Grammaticalization of auxiliary verb constructions: The case of Italo-Romance mixed perfective auxiliation systems and the ‘MIXPAR’ project

Grammaticalization of auxiliary verb constructions: The case of Italo-Romance mixed perfective auxiliation systems and the ‘MIXPAR’ project Pavel Štichauer (Charles University, Prague) ABSTRACT This paper deals with auxiliary verb constructions in Romance, in particular with those that exhibit two auxiliary verbs ‘have’ and ‘be’ which alternate, in many Italo-Romance varieties, within one and the same paradigm. It is argued that such an intra-paradigmatic distribution represents aspecial kind of grammaticalization, traditionally referred to as morphologization. Two aspects are discussed. First, amorphological approach to such ‘mixed paradigms’ is advocated, the main claim being that in order to explain the distribution of the two auxiliaries within the paradigm, one has to make essential reference to paradigmatic structure rather than to the intrinsic featural composition of the auxiliaries (along the lines of paradigmatic approaches reviewed in Blevins 2016). Second, it is shown that these mixed systems, although they often represent “delicate transitional stages” (Loporcaro 2014: 56, n.8), also display interesting diachronic convergence typical of various stem alternation patterns, famously referred to as morphomes (cf. Maiden 2018). The paper draws on adataset that is currently being put together in order to become, in the future, alarge database of mixed perfective auxiliation systems. Some space is thus devoted to the description of the main parameters of this project, called ‘MIXPAR’. KEYWORDS auxiliary selection, Italo-Romance, lexical splits, mixed paradigms, morphologization, morphomes, periphrasis, person-based systems, MIXPAR project, Romance DOI https://doi.org/10.14712/18059635.2022.1.4 1 INTRODUCTION Auxiliary verb constructions are usually taken to be multi-word expressions in which the auxiliary is afunctional, grammaticalized element with a(languagespecific) range of properties different from full lexical verbs (cf., e.g., Heine 1993; Kuteva 2001; Ledgeway 2012: 121–134). In Romance, we find atypical diachronic path which leads from an original point of departure with two auxiliaries, ‘have’ (henceforth H for Latin habere) and ‘be’ (henceforth Efor Latin esse), down to asingleauxiliary system, as is the case of Spanish. While the binary auxiliary alternation may depend on arange of criteria, the most common pattern being the active/stative distinction (cf., e.g, Bentley 2006), the single-auxiliary pattern simply lacks any external motivation altogether (all verbs, be they reflexive, transitive or unaccusative, select one and the same auxiliary). However, such agrammaticalization path is far from being the only way things can happen diachronically. In alarge number of OPEN ACCESS 60 LINGUISTICA PRAGENSIA 1/2022 Italo-Romance varieties, in fact, we find an intricate system of mixed perfective auxiliation patterns in which the two auxiliaries, H and E, alternate inside one and the same paradigm (cf.Loporcaro 2001; 2007; 2014; Manzini and Savoia 2005, Štichauer 2018; 2019, among others). In this paper,1 Iwill argue that this scenario represents aspecial kind of grammaticalization, traditionally referred to as morphologization, and Iintend to show two important aspects of this phenomenon. First, Iwill put forward astrictly morphological approach to such mixed auxiliary systems, the main claim being that in order to explain the distribution of the two auxiliaries within the paradigm, one has to make essential reference to paradigmatic structure rather than to the intrinsic featural composition of the auxiliaries (along the lines of paradigmatic approaches reviewed in Blevins 2016). Second, Iwill demonstrate that these mixed systems, although they often represent “delicate transitional stages” (Loporcaro 2014: 56, n.8), also display an interesting diachronic convergence typical of various stem alternation patterns, famously referred to as morphomes (cf. Maiden 2018). Iwill draw on adataset that is currently being put together in order to develop it, in the future, into alarge database of mixed perfective auxiliation systems. Some space is thus devoted to the description of the main parameters of this project, called ‘MIXPAR’.2 The paper is organized as follows. Section 2 presents abrief introduction to auxiliary verb constructions in Romance along with an overview of mixed perfective auxiliation systems in Italo-Romance. In Section 3, Idwell on theoretical considerations which advocate astrictly morphological approach. This approach rests upon the crucial notion of paradigm and upon the notion of ‘lexical split’. Section 4 is devoted to the diachronic implications of such an approach. Ishall address issues such as suppletion and periphrastic morphomic patterns, and Iwill deal with acase of areceding pattern in one particular variety. Section 5 brings concluding remarks and Section 6 is— by way of an appendix to the present paper— devoted to the description of the ‘MIXPAR’ project. 1 This paper is based on two different presentations. It was first aired as aseminar paper within the seminars organized by Martin Maiden at the Oxford Research Centre for Romance Linguistics on 25October 2018. Iam much indebted to Martin Maiden for his generous invitation and to the friendly audience for anumber of important remarks (many thanks in particular to Xavier Bach, Béatrice Rea, and Serena Romagnoli). The paper was then presented, with afocus on different aspects of the phenomenon of mixed paradigms, at the 4th American International Morphology Meeting (AIMM4), held at Stony Brook University (NY) on 3–5May 2019. Iwish to thank the audience of this wonderful conference, in particular Olivier Bonami and Mark Aronoff. Finally, Iam indebted to Xavier Bach, Martin Maiden, and Mirjam Fried for having read and commented on the first draft of this paper. This work is supported by the European Regional Development Fund-Project “Creativity and Adaptability as Conditions of the Success of Europe in an Interrelated World” (No. CZ.02.1.01/0.0/0.0/16_019/0000734). 