically, including works by Ahrenholz (2007), Anderson and Keenan (1985), Bisle-Müller (1991), and Diessel (1999). As mentioned earlier, the literature posits three functions of demonstrative determiners in complex demonstratives: (a) contrast/non- uniqueness, (b) topic shift, and (c) topic persistence. These functions can be partially extended to https://www2.ids-mannheim.de/cosmas2/projekt/referenz/archive.html GERMAN DEMONSTRATIVE PRONOUNS 67 the function of demonstrative pronouns. In German, dieser seems to specifically specialize in topic persistence (both online and offline study) (see Fuchs & Schumacher, 2020, for evidence supporting the topic-persistent function of dieser). In addition, in Experiment 2, writers referred to the first mentioned entity, which is in the subject/topic position, but not the second NP, which is in a prepositional phrase (e.g., cellist/NP1 in contrast to flautist/NP2 or cellist/NP1 together with flautist /NP2). Our findings extend previous studies that have shown demonstratives’ preference for a non-topic antecedent, object, last- mentioned entity, or recent entity (Bader & Portele, 2019; Bosch, 2003; Fuchs & Schumacher, 2020; Hinterwimmer & Bosch, 2018; Kaiser, 2011; Patil et al., 2020; Zifonun et al., 1997). The recency effect observed in previous studies may be attributed to either (a) the short distance between demonstratives and their recent antecedents/NP2, since demonstratives immediately followed NP2, or (b) the thematic roles (agent vs. patient) explaining the recent antecedents of these two demonstratives. In our experimental design, we maintained a consistent distance between NP1 and NP2 in relation to the demonstratives. In addition, the thematic roles of antecedents remained the same across conditions. Both demonstratives (particularly dieser in both online and offline processing) can also serve as a topic reference in sentence completion tasks (i.e., NP1 references instead of NP2 references). These findings extend Repp and Schumacher’s (2023) corpus analysis of the informal language in the crime novel Tschick, where they observed that der can refer to antecedents in the subject position and antecedent with the thematic role of agent. In essence, Repp and Schumacher (2023) found that both der and the personal pronoun er tend to prefer subject and agent references (see Figure 2 in Repp & Schumacher, 2023). It should be noted that in the EEG study, Repp and Schumacher (2023) discovered that during an experiment in which participants listened to stories from audiobooks, the demonstrative der elicited a more pronounced N400 response compared to personal pronouns. Consequently, Repp and Schumacher (2023) argue that der, when contrasted with personal pronouns, is perceived as a less expected and more distinct form of reference. In their study, due to the design of the EEG experiment, they were unable to distinguish between the determiner and pronoun uses of der in the stories. However, our study specifically hypothesizes that dieser would cause less processing difficulty in reading compared to der. Our findings point to subtle distinctions within the d-pronoun and determiner systems of demonstratives. Further research is needed to differentiate demonstrative forms in online processing, encompassing both EEG and eye-tracking studies. A relevant direction for such further research is to investigate the online processing and production of both determiner and pronoun forms of these demonstratives to gain a broader understanding of when and how they are disambiguated. Additionally, there is a need to move beyond the recency and spatial approaches to information structures and semantics. A limitation of the current study is the absence of