This introductory chapter sets out to provide a concise overview of the approach to inflectional morphology based on Harmonic Serialism (a local, derivational version of Optimality Theory) developed in Müller (2020; 2024) and Gleim, Müller, Privizentseva & Tebay (2021; 2023). Furthermore, the eight contributions to the present volume are briefly sketched, and the arguments for adopting an approach based on Harmonic Serialism (rather than one based on Standard Parallel Optimality Theory) that emerge from the analyses collected here are laid out.

In Potawatomi transitive animate (TA) verb forms, the first person plural object marker /nan/ occurs in the context of a third person subject but not in the context of a second person subject (see Hockett (1948), Stump (2001)). At first sight, this suggests that /nan/ must be a portmanteau agreement marker, i.e., lexically specified for φ-features of both arguments. However, portmanteau agreement gives rise to problems in some existing approaches to morphological exponence. In this paper I show that in a derivational presyntactic merge-based morphological theory like Müller’s (2020) Inflectional Morphology in Harmonic Serialism, one can make do without this assumption by deriving the distribution of /nan/ via the timing of exponent Merge operations relative to the end of the derivation. Building on Andermann (2022, 2023), I assume that in Potawatomi, exponents realising a third person argument are always merged before exponents realising a speech act participant (SAP, i.e., first or second person), and if both arguments are SAP, exponents realising the object are merged before exponents realising the subject. I furthermore assume that the 1pl object marker /nan/ can only be merged if no structure-building feature is unchecked, i.e., if the subject has already been realised, which is the case in the context of a 3rd person subject but not in the case of a 2nd person subject. Thus, the distribution of /nan/ is an epiphenomenon of the order of morphological operations. Moreover, by parametrising the order of merge operations, the analysis can be extended to portmanteau agreement in Lakota.

In Itelmen, subject agreement surfaces as a prefix and object agreement in the form of a suffix. When it comes to third person direct object agreement, these particular suffixes additionally encode the features of the subject and are hence called (person) portmanteaus. Such portmanteaus only surface in the suffix slot and never in the prefix slot, yielding a subject-object asymmetry. Bobaljik (2000) derives this asymmetry by cyclic morphological realization, with Itelmen portmanteaus emerging as object suffixes with contextual subject features. This assumption turns out to be problematic from the perspective of partially superfluous extended exponence (Caballero and Harris 2012). The present reanalysis is based on the pre-syntactic morphological theory of Inflectional Morphology in Harmonic Serialism (IMHS; Müller 2020). Assuming the order of object>subject, the derivation firstly selects the most specific compatible object exponent; portmanteaus are also available for selection and are only selected within this sequence. When the derivation continues to the subject exponent, all possible portmanteaus have already been considered for selection. This prevents a portmanteau from taking the place of a subject exponent.

Western Keresan intransitives fall into three categories, which mark their only subject like subjects of transitive verbs, objects of transitive verbs and a mixture of the two respectively. These three surface types are reducible to two underlying structures, corresponding to whether the subject is an internal or external argument. The third, mixed, type results from the lexically induced addition of a deponent morphosyntactic feature to either of the other types without deleting or suppressing a potentially conflicting feature (additive deponency). The surface forms are derived with Serial Distributed Optimality (SDO), where (if present) number agreement is inserted before person agreement. Number agreement crucially influences the competition for insertion between adequate person agreement prefixes. On the other hand, the selection of the person prefix counterbleeds the insertion of number agreement, which is taken as an argument against a fully parallel approach.

The present paper addresses extended exponence and stem allomorphy in Modern Greek past Tense morphology and provides a new analysis of these phenomena in the framework of Inflectional Morphology in Harmonic Serialism (Müller (2020)). The paper focuses on the active past Tense augment -e (whose phonological realization depends on whether or not an antepenultimate stress pattern can be established), and the passive past augment -ik (see,e.g., Joseph and Janda (1993), Spyropoulos and Revithiadou (2009)). Past Tense is not only realized by the augment but also by a suffix that additionally encodes person and number. This is an instance of partially superfluous extended exponence in the sense of Caballero and Harris (2012): The more generic augment realizing past Tense would strictly speaking seem to be superfluous, since it cooccurs with a more specific exponent that realizes past Tense as well as φ-features. The analysis presented in this paper is based on the approach to extended exponence in (Müller (2020)), which relies on the highly ranked constraint MinSat (‘Minimize Satisfaction’). MinSat ensures that the more specific exponent is blocked at an early stage in the derivation, such that the more generic exponent can show up, and the more specific exponent can only be added later (to realize the φ-features that the augment does not cover). In the second part of the paper, the analysis is extended to stem allomorphy. It is argued that there are two different types of stem allomorphy in Modern Greek verb inflection: The morpho-syntactic feature on a stem may, or may not, satisfy the Max constraint that holds for it; if it does, affixal inflection is blocked, and if it does not, inflection is possible.

