Papers by Marko Simonovic
Glossa: a journal of general linguistics, 2023
Slovenian verbs have been analyzed as displaying non-local root allomorphy which gets triggered a... more Slovenian verbs have been analyzed as displaying non-local root allomorphy which gets triggered across the theme vowel (Božič 2016; 2019). We reevaluate the data (using a larger set of verbs) and reconsider the structure of the verb in Slovenian (focusing primarily on the position of the theme vowel). We show that root allomorphy in Slovenian only occurs in a very limited number of theme-vowel classes, making the inventory of theme-vowel classes crucial for the correct analysis of root allomorphy. We further show that an overwhelming majority of verbs with root allomorphy is ambiguous in that these verbs can be analyzed as belonging to different theme-vowel classes. We propose a way of resolving the ambiguity and present an analysis of root allomorphy in terms of phonologically conditioned allomorph selection. In the proposed model, which combines Distributed Morphology and Optimality Theory, a single Vocabulary Item with a complex phonological representation gets inserted and then ...
Glossa: a journal of general linguistics, 2022
This paper focuses on the e/i theme vowel class of verbs in Slovenian to bring together two seemi... more This paper focuses on the e/i theme vowel class of verbs in Slovenian to bring together two seemingly unrelated debates: (i) the debate on the correlation between theme vowel classes with certain argument structures and (ii) the debate on the status of derivational affixes within the framework of Distributed Morphology. Our core data come from a list of 108 unaccusative verbs obtained using the adjectival active l-participles as an unaccusativity diagnostic. We show that (i) no unaccusative verbs belong to the two largest theme vowel classes in Slovenian (a/a and i/i), whereas (ii) the two big theme vowel classes tend to get accusative arguments quite frequently. Most importantly, (iii) the e/i-class stands out since more than one half of the unaccusative sample falls into it. The e/i-class is furthermore exceptional in that its theme vowel surfaces in adjectival l-participles, it is the theme vowel class to which inchoatives in inchoative-causative pairs belong and it behaves unifo...
Journal of Slavic linguistics (published, this is a preprint draft), 2022
Several recent accounts (Lowenstamm, 2014; Nevins, 2015; Creemers et al., 2017) couched in the fr... more Several recent accounts (Lowenstamm, 2014; Nevins, 2015; Creemers et al., 2017) couched in the framework of Distributed Morphology (Halle and Marantz, 1993, 1994) argue for extending the separation between roots and categorial heads to derivational affixes. Such approaches offer a straightforward account of affixes which surface under different categorial embeddings (e.g.,-ant, both in the noun defendant and in the adjective defiant) by viewing these affixes as roots. In this article, the affixes-as-roots approach is applied to Slovenian affixes. An account is proposed of the variable prosodic behaviour of Slovenian derivational affixes, which behave as either stress-attracting or stress-neutral. It is shown that Slovenian derivational affixes have no lexical stress and all their prosodic effects follow from the structures in which they occur. Specifically, stress-attracting behaviour is a result of the fact that sequences of roots with no intermediate functional structure (the so-called radical cores) are spelled out to phonology without any prosodic specification. Phonology then assigns the default final prosody to such sequences, creating the illusion of accented derivational affixes. The proposed account is applied to two affixes av and ov, which occur across categorial embeddings (nominal, verbal, adjectival).
Journal of Slavic linguistics (Special FASL 29 issue, accepted, this is a preprint draft), 2023
lavic verbal prefixes have been divided into lexical, argued to be basegenerated below the verb, ... more lavic verbal prefixes have been divided into lexical, argued to be basegenerated below the verb, and superlexical prefixes, analyzed as basegenerated in the verb’s functional projection. This falsely predicts their
linearization on different sides of the verb. We argue that all verbal prefixes
correspond to various argument-structural effects, and are base generated
as heads of a voice projection. On this basis, we formulate an analysis at the
PF interface in terms of Embick & Noyer (2001), heads linearizing heads
in a bottom-up fashion, yet every step in the linearization is followed by
local dislocation (LD). LD is by default string-vacuous; inversion occurs
only when the head being linearized partakes in allomorphic realization –
being either allomorphic itself, or part of a context in rules of allomorphic
realization. Strict implementation of this simple algorithm results exactly
in the observed surface order
Linguistica, 2020
In this paper we consider several instances of the Slovenian affix ov, which surfaces in many, ap... more In this paper we consider several instances of the Slovenian affix ov, which surfaces in many, apparently unrelated contexts. Here we focus on (i) ov in verbs, where it can act as an imperfectivizer or a verbalizer, (ii) ov found in possessive adjectives and kind adjectives derived from nouns, (iii) ov which precedes the adjectiviser (e)n in denominal adjectives, and (iv) ov in nominal declension (acting as a genitive case ending in dual and plural or as a dual/plural augment). Building on the observation that certain affixes function either as inflectional or as derivational (see Simonović and Arsenijević 2020), and working within a Distributed Morphology approach which postulates that derivational affixes should be analyzed as roots (e.g. Lowenstamm 2014), we argue for a single multifunctional ov. This ov is a potentially meaningless root that can take as a complement other roots (thus forming a “radical core”) or phrases, resulting in different structures and consequently differe...
Open Linguistics 4(1), 2018
In a number of Slavic and Germanic languages, various derivational affixes and morphological patt... more In a number of Slavic and Germanic languages, various derivational affixes and morphological patterns of Latin origin are relatively common, and bear effects as abstract as deriving event nouns from verbs and property nouns from adjectives. This seems to contradict the general observation that abstract morphology typically is not subject to borrowing. We discuss the status of two Serbo-Croatian (S-C) nominalizing Latinate suffixes, –cija and –itet, complemented by one Germanic suffix, –er. On our analysis, these are not borrowed suffixes and derivational patterns, in the sense that they were present in another language and got copied into S-C, but rather suffixes and patterns which emerged within S-C, more specifically in the borrowed stratum of the S-C lexicon. Crucial factors in their emergence were the shared semantic properties of the nouns ending in the respective sequences (–cija, –itet and –er), and the quantitative properties of these sequences closely matching those of native derivational suffixes. Pragmatic, phonological and prosodic constraints apply to these derivations to the effect that the suffixes that have emerged in the borrowed domain of the lexicon never enter a competition with the native nominalization patterns.
Sinfonija 6, 2014
SerBoCroatian (henceforth SBC) displays a zero:a alternation within paradigms, which always has t... more SerBoCroatian (henceforth SBC) displays a zero:a alternation within paradigms, which always has the effect of disrupting consonant clusters in certain forms of a word (e.g. in koverat~kovert-a ‘envelop NOMSG~GENSG’). This alternation is traditionally termed nepostojano a ‘wobbly a’. We present an overview of the contexts in which the ‘wobbly a’ alternation is attested and argue for an OT-analysis which requires no special ‘wobbly’ segments, and distinguishes between an epenthetic and a metathetic wobbly a, which surface in different phonological contexts triggered by different morphemes. We further account for the observed underapplication of the two types of ‘wobbly a’ using a special type of Faithfulness indexed to loanwords (DEP(LOAN)) and Lexical Conservatism, respectively. We finally discuss dynamics of the present-day SBC lexicon, in which the class of words which allow wobbly a acts like a closed class and the traditionally inseparable ST clusters (which were originally treated as single segments) show different signs of reanalysis into regular clusters.
I discuss productive verb derivation patterns with special attention for allowed preservation of ... more I discuss productive verb derivation patterns with special attention for allowed preservation of prosodic contrast. A preliminary study yielded two sets of extremely productive verb derivational patterns, which allow very different amount of prosodic contrast. Specific Serbo-Croatian and Slovenian suffixes will be analysed and compared. In this abstract we illustrate the main generalisations using Serbo-Croatian examples.
On the one hand, suffixes -ova- and -a- which have been used for denominal verbs and are currently mostly employed for the integration of English verbs allow the preservation of prosodic patterns, even some which are not attested elsewhere in the verbal system. For instance, S-C verbs generally follow the pattern where the stem-final syllable has a high tone in the infinitive form. However, denominal verbs such as uHžin-a-ti ‘to snack’, based on the noun uHžina ‘snack’, form an exception. Similarly, non-initial high tones have to spread to the left in native verbs, forming rising accents (čeHk-a-ti derives doH-čeHk-a-ti and not *do-čeHk-a-ti), but loanwords allow patterns like rikveHst-a-ti *riHkveHst-a-ti ‘ to request’ based on the English verb.
On the other hand, suffixes -i:va-, -a:va- and -V:.a- which are involved in co-called secondary imperfectivisation (derivation of an imperfective verb from a verb which was made perfective by prefixation) all have predictable prosodic patterns, thereby allowing no prosodic contrast, as can be seen from the same pattern in the rightmost column below. This pattern is analysable in terms of the prosodic dominance of the suffix, but it is also worth considering an option in which the surfacing pattern is post-lexical (see Simonović & Arsenijević 2014, for a similar account).
Original IMP
Prefixed PERF
Secondary IMP
pi:Hs-aH-ti
‘to write’
pre-pi:Hs-aH-ti
‘to copy’
pre-pis-i:Hv-aH-ti
‘to copy repeatedly’
čeHk-a-ti
‘to wait’
doH-čeHk-a-ti
‘to meet’
do-ček-i:Hv-aH-ti
‘to meet repeatedly’
Finally, I observe that the two kinds of derivational patterns seem to be strictly separated in the dynamics of verb aspect encoding in the lexicon, so that hardly any denominal and borrowed verbs ever derive secondary imprefectives.