eye movement and reading time data for the determiner use of these two demonstratives when referring to a protagonist in both contrast and no- contrast contexts. Our corpus analysis indicates there would be less of a penalty if the determiner der was used (note that this is the definite article in most instances). Another limitation is that our design is clearly biased towards NP1 to test topic persistency. However, our offline sentence completion experiment results suggest participants predominantly had NP1 references with der and dieser, and fewer NP2 references occurred in a contrast context compared to a no-contrast context. In conclusion, our results provide novel insights regarding demonstrative pronoun resolution in online reading and offline sentence production tasks. The following findings also contribute to previous research: (i) dieser was preferred when referring to the topic of a sentence in both contrast and no-contrast contexts, (ii) der can be used for NP1 references as well. We feel our findings move the state-of-the-art in psychological and theoretical modelling of anaphora resolution forward beyond the recency and/or thematic role approach to demonstrative pronoun in cases requiring such inferences. ÇOKAL AND VON HEUSINGER 68 Acknowledgements This research was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project-ID 281511265 – SFB 1252 Prominence in Language – in the project C04 “Conceptual and referential activation in discourse” at the University of Cologne, Department of German Language and Literature I, Linguistics. We would like to express our gratitude to Robert Voigt and Nagihan Gökben Konuk for their invaluable assistance in crafting the stimuli and conducting piloting sessions with native speakers of German. GERMAN DEMONSTRATIVE PRONOUNS 69 Appendix A: Right-bound rereading across regions for Experiment 1 Anaphora Disambiguation Spillover β SE t β SE t β SE t Intercept 5.529 .028. 194 5.88 .057. 102 5.642 .034 162 Contrast -0.05 .020 -0.267 -0.018 .019 -0.944 0.013 .016 .803 Anaphora -0.021 .021. 1.044 -0.046 .019. -2.326 0.013 .016 -.508 Contrast * Anaphora -0.030 .041. 0.720 -0.001 .039. -0.015 0.008 .033 0.268 Table A1. Right-bound rereading across regions for Experiment 1. Pronoun Disambiguation Spillover Right-bound rereading M (SE) M (SE) M (SE) Der, contrast 144 (11) 415 (24) 296 (11) Dieser, contrast 225 (12) 395 (20) 286 (12) Der, no-contrast 155 (11) 421 (23) 295 (12) Dieser, no-contrast 218 (11) 387 (18) 286 (12) Table A2. Means (and standard errors) across regions for right-bound rereading in Experiment 1 ÇOKAL AND VON HEUSINGER 70 Appendix B: Odd Ratios in Experiment 2 Predictors Odds Ratios std. Error P-Value (Intercept) 9.77 (5.89 – 16.22) 2.53 <0.001* Contrast 0.69 (0.49 – 0.99) 0.13 0.041* Anaphora 0.67 (0.47 – 0.95) 0.12 0.024* Contrast * Anaphora 1.17 (0.58 – 2.35) 0.42 0.659 Random Effects σ2 3.29 τ00 subj 1.55 τ00 item 0.12 ICC 0.34 Nsubj 32 Nitem 40 Observations 1257 Marginal R2 / Conditional R2 0.015/ 0.346 Table A3. Odd ratios of der and dieser in referring to NP1 and NP2 in Experiment 2. GERMAN DEMONSTRATIVE PRONOUNS 71 Appendix C: Cases coded as ‘other cases’ were excluded in the statistical analysis in Experiment 2 (sentence completion experiment) Der and dieser are used to refer to other entities, as in (1) below: (1) Bei einer Verhandlung im Kölner Landgericht habe ich gestern einen Richter und einen Anwalt intensiv diskutieren gehört. Der Richter zusammen mit dem Anwalt empörte sich über eine Zeugenaussage. Es herrschte eine angespannte Atmosphäre. Der hat scheinbar einfach gelogen. (During a trial at the Cologne District Court, I heard a judge and a lawyer engaged in an intense discussion yesterday. The judge, along