Jump to ContentJump to Main Navigation
The Oxford Guide to the Romance Languages$

Adam Ledgeway and Martin Maiden

Print publication date: 2016

Print ISBN-13: 9780199677108

Published to Oxford Scholarship Online: August 2016

DOI: 10.1093/acprof:oso/9780199677108.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

The dialects of northern Italy

The dialects of northern Italy

Chapter:
(p.185) Chapter 13 The dialects of northern Italy1
Source:
The Oxford Guide to the Romance Languages
Author(s):

Paola Benincà

Mair Parry

Diego Pescarini

Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780199677108.003.0013

Abstract and Keywords

After a brief outline of the principal episodes in the external history, the chapter surveys the principal aspects of phonology, morphology and syntax of the dialects of northern Italy, including both Gallo-Italic (Ligurian, Piedmontese, Lombard, Emilian-Romagnol) and Venetan dialects. Topics dealt with include: characteristics of northern Italian dialects; suprasegmental phonology; segmental phonology; morphology of nouns and adjectives; verb morphology: tenses, person endings, root alternations, past participles; word formation; pronouns; articles; demonstratives; possessives; agreement in the nominal group; subject clitic pronouns; wh-movement constructions; negation; auxiliaries and double compound forms; restructuring and clitic climbing; clitic syntax.

Keywords:   northern Italian dialects, phonology, morphology, syntax

13.1 External and linguistic history

The dialects of northern Italy

Map 13.1 Dialects of northern Italy

The linguistic sub-areas that can be identified in northern Italy since the very first records are Piedmontese, Occitan, Lombard, Venetan, Friulian, Ligurian, and Emilian-Romagnol (Map 13.1). Ladin (Ch. 11), Romansh (Ch. 12), and Occitan (Ch. 19) have also been considered to belong to this area in some respects, but their status is marginal; they are treated in separate chapters and are not included in our comparison.2

Since the earliest studies on Italian dialects, starting from Dante’s De vulgari eloquentia (DVE) down to Ascoli (1873), northern Italy has been seen as a coherent linguistic sub-area of Romance, a linguistic and cultural area that, despite the rich dialectal variation, also has a unitary character.3 There are geographical reasons that motivate this conception, which are also among the factors that produced the relative grammatical consistency of the languages found there. The area corresponds to the basin of the river Po and its tributaries, and is bordered on the northern side by the arc of the Alps from west to east, and the Apennines and the sea to the south.

This area has never enjoyed political status, except for a short period in the first century BC, when it was a Roman province, Gallia Cisalpina, the ‘Gaul on this side of the Alps’ (from the Roman perspective). Soon after, in 42 BC, it was incorporated into Italy proper and, with Augustus’s reorganization of the Empire, was subdivided into four regions (within a total of eleven).

In the following centuries, only Napoleon gave—for a short period—a political identity to part of northern Italy (excluding Piedmont, Liguria, and Veneto), when he established the Cisalpine Republic (1797–1815) during the Italian campaign.

Despite very marginal and ephemeral political recognition, this area has preserved the character of both a cultural and a linguistic community, albeit of a very special type.

Until Roman colonization, the area was mostly inhabited by Celtic speakers, apart from a few areas where other, Italic languages were spoken, most of which were celticized to a greater or lesser extent before the adoption of Latin; this Celtic substratum was considered the first cause of the indisputable unity of this linguistic area (Biondelli 1853; Ascoli 1873).4

Dante (DVE I, xix) refers explicitly to a ‘common language’ of northern Italy, a superdialectal vernacular, which he called semilatium ‘(the language) of a half of Italy’.5 Mussafia (1864), after a detailed examination of medieval texts, concluded that northern authors adopted languages that had just some features of the local varieties but mainly showed a homogenized form, where the most local features were deleted and many morphological and lexical features extended beyond their original location.

The debate that has developed over the years to oppose or support Mussafia’s position is due first of all to the fact that the term ‘koiné can be taken in a strong or in a lax sense, referring either to a stable and unitary language, or (as we prefer) to an unstable set of varieties showing clear traces of merging and simplification; it is well known, for example, that it is sometimes impossible to localize (p.186) (p.187) precisely—on the basis of linguistic characteristics—a medieval text from northern Italy.