I discuss two possible approaches to accounting for the observed grouping and separation of the productive derivational patterns. On the one hand, the two kinds of derivation involve radically different syntactic structures, so I test an account couched in Distributed Morphology (Halle & Maranz 1994, Mavrin 2003), which allows syntax to impose or delete prosody of the stem. On the other hand, representational phonological approaches will be tested, which involve the well-known mechanisms of OO-FAITH (Benua 1997) and Lexical Conservatism (Steriade 1997). I conclude that while phonological approaches seem to be suited to account for the exceptional preservation of contrast, a more syntactic approach is required to account for the loss of contrast in secondary imperfectives.
In a number of Slavic and Germanic languages, various derivational affixes and morphological patt... more In a number of Slavic and Germanic languages, various derivational affixes and morphological patterns of Latin origin are relatively common, and bear effects as abstract as deriving event nouns from verbs and property nouns from adjectives. This seems to contradict the general observation that abstract morphology typically is not subject to borrowing. We discuss the status of two Serbo-Croatian (S-C) nominalizing Latinate suffixes, -cija and -itet, which display relatively high values of productivity and frequency. On our analysis, these are not borrowed suffixes and derivational patterns, in the sense that they were present in another language and got copied into S-C, but rather suffixes and patterns which emerged within S-C, more specifically in the Latinate stratum of the S-C lexicon. In other words, we argue that members of these two classes of nominalizations were first borrowed as simplex words. After demonstrating that the sequences -cija and -itet display quantitative patterns characteristic of the productive native suffixes, we argue that these quantitative effects, in conspiracy with the shared semantic properties of the nouns ending in these sequences, have led to a reanalysis of the initial simplex borrowed forms as derived nouns involving a stem and the suffix -cija or -itet, respectively. Pragmatic, phonotactic and prosodic constraints apply to these derivations to the effect that the suffixes that have emerged in the borrowed domain of the lexicon never enter a competition with the native nominalization patterns.
: Rad se bavi upotrebom glagola trebati, od njegove težnje ličnoj upotrebi u pojedinim značenji... more : Rad se bavi upotrebom glagola trebati, od njegove težnje ličnoj upotrebi u pojedinim značenjima, preko rigidne verzije norme koja predviđa isključivo bezličnu upotrebu, do hibridnih upotreba kao posledice sukoba prethodna dva parametra i pokušaja liberalizacije norme. Osnovna pitanja su vezana za visok stepen svesti prosečnog govornika o normi u ovom domenu, pojavu hibridnih varijanata, mehannizme upotrebe kojima se govornici upravljaju, gde prepoznajemo primenu odgovarajućeg post-gramatičkog filtera, te sintaksičkih i semantičkih karakteristika glagola trebati u savremenom kolokvijalnom jeziku, za koje pokazujemo da su daleko bliže obrascu upotrebe koji dozvoljava lične oblike.
We examine two types of deverbal nominalisations in Serbo-Croatian, both derived by the same suff... more We examine two types of deverbal nominalisations in Serbo-Croatian, both derived by the same suffix -VV.je (suffix -je which lengthens the final syllable of the stem). One type is derived from imperfective verbs; it involves productive nominalizations with gerund semantics, and it is characterized by a prosodic pattern matching at least one paradigmatic form of the verb. The other is derived from perfective verbs; it involves idiosyncratic nominalisations, both in terms of the verbs which can or cannot derive them, and in terms of the semantic interpretation, and they are characterized by a prosodic pattern that never fits any of the forms in the paradigm of the verb: the high tone on the two final syllables, and the stress on the penultimate.
We propose that the notion of paradigm should be defined as the domain of systematic productivity of a stem, under full semantic transparency. Our analysis of the facts observed is that members of the paradigm - in our case the imperfective nominalisations - are subject to Steriade's Lexical Conservatism, which presses them to use a prosodic pattern available within the paradigm. When the result of the morphological process (adding the suffix -VV.je to the stem) is a new stem with an own paradigm, the new stem is fused and treated as a simplex without a prosodic specification in the lexicon. While it has been shown that the post-lexical, i.e. default, prosodic pattern in S-C is that with the high tone on the initial syllable (Zec 1999, Simonović 2012), we argue that lexical items with a long vowel select for a different post-lexical default, namely the one with the high tone on the final syllable. The surface prosody is then the result of the general rule of tone spreading in S-C (e.g. Zec 1999).
We argue that this picture fits well a more general view that the lexicon has the structure of a coconut: a soft core with productive elements deriving large paradigms connected by shared lexical prosodic shapes, a hard outer layer of extreme idiosyncratic items in respect of productivity, semantics and prosody, and a middle semi-hardened layer of derived words exiled from the paradigms of their stems, with both semantics and prosody somewhere between the two extremes.
We present and analyse a set of interface phenomena showing important correlations between certai... more We present and analyse a set of interface phenomena showing important correlations between certain phonological regularities on the one hand and a set of syntactic and semantic properties of the respective expressions on the other. Serbo-Croatian deadjectival nominalizations typically exhibit two different prosodic patterns: 1) prosody faithful to the base i.e. surface prosody of the lexical adjective (e.g. Ispraavnoost ‘correctness’, derived from Ispraavan ‘correct’), 2) a rising span over a long closed penultimate syllable and the syllable following it (e.g. isprAAvnOOst ‘correctness’). We formulate a generalization where, all things being equal, nominalised predicational structures correspond to (1), while nominalised stems correspond to (2). The paper provides a formal model of the syntactic and semantic as well as the phonological reality of these nominalisations, and an attempt at explaining these facts.
We present and analyse a set of interface phenomena showing important correlations between certai... more We present and analyse a set of interface phenomena showing important correlations between certain phonological regularities on the one hand and a set of syntactic and semantic properties of the respective expressions on the other. Serbo-Croatian deadjectival nominalizations typically exhibit two different prosodic patterns: 1) prosody faithful to the base i.e. surface prosody of the lexical adjective (e.g. Ispraavnoost 'correctness', derived from Ispraavan 'correct'), 2) a rising span over a long closed penultimate syllable and the syllable following it (e.g. isprAAvnOOst 'correctness'). We formulate a generalization where, all things being equal, nominalised predicational structures correspond to (1), while nominalised stems correspond to (2). The paper provides a formal model of the syntactic and semantic as well as the phonological reality of these nominalisations, and an attempt at explaining these facts.
We present and analyse a set of interface phenomena showing important correlations between syntac... more We present and analyse a set of interface phenomena showing important correlations between syntactic and semantic aspects on the one hand and phonological effects on the other. Serbo-Croatian deadjectival nominalizations exhibit two different prosodic patterns: 1) prosody faithful to the base (the prosody of the lexical adjective), 2 a rising span over a long closed penultimate syllable and the syllable following it. We formulate a generalization where, all things being equal, nominalised predicational structures correspond to (1) (also when the base has postcyclic prosody), while nominalised stems correspond to (2). The paper provides a formal model of the syntactic and semantic as well as the phonological reality of these nominalizations, and an attempt at explaining these facts.
This contribution is challenging the traditional concept that languages in contact are separate (... more This contribution is challenging the traditional concept that languages in contact are separate (or separable) entities. An alternative model is proposed, inspired by an approach from science and technology studies: agential realism (Barad 2007). The central concept is that of inter-language mappings, the correspondences between structures of languages which get stabilised in the contact speakers’ community and make whole subsets of the source language lexicon in principle borrowable and potentially already borrowed. Importantly, the inter-language mappings cannot be properly assigned to any of the languages in contact and they enforce a profound rethinking of the way the most important processes of language contact - borrowing and code-switching - are conceptualised. Using examples from a variety of languages, I am showing how borrowing needs to be conceptualised not as adaptation to a new system, but as creation of a relation which cannot be properly situated in any of the varieties in contact.
ANTROPLOGIA - Zbornik radova Odeljenja humanističkih nauka Istraživačke stanice Petnica, 2007
Standard language awareness of the language community – the
case of Serbian in Serbia
The obj... more Standard language awareness of the language community – the
case of Serbian in Serbia
The objective of this research is to map out the effects of the standardization
process in terms of the standard language awareness created in the language
community. Standardization is regarded as a type of communication, whereby the
standardization organ (the authorized institution) communicates the norm to the
language community by means of the channel of standardization, which can be
direct (e.g. written prescriptive work) and indirect (through reference to «model»
local dialects or concrete language use).
There were three groups
of subjects: students of non-philological disciplines from: 1) Belgrade,
2) Vojvodina, 3) other parts of Serbia. It was relevant to analyze results of the
research in reference to the fact that there are two prestige centers (in accordance
with two mentioned Universities) and two different (competitive) prestige regions
(standards).
Zbornik "Jezik, književnost, marginalizacija"
Arsenijević, Boban & Sabina Halupka (eds.) Srpski jezik u savremenoj lingvističkoj teoriji.
"Fleeting or Simply not There?: A Generative Analysis of the „Fleeting a“ Alternations in Serbo-C... more "Fleeting or Simply not There?: A Generative Analysis of the „Fleeting a“ Alternations in Serbo-Croatian
In this contribution, the Serbo-Croatian vowel:zero alternation, traditionally termed Fleeting/Wobbly a (nepostojano a) is considered. The proposed analysis, de-veloped for nouns and adjectives and couched in terms of OptimalityTheory, brings a conceptual separation of the fleeting a which surfaces in bare forms (e.g. članak „article.Nom.Sg“and tanak „thin.Masc.Nom.Sg“) from that in genitive plural forms (e.g. članaka „article.Gen.Pl“ and banaka „bank.Gen.Pl“). Although both a's are inserted into stems (which do not contain them: /člank/, /tank/, /bank/), this happens under different conditions for both. The first a is epenthetic, inserted to prevent word-final complex codas from surfacing. This epenthesis is entirely phonological, occurs in all native stems and all the ex-ceptions can be accounted for using a special FAITHFULNESS constraint for loanwords:FAITH(LOAN). The other a is methatetic and gets inserted because the genitive plural suffix has the form /aːaː/, which makes metathesis the preferred outcome in combinations of the type /člank+aː aː/.
