Agreement is often analyzed in terms of syntactic versus semantic agreement (Corbett 2006: 155–160; Corbett 2023). “Syntactic agreement”, as in family has traveled..., follows straightforward rules of syntax. Family is in a singular form and denotes a unit, and hence we find singular agreement. In contrast, “semantic agreement” involves additional information, in this case the lexical semantics. When family is understood as a collection of individuals, access to this lexical information gives rise to semantic agreement, as in family have traveled... (Corbett 2023: 30). Building on this distinction between syntactic and semantic agreement, this dissertation investigates a set of Norwegian agreement patterns in which predicative adjectives diverge from canonical morphosyntactic agreement.
This dissertation focuses on three well-attested non-canonical agreement patterns in Norwegian: pancake agreement, anti-pancake agreement, and neuter collective agreement. In the canonical Norwegian agreement pattern, predicative adjectives agree with subjects in terms of number (singular/plural) and gender (masculine/feminine/neuter), as in huset er fint =house:N:DEF:SG be:PRS nice:N:SG‘The house is nice’. However, there are observed exceptions to this rule. In the so-called “pancake agreement”, the predicative adjective unexpectedly takes neuter singular agreement when the subject is not a neuter singular NP, as in Pannekaker er godt =pancake(F):INDF:PL be:PRS good:N:SG ‘Pancakes are good.’ In what I call “anti-pancake agreement”, the predicative adjective unexpectedly takes common gender singular when the subject is a neuter singular NP, as in Postbudet er syk =post.officer:N:DEF:SG be:PRS sick:C:SG ‘The post officer is sick.’ In Norwegian, “common gender” is used when the distinction between masculine and feminine nouns is not made on adjectives. Finally, in what I call “neuter collective agreement”, the predicative adjective unexpectedly takes common singular or plural agreement when the subject is a neuter collective noun, as in Personalet er omsorgsfull(e) = personnel:N:DEF:SG be:PRS caring:C:SG/PL ‘The personnel is/are caring.’
These phenomena, especially pancake agreement, have attracted significant attention in Scandinavian linguistics and the typology of agreement (Widmark 1966; Faarlund 1977; Faarlund, Lie & Vannebo 1997; Enger 2004; Enger 2013; Enger 2020; Corbett 2006; Corbett 2023; Josefsson 2009; Wechsler 2013; Josefsson 2014; Haugen & Enger 2014; Haugen & Enger 2019; Åkerblom 2020; Åfarli & Vangsnes 2021; Nesset, Enger & Janda 2022, etc.). In Scandinavian linguistics, exploring when and why these patterns occur is essential for understanding the Scandinavian agreement system as a whole. From a typological perspective, these phenomena provide interesting case studies of intralinguistic and crosslinguistic variations of agreement patterns observed in languages in the world.
Methodologically (Chapter 2), I explore the non-canonical agreement patterns found in Norwegian by adopting a holistic and usage-based approach. I adopt a holistic approach to illuminate the reasons behind and mechanisms of non-canonical agreement patterns in the Norwegian agreement system as a whole. Each agreement phenomenon has attracted considerable interest in the literature, but a study that brings together these phenomena under a unified analytical framework is needed. Such a study would not only enhance our understanding of each agreement pattern but also provide insights into broader mechanisms governing agreement variations observed in Norwegian.
Moreover, the methodology adopted in this study follows the principles of the usage-based approach. In linguistics, there is a shift from traditional qualitative and introspective methods toward empirical, data-drive, and statistically informed approaches, called the “quantitative turn” (Janda 2019). Reflecting solely on linguistic intuitions can pose challenges for linguistic analyses because there are disagreements between speakers and linguists (Carden and Dieterich 1980; Cowart 1997; Janda 2019). In this study, I conduct systematic corpus investigations on Norwegian agreement phenomena, using the “noTenTen17 Bokmål” corpus run by Sketch Engine. This corpus consists of Norwegian texts written in bokmål (the major written standard of Norwegian). It belongs to the TenTen Corpus Family (TenTen corpora), which is a set of web corpora built using the same method with a target size of 10+ billion words per language (Jakubíček et al. 2013; Kilgarriff et al. 2014). Separate searches were conducted for each phenomenon. For each sentence, semantic (e.g., animacy and definiteness) and morphosyntactic properties (e.g., distance between subjects and adjectives) were coded manually. The effects of such variables were corroborated by the utilization of mixed-effects logistic regression and, for neuter collective agreement, conditional inference trees, as well.