2 MIXPAR for ‘MIXed PARadigms’. OPEN ACCESS PAVEL ŠTICHAUER 61 2 PERFECTIVE AUXILIATION SYSTEMS IN ITALO-ROMANCE In the Romance languages, we witness aremarkable variation in how the periphrastic tenses, formed through an auxiliary and apast participle, have evolved. Although we could say that two major patterns predominate, namely two auxiliaries, H and E, distributed across two or three more or less well-defined classes of verbs (as in Standard French or Italian), and asingle-auxiliary system with H or Eused for all verbs (as in Spanish), there is still alarge amount of variation (for arecent overview, see Pescarini and Loporcaro 2022). In what follows, Ishall briefly describe these major patterns focusing then on mixed auxiliation systems henceforth referred to as ‘mixed paradigms’. 2.1 AUXILIARY VERB CONSTRUCTIONS IN ROMANCE Although the facts are well known, Ishall nonetheless briefly describe the two major auxiliary selection strategies in the Romance languages, so that the peculiarity of mixed auxiliation systems can be clearly apparent. In Standard Italian (and, with some differences, in Standard French), for all compound tenses, the selection between H and Efollows an active/stative distinction, (with all reflexives aligning with E). Thus, transitive and unergative verbs trigger the choice of H, whereas unaccusatives align with E, as in the examples (1) with dormire ‘sleep’ and (2) arrivare ‘arrive’.3 (1) sg pl 1ho dormito I.have sleep.ptcp abbiamo dormito we.have sleep.ptcp 2hai dormito you.have.sg sleep.ptcp avete dormito you.have.pl sleep.ptcp 3ha dormito s/he.has sleep.ptcp hanno dormito they.have sleep.ptcp 3 It is important to note, as one reviewer points out, that the distinction between unergatives and unaccusatives is not without controversy. As is well known, the distinction, nowadays usually referred to as ‘split intransitivity’, stems from Perlmutter’s work and has given rise to anumber of semantic or syntactic approaches which all point to acertain degree of variation or gradience (cf., e.g., Bentley 2006: 1–7). Iadopt this distinction mainly for practical reasons (see Appendix in Sect. 6): in order to arrive at atypology of mixed systems Ineed to classify the verbs at different levels of semantic and syntactic “granularity”. The labels ‘transitives’, ‘unergatives’, and ‘unaccusatives’ thus provide the first macro-level of classification. There will be afurther, more fine-grained classification of reflexive verbs as these cannot be uniformly classified with any of the available labels. OPEN ACCESS 62 LINGUISTICA PRAGENSIA 1/2022 (2) sg pl 1sono arrivato/a I.am arrive.ptcp.m/f siamo arrivati/e we.are arrive.ptcp.m/f 2sei arrivato/a you.are.sg arrive.ptcp.m/f siete arrivati/e you.are.pl arrive.ptcp.m/f 3è arrivato/a he/she.is arrive.ptcp.m/f sono arrivati/e they.are arrive.ptcp.m/f As for reflexive verbs, they all select Eregardless of their argument structure so that unaccusative reflexives (also called inherent reflexives, or reflexiva tantum), such as accorgersi ‘notice’, and, for example, indirect transitive reflexives, in constructions such as lavarsi le mani ‘wash self’s hands’, behave uniformly. It might be interesting to note that under aclass-based account, we could go so far as to claim that we have two (or, perhaps, three) inflectional classes differentiated according to the different periphrastic realization (for such aradical view, see Bonami 2015; Bach 2019; Bach and Štichauer 2019a; 2019b; 2022). In contrast, in Spanish, this original two-auxiliary system has long been superseded by asystematic single-auxiliary system with H selected for all verbs and in all compound tenses, as can be seen in the examples (3) and (4), where the present perfect of the verb comer ‘eat’ and the pluperfect of the verb ir ‘go’ are given. (3) sg pl 1he comido I.have eat.ptcp hemos comido we.have eat.ptcp 2has comido you.have.sg eat.ptcp habéis comido you.have.pl eat.ptcp 3ha comido s/he.has eat.ptcp han comido they.have eat.ptcp (4) sg pl 1había ido I.had go.ptcp habíamos ido we.had go.ptcp 2habías ido you.had.sg go.ptcp habíais ido you.had.pl go.ptcp 3había ido s/he.had go.ptcp habían ido they.had go.ptcp Although these facts, as already mentioned, are familiar, two remarks are in order before embarking on the discussion of the ‘wildly’ behaving mixed paradigms. OPEN ACCESS PAVEL ŠTICHAUER 63 First, once the choice of one or the other auxiliary is made— in those languages where we have the two-auxiliary system, of course— the selected auxiliary is systematically used throughout the whole periphrastic part of the conjugation, both within one and the same tense-aspect-mood (henceforth TAM) paradigm and across all these TAM paradigms. There is thus what we might dub intra-paradigmatic and inter-paradigmatic uniformity. Second, as already alluded to above, all reflexive verbs generally follow the same auxiliary strategy in that all reflexives— no matter which kind of reflexive verb we are dealing with— select E.This also looks like asystematic uniformity, based here on the morphological (and syntactic) condition of the presence of areflexive clitic (regardless of its syntactic nature— i.e. whether it corresponds to an internal argument, or is just an inherent part of the verb as in reflexiva tantum). But these two apparently robust generalizations break down once we move to awide range of non-standard Italo-Romance varieties where, firstly, we witness afascinating intra-paradigmatic and inter-paradigmatic variation, and, secondly, we also find different auxiliation strategies within the group of reflexive verbs. Ithus move to these varieties in the next section. 