The metathetic a, unlike the epenthetic counterpart, occurs in contexts defined by lexical and morphological factors, mainly the shape of the paradigm in which the form occurs. The discussion of the specific con-texts in which the two types of fleeting a occur, will demonstrate the usefulness of the lexical constraint family LEXICAL CONSERVATISM (Steriade 1997), which militates against the introduction of new allomorphs into the paradigm. Moreover, the discussion of some specific issues will lead us to register at least two phenomena which could show restructuring with respect to the situation described in the standard grammars. First, the option for the suffix /aːaː/ to have the just the exponent –a seems to have disappeared for feminine nouns (e.g. forms such as svâdbā „wedding.Gen.Pl“, vóžnjā „drive.Gen.Pl“ have become ungrammatical). Second, the sequences which the grammar previously treated as single segments ([st], [ʃt], [zd] and [ʒd]) are losing the single segment status and becoming real clusters. "
"The goal of this contribution is twofold. On the one hand, I am looking at the “normative” conti... more "The goal of this contribution is twofold. On the one hand, I am looking at the “normative” continuum (comparable to crazy – awkward/weird – strange – peculiar – normal – common in English) in Serbian and Dutch, in an attempt to identify the main similarities and differences using the Natural Semantic Metalanguage (Goddard & Wierzbicka 2002). On the other hand, I am proposing moving away from the comparison paradigm. Instead, I am developing an account approaching languages diffractively (Barad 2007), as an ongoing intra-action. Under such an approach, the role of the practices of the (broadly defined) bilingual speaker changes radically: the speaker is invited to live the difference productively and to overcome the ideology of sameness and representationalism. The bilingual speaker is always consigned to being more-than-normal and accountable for how she speaks the constitutive boundary.
But there is more. The goal of this contribution is to spoil othering/normativity/universality for you, strategically using the insight that not only are different things “crazy” in different discourses, but also the very scale of measuring “crazy” is discourse/language-specific and ever-becoming. In this sense, there is no transcendental norm(ality) to measure against, only what we make of what has been entrusted to us."
Uploads
Papers by Marko Simonovic
linearization on different sides of the verb. We argue that all verbal prefixes
correspond to various argument-structural effects, and are base generated
as heads of a voice projection. On this basis, we formulate an analysis at the
PF interface in terms of Embick & Noyer (2001), heads linearizing heads
in a bottom-up fashion, yet every step in the linearization is followed by
local dislocation (LD). LD is by default string-vacuous; inversion occurs
only when the head being linearized partakes in allomorphic realization –
being either allomorphic itself, or part of a context in rules of allomorphic
realization. Strict implementation of this simple algorithm results exactly
in the observed surface order
On the one hand, suffixes -ova- and -a- which have been used for denominal verbs and are currently mostly employed for the integration of English verbs allow the preservation of prosodic patterns, even some which are not attested elsewhere in the verbal system. For instance, S-C verbs generally follow the pattern where the stem-final syllable has a high tone in the infinitive form. However, denominal verbs such as uHžin-a-ti ‘to snack’, based on the noun uHžina ‘snack’, form an exception. Similarly, non-initial high tones have to spread to the left in native verbs, forming rising accents (čeHk-a-ti derives doH-čeHk-a-ti and not *do-čeHk-a-ti), but loanwords allow patterns like rikveHst-a-ti *riHkveHst-a-ti ‘ to request’ based on the English verb.
On the other hand, suffixes -i:va-, -a:va- and -V:.a- which are involved in co-called secondary imperfectivisation (derivation of an imperfective verb from a verb which was made perfective by prefixation) all have predictable prosodic patterns, thereby allowing no prosodic contrast, as can be seen from the same pattern in the rightmost column below. This pattern is analysable in terms of the prosodic dominance of the suffix, but it is also worth considering an option in which the surfacing pattern is post-lexical (see Simonović & Arsenijević 2014, for a similar account).
Original IMP
Prefixed PERF
Secondary IMP
pi:Hs-aH-ti
‘to write’
pre-pi:Hs-aH-ti
‘to copy’
pre-pis-i:Hv-aH-ti
‘to copy repeatedly’
čeHk-a-ti
‘to wait’
doH-čeHk-a-ti
‘to meet’
do-ček-i:Hv-aH-ti
‘to meet repeatedly’
Finally, I observe that the two kinds of derivational patterns seem to be strictly separated in the dynamics of verb aspect encoding in the lexicon, so that hardly any denominal and borrowed verbs ever derive secondary imprefectives.
I discuss two possible approaches to accounting for the observed grouping and separation of the productive derivational patterns. On the one hand, the two kinds of derivation involve radically different syntactic structures, so I test an account couched in Distributed Morphology (Halle & Maranz 1994, Mavrin 2003), which allows syntax to impose or delete prosody of the stem. On the other hand, representational phonological approaches will be tested, which involve the well-known mechanisms of OO-FAITH (Benua 1997) and Lexical Conservatism (Steriade 1997). I conclude that while phonological approaches seem to be suited to account for the exceptional preservation of contrast, a more syntactic approach is required to account for the loss of contrast in secondary imperfectives.
We propose that the notion of paradigm should be defined as the domain of systematic productivity of a stem, under full semantic transparency. Our analysis of the facts observed is that members of the paradigm - in our case the imperfective nominalisations - are subject to Steriade's Lexical Conservatism, which presses them to use a prosodic pattern available within the paradigm. When the result of the morphological process (adding the suffix -VV.je to the stem) is a new stem with an own paradigm, the new stem is fused and treated as a simplex without a prosodic specification in the lexicon. While it has been shown that the post-lexical, i.e. default, prosodic pattern in S-C is that with the high tone on the initial syllable (Zec 1999, Simonović 2012), we argue that lexical items with a long vowel select for a different post-lexical default, namely the one with the high tone on the final syllable. The surface prosody is then the result of the general rule of tone spreading in S-C (e.g. Zec 1999).
We argue that this picture fits well a more general view that the lexicon has the structure of a coconut: a soft core with productive elements deriving large paradigms connected by shared lexical prosodic shapes, a hard outer layer of extreme idiosyncratic items in respect of productivity, semantics and prosody, and a middle semi-hardened layer of derived words exiled from the paradigms of their stems, with both semantics and prosody somewhere between the two extremes.
case of Serbian in Serbia
The objective of this research is to map out the effects of the standardization
process in terms of the standard language awareness created in the language
community. Standardization is regarded as a type of communication, whereby the
standardization organ (the authorized institution) communicates the norm to the
language community by means of the channel of standardization, which can be
direct (e.g. written prescriptive work) and indirect (through reference to «model»
local dialects or concrete language use).
There were three groups
of subjects: students of non-philological disciplines from: 1) Belgrade,
2) Vojvodina, 3) other parts of Serbia. It was relevant to analyze results of the
research in reference to the fact that there are two prestige centers (in accordance
with two mentioned Universities) and two different (competitive) prestige regions
(standards).
In this contribution, the Serbo-Croatian vowel:zero alternation, traditionally termed Fleeting/Wobbly a (nepostojano a) is considered. The proposed analysis, de-veloped for nouns and adjectives and couched in terms of OptimalityTheory, brings a conceptual separation of the fleeting a which surfaces in bare forms (e.g. članak „article.Nom.Sg“and tanak „thin.Masc.Nom.Sg“) from that in genitive plural forms (e.g. članaka „article.Gen.Pl“ and banaka „bank.Gen.Pl“). Although both a's are inserted into stems (which do not contain them: /člank/, /tank/, /bank/), this happens under different conditions for both. The first a is epenthetic, inserted to prevent word-final complex codas from surfacing. This epenthesis is entirely phonological, occurs in all native stems and all the ex-ceptions can be accounted for using a special FAITHFULNESS constraint for loanwords:FAITH(LOAN). The other a is methatetic and gets inserted because the genitive plural suffix has the form /aːaː/, which makes metathesis the preferred outcome in combinations of the type /člank+aː aː/.
The metathetic a, unlike the epenthetic counterpart, occurs in contexts defined by lexical and morphological factors, mainly the shape of the paradigm in which the form occurs. The discussion of the specific con-texts in which the two types of fleeting a occur, will demonstrate the usefulness of the lexical constraint family LEXICAL CONSERVATISM (Steriade 1997), which militates against the introduction of new allomorphs into the paradigm. Moreover, the discussion of some specific issues will lead us to register at least two phenomena which could show restructuring with respect to the situation described in the standard grammars. First, the option for the suffix /aːaː/ to have the just the exponent –a seems to have disappeared for feminine nouns (e.g. forms such as svâdbā „wedding.Gen.Pl“, vóžnjā „drive.Gen.Pl“ have become ungrammatical). Second, the sequences which the grammar previously treated as single segments ([st], [ʃt], [zd] and [ʒd]) are losing the single segment status and becoming real clusters. "
But there is more. The goal of this contribution is to spoil othering/normativity/universality for you, strategically using the insight that not only are different things “crazy” in different discourses, but also the very scale of measuring “crazy” is discourse/language-specific and ever-becoming. In this sense, there is no transcendental norm(ality) to measure against, only what we make of what has been entrusted to us."
linearization on different sides of the verb. We argue that all verbal prefixes
correspond to various argument-structural effects, and are base generated
as heads of a voice projection. On this basis, we formulate an analysis at the
PF interface in terms of Embick & Noyer (2001), heads linearizing heads
in a bottom-up fashion, yet every step in the linearization is followed by
local dislocation (LD). LD is by default string-vacuous; inversion occurs
only when the head being linearized partakes in allomorphic realization –
being either allomorphic itself, or part of a context in rules of allomorphic
realization. Strict implementation of this simple algorithm results exactly
in the observed surface order
On the one hand, suffixes -ova- and -a- which have been used for denominal verbs and are currently mostly employed for the integration of English verbs allow the preservation of prosodic patterns, even some which are not attested elsewhere in the verbal system. For instance, S-C verbs generally follow the pattern where the stem-final syllable has a high tone in the infinitive form. However, denominal verbs such as uHžin-a-ti ‘to snack’, based on the noun uHžina ‘snack’, form an exception. Similarly, non-initial high tones have to spread to the left in native verbs, forming rising accents (čeHk-a-ti derives doH-čeHk-a-ti and not *do-čeHk-a-ti), but loanwords allow patterns like rikveHst-a-ti *riHkveHst-a-ti ‘ to request’ based on the English verb.