with the lawyer, expressed outrage over a witness statement. There was a tense atmosphere. He...apparently just lied.) Der and dieser are used to refer to two NPs, as in 2 below: (2) Beim morgendlichen Einkaufen in der überfüllten Markthalle des großen Einkaufscenters war ich bei einem Metzger und einem Gemüsehändler. Der Metzger zusammen mit dem Gemüsehändler hat die Leute durchgehend bedient. Es gab einige aufmerksame Fragen und lustige Bemerkungen. Der hat den Leuten einen guten Morgen beschert. (During my morning shopping in the crowded market hall of the large shopping center, I was at a butcher's shop and a vegetable vendor's stall. The butcher, along with the vegetable vendor, served people continuously. There were some attentive questions and amusing remarks. Der made people's morning enjoyable.) ÇOKAL AND VON HEUSINGER 72 References Werner Abraham (2002). Pronomina im Diskurs: deutsche Personal- und Demonstrativpronomina unter ‘Zentrierungsperspektive’. Grammatische Überlegungen zu einer Teiltheorie der Textkohärenz. Sprachwissenschaft, 27(4):447–491. Dorothy Ahn (2019). THAT thesis: A competition mechanism for anaphoric expressions. PhD thesis, Harvard University, Cambridge, Massachusetts. Bernt Ahrenholz (2007). Verweise mit Demonstrativa im gesprochenen Deutsch. Grammatik, Zweitspracherwerb und Deutsch als Fremdsprache. de Gruyter. doi: https://doi.org/10.1515/9783110894127. Stephen R. Anderson and Edward L. Keenan (1985). Deixis. In Timothy Shopen, editor, Language, Typology, and Syntactic Description, 2: 259–308. Cambridge University Press, Cambridge. doi: doi.org/10.1017/CBO9780511619434. Mira Ariel (1990). Accessing noun-phrase antecedents. Routledge, London. doi: https://doi.org/10.4324/9781315857473. Jennifer E. Arnold, Janet G. Eisenband, Sarah Brown-Schmidt, and John Trueswell (2000). The rapid use of gender information: Evidence of the time course of pronoun resolution from eyetracking. Cognition, 76:13–26. doi: https://doi.org/10.1016/S0010-0277(00)00073-1. Harald Baayen (2008). Analyzing linguistic data. A practical introduction to statistics using R. Cambridge University Press, Cambridge. doi: doi.org/10.1017/CBO9780511801686. Sarah, Brown-Schmidt, Donna, Byron, and Michael K. Tanenhaus (2005). Beyond salience: Interpretation of personal and demonstrative pronouns. Journal of Memory and Language, 53(2): 292–313. https://doi.org/10.1016/j.jml.2005.03.003 Markus Bader and Yvonne Portele (2019). The interpretation of German personal pronouns and d-pronouns. Zeitschrift für Sprachwissenschaft, 38(2):155-190. doi: https://doi.org/10.1515/zfs-2019-2002. Markus Bader, Yvonne Portele, and Alice Schäfer. Semantic bias in the interpretation of German personal and demonstrative pronouns (2022). In Robin Hörnig, Sophie von Wietersheim, Andreas Konietzko, and Sam Featherston, editors, Proceedings of Linguistic Evidence: Linguistic Theory Enriched by Experimental Data, pages 399–419. University of Tübingen, Tübingen. doi: https://publikationen.uni- tuebingen.de/xmlui/handle/10900/119301. Douglas Bates, Martin Mächler, Ben Bolker, and Steve Walker (2015). Fitting Linear Mixed- Effects Models. Using lme4. Journal of Statistical Software, 67: 1–48. 