2.2 MIXED AUXILIATION SYSTEMS IN ITALO-ROMANCE As anticipated above, we find in alarge number of Italo-Romance varieties mixed perfective auxiliation patterns in which the two auxiliaries, H and E, are distributed not across different groups of lexemes, but intra-paradigmatically, within one single TAM paradigm giving rise to what is usually described in the literature as ‘mixed paradigms’ (cf. Bentley and Eythórsson 2001) or ‘person-driven systems’ (cf., among others, D’Alessandro and Roberts 2010; Ledgeway 2019).4 Let us consider, in away parallel to example (1), the present perfect of the unergative verb dormire ‘sleep’ in example (5) which comes from the variety of Popoli (Abruzzo, province of Pescara, cf. Manzini and Savoia 2005: 688–689): (5) sg pl 1sɔ ddurˈmoitə I.am sleep.ptcp aˈvemmə durˈmoitə we.have sleep.ptcp 2ʃi ddurˈmoitə you.are.sg sleep.ptcp aˈve:tə durˈmoitə you.have.pl sleep.ptcp 3adurˈmoitə s/he.has sleep.ptcp annə durˈmoitə they.have sleep.ptcp 4 Ionly add, for the sake of completeness, that anumber of Italo-Romance varieties also follow the single-auxiliary strategy where only H or Eis selected in all compound tenses. More varieties are attested with generalized H than with generalized E(cf., e.g., Manzini and Savoia 2005: 779–809, 759–778). For amore detailed geographical overview cf. in particular Loporcaro 2001; 2007; 2014; 2016. OPEN ACCESS 64 LINGUISTICA PRAGENSIA 1/2022 Unlike example (1), where one auxiliary, H, is used throughout the paradigm, here we have acase of intra-paradigmatic mixing of the two auxiliaries since, as the grey shading indicates,5 the auxiliary H is selected in asubset of paradigm cells, namely the 3rd pers. sg. and in all cells of the plural, whereas the 1st and 2nd pers. sg. are realized with E. Now let us look at adifferent pattern, parallel to example (2), the unaccusative verb ʃiˈ (< īre) ‘go’ in the southern variety of Bari (cf. Andriani 2017: 156):6 (6) sg pl 1so ʃʃutə I.am go.ptcp simə ʃutə we.are go.ptcp 2si ʃʃutə you.are.sg go.ptcp sitə ʃutə you.are.pl go.ptcp 3aʃʃutə he/she.has go.ptcp annə ʃutə they.have go.ptcp Here, the auxiliary Eis used in adifferent subset of cells, namely the 1st and 2nd pers. of both number values, leaving the auxiliary H in the 3rd persons. This pattern, although it represents one of the most widespread distributions within mixed systems, is only one of many other patterns (see, e.g., Loporcaro 2007; 2014). Moreover, once we move from one TAM paradigm, the present perfect that we have been considering here, to another one, the pattern changes crucially, thus disrupting what we called above inter-paradigmatic uniformity. Thus, the same verb ʃi in the variety of Bari, displays the generalized auxiliary H throughout the paradigm of the pluperfect subjunctive (which overlaps, in this variety, with the past conditional serving to express counterfactuals; see, for details, Andriani 2017: 163–164), as example (7) illustrates: (7) sg pl 1avèssə ʃutə I.had.sbjv go.ptcp avèssəmə ʃutə we.had.sbjv go.ptcp 2avìssə ʃutə you.had.sbjv.sg go.ptcp avìssə(və) ʃutə you.had.sbjv.pl go.ptcp 3avèssə ʃutə s/he.had.sbjv go.ptcp avèssərə ʃutə they.had.sbjv go.ptcp 5 Henceforth the grey shading will always indicate those paradigm cells where the auxiliary Eis selected as opposed to the cells with H left in blank. Note that this does not imply any ‘markedness’ considerations on the selection of either auxiliary. 6 Iadapt Andriani’s simplified IPA-based transcription so that the IPA characters are used here consistently (Ithus rewrite, for instance, so ssciùtə as so ʃʃutə). At the same time, the OPEN ACCESS PAVEL ŠTICHAUER 65 As already hinted at above, the phenomenon of mixed paradigms can also be limited to only aspecific group of verbs, such as reflexives. This situation is far from rare, and there are indeed alarge number of varieties with a‘standard’ split between transitives/unergatives with H and unaccusatives with E, but with amixed auxiliation system just within the class of reflexive verbs. This situation is found not only in northern Italian dialects (Piedmont, Veneto, Friuli), as Loporcaro (2007: 200) and Benincà, Parry, and Pescarini (2016: 203–204) point out, but also elsewhere (e.g. Calabria, Marche). It is also important to note that ‘reflexive’ represents an overarching label for aseries of distinct types of reflexive verbs. That these distinct types of reflexivity are crucial for the auxiliary selection has been demonstrated by Loporcaro (2007; 2014; 2016). As an illustration, let us consider the examples (8) and (9), from the variety of Macerata (central Marchigian dialects, cf. Paciaroni 2009: 49–50). (8) sg pl 1io me sɔ ʒveʝʝatu/a Imyself= I.am wake.ptcp.m/f nuˈa tʃe simo ʒveʝʝati/e we ourselves= we.are wake.ptcp.pl.m/pl.f 2tu te si ʒveʝʝatu/a you.sg yourself= you.are.sg wake.ptcp.m/f vuˈa ve sete ʒveʝʝati/e you.pl yourselves= you.are.pl wake.ptcp.pl.m/pl.f 3issu/essa sɛ ʒveʝʝatu/a he/she oneself= is wake.ptcp.m/f issi/esse sɛ ʒveʝʝati/e they.m/f oneself= is wake.ptcp.pl.m/pl.f Apparently, this example of the middle reflexive verb— corresponding to the Standard Italian svegliarsi ‘wake up’— does not represent any particular deviation from the general pattern with the systematic Ethroughout the paradigm. Indeed, as Paciaroni (2009: 49–50) reports, the generalized Eholds for direct transitive reflexives as well as for unergative reflexives. However, the other types of reflexive verbs may follow adifferent— mixed— pattern with, for