This paper discusses the position of the reflexive morpheme in Lithuanian and other Baltic languages. In Lithuanian, the reflexive appears at the right edge of unprefixed verbs, but is placed between the rightmost prefix and the root of the verb otherwise. I propose an analysis of the observed pattern in which the reflexive always starts out to the left of the verbal root, i.e. in the position where it surfaces in prefixed verbs. The word-final position in unprefixed verbs is claimed to be secondary, resulting from a constraint prohibiting the reflexive from being the initial element within a prosodic domain. This constraint causes rightward movement of the reflexive. The analysis developed here employs the Harmonic Serialism framework, in line with recent research applying this framework to the domain of inflectional morphology (Müller 2020, Gleim et al. 2023, Andermann 2023). The article also identifies some interesting patterns found in Lithuanian participles and in prefixed verbal forms in Curonian Latvian. An extension of the analysis is proposed to account for the data found in Curonian, and possible ramifications for Lithuanian are discussed.

Urarina exhibits intriguing affix orders, where number affixes precede and follow person and negation affixes. In this article, I show that affix predictability reflects the order of number and person affixes in Urarina. This result is based on a corpus study that determines the predictability of Urarina person, number and negation affixes. It shows that less predictable affixes are positioned closer to the stem than more predictable affixes. However, the position of negation with respect to person and number does not pattern the same way. From this study, I concur that affix predictability reflects and directly determines affix ordering but only to a certain degree. I implement this result into Harmonic Serialism in Morphology using gradient symbolic representations to encode affix predictability. The theory derives all person, number and negation affix orders in Urarina by allowing affix predictability to interact with category-specific alignment constraints in cyclic morphological structure building.

In the Austronesian language Irarutu (spoken in North-West New Guinea), alienable possession is expressed by a person/number prefix, whereas inalienable possession is realized by (what may at first sight look like) a circumfix, i.e., there is a co-occurrence of the prefix used for alienable possession and an additional person suffix (Voorhoeve (1989), van den Berg & Matsumara (2008), Jackson (2014)). Interestingly, there is a subclass of inalienable kinship terms where the exponent that is otherwise realized as a suffix emerges as an inner prefix. This phenomenon lends itself to an analysis in terms of morphological movement in an approach to inflectional morphology based on Harmonic Serialism.

The argument for morphological movement consists of three steps. First, it is shown that prefix/suffix status cannot be an inherent property of possessive exponents but follows from interacting alignment constraints. Second, closer inspection reveals that the exponents in question are not in fact specified for possession marking; this in turn implies that they are in a partially superfluous extended exponence relation, where the more general exponent enters the structure before the more specific one. And third, this means that the more general exponent has to have undergone morphological movement from a suffix to a prefix position in surface representations where both exponents are prefixes.

This paper presents a study of personal pronouns in Finno-Ugric languages. Personal pronouns in these languages show a curious phenomenon of possessive exponents that duplicate φ-features of the pronominal root. While such duplication is attested in a vast majority of Uralic languages, it does not apply to all pronominal forms. The paper investigates the distribution of possessive marking within the paradigms and shows that it is governed by the case hierarchy (Blake 1994, Caha 2009). I suggest that the pattern instantiates partially superfluous extended exponence and requires a morphological analysis. I show that the data are straightforwardly derived under the harmonic serialist approach to inflectional morphology by employing the MinSat constraint proposed in Müller (2020) and the observable variation within Uralic can be accounted for by invoking harmonic alignment of scales and subsequent constraint conjunction, as proposed by Aissen (1999, 2003).