10.18637/jss.v067.i01 Dale J. Barr, Roger Levy, Christoph Scheepers, and Harry J. Tily (2013). Random effects structure for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language, 68(3): 0.1016/j.jml.2012.11.001. doi: 10.1016/j.jml.2012.11.001 David I. Beaver (2004). The optimization of discourse anaphora. Linguistics and Philosophy, 27:3–56. doi: https://doi.org/10.1023/B:LING.0000010796.76522.7a. Hansjörg Bisle-Müller (1991). Artikelwörter im Deutschen: Semantische und pragmatische Aspekte ihrer Verwendung. Niemeyer, Tübingen. Peter Bosch (1988). Representing and accessing focussed referents. Language and Cognitive Processes, 3(3):207–231. doi: https://doi.org/10.1080/01690968808402088. Peter Bosch and Stefan Hinterwimmer (2016). Anaphoric reference by demonstrative pronouns in German. In search of the relevant parameters. In Anke Holler and Kaja Suckow, editors, Empirical Perspectives on Anaphora Resolution, pages 193–212. de Gruyter, Berlin. doi: https://doi.org/10.1515/9783110464108-010. Peter Bosch and Carla Umbach. Reference determination for demonstrative pronouns (2007). ZAS Papers in Linguistics, 48:39-51. doi: https://doi.org/10.21248/zaspil.48.2007.353. Peter Bosch, Tom Rozario, and Yufan Zhao (2003). Demonstrative pronouns and personal pronouns. German der vs. er. In Proceedings of the EACL, Workshop on The Computational Treatment of Anaphora, pages 61–68, Budapest. doi: https://aclanthology.org/W03-2609. https://doi.org/10.1515/9783110894127 https://psycnet.apa.org/doi/10.1016/j.jml.2005.03.003 https://doi.org/10.1023/B:LING.0000010796.76522.7a GERMAN DEMONSTRATIVE PRONOUNS 73 Gerlof Bouma and Holger Hopp (2007). Coreference preferences for personal pronouns in German. In Dagmar Bittner and Natalia Gargarina, editors, Proceedings of the conference on Intersentential Pronominal Reference in Child and Adult Language, (ZAS Papers in Linguistics 48), pages 53–74. ZAS, Berlin. doi: https://doi.org/10.21248/zaspil.48.2007.354. Wallace Chafe (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point of view. In Charles N. Li, editor, Subject and Topic, pages 25–55. Academic Press, New York. Derya Çokal, Patrick Sturt, and Fernanda Ferreira (2014). Deixis: This and that in written narrative discourse. Discourse Processes, 51(3):201–229. doi: https://doi.org/10.1080/0163853X.2013.866484. Derya Çokal, and Fernanda Ferreira (2015). Online processing of noisy input from native and non-native language users. Poster presented at the 28th Annual Conference on Human Sentence Processing (CUNY), University of Southern California, 19-21 March, LosAngeles, CA. Derya Çokal, Patrick Sturt, and Fernanda Ferreira (2018). Processing of it and this in written narrative discourse. Discourse Processes, 55(3):272–289. doi: https://doi.org/10.1080/0163853X.2016.1236231. Derya Çokal, Ruth Filik, Patrick Sturt, and Massimo Poesio (2023). Anaphoric reference to mereological entities. Discourse Processes, 60(3):202–223. doi: https://doi.org/10.1080/0163853X.2023.2197682. Rosalind A. Crawley, Rosemary Stevenson, and David Kleinman (1990). The use of heuristic strategies in the interpretation of pronouns. Journal of Psycholinguistic Research, 19:245–264. doi: https://doi.org/10.1007/BF01077259. Annika Deichsel and Klaus von Heusinger (2011). The cataphoric potential of indefinites in German. In Iris Hendrickx, Sobha Lalitha Devi, António Branco, and Ruslan Mitkov, editors, Anaphora Processing and Applications. DAARC 2011, pages 144–156. Springer, Heidelberg. Holger Diessel (1999). Demonstratives: Form, function and grammaticalization. John Benjamins, Amsterdam. Miriam Ellert (2013). Resolving ambiguous pronouns in a second language: A visual-world eye- tracking study with Dutch learners of German. International Review of Applied Linguistics in Language Teaching, 51(2):171–197. doi: https://doi.org/10.1515/iral-2013-0008. Marion Fossard, Alan Garnham, and H. Wind Cowles (2012). Between anaphora and deixis... The resolution of the demonstrative noun phrase “that N”. Language and Cognitive Processes, 27(9):1385–1404. doi: https://doi.org/10.1080/01690965.2011.606668. Melanie Fuchs and Petra B. Schumacher (2020). Referential shift potential of demonstrative pronouns – evidence from text continuation. In