In the fourteenth and fifteenth centuries, a literary current connected to the popularization of the French romances in northern Italy used a language (in fact a set of linguistic varieties closely linked to each other and to French) called ‘franco-italiano’, ‘franco-veneto’, or ‘franco-lombardo’. This language was a mixture of lexical and morphological features from northern dialects and literary and regional French, unstable but quite well defined; each author could adapt this basic instrument, emphasizing some French (or some local) aspects, with the certainty of being understood in the whole territory, albeit within a restricted set of social and cultural strata (see Pellegrini 1956; Renzi 1970). The northern Italian dialect repertoire included a choice of morphemes and functional words: for example, Pellegrini (1956:136–9) illustrates the alternation, even within the same text with the same verb, of the endings -on/‑emo with the sole function of marking first person plural. Originally these morphemes belonged to different varieties; later, for a long period, they were part of this incipient koiné, whereas more recently they have again become distinctive of different dialects or dialectal areas.

The basis for the evolution towards a sort of unity was the grammatical and lexical features that the languages of the area had in common. Moreover, the diachronic phonological rules that characterized the sub-areas were part of a unitary system of rules which bound these dialects, including French varieties. As a consequence, the shared Latin stock was recognizable even if it had been modified by diachronic phonological rules. In situations like these, one could even say that the development of a koiné is unavoidable, the necessary product of cultural and commercial contacts of people speaking very similar languages.

The medieval situation calls to mind what Trumper (1977) described as ‘macro-diglossia’, which obtains in many Italian regions in modern times: speakers have at their disposal the local dialect and one or more varieties obtained by undoing the operation of the more specific phonetic and phonological rules, thus obtaining a less marked language. In macro-diglossic areas, the local dialects survive longer, ‘protected’ by the existence of dialectal varieties of higher level and larger circulation. Conversely, in a micro-diglossic area, the local dialect is directly opposed to a regional variant of Italian, and it disappears more rapidly.6

This process produces a series of varieties progressively less marked and with a wider circulation, which are in fact a (kind of) koiné, or else the precondition for the birth of a koiné.7 Note that even the most prestigious centre of a sub-area (such as Milan, Venice, or Florence) did not export into their respective region’s variety their most distinctive features.8

13.1.1 Phenomena characterizing northern Italian dialects

To conclude this introduction, let us sum up those characteristics that are shared by all northern Italian varieties, marking northern Italy as part of an innovative sub-area of Romance comprising French, the dialects of northern Italy, Friulian, Ladin, and Romansh. From now on we will term this sub-area ‘northern Romance’.

Looking at diachronic phonological phenomena, we have:

  • A common, but not exclusive, Romance vowel system (cf. §25.1.1) underlying all the modern systems of this area (Tagliavini 1969:237; Lepschy and Lepschy 1977:45). Final unstressed vowels, except -a, are dropped; this rule affects all northern Italian dialects, even though it applies to different sets of contexts, depending on the dialect; in particular, in southern Veneto and Ligurian the contexts are very limited. All unstressed vowels are weak, and tend to be neutralized or to disappear; this is particularly evident in Emilian and Piedmontese but, again, less prominent in southern Veneto and Liguria.

  • All double consonants (in many cases produced by the assimilation of Latin obstruent consonant clusters) are phonetically shortened (cf. §25.2.5). Intervocalic voiceless obstruents are voiced; in some dialects some obstruents in this context disappear. Putting together these two rules, one notices that no voiceless obstruent (p.188) resulting from simplification is voiced in intervocalic position in any dialect.9

  • Finally, a diachronic rule that today shows its effects only in French varieties, Ladin, and Romansh (see §39.3.1.4), namely the palatalization of [ka/ɡa], originally affected more or less the whole of northern Italy, for clear traces of it have been painstakingly collected and discussed throughout the area (see Pellegrini 1991:35f.; Vigolo 1986; Tuttle 1997b).

All Romance languages show the general tendency to substitute synthetic forms by analytic forms, a process that can be seen as part of a more abstract and generalized diachronic drift (§46.1), leading to a progressive reduction of syntactic movement (Benincà and Poletto 2005); this tendency is decidedly more widespread in northern Italian dialects than in other Italian varieties.

  • Similarly, negation in northern Italy appears with various types of negative marker, which have different locations in the sentence structure (cf. §51.2.1).

  • The most obvious, and best-studied phenomenon that characterizes northern Italian dialects (and northern Romance including Florentine) is the presence of subject pronouns with variable degrees of obligatoriness, since the Middle Ages: see §13.4.6 and Ch. 47 for the details of this characteristic, which interacts with many fundamental aspects of sentence structure.