On the other hand, suffixes -i:va-, -a:va- and -V:.a- which are involved in co-called secondary imperfectivisation (derivation of an imperfective verb from a verb which was made perfective by prefixation) all have predictable prosodic patterns, thereby allowing no prosodic contrast, as can be seen from the same pattern in the rightmost column below. This pattern is analysable in terms of the prosodic dominance of the suffix, but it is also worth considering an option in which the surfacing pattern is post-lexical (see Simonović & Arsenijević 2014, for a similar account).
Original IMP
Prefixed PERF
Secondary IMP
pi:Hs-aH-ti
‘to write’
pre-pi:Hs-aH-ti
‘to copy’
pre-pis-i:Hv-aH-ti
‘to copy repeatedly’
čeHk-a-ti
‘to wait’
doH-čeHk-a-ti
‘to meet’
do-ček-i:Hv-aH-ti
‘to meet repeatedly’
Finally, I observe that the two kinds of derivational patterns seem to be strictly separated in the dynamics of verb aspect encoding in the lexicon, so that hardly any denominal and borrowed verbs ever derive secondary imprefectives.
I discuss two possible approaches to accounting for the observed grouping and separation of the productive derivational patterns. On the one hand, the two kinds of derivation involve radically different syntactic structures, so I test an account couched in Distributed Morphology (Halle & Maranz 1994, Mavrin 2003), which allows syntax to impose or delete prosody of the stem. On the other hand, representational phonological approaches will be tested, which involve the well-known mechanisms of OO-FAITH (Benua 1997) and Lexical Conservatism (Steriade 1997). I conclude that while phonological approaches seem to be suited to account for the exceptional preservation of contrast, a more syntactic approach is required to account for the loss of contrast in secondary imperfectives.
We propose that the notion of paradigm should be defined as the domain of systematic productivity of a stem, under full semantic transparency. Our analysis of the facts observed is that members of the paradigm - in our case the imperfective nominalisations - are subject to Steriade's Lexical Conservatism, which presses them to use a prosodic pattern available within the paradigm. When the result of the morphological process (adding the suffix -VV.je to the stem) is a new stem with an own paradigm, the new stem is fused and treated as a simplex without a prosodic specification in the lexicon. While it has been shown that the post-lexical, i.e. default, prosodic pattern in S-C is that with the high tone on the initial syllable (Zec 1999, Simonović 2012), we argue that lexical items with a long vowel select for a different post-lexical default, namely the one with the high tone on the final syllable. The surface prosody is then the result of the general rule of tone spreading in S-C (e.g. Zec 1999).
We argue that this picture fits well a more general view that the lexicon has the structure of a coconut: a soft core with productive elements deriving large paradigms connected by shared lexical prosodic shapes, a hard outer layer of extreme idiosyncratic items in respect of productivity, semantics and prosody, and a middle semi-hardened layer of derived words exiled from the paradigms of their stems, with both semantics and prosody somewhere between the two extremes.
case of Serbian in Serbia
The objective of this research is to map out the effects of the standardization
process in terms of the standard language awareness created in the language
community. Standardization is regarded as a type of communication, whereby the
standardization organ (the authorized institution) communicates the norm to the
language community by means of the channel of standardization, which can be
direct (e.g. written prescriptive work) and indirect (through reference to «model»
local dialects or concrete language use).
There were three groups
of subjects: students of non-philological disciplines from: 1) Belgrade,
2) Vojvodina, 3) other parts of Serbia. It was relevant to analyze results of the
research in reference to the fact that there are two prestige centers (in accordance
with two mentioned Universities) and two different (competitive) prestige regions
(standards).
In this contribution, the Serbo-Croatian vowel:zero alternation, traditionally termed Fleeting/Wobbly a (nepostojano a) is considered. The proposed analysis, de-veloped for nouns and adjectives and couched in terms of OptimalityTheory, brings a conceptual separation of the fleeting a which surfaces in bare forms (e.g. članak „article.Nom.Sg“and tanak „thin.Masc.Nom.Sg“) from that in genitive plural forms (e.g. članaka „article.Gen.Pl“ and banaka „bank.Gen.Pl“). Although both a's are inserted into stems (which do not contain them: /člank/, /tank/, /bank/), this happens under different conditions for both. The first a is epenthetic, inserted to prevent word-final complex codas from surfacing. This epenthesis is entirely phonological, occurs in all native stems and all the ex-ceptions can be accounted for using a special FAITHFULNESS constraint for loanwords:FAITH(LOAN). The other a is methatetic and gets inserted because the genitive plural suffix has the form /aːaː/, which makes metathesis the preferred outcome in combinations of the type /člank+aː aː/.
The metathetic a, unlike the epenthetic counterpart, occurs in contexts defined by lexical and morphological factors, mainly the shape of the paradigm in which the form occurs. The discussion of the specific con-texts in which the two types of fleeting a occur, will demonstrate the usefulness of the lexical constraint family LEXICAL CONSERVATISM (Steriade 1997), which militates against the introduction of new allomorphs into the paradigm. Moreover, the discussion of some specific issues will lead us to register at least two phenomena which could show restructuring with respect to the situation described in the standard grammars. First, the option for the suffix /aːaː/ to have the just the exponent –a seems to have disappeared for feminine nouns (e.g. forms such as svâdbā „wedding.Gen.Pl“, vóžnjā „drive.Gen.Pl“ have become ungrammatical). Second, the sequences which the grammar previously treated as single segments ([st], [ʃt], [zd] and [ʒd]) are losing the single segment status and becoming real clusters. "
But there is more. The goal of this contribution is to spoil othering/normativity/universality for you, strategically using the insight that not only are different things “crazy” in different discourses, but also the very scale of measuring “crazy” is discourse/language-specific and ever-becoming. In this sense, there is no transcendental norm(ality) to measure against, only what we make of what has been entrusted to us."
which loanword integration relies on inter-language mappings, defined as correspondences
between source-language and recipient-language structures, established in the bilingual
community and productively applied to new items. I furthermore argue that every instance of
regular borrowing strengthens the mapping, rendering it more salient in the contact.
I demonstrate that both phonological and phonetic theories of loanword integration
systematically disprefer taking inter-language knowledge into account, as they assume
monolingual knowledge to be pre-given and inter-language knowledge to require additional
assumptions. This rather contingent tendency has led to the generalised assumption that each
loanword is integrated without any relation to other loanwords from the same source language.
I propose a novel approach which capitalises on the bilingual speakers’ knowledge of cognates
and acknowledges inter-language mappings as concomitant to lasting contact and regular
borrowing. I show how such an approach would lead to the repositioning and reconciliation of
the traditional phonological and phonetic stances. Furthermore, I suggest that such an approach
would resolve the traditional dispute on the distinction between code-switching and borrowing.
On the one hand, suffixes -ova- and -a- which have been used for denominal verbs and are currently mostly employed for the integration of English verbs allow the preservation of prosodic patterns, even some which are not attested elsewhere in the verbal system. For instance, S-C verbs generally follow the pattern where the stem-final syllable has a high tone in the infinitive form. However, denominal verbs such as uHžin-a-ti ‘to snack’, based on the noun uHžina ‘snack’, form an exception. Similarly, non-initial high tones have to spread to the left in native verbs, forming rising accents (čeHk-a-ti derives doH-čeHk-a-ti and not *do-čeHk-a-ti), but loanwords allow patterns like rikveHst-a-ti *riHkveHst-a-ti ‘ to request’ based on the English verb.
On the other hand, suffixes -i:va-, -a:va- and -V:.a- which are involved in co-called secondary imperfectivisation (derivation of an imperfective verb from a verb which was made perfective by prefixation) all have predictable prosodic patterns, thereby allowing no prosodic contrast, as can be seen from the same pattern in the rightmost column below. This pattern is analysable in terms of the prosodic dominance of the suffix, but it is also worth considering an option in which the surfacing pattern is post-lexical (see Simonović & Arsenijević 2014, for a similar account).
Original IMP
Prefixed PERF
Secondary IMP
pi:Hs-aH-ti
‘to write’
pre-pi:Hs-aH-ti
‘to copy’
pre-pis-i:Hv-aH-ti
‘to copy repeatedly’
čeHk-a-ti
‘to wait’
doH-čeHk-a-ti
‘to meet’
do-ček-i:Hv-aH-ti
‘to meet repeatedly’
Finally, I observe that the two kinds of derivational patterns seem to be strictly separated in the dynamics of verb aspect encoding in the lexicon, so that hardly any denominal and borrowed verbs ever derive secondary imprefectives.