For pancake agreement (Chapter 3), the results of the systematic corpus investigation show that pancake agreement is observed frequently, especially when (i) the subject is inanimate, (ii) the subject is in a bare form or an indefinite form, and/or (iii) the verb is in the past participle. This finding lends empirical support to the semantic approach to pancake agreement. To be more specific, the analysis based on the results of the corpus research supports the effectiveness of the semantic concepts of virtuality, unboundedness, and ungroundedness that have been argued by Haugen and Enger (2019) to be important in this agreement pattern. Furthermore, this empirical study elaborates on the semantic approach by proposing the following two conclusions. First, the results of my corpus investigation indicate that virtuality is more crucial than unboundedness. Indefinite NPs were associated with pancake agreement, whether they were singular or plural. Second, I highlight the importance of paying more attention to the verbal properties of the copular verb, være ‘be’. I propose that the formal preferences of the copular verb in the pancake construction, i.e., past participle > infinitive > present > past, reflect the importance of processual ungrounding, which plays an important role in the diachrony of the construction (Haugen and Enger 2019).
For anti-pancake agreement (Chapter 4), the results of the corpus investigation show that this agreement pattern occurs more frequently when the subject refers to a non-age-specific human being and/or when there is a greater distance between the subject and the predicative adjective. Based on these findings, I argue that the choice between anti-pancake and canonical neuter agreement is influenced by the two competing factors working against each other. On the one hand, Norwegian speakers tend to resolve the semantic mismatch between the high degree of individuation of non-age-specific subjects and the lower degree of individuation typically associated with the neuter gender. On the other hand, Norwegian speakers tend to resolve the morphosyntactic mismatch between the subject and the predicative adjective when the mismatch is evident in terms of distance. Furthermore, I argue that these two factors are important in Norwegian agreement phenomena more broadly. These findings suggest the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian.
For neuter collective agreement (Chapter 5), the results of the corpus investigation show that neuter agreement occurs frequently when the subject is expressed without a modifier, while it occurs rarely when the subject is expressed with a modifier and the distance between the subject and the predicative adjective is more than 2. Conversely, plural agreement occurs most frequently when the subject is expressed with a modifier and the distance between the subject and the predicative adjective is more than 2. Differently from neuter singular and plural agreement, common singular agreement is observed equally frequently regardless of (non)-cooccurrence with a modifier and distance between subjects and predicative adjectives. Based on these results, I confirm that neuter agreement is observed when the subject refers to an undifferentiated group, while plural agreement is observed when the subject refers to a number of individuals (cf. Åfarli & Vangsnes [2021: 531]). Moreover, I propose that the distance between the subject and the predicative adjective also influences the choice between agreement patterns. These findings on neuter collective agreement also highlight the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian.
Across all three agreement patterns, the findings from this dissertation converge on two central points. First, the findings from this study support the argument that non-canonical agreement patterns in Norwegian can be analyzed as types of semantic agreement (Chapter 6.1). In the literature, there have been three major approaches to tackling non-canonical agreement phenomena in Norwegian: the syntactic approach (Faarlund 1977; Josefsson 2014a, etc.), the default approach (Åkerblom 2020; Wechsler 2013, etc.), and the semantic approach ( Enger 2004, 2013, 2022; Haugen and Enger 2014, 2019; Widmark 1966, etc.). The syntactic approach locates the use of the neuter singular form of pancake agreement in syntax. In the default approach, it has been argued that the neuter singular, which is considered to be default, is used in the pancake construction because there is no positive motivation (Åkerblom 2020). The findings from this dissertation regarding pancake agreement, anti-pancake agreement, and neuter collective agreement all demonstrate that the choices between canonical and non-canonical agreement patterns are influenced by the positive semantic motivations, such as animacy/individuation, virtuality, and unboundedness.