instance, free variation between the two auxiliaries (signaled here and throughout the paper by the symbol ‘≈’) in asubset of cells, as illustrated in example (9), capturing the indirect transitive reflexive construction ‘wash self’s hands’: phonosyntactic doubling, rafforzamento fonosintattico (RF), is systematically signaled where it obtains. For ageneral overview of RF, see Loporcaro 1997, and for the morphological role of RF in mixed paradigms, see Torcolacci 2015; Štichauer 2017. OPEN ACCESS 66 LINGUISTICA PRAGENSIA 1/2022 (9) sg pl 1io me sɔ llaatu/ale ma Imyself= I.am wash.ptcp.m/f the hands nuˈa tʃe simo laati/ele ma we ourselves= we.are wash.ptcp.pl.m/pl.f the hands 2tu te si llaatu/ale ma you.sg yourself= you.are.sg wash.ptcp.m/f the hands vuˈa ve sete laati/ele ma you.pl yourselves= you.are.pl wash.ptcp.pl.m/pl.f the hands 3issu/essa sɛ llaatu/a≈ salaato le ma he/she oneself= is wash.ptcp.m/f ≈ has wash.ptcp the hands issi/esse sɛ llaati/e≈ salaato le ma they.m/f oneself= is wash.ptcp.pl.m/pl.f ≈ has wash the hands As can be seen, indirect transitive reflexives (along with antipassives) allow for afree choice of either auxiliary in the 3rd persons, thus disrupting not only the intra-paradigmatic uniformity as defined above, but also creating aself-contained subgroup of reflexive verbs which must be taken separately. As is now clear, the variation in mixed perfective auxiliation systems across the wide array of Italo-Romance varieties is so immense that it is extremely difficult to do justice to the full range of data without carefully considering each paradigm in its entirety. For this reason, Ihave decided to pursue aproject comprising alarge database the aim of which is to put together all the attested data. Even though the database is only at its incipient stage, the detailed description of its parameters might be useful here and can be found, as an appendix to this paper, in Section 6.7 3 THEORETICAL ACCOUNTS: AMORPHOLOGICAL APPROACH In this section, Iadvocate aparadigmatic approach to mixed paradigms which amounts to claiming that in order to understand the distribution of the two auxiliaries within the paradigm, it is necessary to take into account the paradigmatic structure rather than positing afunctionally motivated association between one or the other auxiliary and aspecific person/number value. 3.1 APARADIGMATIC APPROACH In one of the most widespread patterns, EEH-EEH, as in example (6) above, the first and second persons are opposed to the third persons. One may thus say that the se7 The idea of creating such adatabase has arisen on the basis of ongoing joint work with Xavier Bach whom Ithank here for anumber of important remarks. The data from the database have already been exploited in acouple of joint presentations (cf. Bach and Štichauer 2019a; 2019b) and arecent publication (cf. Bach and Štichauer 2022). For technical advice Iam indebted to Ondřej Tichý and Tomáš Bořil. OPEN ACCESS PAVEL ŠTICHAUER 67 lection of Eis triggered by the (binary) discourse-participant feature,8 whereby the positive value identifies the first and second persons, whereas the third persons stand out as having the negative value, thus triggering H.Such aperson-based split (cf., e.g., Ledgeway 2019: 354 ff.) is thus accounted for by adopting asyntagmatic— as opposed to paradigmatic — view typical of the vast majority of current approaches (see, e.g., Torcolacci 2015 for aDM-based account, and Andriani 2017; 2018; Ledgeway 2019, among others, for amicroparametric approach).9 Under such asyntagmatic account, the selection of either auxiliary is dictated by the specific value of the person/ number combination. Thus, for instance, it might be said that one auxiliary is triggered by the ‘non-discourse participant’, while the other is reserved for other values (discourse participant, but also other values, see below in example 10), without any “blind” reference to the paradigm structure which would be simply defined as “four cells” versus “two cells” in the present perfect paradigm. Let us now consider one example that represents yet another frequent pattern in which only one person/number combination (i.e. one paradigm cell) is selected by one or the other auxiliary. In example (10), from the variety of Miglionico (the southern region of Basilicata, prov. Matera; cf. Manzini and Savoia 2005: 726), the verb dormire ‘sleep’ leaves only the 3rd pers. pl. for the auxiliary H, while the rest of the paradigm, i.e. acollection of five (irreconcilable) cells, is realized with E. (10) sg pl 1sɔ ddərˈmutə I.am sleep.ptcp simə dərˈmutə we.are sleep.ptcp 2si ddərˈmutə you.are.sg sleep.ptcp sitə dərˈmutə you.are.pl sleep.ptcp 3jɛ ddərˈmutə s/he.is sleep.ptcp vɔnnə dərˈmutə they.have sleep.ptcp It is extremely difficult to account for such apattern in asyntagmatically based way, i.e. attempting to define acommon morphosyntactic or semantic value for the collection of five cells (or, conversely, for the only one cell), which would justify the selection of one or the other auxiliary. Moreover, there are other such patterns, where only one cell stands out as being realized with the different auxiliary, as for instance, EHH-HHH or HHE-HHH. Invoking the featural composition of the individual sequences (such as, for instance, jɛ ddərˈmutə lit. ‘he/she is slept’ versus vɔnnə dərˈmutə lit. ‘they have slept’) would be 8 Note that even this proposal is very problematic in that there is acrucial difference between, say, the 1st pers. singular and 1st pers. plural. See the insightful discussion in Cysouw 2011: 434 ff. 9 Amore restricted pattern, which can be said to rely on this feature, is reported in example (5) above, where only the first and second pers. singular are realized with E, while the rest of the paradigm