I discuss two possible approaches to accounting for the observed grouping and separation of the productive derivational patterns. On the one hand, the two kinds of derivation involve radically different syntactic structures, so I test an account couched in Distributed Morphology (Halle & Maranz 1994, Mavrin 2003), which allows syntax to impose or delete prosody of the stem. On the other hand, representational phonological approaches will be tested, which involve the well-known mechanisms of OO-FAITH (Benua 1997) and Lexical Conservatism (Steriade 1997). I conclude that while phonological approaches seem to be suited to account for the exceptional preservation of contrast, a more syntactic approach is required to account for the loss of contrast in secondary imperfectives.
spell-out of a morphological form depends not only on the features that the morphological form
involves, but also on the way other forms within the paradigm are spelled out. The constraints
observed, which from the functional perspective militate against proliferation of syncretism, call
for a model of morphology which is sensitive to the inventory of the paradigm.
In Serbo-Croatian, feminine singular (FSg), neuter plural (NPl), paucal (Pauc) and collective
(Coll) forms all display homophonous nominative (Nom) suffixes, surfacing as -a. Based on
their syntactic, semantic and morphological relations, Arsenijević (2016) argues that the three
semantically plural ones among these forms: Coll, NPl and Pauc, all involve a paucal classifier,
which is marked by the same ending -a. This view is taken as the departure point of the present
paper, which focuses on the question why these four categories take different endings in other
case forms, i.e. why FSg and Coll forms take the case endings traditionally referred to as the FSg
declension, and NPl and Pauc the ones traditionally labeled MPl declension, as illustrated in (1).
(1) FSg, Coll, NPl and Pauc in –a through case forms
FSg Coll NPl Pauc
Nom bub-a gospod-a sel-a sel-a
Gen bub-e: gospod-e: se:l-a: se:l-a:
Dat, Loc bub-i gospod-i sel-ima sel-ima
Acc bub-u gospod-u sel-a sel-a
Inst bub-o:m gospod-o:m sel-ima sel-ima
We argue that in fact in all the four classes above, the suffix -a is the same suffix which belongs
to the FSg declension. The marked and more semantically restricted ’MPl’ declension is forced
in the NPl and Pauc cases not by their plural semantics, which is also present in Coll, but rather
as the plural-marked default. This is a last resort solution resulting from the fact that NPl and
Pauc are part of a paradigm in which the singular endings are taken by the semantically singular
forms, against which a contrast in number needs to be established. Collective nouns such as
gospoda ’gentry’ lack forms with singular reference in their paradigm, and derive singulatives to
refer to singular entities (godpod-in ’gentleman’, where -in is a singulative suffix), which
licenses them to take the less unmarked set of endings.
We support this view by the behavior of these nouns in respect of number agreement. While
nouns which lack any plural semantics, traditional FSg, never trigger plural agreement, the other
three classes, Coll, NPl and Pauc, when heading the clausal subject, are all compatible with a
plural finite verb. Coll forms, which morphosyntactically group with FSg, also allow for a
morphosyntactic singular agreement. This indicates that Coll shares the feature specification of
Pauc and NPl, due to which it licenses plural agreement, but its singular declension in addition
licenses singular agreement.
(2) a. Buba leti/*lete.
bug.FSg fly.3Sg/3Pl
’The bug is flying.’
b. Gospoda jede/jedu.
gentry eat.3Sg/3Pl
‘The gentlemen are eating.’
c. Sela *gori/gore.
village.NPl burn.3Sg/3Pl
‘The villages are burning.’
d. Dva sela *gori/gore.
2 villages burn.3Sg/3Pl
‘Two villages are burning.’
Our crucial piece of evidence, both for the claim that we are dealing with one and the same
suffix -a and for the interdependencies between the different forms within the same paradigm
comes from phonology. Inflectional suffixes in SC typically carry no prosody: they attach to the
stem which either preserves its lexically specified contour or receives postlexical stress (Zec
1999, Simonović & Arsenijević 2014). The ending -a belongs to a small class of exceptions from
this generalization. It carries tone, which spreads onto the base it attaches to, in the three stable
classes out of the four in
(1): in FSg, NPl and Coll (most contemporary speakers do not decline Pauc in oblique cases,
using instead the form in -a in all case contexts, which is probably part of the answer why in a
majority of speakers Pauc fails to generalize the prosodic shape of the suffix -a). Moreover, the
suffix -a imposes its prosodic pattern onto all the oblique suffixes in the declension, be it the
regular FSg declension as in FSg and Coll or the last resort MPl set of endings, as in NPl. The
only exception is the structural accusative case, which preserves the original prosodic shape of
the base when it is not syncretic with the nominative.
(3) The prosodic shape imposed by the suffix -a
FSg NPl
Nom vod+aH > voHdaH ramen+aH > rameHnaH
Gen voHde:H rame:Hna:H
Dat, Loc voHdiH rameHnimaH
Acc voHdu (prosody of the base) rameHnaH
Inst voHdo:mH rameHnaH
The empirical picture described poses a serious problem for most available theories of word
formation and infection morphology, because for forms referring to pluralities, they require a
spell-out mechanism sensitive to whether the paradigm includes forms with singular reference.
We discuss a tentative model of these effects, as well as of the effects of the suffix -a on the
selection of the declension class and on its prosodic shape.
References:
Zec, D. 1999. Footed Tones and Tonal Feet. Phonology 16: 225-264.
Simonović, Marko and Boban Arsenijević. 2014. Regular and honorary membership: On two
kinds of deverbal nouns in Serbo-Croatian. Linguе e Linguaggio 8.1, 185-210.
Arsenijević, Boban. 2016. Gender, like classifiers, marks uniform atomicity: Evidence from
Serbo-Croatian. CLS (Chicago Linguistic Society) 52, University of Chicago, 21-23. 4. 2016.
(1)
Base word Paradigmatic nominalisation Non-paradigmatic nom.
pUtovati “to travel” pUtovaanje “travelling” putovAAnjE “trip”
Opaasan “dangerous” Opaasnoost “dangerousness” opAAsnOOst “danger”
Simonović & Arsenijević (2014) analyse the non-paradigmatic pattern as the result of what they term Forced Lexicalisation, a process which forces a derived form to lexicalise as a separate item. Furthermore, they argue that this pattern is the result of the non-paradigmatic word having lost its internal structure, due to which the prosodic information from the elements gets deleted, so that the post-lexical prosodic pattern results. In claiming that the rising span is a post-lexical pattern, they go against the grain of most analyses of S-C prosody, according to which the only post-lexical is one with an initial falling accent (as in the base words in (1)). In this presentation, we are comparing the two cases of alleged deaccentuation and adding another one – a very limited number of nouns which display the same rising pattern only in some fixed expressions in the Locative Singular form.
(2)
na svakom korAAku u tom slučAAjU visi u vazdUUhu
at every step.Loc in that case.Loc hangs in air.Loc
“everywhere” “then” “is in the air”
na petom kOraaku/*korAAkU u Petrovom slUčaaju/*slučAAjU amonijak u vAzduuhu/*vazdUUhU
at fifth step.Loc in P’s case.Loc ammonia in air.Loc
“at the fifth step” “in P’s case” “ammonia in the air”
As will be shown, the three classes indeed display deaccentuation, but differ radically in productivity. Deaccentuation is obligatory in all deverbal nouns which are derived from perfective verbs. In deadjectival nominalisations deaccentuation is possible whenever the base is toneless (displays an initial falling accent), and the final syllable of the base is long and closed. In nouns, this pattern is limited to a few toneless nouns which have non-animate referents. We argue that this distribution is expected due to S-C-internal factors. For, instance, the existence of the aspect distinction, in combination with the fact that deverbal nominalisations are derived using the pre-lengthening suffix -VV.je, whose semantics is incompatible with perfective aspect (as argued by Simonović & Arsenijević (2014)) forms a perfect constellation for the obligatorification of deaccentuation based on purely formal conditions. In nouns, on the other hand, there is neither of these conditions. We discuss the mutual interaction of properties from different domains of grammar, such as the morphological (i.e. syntactic) structural complexity, specification for particular syntactic features (animacy), compositionality of interpretation, and the availability of the lexical specification of prosody, and its implication for the grammar of S-C.
References
Arsenijević, Boban and Marko Simonović. 2013. The role of syntax in stress assignment in Serbo-Croatian. In Folli, Raffaella, Christina Sevdali, and Robert Truswell (eds.). Syntax and its Limits. Oxford: Oxford University Press, pp. 190-217.
Simonović Marko & Boban Arsenijević. 2014. Regular and honorary membership: on two kinds of deverbal nouns in Serbo-Croatian, in Lingue e linguaggio 2/2014, pp. 185-210.
The goal of this paper is mapping out the verb prosody of the varieties of Serbo-Croatian (henceforth, SC),
which have undergone a switch from a pitch-accent system to a stress system and consider these data
in the context of prosodic theory. Many if not all stress varieties have developed an innovative system, in
which verbs have no lexical prosody, but surface with a morphologically conditioned stress pattern. This
constellation, which has independently emerged in a number of ‘peripheral’ varieties of standard S-C,
may very well be unique within Slavic. We argue that two typologically unmarked properties of stress
systems emerge when a pitch-accent system collapses into a stress system, viz. (a) stem-based stress and
(b) morphologically conditioned verb stress (e.g. Alderete 2001).
Standard S-C is a pitch-accent system based on the central, Neo-Štokavian dialects, in which the
distribution of tone guides the distribution of stress (for a recent overview, see Werle 2009). High tone is
associated either to one or to two syllables in a word and stress is always on the leftmost syllable
endowed with a high tone. The former prosodic pattern (two adjacent syllables with a high tone, stress on
the first of them) is usually referred to as rising accent, whereas the latter pattern (stress and tone cooccur
on one syllable) is referred to as falling accent. Falling accents can only occur word-initially. In (1)
some examples of nouns with rising (a) and falling (b) accents are provided. Capitalisation us used to
represent the presence of a high tone, double vowels represent vowel length.
