The Morphosyntax of Albanian and Aromanian Varieties

Case, Agreement, Complementation

M. Rita Manzini, Leonardo Savoia

• Publisher: De Gruyter Mouton
• Year: 2018
• Print ISBN:
  9781501514326
• eBook ISBN:
  9781501505065
• Topic: Languages & Linguistics
• Subtopic: Languages
• Index: Languages & Linguistics

This book deals with Albanian, including the dialects spoken in Southern Italy, and with the Aromanian spoken in Southern Albania. These languages are set in the context of current generative research on syntax, morphology, language variation and contact – yielding insights into key morphosyntactic notions of case, agreement, complementation, and into phenomena such as Differential Object Marking, the Person Case Constraint, linkers and control.

1 Introduction: Ext, Agree, Parameters

The title of this monograph highlights its focus on Albanian and Aromanian varieties, and on the empirical domains of nominal inflection, Case and agreement, complementation. The reason why we came to investigate these languages has to do both with their status of relatively little studied languages and with the availability of informants; microvariation data in turn facilitate investigations into domains of syntax closer to the externalization interface. All of this is useful to understand the external circumstances which shaped the monograph; the sources of several of the present chapters in occasional or hard to access publications are dutifully aknowledged at the end of this introduction.

At the same time, both our work on Balkan dialects and our analyses of the morphosyntax of case, agreement and complementation cannot be understood without reference to Chomsky’s minimalist research program (Chomsky 1995, 2000a, 2001, 2008, 2013; Chomsky et al. 2018). In this sense, the themes of the monograph are those enunciated by the title of this chapter, namely the externalization interface EXT, as well as operations such as Agree whose properties have sometimes led theorists to attribute them to the EXT component, despite their strong ties to the computational core. On these topics, we have both objections to offer to some currently prevailing approaches and proposals to advance. The chapters that follow, taken in isolation, may be appreciated as studies of microvariation facts or as original proposals concerning specific phenomena. However it is important to realize that they address a consistent set of themes, and that the conclusions of each chapter not only are consistent with those of other chapters but are meant to cumulatively build a case in favour of a particular set of theoretical conclusions. The aim of this chapter is to bring the theoretical themes and conclusions to the fore. Because the matter is conceptually complex, we begin by designing a brief roadmap of the chapter.

In section 1 of this chapter we focus on EXT, i.e. essentially the matching of PHON and SEM. This is a core problem of linguistics, and one where Chomsky’s model of the computational component underdetermines theoretical choices. We make one main point, which we believe current generative theory and pratice fail to appreciate and which on the contrary could in our opinion restrict the available options considerably. We hold that the role of PHON is not that of opacizing the output of the computational component, but rather of cooperating with the computational component in order to make SEM optimally legible. In other words, approaches such a Distributed Morphology (DM) that stress the internal organization of morphophonology to the detriment of its immediate legibility by syntax (and vice versa) take the wrong approach to the EXT problem. PHON does not increase noise, but rather works to overcome noise that may disturb the efficient work of the language faculty. This way of looking at the EXT interface has simplicity on its side. Another bonus is an insight into the role of functional optimization devices – these have a place precisely at the EXT interface, insuring the best interfacing of the PHON and SEM computational components. The main parts of the monograph that deal with EXT are Part I on the inflectional structure of Nouns and Part III on complementation, which reject morphological opacization and grammaticalization analyses.

In section 2 of the present chapter we address Agree. We make a point which is potentially controversial, face to prevailing wisdom in generative grammar practice and theorizing. One of the questions that Chomsky et al. (2018) leave open is the status of Agree as computational process or as a process belonging to EXT, where its role would be that of repairing underspecified features clusters. We reject the latter approach. The simplest view of φ-feature clusters is that they are all alike, both positively specified for interpretability and valuation, and that the only Probe-Goal asymmetry is introduced by c-command (see the core definition of the operation by Chomsky 2000a). Note that neither Minimal Search, nor of course identity, depend on an intrinsic asymmetry in the definition of the feature cluster (±interpretable, ±valued). Minimal Search is indeed asymmetric but because c-command is. Agree may in fact be best conceived as an interpretive procedure at the SEM interface. SEM cannot return a well-­formed interpretation unless φ-feature clusters denoting a single referent are properly identified as discontinous occurrences of the same item (rather like chains). Part II of the monograph is an extended argument in favour of this view of Agree, based essentially on agreement internal to the DP and in adnominal modification (Suffixaufnahme). Independently of our proposal, the latter are notoriously expensive to deal with in terms of asymmetric Agree of the type devised for sentential agreement with I and v. This part of the monograph in turn builds on the morphosyntactic analysis of nominal inflections in Part I and is relevant for the discussion of sentential complementation in Part III.

In section 3 of this chapter we separately consider the issue of variation, which bears a strong conceptual relation to EXT, but also to the lexicon, hitherto unmentioned. It is the lexicon that carries the weight of variation according to early minimalist theorizing (Chomsky 1995). This is in principle consistent with the view that variation is located at EXT (Berwick and Chomsky 2011), since the lexicon traditionally is the locus for the pairing of PHON and SEM, i.e. EXT. Yet, unresolved questions lurk under such uncontroversial statements. One problem has to do with the exact locus where PHON is paired up with SEM. Notoriously, DM advocates Late Insertion. However, we judge the empirical evidence in favour of such a conclusion (syncretisms, portmanteau morphology etc.) vastly ­overestimated (Manzini and Savoia 2005, 2007). If so, the conclusion that the lexicon, pairing up PHON and SEM, serves as input to the computational system remains possible. Needless to say, if the lexicon is prior to computation then ­different languages, having different lexicons, will have different syntactic ­structures. Late Insertion makes in principle available the idea of an universal abstract lexicon (prior to Vocabulary Insertion at EXT) and therefore effectively of a ­Universal Base – which we reject.