selects H. OPEN ACCESS 74 LINGUISTICA PRAGENSIA 1/2022 Although Andriani offers adiachronic interpretation of these facts in terms of progressive replacement of individual cells, Iwish to look at these paradigms in adifferent way. What is striking, in fact, is that all three patterns are used, as Andriani observes, in parallel, with strong preference for the ‘canonical pattern’ EEH-EEH both by older and middle-aged speakers, while being the first choice for younger speakers. There is thus not really amicro-diachronic path from Pattern 1 to Pattern2, and finally to Pattern 3, but, admittedly, global convergence towards Pattern3 from both of the competing and receding patterns. Indeed, the passage across Pattern 2 would be disrupting in that in Pattern 1 we already have the 1st and 2nd pl. firmly established with E, while it is only the 1st sg. which undergoes ashift from H to E.In Pattern 2, however, this shift would also be accompanied, unexpectedly, by free variation between E≈H in the 1st and 2nd pers. pl., resolved in turn in Pattern 3. Let us appreciate this situation through aschematic representation of the same data in example (15): (15) PATTERN 1 (older generation) PATTERN 2 (middle-aged generation) singular plural singular plural 1H E 1E E≈H 2E E 2E E≈H 3H H 3H H PATTERN 3 (younger generations— the commonest pattern) singular plural 1E E 2E E 3H H Thus what Ifind striking, in Pattern 2, is this kind of disrupting free variation in the two cells which seem to be stable in the other two patterns. Therefore, Iargue— in line with Andriani’s observation about the parallel use of all three patterns— that we have here acase of morphomic convergence or attraction of the most common pattern EEH-EEH already widespread in alarge number of other varieties of the same dialectal area (and of other areas as well).17 Of course, Iam well aware of the fact that 17 Martin Maiden (p.c.) raises an intriguing question. According to his definition of ‘convergence’ and ‘attraction’, the morphomic patterns are already present in the system and ‘replicated’, while here, in the case at hand, what Iam talking about would be acase of ‘pattern borrowing’, as the most common pattern EEH-EEH, so far unavailable in the present perfect paradigm of the variety of Bari but widespread in other varieties, is ‘borrowed’, which is “controversial and certainly rare”. OPEN ACCESS PAVEL ŠTICHAUER 75 this represents only one— speculative— case. But as the MIXPAR database progressively grows, more data will be available to investigate not only the current variation but also some diachronic implications. 5 CONCLUDING REMARKS In this paper, Ihave argued that progressive morphologization is going on. Ipropose that what we witness here is aloss of the original transitivity/unaccusativity marking and the subsequent reorganization of the two auxiliaries which can eventually give rise to inflection classes with no semantic or syntactic motivation. Ihave stressed the morphological nature of the alternation— we do need aparadigmatic account but the morphological behavior of these mixed auxiliary systems does not seem to display typical morphomic behavior associated with the major morphomic patterns described by Maiden. However, Iwould insist on the claim that these patterns, even if unique within the system, are as psychologically real as any other inflectional phenomena in that speakers, rather than figuring out acovert motivation for apattern, simply acquire it as apredictable distribution where the implicational relationships can be established. Ithink that this holds— in line with Maiden’s above-mentioned warning— even for those patterns that happen to be present cross-linguistically and for which extramorphological motivation has always been envisaged, as is the case with the widespread pattern EEH-EEH. Iconclude by offering an audacious comparison of two examples which are linguistically unrelated, as they come from two entirely different languages, but closely connected on theoretical grounds since they both assume, Iargue, the same abstract pattern.18 In example (16), Ionly repeat the much discussed instance of the EEH-EEH pattern reported in example (6) above, while in example (17), Czech periphrastic past of the verb číst ‘read’ is reported. 18 There has been adebate about whether ‘typological uniqueness’, i.e. the fact that some patterns are cross-linguistically rare or, on the other hand, frequent, should be taken as an important diagnostic for genuine ‘morphomehood’ (cf. Herce 2019; Maiden 2021: 93–95). Ido not wish to enter into the debate directly because Iam not discussing the traditionally defined L-, N-, and U-patterns but simply describing asuperficial, synchronically active identity between these two patterns. However, what is interesting to note is that the two patterns, Italo-Romance EEH-EEH and the Czech periphrastic past, have an entirely different diachronic origin. While the former arises from the mixing of two auxiliaries, the latter is the result of a(long attested) loss of the 3rd sg. and pl. auxiliary forms. Such completely different diachronic scenarios of how two presumably identical patterns came into being might indicate that ahidden, motivated pattern could lie behind them (for asimilar point, cf. Maiden 2018: 22; 2021: 93). OPEN ACCESS 76 LINGUISTICA PRAGENSIA 1/2022 (16) sg pl 1so ʃʃutə I.am go.ptcp simə ʃutə we.are go.ptcp 2si ʃʃutə you.are.sg go.ptcp sitə ʃutə you.are.pl go.ptcp 3aʃʃutə he/she.has go.ptcp annə ʃutə they.have go.ptcp (17) sg pl 1četl/ajsem read.ptcp.m/f I.am četl-i/yjsme read.ptcp.m/f we.are 2četl/ajsi read.ptcp.m/f you.are.sg četl-i/yjste read.ptcp.m/f you.are.pl 3četl/a(*je) read.ptcp.m/f (*he/she is) četl-i/y(*jsou) read.ptcp.m/f (*they are) In Czech as well as in Slovak (as discussed by Corbett 2013: 173), the auxiliary být ‘be’ only surfaces in the 1st and 2nd persons, while being obligatorily absent in the 3rd persons.19 As Bonami and Webelhuth (2021: 88) point out, the absence of the auxiliary is inherent in this particular paradigm since the 3rd. pers. forms je ‘she/he/it is’ and jsou ‘they are’ are obligatorily used in copular constructions. The paradigm is thus striking precisely for the fact that it follows aunique pattern of alternation which can ultimately be, on an abstract morphomic level, just the same as what we have discussed in the case of the Italo-Romance varieties. 