All accounts of S-C prosody agree that the prosody (i.e. tone and vowel length) is lexically encoded, so
that stems like tema contrast underlyingly with stems like škola and stems like muzika contrast
underlyingly with stems like kapara or terasa.
In a considerable part of the S-C language area the local dialect had a rather different prosodic system
and, as a consequence of the standardisation process, in these areas the standard prosody has undergone
significant simplifications. In quite a few regional centres a stress system without tone and vowel length
has emerged. Importantly, this happened both in those areas where a stress system without vowel length
had existed in the local dialect (e.g. in Niš), but also in those areas where all local dialects have a pitchaccent
system (e.g. in Bor, Zagreb and Rijeka), in both cases with very similar results. We term such
stress varieties spoken in urban areas “creolised standard varieties” (henceforth CSV’s). In the nouns in
(1), the new stress pattern can be derived relatively straight-forwardly from the standard system, by
removing all the tone and length information from the phonological representation.
As can be read off the examples in (2), the words which we have seen in (1) now display much less
contrast than in the classical standard system. The disyllables actually allow only one possible pattern,
whereas for trisyllables some lexical encoding is still required, but there is only a two-way contrast
(antepenultimate vs. penultimate). The nominal system of CSV’s therefore has purely stem-based
prosody, unlike that of the classical standard S-C. The simplification pattern is even more prominent in
verbs, where, as illustrated for in the present tense in (3), stress always falls on the syllable before the
tense ending. As the three examples of present tense verb forms in (3) show, the location of stress in the
classical standard language is no longer the predictor of stress in CSV. The pattern in CSV is overall
similar to that attested in Non-Neoštokavian dialects (illustrated in 3c by a simplified representation,
disregarding tonal information and vowel length), with one crucial difference: the type with stress on the
present-tense ending (stoj-í) is missing. Another class in which the standard stress pattern won in CSV’s
is illustrated in the verb forms in (5), prefixed versions of those in (4). In order to appreciate these
examples, the infinitive form (e.g. úb-iti) needs to be taken into account, as its ending also needs to
remain unstressed. For the verbs in (4) there is no avoiding stressed endings (because the stem bcontains
no stressable elements) and the substrate dialect pattern is probably preferred because it keeps
the paradigm uniform. However, in the verbs in (5) the substrate dialect ‘unnecessarily’ keeps the stress
on the ending, so that CSV goes with the standard in the present tense form and introduces a relative
prosodic innovation in the negated present tense form. In sum, the emergent CSV system has a
morphologically conditioned verb stress, which is the second unmarked property of stress systems we
observe.
2
(1a) Rising accent tEEmA mUzIka terAsA
‘theme’ ‘music’ ‘terrace’
(1b) Falling accent škOOla kApara
‘school’ ‘down payment’
(2) téma múzika terása
škóla kápara
(3a) Classical standard stOj-II tOvAr-ii UrAAd-ii
(3b) Creolised standard varieties stój-i továr-i urád-i
(3c) Non-Neoštokavian dialects (stress only) stoj-í továr-i urád-i
‘s/he stands’ ‘s/he loads’ ‘s/he does’
(4a) Classical standard b-Iti b-Ijee nE b-Ijee
(4b) Creolised standard varieties b-íti b-íje ne b-íje
(4c) Non-Neoštokavian dialects b-íti b-íje ne b-íje
‘to beat’ ‘s/he beats’ ‘s/he does not beat’
(5a) Classical standard Ub-Iti Ub-ijee nE ub-ijee
(5b) Creolised standard varieties úb-iti úb-ije ne úb-ije
(5c) Non-Neoštokavian dialects ub-íti ub-íje ne ub-íje
‘to kill’ ‘s/he kills’ ‘s/he does not kill’
References
Werle, Adam. 2009. Word, Phrase, and Clitic Prosody in Bosnian, Serbian, and Croatian. University of
Massachusetts Amherst PhD dissertation. Amherst: GLSA.
Alderete, John (2001). Morphologically governed accent in Optimality Theory. New York & London:
Routledge.
Only slightly crazy! Phonologically restricted root allomorphy in Dutch, Slovenian and Serbo-Croatian
Marko Simonović, University of Nova Gorica
A small group of Dutch nouns and verbs illustrated in (1) display a pattern of root allomorphy in
which allomorphs with long and short stem-nal vowels havea phonologically optimizing distribution:
the long-vowel version shows up in open syllables, the short-vowel version shows up in closed syllables.
(1) Dutch stems with root alternation
a. nouns SG PL
lɔt lo:tǝn ‘lottery ticket(s)’
slɔt slo:tǝn ‘lock(s)
b. 3rd person present tense of the verb komen ‘come’ SG PL
kɔmt ko:mǝn
Kager (2009) proposes an OT analysis with the two root allomorphs stored as an unordered pair of
underlying representations (e.g. the UR for ‘lottery ticket’ is/lɔt~lo:t/)and GEN can choose between
the two when creating candidates (but it cannot combine elements of both in the same candidate). The
distribution is determined by Markedness constraints generally invisible in Dutch evaluations because
they are dominated by Faithfulness. Kager’sanalysisaccounts for the data, but given the very few axes
which constitute the environment for the two versions ([∅]and [t] for the short-vowel version, [ǝn] for
the long-vowel version) it is hard to exclude a reanalysis which views the pattern as tied to specic
morphemes rather than being caused by otherwise invisible Markedness. A reanalysisalong these lines
could assume that the sux [ǝn] has an extra mora, which typically gets erased due to
RFAF (McCarthy & Prince 1995), but surfaces when the UR of the stem isan
unordered pair of vowel-length allomorphs. The dilemma cannot be resolved using Dutch data only.
However, Kager’s analysis makes important and testable cross-linguistic predictions. First, if the
distribution of the allomorphs is purely phonologically determined, there should exist languages in
which allomorphs with the same phonological features (e.g. a long vowel) realise one syntactic feature
in one inectional class and the opposite feature in another inectional class purely depending on the
phonology of the inectional morphemes. Such a language would look something like Dutch′ (2),
having two dierent inectional classes, one like the actual Dutch (2a) and one in which the singular
sux is /ǝ/, while the plural sux is /s/. The long-vowel and short-vowelallomorphs surface either in
the singular or in the plural, depending on the phonological context presented by NUM morphology.
(2) Dutch′ a. Inectional class 1 (SG=∅, PL=[ǝn]) b. Inectional class 2 (SG=[ǝ], PL=[s])
SG PL SG PL
lɔt lo:tǝn ‘lottery ticket(s)’ slo:tǝ slɔts ‘lock(s)’
Second, if Kager’s analysis is correct, the ‘revelatory power’ of ordered pairs of allomorphs should not
stop at Markedness: there should also exist languages in which allomorph selection is inuenced by
otherwise masked Faithfulness constraints. I argue that Slovenian is like Dutch′ and that
Serbo-Croatian has a unordered-pair UR which reveals a Faithfulness/Markedness interaction
otherwise entirely masked by higher-ranked constraints.
Slovenian verb forms have the structure Root(√)-Theme(θ)-Tense & Agreement Morphology().
There is considerable root allomorphy (kl-a-ti ‘to slaughter’ vs. kol-je-m ‘I slaughter’) and a verb can
take dierent theme vowels in dierent tenses (kl-a-ti ‘to slaughter’ vs. kol-je-m ‘I slaughter’).
Traditional grammars therefore analyse the sequences √+θ as single units called bases bases and each
verb is analysed as having two distinct bases: an ininitivaland a present-tense base. I will use θ1 and θ2
for the two theme vowels. There are a dozen possible combinations of θ1 and θ2
, which I consider
inectional classes. One of the reasons for the traditional concept of the base is the generalisation that
in cases of root allomorphy, each root allomorph co-occurs with one of the theme vowels.
In the wake of recent development in Distributed Morphology (Lowenstamm 2014) Iassume that (a)
derivational axes are roots and (b) sequences shared by dierent words can in principle beanalysed as
roots, even in the face of no evident semantic content. Such an analysis resolves atraditional problem
presented in Šekli (2005), where there are two kinds of verbs in -ovati, -ujem: (a) those in which these
sequences can be analysed as containing a derivational sux, e.g. dar-ov-a-ti ‘to present’, dar-u-je-m ‘I
present’ (related to dar ‘present’),and those in which there is only rootallomorphy and no derivational
sux, e.g. kov-a-ti ‘to forge’, ku-je-m ‘I forge’. My reanalysis amounts to allowing k-ov-a-ti ‘to forge’,
k-u-je-m ‘I forge’ even though √ never occurs on its own and √ does not seem to contribute any
meaning. This analysis being in place, we can turn to ov~u allomorphy. The distribution is entirely
predictable by well-established constraints O and NC. Quite spectacularly, since √
surfaces in dierent inectional classes, itsallomorphs can appear within both traditional bases. This is
exactly what we expect from Dutch′.
(3) Slovenian
a. Inectional class 1 (θ1=a, θ2=je) b. Inectional class 2 (θ1=∅, θ2=e)
√-√-θ1
-INF √-√-θ2
-1SG √-√-θ1
-INF √-√-θ2
-1SG
k-ov-a-ti k-u-je-m rj-u-∅-ti rj-ov-e-m
‘to forge’ ‘I forge’ ‘to scream’ ‘I scream’
Since much depends on the analysis of the inectional classes, I illustrate ‘normal’ verbs in these classes.