A final often debated opposition, which our data evoke, pits micro- against macro-variation and hence potentially microparameters against macroparameters. Here we reiterate the position explicitly articulated by Manzini and Savoia (2011a), namely that these oppositions are merely descriptive, because there is of course a single set of parameters, exactly as there is a single set of rules. Specifically parameters are lexicalization choices open in the conceptual workspace. Certainly categorial splits may or may not be realized in a given language/lexicon and if they are, they may be instantiated in different domains and subdomains of the lexicon. The opposition of macro- vs micro-variation depicts the extent to which (the lexicon of) a language is affected by a certain parametric choice. Part IV of the monograph, concentrating on descriptive and theoretical issues arising from language contact, most directly relates to parametrization (and indirectly change).

1 EXT

As foreshadowed by the original DM program (“syntax all the way down”, Halle and Marantz 1993), we assume that the traditional domain of morphological facts is largely to be subsumed by the syntax – in fact entirely, in the ideal case. PHON is a computational system with its own primitives and its own operations. The interface between PHON and CHL/SEM is optimized, so that PHON yields optimum legibility of CHL/SEM. Optimization devices of the type popularized by functionalist and typological studies (but see also Chomsky’s (2005) “third factor”) are likely to have a role precisely in insuring the matching of the two computational systems involved.

This section is organized as follows. First, we will illustrate how we believe the phonology-to-syntax mapping is run by reference to a domain, that of Romance (3rd person) clitics, where a great deal is know about history, variation, phonological, morphological and syntactic restrictions of a easily delimited domain, creating conditions close to those of an experimental set-up. We will then highlight some of the chapter and sections of the monograph that bear most directly to the theses we set out.

1.1 A case study concerning proclisis/enclisis alternations in Romance

In many Romance varieties, proclitic and enclitic position correlate with a number of morphophonological and morphosyntactic alternations. One source of alternations is segmental, involving for instance the presence in proclisis and the absence in enclisis of the l- segment of accusative 3rd person clitics. Thus in Corsican (1), the vocalic proclitic u/a/i alternates with the syllabic enclitic lu/la/li, exemplifying segmental (l-) allomorphy.

Another source of alternations are stress patterns. In Lucanian (2) the same segmental alternation as in (1) is exemplified by the alternation of singular u/a in proclisis with lə in enclisis (final vowels are neutralized in this variety). However, the enclitic shifts the word stress, which in (2b) is on the syllable immediately preceding the clitic, not on the verb root, as in (2a) and also in (1b); for reasons that will become apparent, in (2b) we show that the 1st person clitic has the same effect as the 3rd person one. In the enclitic group in (2c), the word stress is also shifted to the syllable immediately preceding the lə enclitic. The alternations may combine, but may also be observed independently of one another.

Note that there is nothing especially dialectal or microparametric about the relevant facts. A standard Romance language like French in (3) illustrates allomorphy, involving nuclei, e.g. me>moi, and reordering, which Corsican dialects also display, but which is out of the scope of the present discussion. In equally well-known ­standard Romance languages, on the other hand, nothing happens, e.g. Italian in (4). The alternations are Romance-wide and there is no external basis for their distribution.

Manzini and Savoia (2005, 2017a) conclude that both the segmental and stress alternations are allomorphies, i.e. there are two different sets of clitics involved; the reader is referred to those works for detailed data and discussion of the rather extensive literature. Vocalic clitics of the form in (1a), (2a) consist merely of nominal class (gender) and number specifications, which in the appropriate context are sufficient to externalize reference to the 3rd person. In contexts that include the imperatives in (1b), (2b) the class and number specifications must be supported by the l- lexical base, introducing D properties, as in (5).

Assuming what precedes to be...

Table of contents

• Cover
• Title Page
• Copyright
• Contents
• 1 Introduction: Ext, Agree, Parameters
• I Nominal inflections, person and case and their syntactic projection
• 2 Case categories in Albanian
• 3 N morphology and its interpretation: The neuter in Italian and Albanian varieties
• 4 Person splits in the case systems of Geg Albanian and Arbëresh: DOM and the PCC
• II Lkrs, possessors and agreement in the DP
• 5 Lkrs in Aromanian in comparison with Albanian
• 6 Case in Aromanian compared to Romanian: The Person split and agreeing possessives
• 7 Oblique case in Punjabi: Ergativity splits and agreeing possessors
• III Complementation: Particles, Complementizers, Prepositions
• 8 Finite control (and raising) in Albanian: The subjunctive Prt as Lkr
• 9 The finite complementation system of Aromanian, in comparison to other Romance languages and Albanian
• 10 Non-finite complementation in Aromanian and Albanian: prepositional introducers, infinitivals, supines
• IV Linguistic Contact
• 11 Notes on the contact between Italo-Albanian and Romance (Calabrian, Lucanian) varieties: borrowings, code-mixing and convergence
• 12 Causatives, case, passivization and agreement in the variety of Ginestra (Arbëresh): Against VP-movement and monoclausality
• References
• Index