13.2 Phonology

The fundamental unitary nature of the area resulting from the shared application of the rules described above is in fact obscured by local variation, due to the application of more specific rules.

13.2.1 Suprasegmental phonology

Northern Italian dialects are characterized by two generalized diachronic processes: degemination and loss of unstressed vowels. The former is attested in (almost) all northern dialects, although various dialects exhibit the emergence of non-etymological context-determined gemination due to primary stress (see below). Consonantal-length oppositions appear to have survived in some peripheral areas such as Apennine Emilian dialects and more extensively in some Alpine Lombard varieties, e.g. Soglio (Val Bregaglia, Switzerland): [ˈtɛrːɐ] ‘earth’ < TERRAM ~ [ˈkɛːrɐ] ‘dear.FSG’ < CARAM (Loporcaro et al. 2005:601).

Regarding the diachrony of degemination, sonorants were targeted after obstruents (Martinet 1955) and underwent degemination quite recently: the northern dialects spoken by communities settled in Calabria and Sicily around the twelfth century exhibit /ɖ(ɖ)/ from /ll/, e.g. [jaˈɖina] < GALLĪNAM ‘hen’ (Piazza Armerina, Sicily, Rohlfs 1969a:229). Pescarini (2011) noted that in certain thirteenth-century vernaculars such as old Veronese the orthographic alternation between <l> and <ll> was still phonologically relevant, as apocope was not allowed after <ll>, e.g. cavall(**o) ‘horse’, ell(**o) ‘he’ before a consonant. As apocope was not permitted after geminates (see Zamboni 1976), this means that apocope is blocked by the double segment represented by the grapheme <ll> and hence that degemination of sonorants was not complete in the thirteenth century.

Elsewhere, non-etymological geminates emerged after short tonic vowels, arguably to fulfil a generalized requirement that stressed syllables must be heavy (§26.1.1). In Piedmontese this happened after [ə] < late Lat. [e]: [ˈvədːe] ‘see.INF’, [ˈməsːa] ‘mass’, while in Bolognese we observe consonantal lengthening after [a] < late Lat. [e] [sakː] ‘dry’ ~ [saːk] ‘sack’, [ˈfatːa] ‘slice’ ~ [ˈfaːta] ‘made.FSG’ (Coco 1970:88).

Due to the same requirement regarding the heaviness of tonic syllables, non-final vowels in stressed open syllables were lengthened. As a consequence of apocope and degemination, in several dialects vowel length remained the only phonological clue to distinguish minimal pairs such as [peːs] < PĒ(N)SUM ‘weight’ vs [pes] < PĬSCEM ‘fish’ (Mendrisiotto, Lurà 1987); [puˈliːt] < POLITUM ‘clean’ ~ [puˈlit] < PULLI + *ˈetti ‘chicks.DIM’ (Biella area, Grassi 1968:158); [pɑːs] ‘peace’ < PACEM ~ [pɑs] ‘step’ < PASSU(M) (Canavese: Oglianico) vs Tur. [pɑːz] ~ [pɑs]); Cremonese [kaːr] ‘dear.MSG’ < CARUM ~ [kar] ‘cart’ < CARRUM (Oneda 1976).

Some dialects have acquired new distinctive length contrasts in stressed syllables as a result of syncope, e.g. Gen. [ˈpusu] < PUTEUM ‘well (N)’ ~ [ˈpuːsu] < PULSUM ‘wrist’, [ˈpaŋsa] < PANTICEM ‘belly’ ~ [ˈbaːŋsa] < BILANCIAM ‘scales’ (Toso 1997:16f.).

Vanelli (1998a) reviews different theoretical analyses of the development of phonological vowel-length oppositions, arguing that word-final devoicing of an originally intervocalic consonant in Friulian contributed to extra lengthening of the preceding vowel (e.g. in [laːt] < *(al)ˈlatu ‘gone’ (~ [lat] < *ˈlakte ‘milk’). Indeed, these long vowels are significantly longer than those that occur before the intervocalic voiced consonants of their morphological alternants, e.g. [ˈlaːde] ‘gone.FSG’. Acoustic measurements show that the (definitely) devoiced final consonants remain significantly shorter than shortened original Latin geminates. For Milanese, Nicoli (1983:52) also records a difference in (p.189) the quality of the devoiced obstruent, since he describes devoiced final consonants as being intermediate between the voiced and voiceless obstruents.