Ashild Næss, Anna Margetts, and Yvonne Treis, editors, Demonstratives in discourse, pages 185–213. Language Science Press, Berlin. doi: https://langsci- press.org/catalog/book/282. Kumiko Fukumura and Roger P. G. van Gompel (2015). Effects of order of mention and grammatical role on anaphor resolution. Learning, Memory, and Cognition, 41(2):501–525. doi: https://doi.org/10.1037/xlm0000041. Morton Ann Gernsbacher and Suzanne Shroyer (1989). The cataphoric use of the indefinite this in spoken narratives. Memory & Cognition, 17(5):536–540. Talmy Givón (1983). Topic continuity in discourse: An introduction. In Talmy Givón, editor, Topic continuity in discourse: A quantitative cross-language study, pages 1–42. John Benjamins, Amsterdam. doi: https://doi.org/10.1075/tsl.3.01giv. Peter Gordon, Barbara Grosz, and Laura Gilliom (1993). Pronouns, names, and the centering of attention in discourse. Cognitive Science, 17(3):311–346. doi: https://doi.org/10.1207/s15516709cog1703. Barbara Grosz, Aravind Joshi, and Scott Weinstein. (1995). Centering: A framework for modeling the local coherence of discourse. Computational Linguistics, 21:203–226. doi: https://doi.org/10.1.1.14.9312. https://doi.org/10.1007/BF01077259 ÇOKAL AND VON HEUSINGER 74 Jeanette K. Gundel (1988). Universals of topic-comment structure. In Michael Hammond, Edith Moravcsik, and Jessica R. Wirth, editors, Studies in syntactic typology, pages 209–239. John Benjamins, Amsterdam. Jeanette K. Gundel., Nancy, Hedberg, & Ron, Zacharski (1993). Cognitive status and the form of referring expressions in discourse. Language, 69(2), 274–307. doi: https:doi.org/10.2307/416535 Elyse, D., Johnson, & Jennifer, Arnold (2023). The frequency of referential patterns guides pronoun comprehension. Journal of Experimental Psychology: Learning, Memory, and Cognition, 49(8), 1325–1344. doi: https://doi.org/10.1037/xlm0001137 Martina Hielscher and Jochen Musseler (1990). Anaphoric resolution of singular and plural pronouns: The reference to persons being introduced by different co-ordinating structures. Journal of Semantics, 7:347–364. Nikolaus P. Himmelmann (1997). Deiktikon, Artikel, Nominalphrase. Zur Emergenz syntaktischer Struktur. Max Niemeyer Verlag, Tübingen. Stefan Hinterwimmer and Peter Bosch (2018). Demonstrative pronouns and propositional attitudes. In Pritty Patel-Grosz, Patrick Georg Grosz, and Sarah Zobel, editors, Pronouns in embedded contexts at the syntax-semantics interface, pages 105–144. Springer. Jerry R. Hobbs (1979). Coherence and coreference. Cognitive Science, 3(1):67–90. doi: https://doi.org/10.1207/s15516709cog0301_4. Juhani Järvikivi, Roger P. G. van Gompel, and Jukka Hyönä (2017). The interplay of implicit causality, structural heuristics, and anaphor type in ambiguous pronoun resolution. Journal of Psycholinguistic Research, 46:525–550. Elsi Kaiser (2011b). On the relation between coherence relations and anaphoric demonstratives in Ger- man. In Ingo Reich, Eva Horch, and Dennis Paul, editors, Proceedings of Sinn und Bedeutung 15, pages 337–351. Universaar – Saarland University Press, Saarbrücken. doi: https://ojs.ub.uni- konstanz.de/sub/index.php/sub/article/view/385. Elsi Kaiser and John Trueswell (2004). The referential properties of Dutch pronouns and demonstratives: Is salience enough? In Proceedings of Sinn und Bedeutung 8, pages 137–150. Universität Konstanz, Konstanz. doi: https://doi.org/10.18148/sub/2004.v8i0.754. Jeffrey King (2001). Complex demonstratives- A quantificational account. MIT Press, Cambridge, Massachusetts. doi: doi.org/10.7551/mitpress/1990.003.0011. Sungryong Koh and Charles Clifton Jr (2002). Resolution of the antecedent of a plural pronoun: Ontological categories and predicate symmetry. Journal