This introductory chapter sets out to provide a concise overview of the approach to inflectional morphology based on Harmonic Serialism (a local, derivational version of Optimality Theory) developed in Müller (2020; 2024) and Gleim, Müller, Privizentseva & Tebay (2021; 2023). Furthermore, the eight contributions to the present volume are briefly sketched, and the arguments for adopting an approach based on Harmonic Serialism (rather than one based on Standard Parallel Optimality Theory) that emerge from the analyses collected here are laid out.

In Potawatomi transitive animate (TA) verb forms, the first person plural object marker /nan/ occurs in the context of a third person subject but not in the context of a second person subject (see Hockett (1948), Stump (2001)). At first sight, this suggests that /nan/ must be a portmanteau agreement marker, i.e., lexically specified for φ-features of both arguments. However, portmanteau agreement gives rise to problems in some existing approaches to morphological exponence. In this paper I show that in a derivational presyntactic merge-based morphological theory like Müller’s (2020) Inflectional Morphology in Harmonic Serialism, one can make do without this assumption by deriving the distribution of /nan/ via the timing of exponent Merge operations relative to the end of the derivation. Building on Andermann (2022, 2023), I assume that in Potawatomi, exponents realising a third person argument are always merged before exponents realising a speech act participant (SAP, i.e., first or second person), and if both arguments are SAP, exponents realising the object are merged before exponents realising the subject. I furthermore assume that the 1pl object marker /nan/ can only be merged if no structure-building feature is unchecked, i.e., if the subject has already been realised, which is the case in the context of a 3rd person subject but not in the case of a 2nd person subject. Thus, the distribution of /nan/ is an epiphenomenon of the order of morphological operations. Moreover, by parametrising the order of merge operations, the analysis can be extended to portmanteau agreement in Lakota.

In Itelmen, subject agreement surfaces as a prefix and object agreement in the form of a suffix. When it comes to third person direct object agreement, these particular suffixes additionally encode the features of the subject and are hence called (person) portmanteaus. Such portmanteaus only surface in the suffix slot and never in the prefix slot, yielding a subject-object asymmetry. Bobaljik (2000) derives this asymmetry by cyclic morphological realization, with Itelmen portmanteaus emerging as object suffixes with contextual subject features. This assumption turns out to be problematic from the perspective of partially superfluous extended exponence (Caballero and Harris 2012). The present reanalysis is based on the pre-syntactic morphological theory of Inflectional Morphology in Harmonic Serialism (IMHS; Müller 2020). Assuming the order of object>subject, the derivation firstly selects the most specific compatible object exponent; portmanteaus are also available for selection and are only selected within this sequence. When the derivation continues to the subject exponent, all possible portmanteaus have already been considered for selection. This prevents a portmanteau from taking the place of a subject exponent.

Western Keresan intransitives fall into three categories, which mark their only subject like subjects of transitive verbs, objects of transitive verbs and a mixture of the two respectively. These three surface types are reducible to two underlying structures, corresponding to whether the subject is an internal or external argument. The third, mixed, type results from the lexically induced addition of a deponent morphosyntactic feature to either of the other types without deleting or suppressing a potentially conflicting feature (additive deponency). The surface forms are derived with Serial Distributed Optimality (SDO), where (if present) number agreement is inserted before person agreement. Number agreement crucially influences the competition for insertion between adequate person agreement prefixes. On the other hand, the selection of the person prefix counterbleeds the insertion of number agreement, which is taken as an argument against a fully parallel approach.

The present paper addresses extended exponence and stem allomorphy in Modern Greek past Tense morphology and provides a new analysis of these phenomena in the framework of Inflectional Morphology in Harmonic Serialism (Müller (2020)). The paper focuses on the active past Tense augment -e (whose phonological realization depends on whether or not an antepenultimate stress pattern can be established), and the passive past augment -ik (see,e.g., Joseph and Janda (1993), Spyropoulos and Revithiadou (2009)). Past Tense is not only realized by the augment but also by a suffix that additionally encodes person and number. This is an instance of partially superfluous extended exponence in the sense of Caballero and Harris (2012): The more generic augment realizing past Tense would strictly speaking seem to be superfluous, since it cooccurs with a more specific exponent that realizes past Tense as well as φ-features. The analysis presented in this paper is based on the approach to extended exponence in (Müller (2020)), which relies on the highly ranked constraint MinSat (‘Minimize Satisfaction’). MinSat ensures that the more specific exponent is blocked at an early stage in the derivation, such that the more generic exponent can show up, and the more specific exponent can only be added later (to realize the φ-features that the augment does not cover). In the second part of the paper, the analysis is extended to stem allomorphy. It is argued that there are two different types of stem allomorphy in Modern Greek verb inflection: The morpho-syntactic feature on a stem may, or may not, satisfy the Max constraint that holds for it; if it does, affixal inflection is blocked, and if it does not, inflection is possible.