Furthermore, the length distinction may be accompanied by differences in the quality of the vowel and/or following consonant: Mil. [loˈdɒː] < LAUDATUM ‘praised.PST.PTCP’ ~ [loˈda] < LAUDARE ‘praise.INF’, [peːz] < PENSUM ‘weight’ ~ [pɛs] < PISCEM ‘fish’ (Nicoli 1983:45f.). In Piedmont, Liguria, and Emilia-Romagna especially, long close-mid vowels [e] and [o] in open syllables evolved into falling diphthongs, e.g. Pie. [ˈtɛjla], Lig. [ˈtɛj(ɹ)a], < TĒLAM ‘canvas’. Emilian-Romagnol varieties show a tendency towards opening of the diphthong’s first element [ˈtajla], and in several Modenese dialects /aː/ in open syllables underwent fronting to /æ/ e.g. [mɛːr] < MARE ‘sea’. In Piedmontese (but not Canavese varieties) this phenomenon only affected the tonic vowel of first conjugation infinitives.

As previously mentioned, northern vernaculars exhibit cases of apocope, namely loss of final vowels. As in Tuscan, in central Veneto apocope is allowed only after single sonorants and it targets only [o] and [e]. The latter is not subject to apocope when expressing feminine plural features, as the feminine plural ending was not -e at the time when apocope occurred (Maiden 1996a). In dialects of Lombardy, Piedmont, and Emilia-Romagna, by contrast, apocope extended to other phonological contexts (i.e. after any kind of consonant) and to any final vowel but -[a]. Regarding the interplay of degemination and apocope, Zamboni (1976b) observed that apocope was originally sensitive to the length of the preceding consonant: in Venetian apocope is thereby mandatory after -L-, e.g. [mjel] ‘honey’ vs **[ˈmjele] < *ˈmɛle, while it is not allowed after -LL-, e.g. [ˈkae] ‘alley’ vs **[kal] < CALLEM. Such a synchronic distribution throws light on the chronological order of the phonological processes, as it entails apocope having taken place before degemination. Alternatively, degemination may be deemed a phonetic process without effects on the phonological representation of double segments, which remain intact.

As a consequence of vowel loss, Lombard and, more frequently, Piedmontese and Emilian-Romagnol dialects allow obstruents in coda position, e.g. [ˈkød.ɡa] ‘pig skin’ < *ˈkoteka (Frignano, Modena: Uguzzoni 1971) and complex consonantal clusters in word-initial position, e.g. Tur. [fnuj] ‘fennel’ < FENUCULUM, Eml. [zbdɛl] ‘hospital’ < HOSPITALEM. Such consonantal clusters are sometimes avoided by either cluster reduction (e.g. Frl. [fɔːr] ‘oven’ < FURNUM) or by inserting a non-etymological vowel. In word-initial position, a prosthetic segment allows the leftmost segment to syllabify as the coda of a newly formed syllable (Sampson 2010), e.g. Pie. [avˈziŋ] ‘neighbour’ < UICINUM. The same prosthetic vowel is used to syllabify clitic elements (see §13.4.4), which in dialects with generalized apocope are reduced to single consonants such as m (< ME, ‘me’), t (< TE, ‘you.SG’), l (< ILLUM, ‘him’), e.g. Tur. **it mit am ‘you to.me’ (Vanelli 1984b).

In word-final position, syllabification is obtained by means of a paragogic vowel, as in southern Romagnol, e.g. [ˈforne] < FURNUM ‘oven’. Elsewhere, as in northern Romagnol and Emilian, an epenthetic vowel is inserted between the two consonants, e.g. [ˈforɐn].

The distribution of such repairs is subject to cross-linguistic and intralinguistic variation, the latter depending ultimately on the nature of the cluster: sequences formed by segments that are close on the sonority scale, e.g. a glide and a sonorant, are more readily repaired than others (Bafile 2003b). The same syllabic principle is arguably responsible for the insertion of epenthetic consonants in sonorant + sonorant sequences, e.g. nombre < NUMERUM ‘number’ (Brisighella (Ravenna), AIS map 476). Elsewhere, syllabic constraints trigger the insertion of epenthetic consonants or cause the hardening of a glide to obtain a CV syllable, e.g. PAUOREM > [paˈgyra] ‘fear’ (Travo, Piacenza), cf. It. paura (Zörner 1989).