(4) a. Inectional class 1 (θ1=a, θ2=je) b. Inectional class 2 (θ1=∅, θ2=e)
√-θ1
-INF √-θ2
-1SG √-θ1
-INF √-θ2
-1SG
or-a-ti or-je-m nɛs-∅-ti nɛs-e-m
‘to plough’ ‘I plough’ ‘to carry’ ‘I carry’
kl-a-ti kol-je-m i-∅-ti gr-e-m
‘to slaughter’ ‘I slaughter’ ‘to go’ ‘I go’
Turning to Serbo-Croatian, there is one noun which is traditionally described as displaying progressive
voicing assimilation. It is illustrated in (5) and the assumed UR would be /mozk/.
(5) Serbo-Croatian SG PL
NOM mozak mozg-ov-i
GEN mozg-a mozg-ov-a
DAT/LOC mozg-u mozg-ov-ima
INS mozg-om mozg-ov-ima
I argue that the UR is the unordered pair /mozk~mozg/ and that the evaluation also reveals an
interaction between VC and IIO
References
Kager, René. 2009. Lexical irregularity and the typology of contrast. In Hanson, Kristin and Sharon
Inkelas (eds.) The Nature of the Word: Studies in Honor of Paul Kiparsky, 397–432. Cambridge, MA:
MIT Press.
Lowenstamm, Jean. 2014. Derivational axes as roots: Phasal spell-out meets English stress shift. In
Artemis Alexiadou, Hagit Borer,and Florian Schafer (eds.) The syntax of rootsand the roots of syntax,
Oxford University Press, Oxford, 230–259.
McCarthy John and Alan Prince. 1995. Faithfulness and Reduplicative Identity. ROA-60-0000.
Šekli, Matej. 2005. Tonemski naglasni tipi glagola v (knjižni) slovenščini [Prosodic types in Standard
Slovenian verbs]. Jezikoslovni zapiski 11/2: 31–61.
nouns” generated a long-standing discussion within language contact research (see e.g. Wohlgemuth
2009 for a recent summary). More precisely, the claim was that “the borrowing language employs its
own means of denominal verbalization to turn the borrowed forms into verbs”. This can be interpreted
either as a statement about the integration pattern (which may not be mentally represented in
monolinguals) or as a claim about the syntactic representation of borrowed verbs in general, whereby
borrowed verbs contain an nP embedded under the vP. Both interpretations constitute important
hypotheses, which can serve as useful windows into the relation between morphology and phonology.
The rst
interpretation is compatible with the claim by Simonović (2015) that the integration pattern
essentially gets selected by Lexical Conservatism (Steriade 1997): the pattern with most preservation of
the properties of the base and least stem allomorphy integrates loanwords. The second interpretation
makes important predictions whose implementation is highly dependent on the theory of morphology
employed. In this presentation I use a recent elaboration of Distributed Morphology in which the
separation between roots and categorial heads is extended to derivational suxes
(Lowenstamm 2015)
and put it to use in accounting for verb borrowing and denominal verbalisation in three Western South
Slavic varieties: Slovenian, Western Serbo-Croatian (henceforth Croatian) and Eastern Serbo-Croatian
(henceforth Serbian). All three varieties have lexical prosody. Slovenian has lexically determined stress.
In Serbo-Croatian each word has a lexically determined H, and stress assignment follows from its
distribution: if the syllable with a H is initial, italso gets stress; if the syllable with a H is non-initial, the
stress goes to the preceding syllable, forming a disyllabic rising accent (Zsiga & Zec 2013).
Simonović (2018) discusses exceptional preservation of base prosody in Western South Slavic verbs,
showing that WSS verbsallow only two prosodic shapes: stress/H stem-nally
(1a)and stress/H on the
theme vowel (1b), analysed as the contrast between accented and accentless roots. The only verbs
which ever display more contrast are borrowed and denominal verbs (2). Since nouns generally allow
more prosodic contrast than verbs (Smith 2011), Simonović (2018)argues that verb prosody should be
viewed as the regular WSS prosody, whereas all the cases of additional contrast should beanalysed asa
consequence of special Faithfulness, and, at least for the classes discussed by Simonović (2018),asingle
type of special Faithfulness seems to be sucient:
NF
Smith
2001).
Against the sketched background, variation within WSS is analysed. All three varieties have two
patterns for denominal verbs which both allow for exceptional preservation of the base prosody: -a-ti
and -ov-a-ti (illustrated in 2a; a isatheme vowel in both cases, ti is the innitive
ending). Tellingly, each
variety now hasastabilised borrowing pattern in which one of the two suxes
is used for English verbs
(illustrated in 2b). The necessity ofa denominal verbalisation analysis is relatively limited for Slovenian
and Croatian, where a large majority of verbs (but not all) become reanalysable as verbalised accented
roots (all the verbs in 3 have a stem-nal
stress/H). For Serbian, however, virtually all borrowed verbs
from the modern contact with English display the intermediate root -ov-, which makes the denominal
verbalisation analysis very attractive. Completing the picture for all three varieties, we turn to older
borrowed verbs, especially those from the contact preceding the one with English, in which alarge class
of international verbs were integrated and in which no prosodic contrast is instantiated (4).
In sum, the deverbal nominalisation analysis seems to beastrong cross-linguistic tendency rather than
an absolute rule and its availability depends both on the phonological makeup of the available
denominal verbalisation patterns and on the amount of prosodic contrast in the source language.
(1) Slovenian Croatian Serbian
a. Accented √ gléd-a-ti ‘to look’ gléHd-a-ti ‘to look’ gléHd-a-ti ‘to look’
b. Unaccented √ kop-á-ti ‘to dig’ kóp-aH
-ti ‘to dig’ kóp-aH
-ti ‘to dig’
(2) Slovenian Croatian Serbian
a. Denominal verbs málic-a-ti ‘to snack’ úH
žin-a-ti úH
žin-a-ti
(cf. málica ‘snack’) (cf. úH
žin-a ‘snack’) (cf. úH
žin-a ‘snack’)
vér-ov-a-ti ‘to believe’ vjéH
r-ov-a-ti ‘to believe’ v(j)éH
r-ov-a-ti ‘to believe’
(cf. vér-a ‘faith’) (cf. vjéH
r-a ‘faith’) (cf. v(j)éH
r-a ‘faith’)
b. Borrowed verbs édit-a-ti ‘to edit’ rikvéH
st-a-ti ‘to request’ rikvéH
st-ov-a-ti ‘to request’
tríger-a-ti ‘to trigger’ inváH
jt-a-ti ‘to invite’ inváH
jt-ov-a-ti ‘to invite’
(3) Borrowed verbs which can be reanalysed as verbalised accented root
Slovenian Croatian
sénd-a-ti ‘to send’ séHnd-a-ti ‘to send’
submít-a-ti ‘to submit’ éHdiH
t-a-ti ‘to edit’
(4) International verbs
Slovenian Croatian Serbian
Innitive
asist-ír-a-ti asist-í:r-aH
-ti asist-í:r-aH
-ti
Present.1Pl asist-ír-a-mo asíst-i:H
r-a:-mo asíst-i:H
r-a:-mo
Innitive
fotograf-ír-a-ti fotograf-í:r-aH
-ti fotográf-iH
s-a-ti
Present.1Pl fotograf-ír-a-mo fotográf-i:H
r-a:-mo fotográf-iH
š-e:mo-mo
Innitive
protest-ír-a-ti protest-í:r-aH
-ti próteH
st-ov-a-ti
Present.1Pl protest-ír-a-mo protést-i:H
r-a:-mo próteH
st-uj-e:-mo
References
Lowenstamm, Jean. 2015. Derivational axes
as roots: Phasal spell-out meets English stress shift. in
Artemis Alexiadou, Hagit Borer,and Florian Schafer (eds.) The syntax of rootsand the roots of syntax,
Oxford University Press, Oxford, 230–259.
Moravcsik, Edith. 1975. Borrowed verbs. Wiener Linguistische Gazette 8.
Simonović, Marko. 2015. Lexicon immigration service - Prolegomena to a theory of loanword
integration. (280 p.). LOT Dissertation Series 393.
Simonović, Marko. 2018. There is Faith and Faith: Prosodic contrast in Slovenian and Serbo-Croatian
verb derivation. Poster presented at the 26th Manchester Phonology Meeting.
Smith, Jennifer. 2001. Lexical Category and Phonological Contrast. In R. Kirchner, J. Pater, and W.
Wikely (eds.) PETL 6: Proceedings of the Workshop on the Lexicon in Phonetics and Phonology.
Edmonton: University of Alberta, 61-72.
Smith, Jennifer. 2011. Category-specic
eects.
In Marc van Oostendorp, Colin Ewen, Beth Hume,
and Keren Rice (eds.), The Blackwell Companion to Phonology, 2439-2463. Malden, MA:
Wiley-Blackwell.
Steriade, Donca. 1997. Lexical Conservatism. In Linguistics in the Morning Calm, Selected Papers
from SICOL 1997, 157-179. Hanshin Publishing House
Wohlgemuth, Jan. 2009. A typology of verbal borrowings. Berlin; New York: Mouton de Gruyter.