Second, a key insight from this study is that Norwegian agreement patterns are not operated solely by morphosyntactic rules but are influenced by an intricate interaction between semantics and morphosyntax (Chapter 6.2). The analyses of non-canonical agreement patterns reveal that both semantics and morphosyntactic environments in which agreement occurs contribute to the choice of agreement patterns. The anti-pancake construction exemplifies this interaction clearly. The choice between anti-pancake and canonical neuter agreement is influenced by the two competing factors working against each other, On the one hand, Norwegian speakers tend to resolve the semantic mismatch between the high degree of individuation of non-age-specific subjects and the lower degree of individuation typically associated with the neuter gender. On the other hand, Norwegian speakers tend to resolve the morphosyntactic mismatch between the subject and the predicative adjective when the mismatch is evident because of the short distance. The analyses from this dissertation highlight the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian. The interplay between semantics and morphosyntax is fundamental to understanding the flexibility and variation observed in Norwegian agreement patterns.
To sum up, by investigating pancake-agreement, anti-pancake agreement, and neuter-collective agreement, this study has shown that Norwegian agreement patterns extend beyond morphosyntactic rules and are strongly influenced by factors such as animacy, individuation, virtuality, and unboundedness. Additionally, this dissertation provides new insights into the nature of non-canonical agreement patterns in Norwegian, demonstrating that these patterns are systematically shaped by an intricate interaction between semantics and morphosyntax.
This dissertation has implications at several levels. Within Scandinavian linguistics, this study lends empirical support to the semantic approach to non-canonical agreement, extending it beyond pancake agreement to anti-pancake and neuter collective agreement. Within Germanic linguistics, the findings contribute to a broader understanding of agreement in Germanic languages, highlighting parallels in other languages and reinforcing the role of semantic motivations in agreement phenomena.
From a typological perspective, the findings from this dissertation contribute to the cross-linguistic picture by showing that semantic factors can override morphosyntactic rules in certain constructional contexts, while morphosyntax prevails in others. This interplay suggests that typological models of agreement should allow for gradual shifts between semantic and syntactic agreement, rather than treating them as mutually exclusive.
Finally, the usage-based approach adopted in this dissertation contributes to research on Norwegian agreement in the following two respects. First, through a corpus-based investigation, this study provides empirical evidence supporting the view that non-canonical agreement patterns in Norwegian should be understood as instances of semantic agreement. Furthermore, the findings presented in this dissertation uncover previously unexplored properties of Norwegian agreement. By systematically analyzing corpus data, this usage-based study not only substantiates the insightful native intuitions proposed in earlier studies but also sheds light on previously overlooked aspects of Norwegian agreement.
This dissertation focuses on three well-attested non-canonical agreement patterns in Norwegian: pancake agreement, anti-pancake agreement, and neuter collective agreement. In the canonical Norwegian agreement pattern, predicative adjectives agree with subjects in terms of number (singular/plural) and gender (masculine/feminine/neuter), as in huset er fint =house:N:DEF:SG be:PRS nice:N:SG‘The house is nice’. However, there are observed exceptions to this rule. In the so-called “pancake agreement”, the predicative adjective unexpectedly takes neuter singular agreement when the subject is not a neuter singular NP, as in Pannekaker er godt =pancake(F):INDF:PL be:PRS good:N:SG ‘Pancakes are good.’ In what I call “anti-pancake agreement”, the predicative adjective unexpectedly takes common gender singular when the subject is a neuter singular NP, as in Postbudet er syk =post.officer:N:DEF:SG be:PRS sick:C:SG ‘The post officer is sick.’ In Norwegian, “common gender” is used when the distinction between masculine and feminine nouns is not made on adjectives. Finally, in what I call “neuter collective agreement”, the predicative adjective unexpectedly takes common singular or plural agreement when the subject is a neuter collective noun, as in Personalet er omsorgsfull(e) = personnel:N:DEF:SG be:PRS caring:C:SG/PL ‘The personnel is/are caring.’
These phenomena, especially pancake agreement, have attracted significant attention in Scandinavian linguistics and the typology of agreement (Widmark 1966; Faarlund 1977; Faarlund, Lie & Vannebo 1997; Enger 2004; Enger 2013; Enger 2020; Corbett 2006; Corbett 2023; Josefsson 2009; Wechsler 2013; Josefsson 2014; Haugen & Enger 2014; Haugen & Enger 2019; Åkerblom 2020; Åfarli & Vangsnes 2021; Nesset, Enger & Janda 2022, etc.). In Scandinavian linguistics, exploring when and why these patterns occur is essential for understanding the Scandinavian agreement system as a whole. From a typological perspective, these phenomena provide interesting case studies of intralinguistic and crosslinguistic variations of agreement patterns observed in languages in the world.