13.2.2 Segmental phonology

The different outcome of Latin single intervocalic consonants has served as a well-known isogloss for typological classification not only within Italo-Romance but within the contiguous Romance-speaking area of Europe; but recent scholarship argues for a more nuanced view that sees northern Italo-Romance (and wider western Romance) phenomena as but a further stage of evolution from what is found in central and insular Italo-Romance (Giannelli and Cravens 1997). In intervocalic position short consonants were prone to extensive lenition, involving voicing, fricativization, and sometimes total loss, with significant variation between and sometimes within regions:

  1. i. -[p]- > -[b]- > -[v]- > Ø, e.g. Rmg. [ʃaˈvoŋ], Mil. [saˈũ],Ven. [saˈoŋ] < SAPONEM ‘soap’;

  2. ii. -[t]- > -[d]- > -[ð]- > Ø (although in the absence of apocope of the following vowel, restoration as [d] is common, especially in Lombard and Venetan, or as a glide in Piedmont), e.g. Gen. [veˈɲyu], [veˈɲya] ‘come.PST.PTCP.M/FSG’ < *veˈnutu/-a, Mil. [senˈti], [senˈtida] ‘heard.PST.PTCP.M/FSG’ < *senˈtitu/-a, Vnz. [maˈɲa], [maˈɲada] ‘eaten.PST.PTCP.M/FSG’< *manduˈkatu/-a, but the nearby dialect of Burano has MSG [maˈɲao] ~ FSG [maˈɲa], while urban Veronese has [maˈɲado] ~ [maˈɲada] (Marcato and Ursini 1998:306);

  3. iii. [k] > [ɡ] > [ɣ] > Ø in Piedmont, if in the vicinity of the front vowels, [i], [y], [ø], e.g. Lmb. [furˈmiɡa], Ven. [forˈmiɡa], Pie. [fyrˈmia] < FORMICAM ‘ant’.

(p.190) As mentioned above, Latin intervocalic geminates were shortened over the whole area, usually remaining distinct from the original short consonants. Significantly, geminates have never been subject to voicing: unvoiced consonants are in fact maintained in intervocalic contexts when derived from geminates, e.g. Ven. [ˈtose] ‘cough’ < TŬSSEM vs [ˈtoze] ‘girls’ < TŌNSAS ‘girls’, while the reflex of -NN- remains distinct from that of short [n], which is weakened in final position: Cai. [ɑn] ‘year’ < ANNUM ~ [aŋ] ‘they have’ < HABENT, Tur. [pɑn] ‘cloth’ < PANNUM ~ [paŋ] ‘bread’ < PANEM (see below for Milanese).

Front rounded vowels [y] < Ū and [ø] < Ŏ are further distinctive northern features, although absent from Romagnol and Venetan varieties, e.g. Pie./Gen. [ˈlyŋa], Lmb. [ˈlyna] ‘moon’ < LUNAM. The fronting of [u] > [y] was often accompanied by the raising of back vowels: [o] > [u], [ɔ] > [o], [ɑ] > [ɔ], e.g. Pie. [ˈura], Gen. [ˈua], but Ven. [ˈora] ‘hour’ < HORAM. Stressed low-mid vowels produced early rising diphthongs over the whole area, as in Italian, probably originating in metaphonic contexts (Maiden 1997b; this volume §38.3.2), i.e. when followed by high vowels, normally [i] in northern Italian dialects.10 These diphthongs usually re-monophthongized, giving [e] < [je], e.g. Pie. [pe], Rmg. [pæ], Ven. [ˈpie] ‘foot’ < PEDEM, and [ø] (<*[ɥo] < *[wo]), where [u] was fronted to [y], e.g. Pie. [œv], Lmb. [øf], Lig. [ˈœvu ], but Ven. [ˈɔvo] ‘egg’ < OUUM ‘egg’. [ø] may also derive from [ɔ] in other palatalizing contexts, such as before a palatalized preconsonantal [k], a development that failed to take place in the Veneto: Pie. [nœjt], Lig. [ˈnøte], Ven. [ˈnɔte] ‘night’ < NOCTEM.

Metaphony (for a definition, see §§25.1.5, 38.1), as a rule triggered by -[i] (a characteristic plural desinence), affected most of the territory, but has long been receding, although morphologized relics abound, e.g. Mil. MSG [kel]/ FSG [ˈkela] ‘this/that’ ~ [ki] ‘these/those’ < ECCU ILLUM/ILLAM/ILLI (Nicoli 1983:176), Cerano (Novara): MSG [ku] ‘that’ ~ [ky] ‘those’ (Manzini and Savoia 2005, III:605), as well as residual areas, especially southern Veneto, around Padua and Vicenza, but excluding Venice, Romagna ([fjawr] ~ [fjur] ‘flower,-s’ < *ˈflore, -i, Piedmont (Biella area): [maŋ] ~ [mɛŋ] < *ˈmanu, -i ‘hand, -s’ (Berruto 1974:30), and peripheral Alpine regions, for instance, in Ticinese (Sanga 1997:254). Metaphony in the Alpine Intemelian group in Liguria appears to be a fairly recent phenomenon (Forner 1997:250).