Zsiga, Elizabeth C. and Draga Zec. 2013. Contextual evidence for the representation of pitch accents
in Standard Serbian. Language and Speech 56;1: 69 – 104.
prosody, is developing a prosodic system in which stressed inflectional suffixes are avoided
and stress is becoming a property of the word stem. Five case studies are shown in which stress
is moving from from inflectional suffixes to stems.
bèžīte >> bežíte (conf. Slovenian bežíte)
In a large class of verbs with a short rising accent in the present tense, the forms for 1PL and
2PL traditionally allowed the stress on the theme vowel: bežímo ‘we run’, bežíte ‘you(PL) run’
(conf. bèž-a-ti ‘to run’, bèž-ī-m ‘I run’). Nowadays this pattern is vanishing from standard
Serbo-Croatian and only bèžīmo/bèžīte is possible. We refer to normative work recognising the
decline of bežímo/bežíte and analyse a recent social media hype after a Serbian politician uttered
‘Gdȅ bežíte?’ ‘Where are you running to?’. An analysis of the ensuing parodies of his statement
shows that modern standard speakers applied the pattern to verbs in which it was never possible
(e.g. zviždíte, for actual zvíždite ‘you(PL) whistle’), which we interpret as evidence that the old
generalisations are not internalised in these speakers and that the modern Serbo-Croatian
grammar only derives bèžīmo, bèžīte.
grȁdōvā >> gradóvā, grȁdovima >> gradòvima (conf. Slovenian gradôvi)
As amply attested in normative work, the paradigm grȁdovi~gradóvā~gradòvima
‘towns.NOM~ACC~DAT.INS.LOC’ has been replaced by grȁdovi~grȁdōvā~grȁdovima,
thereby creating a new accentual paradigm. We also show that for modern speakers the pattern
with the stressed interfix -ov- is not even considered familiar for the DAT.INS.LOC and
familiar but extremely archaic for the GEN form. The pattern in -óvā is now also possible (in
parodies) in words where it was never possible, e.g. robóvā ‘slaves.GEN’ for actual ròbōvā.
kȁmenu >> kamènu, pȍglēdu >> poglédu
The singular locative suffix -u is traditionally considered tone-attracting in accentless nouns,
e.g. kȁmēn ‘stone’ and pȍglēd ‘gaze’. This pattern is unfamiliar to most modern speakers for
the words with a short penult (e.g. kamènu), while for long penults (e.g. poglédu) the pattern
has survived in some fixed expressions, but it is considered impossible outside of them e.g. (u
tvojem pȍglēdu *poglédu ‘in your gaze’). Tellingly, the tone-attracting property of the suffix
has survived in monosyllables, where it never leads to a stress shift (conf. klûb~klúbu
‘club.NOM~LOC’).
pȕtovati >> putòvati
In most standard varieties in Serbia, the denominal verbs of the type putòvati ‘to
travel’~pùtujēm ‘I travel’ are moving to the new type pȕtovati~pùtujēm. The new infinitive
prosody is clearly based on the past participle pȕtovao. Tellingly, these verbs do not join the
existing falling type rȁdovati~rȁdujēm because the present tense pùtujēm already had stress on
the stem.
Formal Description of Slavic Languages 12.5 University of Nova Gorica
December 2017
66
ˈžena >> žeˈna, ˈkonja >> koˈnja, ˈselo >> seˈlo, ˈželim >> žeˈlim
Finally, we present a small case study from the stress standard varieties which developed in
cities outside the Neo-Štokavian area (we present data from Zagreb, Rijeka, Pula, Bor and Niš).
In all these varieties, although final stress is allowed (e.g. in kriˈstal) and stressed inflectional
endings of the type žeˈlim are widely attested in surrounding dialects, stressed inflectional
endings are entirely absent. As a consequence, these varieties are unique in Serbo-Croatian (and
possibly also entire Slavic) for having lost any prosodic prominence on inflectional endings
while having lexical prosody.
Based on the 5 cases studies, we analyse the general move towards stem-stressed prosody as
the final step in a chain of language changes initiated by the Neo-Štokavian retractions, which
were automatic and contrast-preserving, but led to a massive removal of stress from inflectional
endings. We also consider the possibility that the changes in question were accelerated by the
fact that Neo-Štokavian has been the standard since mid-19th century. Finally, we discuss the
general reasons behind this language change in terms of markedness and, more specifically, the
constraints proposed within the Optimality Theory (e.g. in Kager 2000).
References
Kager. René, 2000. Stem stress and peak correspondence in Dutch. In J. Dekkers, F. van der Leeuw & J. van de Weijer (Eds.),
Optimality Theory: Phonology, Syntax, and Acquisition (pp. 121-150) (30 p.). Oxford: Oxford University Press.
word is characterised by a single tonal accent headed by the single stressed syllable of the word.
Classical descriptions distinguish between falling and rising tonal accents. In the falling accents,
stress and high tone co-occur on the same syllable, which is typically the first syllable of the
word. Depending on syllable length, there are long falling and short falling accents (lAAđa
‘boat’ and krAđa ‘stealth’, respectively). The short falling accent is assumed to be the post-
lexical accent, assigned when the lexical accent is erased or not present. Rising accents are tonal
spans with stress on the first syllable and high tone on the second syllable. Rising accents can
also be long or short, depending on the length of the stressed syllable (as in vEEnA ‘vein’ and
žEnA ‘woman’, respectively).
Serbo-Croatian suffixes display different behaviour with respect to prosody. For instance, the
suffixes in (1) behave as accent-bearing, accent-attracting, accent-neutral and accent-erasing,
respectively. Note that in the fourth example, the accent-erasing suffix leads to post-lexical
prosody i.e. initial short falling accent.
(1)
Base šEćEr rAnjIv rAnjIv pOkUš-a-ti
‘sugar’ ‘vulnerable’ ‘vulnerable’ ‘try.INF’
Derivation šećer-AnA ranjIv-Iji rAnjIv-oost pOkuš-aaj
‘sugar factory’ ‘more vulnerable’ ‘vulnerability’ ‘attempt’
In this paper we consider 3 suffixes which appear in different structural contexts and display
different prosodic behaviour in these contexts. The question guiding our analysis is
whether/how the syntactic position of the suffixes correlates with their prosodic behaviour.
The three suffixes which we analyse are -enA, -anA, and -VV.jeN (segmental content: je,
suprasegmental content: vowel length, which gets realised on the previous vowel). All three
suffixes appear in what would traditionally be classified as inflection, derivation and optionally
analysable lexicalised items.
Suffixes -enA and -anA (illustrated by the examples in 2 and 3, respectively) appear in fully
productive past participle formation (first column), relatively productive derivation of
adjectives from nominal bases (second column) and in a few adjectives derived from adjectival
bases (third column). The prosodic behaviour in (2) and (3) represents the typical (but certainly
not the only attested) pattern: the suffix is stress-neutral when used in the context of maximal
productivity, but accent-attracting when used in less productive contexts.
(2)
Base gAz-i-ti gUm-a mAAl-ii
trample.INF’ ‘rubber.N’ ‘small.DEF’
en-Derivation gAž-en gUm-En mAl-En
‘trample.pass.part’ ‘rubber.A’ ‘small’
(3)
Base glEd-a-ti plEh-a mlAAd-i
‘look.INF’ ‘tin.N.Gen.Sg’ ‘young.DEF’
an-Derivation glEd-a-an plEh-An mlAđ-An
‘look.pass.part’ ‘tin.A’ ‘young’
The suffix -VV.jeN is most frequently combined with past participles (illustrated in 4), whereby
the contribution of the suffix to the resulting prosodic pattern is predictable from the aspect of
the verb: the suffix is accent-neutral when combined with imperfective bases but accent-
Formal Description of Slavic Languages 12.5 University of Nova Gorica
December 2017
47
attracting when combined with perfective bases. Here again, the context where the suffix
appears as accent-attracting is characterised by limited productivity, whereas the context in
which the suffix is accent-neutral displays total productivity (see Simonović and Arsenijević
2014 for an analysis).
(4)
Base žRtvovaan pOžRtvovaan
sacrifice.IMP.pass.part’ ‘sacrifice.PERF.pass.part’’
VV.je-Derivation žRtvovaanje požrtvovAAnjE
‘sacrificing’ ‘sacrifice.N’
(5)
Related form(s) ImAAnje Očaajn-i pred sOboom
‘having’ ‘deperate.DEF’ ‘in front of the room’
VV.je-Derivation imAAnjE očajAAnje pREEdsOOblje
‘property’ ‘despair’ ‘antechambre’
A detailed study of the 3 suffixes has been conducted using a corpus, a modern reverse
dictionary and various available descriptions of Serbo-Croatian. Based on the outcomes, we
argue that Serbo-Croatian systematically allows for 3 types stem+suffix constellations: those in
which both elements can contribute to the resulting prosody, those in which only the syntactic
head can influence the resulting prosody, and those in which neither element can influence the
resulting prosody.
We observe that the choice between the possible constellations is restricted by the syntax, but
not entirely determined by it. This matches the fact that some of the combinations seem to have
switched classes (or belong to different classes in different varieties). For instance, for what we
quoted as mlAđ-An in (3), older sources (e.g. Stevanović 1979 and Babić 2002) only mention
mlAđ-an (with accent-erasing prosody).
Finally, Serbo-Croatian generalisations are pitted against recent theories of syntax-prosody
interface, most prominently Revithiadou (1999) and the necessary modifications to the theory
will be proposed.
References
Babić, Stjepan. 2002. Tvorba riječi u hrvatskome književnome jeziku. Zagreb: Nakladni zavod Globus (HAZU).
Revithiadou, Anthi. 1999. Headmost Accent Wins: Head Dominance and Ideal Prosodic Form in Lexical Accent Systems.
Ph.D. thesis, University of Leiden.
Simonović, Marko & Boban Arsenijević. 2014. Regular and honorary membership: On two kinds of deverbal nouns in Serbo-
Croatian. Lingue e Linguaggio.
Stevanovic, Mihailo. 1979. Savremeni srpskohrvatski jezik: Gramatički sistemi i književnojezička norma. Beograd: Naučna
knjiga.