of Memory and Language, 46:830– 844. doi: https://doi.org/10.1006/jmla.2001.2829. Knud Lambrecht (1994). Information structure and sentence form. Topic, focus, and the mental representations of discourse referents. Cambridge University Press, Cambridge. Alfons Maes, Emiel Krahmer, and David Peeters (2022a). Explaining variance in writers’ use of demonstratives: A corpus study demonstrating the importance of discourse genre. Journal of General Linguistics, 7(1):1–36. doi: https://doi.org/10.16995/glossa.5826. Alfons Maes, Emiel Krahmer, and David Peeters (2022b). Understanding demonstrative reference in text: A new taxonomy based on a new corpus. Language and Cognition, 14(2):185–207. doi: https://doi.org/10.1017/langcog.2021.28. Linda M. Moxey, Anthony J. Sanford, and Patrick Sturt (2004). Constraints on the formation of plural reference objects: The influence of role, conjunction, and type of description. Journal of Memory and Language, 51:346–364. Livia Polanyı (1986). The linguistic discourse model: Towards a formal theory of discourse structure. Cambridge, MA: Laboratories Incorp. Umesh Patil, Peter Bosch, and Stefan Hinterwimmer (2020). Constraints on German diese demonstratives: Language formality and subject-avoidance. Glossa: a journal of general linguistics, 5:1–22. doi: https://doi.org/10.5334/gjgl.962. Umesh Patil, Stefan Hinterwimmer, and Petra B. Schumacher (2023). Effect of evaluative GERMAN DEMONSTRATIVE PRONOUNS 75 expressions on two types of demonstrative pronouns in German. Glossa: a journal of general linguistics, 8:1–29. doi: https://doi.org/10.16995/glossa.9577. Clare Patterson and Petra B. Schumacher (2020). The timing of prominence information during the resolution of German personal and demonstrative pronouns. Dialogue & Discourse, 11(1):1–39. doi: https://doi.org/10.5087/dad.2020.101. Clare Patterson and Petra B. Schumacher (2021). Interpretation preferences in contexts with three antecedents: Examining the role of prominence in German pronouns. Cambridge University Press, 42(6):1427–1461. doi: 10.1017/S0142716421000291. Clare Patterson and Petra B. Schumacher (2023). How focus and position affect the interpretation of demonstrative pronouns. Collabra: Psychology, 9:1–23. Ellen F. Prince (1981). On the inferencing of indefinite-this NPs. In Bonnie Webber, Aravind K. Joshi, and Ivan Sag, editors, Elements of Discourse Understanding, pages 231–250. Cambridge University Press, Cambridge. Pirita Pyykkönen and Juhani Järvikivi (2010). Activation and persistence of implicit causality information in spoken language comprehension. Experimental Psychology, 57:5–16. doi: https://doi.org/10.1027/1618- 3169/a000002. Keith Rayner, Timothy J. Slattery, Denis Drieghe, and Simon P. Liversedge (2011). Eye movements and word skipping during reading: Effects of word length and predictability. Journal of Experimental Psychology, 37:514–528. Keith Rayner and Simon P. Liversedge (2012). Linguistic and cognitive influences on eye movements during reading. In Simon P. Liversedge, Iain Gilchrist, and Stefan Everling (eds), The Oxford Handbook of Eye Movements, 752-766. Oxford University Press. doi.org/10.1093/oxfordhb/9780199539789.013.0041. Tanya Reinhart (1981). Pragmatics and linguistics: An analysis of sentence topics. Philosophica, 27:53–94. David, Peeters, Emiel, Krahmer, and Alfons, Maes (2021). A conceptual framework for the study of demonstrative reference. Psychonomic Bulletin & Review, 28(2), 409–433. https://doi.org/10.3758/s13423-020-01822-8 Magdalena Repp and Petra B. Schumacher (2023). What naturalistic stimuli tell us about pronoun resolution in real-time processing. Frontiers in Artificial Intelligence, 6:1–15. doi: 10.3389/frai.2023.1058554. Repp