6 APPENDIX. ‘MIXPAR’: THE PROJECT OF ADATABASE OF MIXED PARADIGMS The overall design of the database is straightforward. In order to get the complete information about concrete paradigms, asimple MS Excel format with rows and columns has been chosen where every single row (or input line) corresponds to one concrete paradigm and the columns capture the relevant types of information. The 19 Hence, it is impossible to say, for example, *četl je lit. ‘he is read’, while it is perfectly acceptable to omit the auxiliary in the 1st pers. when the subject pronoun is spelled out, thus, for instance, pracovali jsme celou noc and my pracovali celou noc ‘we worked all night long’, in which the presence of my ‘we’ admits the auxiliary omission. Interestingly, the future auxiliary, which unlike the present tense form is not aclitic, comes in the complete set, but only for imperfective verbs: budu pracovat, budeš pracovat, bude pracovat ‘Iwill work, you will work, he/she will work’. OPEN ACCESS PAVEL ŠTICHAUER 77 structure is simple enough to allow for any kind of data export (for instance, into the statistical software R for advanced analyses). Thus, the first column hosts the complete paradigm,20 the subsequent columns are labelled Verb/Construction, Class, TAM, Pattern, Region, Province, Place, Dialect classification-1, Dialect-classification-2, Dialect-classification-3, Source, and Notes. One of the concrete input lines, capturing the example (6) discussed above, can be seen in Figures (1) and (2), Figure (2) being the continuation of the same input line. Figure 1.Illustrative input line in the MIXPAR database (lefthand part) Figure 2.Illustrative input line in the MIXPAR database (righthand part, continuation) The column labels thus indicate aseries of factors with anumber of values (statistically speaking, ‘levels’). The Verb/Construction contains the citation form of aconcrete verb whose paradigm is given in the first column. In order to unify this factor, which is obviously more or less open-ended as more and more verbs will be added, the citation forms are given as Standard Italian verbs, with sometimes the etymological point of departure as, for example, the verb ʃiˈ (< īre) subsumed here under the heading of the verb andare ‘go’. This decision will enable us to search for all paradigms of the verb andare in each of the varieties present in the database regardless of the concrete phonological shape. The Class factor is aclosed category based on the argument structure with the levels corresponding to transitives, unergatives, unaccusatives, and to six types of reflexive verbs. For each of the reflexivity types, there is one example in Standard Italian provided in the legend to the database (on the second sheet of the Excel file). Here we find inherent reflexives (accorgersi ‘notice’), transitive direct and indirect reflexives (lavarsi and lavarsi le mani, ‘wash oneself’ and ‘wash self’s hands’, respectively), unergative indirect reflexives (rispondersi ‘answer to oneself’), ‘antipassive’ reflexives expressing the subject’s benefit from the action performed (mangiarsi un 20 Alternatively, as some colleagues propose, it would also be useful to split the first column devoted to full paradigms into six separate columns corresponding to each of the six paradigm cells. OPEN ACCESS 78 LINGUISTICA PRAGENSIA 1/2022 panino ‘eat up asandwich’), and middle reflexives (svegliarsi ‘wake up’). As already alluded to above, this detailed classification of reflexive verbs is crucial as there are differences, across Italo-Romance varieties, between each of these types. The TAM category is also aclosed factor containing the concrete tense-and-mood paradigm. The abbreviations, spelled out in the legend, correspond to those used in the Leipzig glossing rules (thus, for instance, prf = present perfect, plf = pluperfect, etc.). Within the Pattern factor, all attested intra-paradigmatic combinations of the two auxiliaries are captured in the format EEH-EEH, i.e. the three cells of the sg.— the three cells of the plural. Free variation, afrequent situation in many varieties, is signaled by the double tilde. H≈Ethus indicates that in agiven cell, the two auxiliaries can be used interchangeably. There are two points worth mentioning here. The first is the order in which such free variation is indicated. If, in fact, we have clear (diachronic, or theoretical) evidence that the erstwhile choice was H, or E, then the order should also reflect this important observation, so that H≈Eand E≈H capture two different scenarios, and they should be read as ‘the first auxiliary choice is H, being progressively superseded by E’, and, conversely, ‘the first auxiliary choice is E, being progressively superseded by H’. The problem is that we often do not have such clear evidence for positing this order in any reliable way. The second point, probably more serious, is that sometimes, in particular in the pluperfect, the auxiliary forms are clearly the result of amerger of both of the auxiliaries H and E.Thus, for instance, forms such as eva, seva, jeva etc. cannot be readily ascribed to H or E(see Cennamo 2010). In the database, this