Methodologically (Chapter 2), I explore the non-canonical agreement patterns found in Norwegian by adopting a holistic and usage-based approach. I adopt a holistic approach to illuminate the reasons behind and mechanisms of non-canonical agreement patterns in the Norwegian agreement system as a whole. Each agreement phenomenon has attracted considerable interest in the literature, but a study that brings together these phenomena under a unified analytical framework is needed. Such a study would not only enhance our understanding of each agreement pattern but also provide insights into broader mechanisms governing agreement variations observed in Norwegian.
Moreover, the methodology adopted in this study follows the principles of the usage-based approach. In linguistics, there is a shift from traditional qualitative and introspective methods toward empirical, data-drive, and statistically informed approaches, called the “quantitative turn” (Janda 2019). Reflecting solely on linguistic intuitions can pose challenges for linguistic analyses because there are disagreements between speakers and linguists (Carden and Dieterich 1980; Cowart 1997; Janda 2019). In this study, I conduct systematic corpus investigations on Norwegian agreement phenomena, using the “noTenTen17 Bokmål” corpus run by Sketch Engine. This corpus consists of Norwegian texts written in bokmål (the major written standard of Norwegian). It belongs to the TenTen Corpus Family (TenTen corpora), which is a set of web corpora built using the same method with a target size of 10+ billion words per language (Jakubíček et al. 2013; Kilgarriff et al. 2014). Separate searches were conducted for each phenomenon. For each sentence, semantic (e.g., animacy and definiteness) and morphosyntactic properties (e.g., distance between subjects and adjectives) were coded manually. The effects of such variables were corroborated by the utilization of mixed-effects logistic regression and, for neuter collective agreement, conditional inference trees, as well.
For pancake agreement (Chapter 3), the results of the systematic corpus investigation show that pancake agreement is observed frequently, especially when (i) the subject is inanimate, (ii) the subject is in a bare form or an indefinite form, and/or (iii) the verb is in the past participle. This finding lends empirical support to the semantic approach to pancake agreement. To be more specific, the analysis based on the results of the corpus research supports the effectiveness of the semantic concepts of virtuality, unboundedness, and ungroundedness that have been argued by Haugen and Enger (2019) to be important in this agreement pattern. Furthermore, this empirical study elaborates on the semantic approach by proposing the following two conclusions. First, the results of my corpus investigation indicate that virtuality is more crucial than unboundedness. Indefinite NPs were associated with pancake agreement, whether they were singular or plural. Second, I highlight the importance of paying more attention to the verbal properties of the copular verb, være ‘be’. I propose that the formal preferences of the copular verb in the pancake construction, i.e., past participle > infinitive > present > past, reflect the importance of processual ungrounding, which plays an important role in the diachrony of the construction (Haugen and Enger 2019).
For anti-pancake agreement (Chapter 4), the results of the corpus investigation show that this agreement pattern occurs more frequently when the subject refers to a non-age-specific human being and/or when there is a greater distance between the subject and the predicative adjective. Based on these findings, I argue that the choice between anti-pancake and canonical neuter agreement is influenced by the two competing factors working against each other. On the one hand, Norwegian speakers tend to resolve the semantic mismatch between the high degree of individuation of non-age-specific subjects and the lower degree of individuation typically associated with the neuter gender. On the other hand, Norwegian speakers tend to resolve the morphosyntactic mismatch between the subject and the predicative adjective when the mismatch is evident in terms of distance. Furthermore, I argue that these two factors are important in Norwegian agreement phenomena more broadly. These findings suggest the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian.
For neuter collective agreement (Chapter 5), the results of the corpus investigation show that neuter agreement occurs frequently when the subject is expressed without a modifier, while it occurs rarely when the subject is expressed with a modifier and the distance between the subject and the predicative adjective is more than 2. Conversely, plural agreement occurs most frequently when the subject is expressed with a modifier and the distance between the subject and the predicative adjective is more than 2. Differently from neuter singular and plural agreement, common singular agreement is observed equally frequently regardless of (non)-cooccurrence with a modifier and distance between subjects and predicative adjectives. Based on these results, I confirm that neuter agreement is observed when the subject refers to an undifferentiated group, while plural agreement is observed when the subject refers to a number of individuals (cf. Åfarli & Vangsnes [2021: 531]). Moreover, I propose that the distance between the subject and the predicative adjective also influences the choice between agreement patterns. These findings on neuter collective agreement also highlight the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian.