This paper discusses the position of the reflexive morpheme in Lithuanian and other Baltic languages. In Lithuanian, the reflexive appears at the right edge of unprefixed verbs, but is placed between the rightmost prefix and the root of the verb otherwise. I propose an analysis of the observed pattern in which the reflexive always starts out to the left of the verbal root, i.e. in the position where it surfaces in prefixed verbs. The word-final position in unprefixed verbs is claimed to be secondary, resulting from a constraint prohibiting the reflexive from being the initial element within a prosodic domain. This constraint causes rightward movement of the reflexive. The analysis developed here employs the Harmonic Serialism framework, in line with recent research applying this framework to the domain of inflectional morphology (Müller 2020, Gleim et al. 2023, Andermann 2023). The article also identifies some interesting patterns found in Lithuanian participles and in prefixed verbal forms in Curonian Latvian. An extension of the analysis is proposed to account for the data found in Curonian, and possible ramifications for Lithuanian are discussed.

Urarina exhibits intriguing affix orders, where number affixes precede and follow person and negation affixes. In this article, I show that affix predictability reflects the order of number and person affixes in Urarina. This result is based on a corpus study that determines the predictability of Urarina person, number and negation affixes. It shows that less predictable affixes are positioned closer to the stem than more predictable affixes. However, the position of negation with respect to person and number does not pattern the same way. From this study, I concur that affix predictability reflects and directly determines affix ordering but only to a certain degree. I implement this result into Harmonic Serialism in Morphology using gradient symbolic representations to encode affix predictability. The theory derives all person, number and negation affix orders in Urarina by allowing affix predictability to interact with category-specific alignment constraints in cyclic morphological structure building.

In the Austronesian language Irarutu (spoken in North-West New Guinea), alienable possession is expressed by a person/number prefix, whereas inalienable possession is realized by (what may at first sight look like) a circumfix, i.e., there is a co-occurrence of the prefix used for alienable possession and an additional person suffix (Voorhoeve (1989), van den Berg & Matsumara (2008), Jackson (2014)). Interestingly, there is a subclass of inalienable kinship terms where the exponent that is otherwise realized as a suffix emerges as an inner prefix. This phenomenon lends itself to an analysis in terms of morphological movement in an approach to inflectional morphology based on Harmonic Serialism.

The argument for morphological movement consists of three steps. First, it is shown that prefix/suffix status cannot be an inherent property of possessive exponents but follows from interacting alignment constraints. Second, closer inspection reveals that the exponents in question are not in fact specified for possession marking; this in turn implies that they are in a partially superfluous extended exponence relation, where the more general exponent enters the structure before the more specific one. And third, this means that the more general exponent has to have undergone morphological movement from a suffix to a prefix position in surface representations where both exponents are prefixes.

This paper presents a study of personal pronouns in Finno-Ugric languages. Personal pronouns in these languages show a curious phenomenon of possessive exponents that duplicate φ-features of the pronominal root. While such duplication is attested in a vast majority of Uralic languages, it does not apply to all pronominal forms. The paper investigates the distribution of possessive marking within the paradigms and shows that it is governed by the case hierarchy (Blake 1994, Caha 2009). I suggest that the pattern instantiates partially superfluous extended exponence and requires a morphological analysis. I show that the data are straightforwardly derived under the harmonic serialist approach to inflectional morphology by employing the MinSat constraint proposed in Müller (2020) and the observable variation within Uralic can be accounted for by invoking harmonic alignment of scales and subsequent constraint conjunction, as proposed by Aissen (1999, 2003).