conflation represents apractical obstacle to assigning such paradigms to specific patterns of H/Ealternation. The next series of factors all involve detailed geographical information. The first three, Region, Province, and Place provide simple geographical coordinates based on the administrative division of Italy (regions, provinces) with the concrete name of the locality. The three dialect classification factors, Dialect-classification-1/2/3, follow the traditional and still valid dialectal divisions and subdivisons based on Pellegrini 1977. The first ‘macro-classification’ reports the basic dialect area, e.g. Meridionale intermedio, the second follows the narrower area, indicated in Pellegrini’s map as I, II, III, etc., e.g., Pugliese, and the third ‘micro-classification’, indicated as, for example, Ia, Ib, Ic, etc., indicates the most restricted area defined by Pellegrini 1977, e.g., in the case at hand, Apulo-barese. The last two columns, Source and Notes, contain the precise bibliographical reference from where the paradigm is taken, and possibly anote on any additional aspect which might be relevant for the data in the input line. Thus, for instance, in the example reported in Figures (1) and (2), aremark on areceding or competing pattern is present. In other cases, particular comments on the data can also be found, such as anote on some inconsistency or aclassification error (such as when the references identify— often on the basis of aslightly different dialect classification— avariety as belonging to one group rather than another one). This is, then, the overall design of the MIXPAR database which is, as already said, at the very beginning of development (as of August 2022, it contains 100 full paradigms), but has been growing progressively. Once all the attested data have been put together, the database will be ready for online publication. OPEN ACCESS PAVEL ŠTICHAUER 79 REFERENCES Ackerman, F. and G.T.Stump (2004) Paradigms and periphrastic expression. In: Sadler, L. and A.Spencer (eds), Projecting Morphology, 111–157. Stanford: CSLI. Andriani, L. (2017) The Syntax of the Dialect of Bari. University of Cambridge (UK): Unpublished PhD thesis. Available at https:// doi.org/10.17863/CAM.15625. Andriani, L. (2018) Instability and Change: AParametric Approach to Barese Auxiliary Selection. In: Dindelegan, G.P., A.Dragomirescu, I.Nicula and A.Nicolae (eds) Comparative and Diachronic Perspectives on Romance Syntax, 361–397. Cambridge: Cambridge Scholars Publishing. Bach, X. (2019) The Origins of Inflectional Classes. University of Oxford: unpublished PhD thesis. Bach, X. and P.Štichauer (2019a) Mixed paradigms in Italo-Romance as noncanonical inflectional classes. Paper given at the Workshop on non-canonicity in inflection, University of Surrey, Guildford, UK, 21–22June 2019. Bach, X. and P.Štichauer (2019b) Auxiliary selection in Italo-Romance and inflectional classes. Poster presented at the International Symposium of Morphology (ISMo 2019), Université Paris-Diderot, Paris, France, 25–27September 2019. Bach, X. and P.Štichauer (2022) Auxiliary selection in Italo-Romance and inflexional classes. In: Ledgeway, A., J.C.Smith and N.Vincent (eds) Periphrasis and Inflexion in Diachrony. AView from Romance, 193–212. Oxford: Oxford University Press. Baerman, M., D.Brown and G.Corbett (2017) Morphological Complexity. Cambridge: Cambridge University Press. Benincà, P., M.Parry and D.Pescarini (2016) The dialects of northern Italy. In: Ledgeway, A. and M.Maiden (eds) The Oxford Guide to the Romance Languages, 185–205. Oxford: Oxford University Press. Bentley, D. and T.Eythórsson (2001) Alternation according to person in Italo-Romance. In: Brinton, L.J. (ed) Historical Linguistics 1999. Selected papers from the 14th International Conference on Historical Linguistics, Vancouver, 9–13August 1999, 63–74. Amsterdam: John Benjamins. Bentley, D. (2006) Split intransitivity in Italian. Amsterdam: John Benjamins. Bjorkman, B.M. (2016) Go get, come see. Motion verbs, morphological restrictions, and syncretism. Natural Language and Linguistic Theory 34/1, 53–91. Blevins, J.P. (2016) Word and Paradigm Morphology. Oxford: Oxford University Press. Bonami, O. (2015) Periphrasis as collocation. Morphology 25/1, 63–110. Bonami, O. and G.Webelhuth (2021) Periphrasis and morphosyntatic mismatch in Czech. In: Crysmann, B. and M.Sailer (eds) One-to-many relations in morphology, syntax, and semantics, 85–115. Berlin: Language Science Press. DOI: 10.5281/zenodo.4729795. Brown, D., M.Chumakina, G.G.Corbett, G.Popova and A.Spencer (2012) Defining ‘periphrasis’: key notions. Morphology 22/2, 233–275. Cennamo, M. (2010) Perfective auxiliaries in the pluperfect in some southern Italian dialects. In: D’Alessandro, R., A.Ledgeway and I.Roberts (eds) Syntactic Variation: The dialects of Italy, 210–224. Cambridge: Cambridge University Press. Corbett, G.G. (2013) Periphrasis and possible lexemes. In: Chumakina, M. and G.G.Corbett (eds) Periphrasis. The Role of Syntax and Morphology in Paradigms, 169–189. Proceedings of the British Academy, 180. Oxford: Oxford University Press. Corbett, G.G. (2015) Morphosyntactic Complexity: ATypology of Lexical Splits. Language 91/1, 145–193. Corbett, G.G. (2016) Morphomic splits. In: Luís, A.R. and R.Bermúdez-Otero (eds) The morphome debate. Diagnosing and analysing morphomic patterns, 64–88. Oxford: Oxford University Press. Corbett, G.G. (submitted) The typology of external splits. OPEN ACCESS 80 LINGUISTICA PRAGENSIA 1/2022 Cruschina, S. (2013) Beyond the stem and inflectional morphology: an irregular pattern at the level of periphrasis. In: Cruschina, S., M.Maiden and J.C.Smith (eds) The Boundaries of Pure Morphology, 262–283. Oxford: Oxford University Press. Cysouw, M. (2011) The