Across all three agreement patterns, the findings from this dissertation converge on two central points. First, the findings from this study support the argument that non-canonical agreement patterns in Norwegian can be analyzed as types of semantic agreement (Chapter 6.1). In the literature, there have been three major approaches to tackling non-canonical agreement phenomena in Norwegian: the syntactic approach (Faarlund 1977; Josefsson 2014a, etc.), the default approach (Åkerblom 2020; Wechsler 2013, etc.), and the semantic approach ( Enger 2004, 2013, 2022; Haugen and Enger 2014, 2019; Widmark 1966, etc.). The syntactic approach locates the use of the neuter singular form of pancake agreement in syntax. In the default approach, it has been argued that the neuter singular, which is considered to be default, is used in the pancake construction because there is no positive motivation (Åkerblom 2020). The findings from this dissertation regarding pancake agreement, anti-pancake agreement, and neuter collective agreement all demonstrate that the choices between canonical and non-canonical agreement patterns are influenced by the positive semantic motivations, such as animacy/individuation, virtuality, and unboundedness.
Second, a key insight from this study is that Norwegian agreement patterns are not operated solely by morphosyntactic rules but are influenced by an intricate interaction between semantics and morphosyntax (Chapter 6.2). The analyses of non-canonical agreement patterns reveal that both semantics and morphosyntactic environments in which agreement occurs contribute to the choice of agreement patterns. The anti-pancake construction exemplifies this interaction clearly. The choice between anti-pancake and canonical neuter agreement is influenced by the two competing factors working against each other, On the one hand, Norwegian speakers tend to resolve the semantic mismatch between the high degree of individuation of non-age-specific subjects and the lower degree of individuation typically associated with the neuter gender. On the other hand, Norwegian speakers tend to resolve the morphosyntactic mismatch between the subject and the predicative adjective when the mismatch is evident because of the short distance. The analyses from this dissertation highlight the importance of paying attention to both semantic and morphosyntactic properties and the interplay between them when dealing with agreement phenomena in Norwegian. The interplay between semantics and morphosyntax is fundamental to understanding the flexibility and variation observed in Norwegian agreement patterns.
To sum up, by investigating pancake-agreement, anti-pancake agreement, and neuter-collective agreement, this study has shown that Norwegian agreement patterns extend beyond morphosyntactic rules and are strongly influenced by factors such as animacy, individuation, virtuality, and unboundedness. Additionally, this dissertation provides new insights into the nature of non-canonical agreement patterns in Norwegian, demonstrating that these patterns are systematically shaped by an intricate interaction between semantics and morphosyntax.
This dissertation has implications at several levels. Within Scandinavian linguistics, this study lends empirical support to the semantic approach to non-canonical agreement, extending it beyond pancake agreement to anti-pancake and neuter collective agreement. Within Germanic linguistics, the findings contribute to a broader understanding of agreement in Germanic languages, highlighting parallels in other languages and reinforcing the role of semantic motivations in agreement phenomena.
From a typological perspective, the findings from this dissertation contribute to the cross-linguistic picture by showing that semantic factors can override morphosyntactic rules in certain constructional contexts, while morphosyntax prevails in others. This interplay suggests that typological models of agreement should allow for gradual shifts between semantic and syntactic agreement, rather than treating them as mutually exclusive.
Finally, the usage-based approach adopted in this dissertation contributes to research on Norwegian agreement in the following two respects. First, through a corpus-based investigation, this study provides empirical evidence supporting the view that non-canonical agreement patterns in Norwegian should be understood as instances of semantic agreement. Furthermore, the findings presented in this dissertation uncover previously unexplored properties of Norwegian agreement. By systematically analyzing corpus data, this usage-based study not only substantiates the insightful native intuitions proposed in earlier studies but also sheds light on previously overlooked aspects of Norwegian agreement.