Sophie and Spalek Katharina (2021). The role of alternatives in language. Frontiers in Communication, 6. doi: https://doi.org/10.3389/fcomm.2021.682009. Petra B. Schumacher, Jana Backhaus, and Manuel Dangl (2015). Backward- and forward-looking potential of anaphors. Frontiers in Psychology, 6:1–14. doi: https://doi.org/10.3389/fpsyg.2015.01746. Petra B. Schumacher, Manuel Dangl, and Elyesa Uzun (2016). Thematic role as prominence cue during pronoun resolution in German. In Anke Holler and Kaja Suckow, editors, Empirical Perspectives on Anaphora Resolution, pages 213–240. de Gruyter, Berlin. doi: https://doi.org/10.1515/9783110464108-011. Petra B. Schumacher, Leah Roberts, and Juhani Järvikivi (2017). Agentivity drives real-time pronoun resolution: Evidence from German er and der. Lingua, 185:25–41. doi: https://doi.org/10.1016/j.lingua.2016.07.004. Petra B. Schumacher, Clare Patterson, and Magdalena Repp (2022). Die diskursstrukturierende Funktion von Demonstrativpronomen. In Chiara Gianollo, Lukasz Jedrzejowski, and Sofiana I. Lindemann, editors, Paths through meaning and form. Festschrift offered to Klaus von Heusinger on the occasion of his 60th birthday, pages 216–220. Universitäts- und Stadtbibliothek Köln, Köln. doi: https://doi.org/10.18716/omp.3. Rosemary J. Stevenson, Rosalind Crawley, and David Kleinman (1994). Thematic roles, focus, and the representation of events. Language and Cognitive Processes, 9(4):519–548. doi: https://doi.org/10.1080/01690969408402130. ÇOKAL AND VON HEUSINGER 76 Rosemary J. Stevenson, Alistair Knott, Jon Oberlander, and Sharon McDonald (2000). Interpreting pronouns and connectives: Interactions among focusing, thematic roles and coherence relations. Language and Cognitive Processes, 15(3):225–262. doi: https://doi.org/10.1080/016909600386048. Rosemary J. Stevenson, Alexander W. R. Nelson, and Keith Stenning (1995). The role of parallelism in strategies of pronoun comprehension. Language and Speech, 38:393–418. doi: https://doi.org/10.1177/002383099503800404. Patrick Sturt and Nayoung Kwon (2018). Processing information during regressions: An application of the reverse boundary-change paradigm. Frontiers in Psychology, 9:1630. doi: https://doi.org/10.3389/fpsyg.2018.01630. Emil van den Hofen and Evelyn C. Ferstl (2018). Discourse context modulates the effect of implicit causality on rementions. Language and Cognition, 10(4):561–594. doi: https://doi.org/10.1017/langcog.2018.17. Jacqueline van Kampen (2007). Anaphoric topic-shift devices. Paper presented at the ‘Workshop on anaphoric uses of demonstrative expressions’, Siegen. Titus von der Malsburg and Bernhard Angele (2017). False positives and other statistical errors in standard analyses of eye movements in reading. Journal of Memory and Language, 94:119– 133. Klaus von Heusinger and Petra B. Schumacher (2019). Discourse prominence: Definition and application. Journal of Pragmatics, 154:117–127. doi: https://doi.org/10.1016/j.pragma.2019.07.025. Regina Weinert (2007). Demonstrative and personal pronouns in formal and informal conversations. In Regina Weinert, editor, Spoken language pragmatics: An analysis of form- function relations, pages 1–28. Continuum, London. Frances Wilson (2009). Processing at the syntax-discourse interface in second language acquisition. PhD thesis, University of Edinburgh, Edinburgh. Angelika Wöllstein (2022). die Dudenredaktion. Duden. Die Grammatik. Dudenverlag, Berlin. Lynsey Wolter (2006). That’s that: The semantics and pragmatics of demonstrative noun phrases. PhD thesis, University of California, Santa Cruz, CA. Gisela Zifonun, Ludger Hoffmann, and Bruno Strecker (1997). Grammatik der deutschen Sprache.10., völlig neu verfasste Auflage. de Gruyter, Berlin.