expression of person and number: atypologist’s perspective. Morphology 21/2, 419–443. D’Alessandro, R. and I.Roberts (2010) Past participle agreement: split auxiliary selection and the null-subject parameter. Natural Language and Linguistic Theory 28/1, 41–72. Heine, B. (1993) Auxiliaries. Oxford: Oxford University Press. Herce, B. (2019) Morphome interactions. Morphology 29/1, 109–132. Fresu, R. (2006) La Cronaca teramana del canonico Angelo de Jacobis. Edizione critica con studio introduttivo eglossario acura di Rita Fresu. L’Aquila: Edizioni Libreria Colacchi. Kuteva, T. (2001) Auxiliation. Oxford: Oxford University Press. Ledgeway, A. (2009) Grammatica diacronica del napoletano. Tübingen: Niemeyer. Ledgeway, A. (2012) From Latin to Romance. Oxford: Oxford University Press. Ledgeway, A. (2016) From Coordination to Subordination: The Grammaticalisation of Progressive and Andative Aspect in the dialects of Salento. In: Pratas, F., S.Pereira and C.Pinto (eds) Coordination and Subordination. Form and Meaning: Selected papers from CSI LIsbon 2014, 157–184. Newcastle: Cambridge Scholars Publishing. Ledgeway, A. (2019) Parameters in the development of Romance perfective auxiliary selection. In: Cennamo, M. and C.Fabrizio (eds) Historical Linguistics 2015. Selected papers from the 22nd International Conference on Historical Linguistics, Naples, 27–31July 2015, 343–384. Amsterdam: Benjamins. Ledgeway, A. and N.Vincent (2022) Periphrasis and Inflexion. Lessons from Romance. In: Ledgeway, J.C.Smith and N.Vincent (eds) Periphrasis and Inflexion in Diachrony. AView from Romance, 11–60. Oxford: Oxford University Press. Ledgeway, A., J.C.Smith and N.Vincent (2022) Periphrasis and Inflexion. AView from Romance. Oxford: Oxford University Press. Loporcaro, M. (1997) L’origine del raddoppiamento fonosintattico. Basel-Tübingen: Francke. Loporcaro, M. (2001) La selezione dell’ausiliare nei dialetti italiani: dati eteorie. In: Sornicola, R., E.Stenta Krosbakken and C.Stromboli (eds) Dati empirici eteorie linguistiche: Atti del XXXIII Congresso della Società di Linguistica Italiana, Napoli, 28–30 ottobre 1999, 455–476. Rome: Bulzoni. Loporcaro, M. (2007). On triple auxiliation in Romance. Linguistics 45/1, 173–222. Loporcaro, M. (2014) Perfective auxiliation in Italo-Romance. The complementarity of historical and modern cross-dialectal evidence. In: Benincà, P., A.Ledgeway and N.Vincent (eds) Diachrony and Dialects. Grammatical Change in the Dialects of Italy, 48–70. Oxford: Oxford University Press. Loporcaro, M. (2016): Auxiliary selection and participial agreement. In: Ledgeway, A. and M.Maiden (eds) The Oxford Guide to the Romance Languages, 802–818. Oxford: Oxford University Press. Maiden, M. (2005) Morphological autonomy and diachrony. In: Booij, G. and J. van Marle (eds) Yearbook of Morphology 2004, 137–175. Dordrecht: Springer. Maiden, M. (2011) Morphophonological innovation. In: Maiden, M., J.C.Smith and A.Ledgeway (eds) The Cambridge History of the Romance Languages. Volume I.Structures, 216–267. Cambridge: Cambridge University Press. Maiden, M. (2016a) Some lessons from history. Morphomes in diachrony. In: Luís, A.R. and R.Bermúdez-Otero (eds) The morphome debate. Diagnosing and analysing morphomic patterns, 33–63. Oxford: Oxford University Press. Maiden, M. (2016b) Morphomes. In: Ledgeway, A. and M.Maiden (eds) The Oxford Guide to the Romance Languages, 708–721. Oxford: Oxford University Press. Maiden, M. (2018) The Romance Verb. Morphomic Structure and Diachrony. Oxford: Oxford University Press. OPEN ACCESS PAVEL ŠTICHAUER 81 Maiden, M. (2021) The Morphome. Annual Review of Linguistics 7, 89–108. Manzini, M.R. and L.M.Savoia (2005) Idialetti italiani eromanci: Morfosintassi generativa Volume II. Alessandria: Edizioni dell’Orso. Pellegrini, G.B. (1977) Carta dei dialetti italiani. Pisa: Pacini. Pescarini, D. and M.Loporcaro (2022) Variation in Romance. In: Ledgeway, A. and M.Maiden (eds) The Cambridge Handbook of Romance Linguistics, 150–180. Cambridge: Cambridge University Press. Pirrelli, V. (2000) Paradigmi in morfologia. Un approccio interdisciplinare alla flessione verbale italiana. Pisa: Ist. Editoriali ePoligrafici. Pirrelli, V. and M.Battista (2000) The paradigmatic dimension of stem allomorphy in Italian verb inflection. Italian Journal of Linguistics / Rivista di Linguistica 12/2, 307–380. Rensch, K.-H. (1964) Beiträge zur Kenntnis nordkalabrischer Mundarten. Münster: Aschendorff. Savini, G. (1881) La grammatica ed il lessico del dialetto teramano. Torino: Loescher. Schädel, B. (1903) Die Mundart von Ormea. Halle: Niemeyer. Štichauer, P. (2017) Italské rafforzamento fonosintattico ajeho role vtzv.smíšených paradigmatech [Italian rafforzamento fonosintattico and its role in mixed paradigms]. Časopis pro moderní filologii 99/1, 29–39. Štichauer, P. (2018) Lexical splits within periphrasis: mixed perfective auxiliation systems in Italo-Romance. Morphology 28/1, 1–23. Štichauer, P. (2019) Mixed paradigms in ItaloRomance: acase of morphologization of auxiliary selection? In: Cruschina, S., A.Ledgeway and E.-M. Remberger (eds) Italian Dialectology at the Interfaces, 79–100. Amsterdam: John Benjamins. Torcolacci, G. (2015) Marking the Default. Auxiliary selection in Southern Italian dialects. PhD dissertation, University of Leiden. Utrecht: LOT. Vincent, N. (2011) Non-finite Forms, Periphrases, and Autonomous Morphology in Latin and Romance. In: Maiden, M., J.C.Smith, M.Goldbach and M.-O. Hinzelin (eds) Morphological Autonomy. Perspectives from Romance Inflectional Morphology, 417–435. Oxford: Oxford University Press. Pavel Štichauer Department of Romance Studies Faculty of Arts, Charles University Nám. Jana Palacha 2, 116 38 Praha 1 ORCID ID: 0000-0002-6777-1104 [email protected] OPEN ACCESS