In Venetan dialects, only mid vowels originating from Latin tense vowels are affected by metaphony (even though the tense/lax distinction is now neutralized for back vowels, while front mid vowels maintain the Latin distinction). Hence, metaphony never targets reflexes of Lat. Ŏ (e.g. SG [ˈloɡo, ˈfoɡo, boŋ] ‘place, fire, good’ ~ PL [ˈloɡi, ˈfoɡi, ˈboni]) and Ě (e.g. [ˈfɛro, ˈvɛro, ˈbɛlo] ‘tool, true, nice’ ~ PL [ˈfɛri, ˈvɛri, ˈbɛj]). Reflexes of Ē, Ō do undergo metaphony: [ˈtozo, ˈtuzi] ‘boy, -s’, [ˈfjore, ˈfjuri] ‘flower, -s’, [ˈvero, ˈviri] ‘glass, -es’, [ˈmeto, ˈmiti] ‘I put, you put’. In old Paduan (e.g., 16th-c. Ruzante) mid lax vowels in metaphonic contexts diphthongized (as in southern Italian dialects: see Calabrese 1985): MSG bon ‘good ’~ PL buoni, MSG vegio ‘old’~ PL viegi.

Vowel harmony is attested in Lombardy (Sanga 1997:256) and Piedmont. In Piverone (Piedmont), the vowel height of the stressed vowel determines the height of the final vowel (Flechia 1896–8; Canalis 2010): if the stressed vowel is high (/ɪ i u y/), only [a i u] appear as final vowels, otherwise only [a e o] are possible. This harmonization leads automatically to [i/e] allomorphs for feminine plural and second singular endings, and [u/o] for third plural, e.g. [ˈlyva] ‘she-wolf’< LUPAM ~ PL [ˈlyvi], but [ˈpɛra] < PETRA(M) ‘stone’~ PL [ˈpɛre]; [a ˈskrivu] ‘(they) write’ < SCRIBUNT but [a ˈkrɛdo] ‘(they) believe’ < CREDUNT.

Final obstruent devoicing is characteristic of Lombard, Emilian, Romagnol, and some Piedmontese varieties. Among the latter, Canavesano, Astigiano, and Biellese tend to devoice final consonants: Asti [œf] ‘egg’ < ˈɔvu, Biellese [raʧ] ‘ray’ < RADIUM vs Tur. [raʤ] (Berruto 1970:29), but western and southern varieties maintain voiced consonants, which remain phonemically distinct from their voiceless counterparts: Tur. [lyz] ‘light’ LUCEM ~ [lys] ‘pike’ < LUCIUM, [nɑz] ‘nose’ < NASUM ‘nose’ ~ [nɑs] ‘(it/he/she) is born’ < *ˈnaske. Similar contrasts may also be found in Emilia-Romagna: Bol. [frap] ‘rags’ ~ [frab] ‘blacksmith’ < FABRUM, [lot] ‘I struggle’ < LUCTOR ~ [lod] ‘I praise’ < LAUDO (Coco 1970:94, 96); Lugo [mats] ‘bunch’ ~ [maʣ] ‘May’ (Pelliciardi 1977:24f.).

While Milanese exhibits distinctive vowel length in the presence of vowel apocope and final obstruent devoicing (e.g. [ɡœp] [ˈɡœːbɒ] ‘hunchback’ < *ˈɡɔbbu, *ˈɡɔbba, [maʧ] ‘May’ < MAIUM), apocope and devoicing are not crucial for the development of phonologically relevant vowel length in Genoese (see the examples above) and various Emilian Apennine dialects, e.g. Lizzano in Belvedere: [kaŋˈta] < CANTATUM ‘sung’ ~ [kaŋˈtaː] CANTATIS ‘you.PL sing’ (Loporcaro et al. 2006).

Vowel nasalization is another feature that distinguishes the development of northern from that of central and southern Italo-Romance, linking it again with transalpine (p.191) Romance. In some cases this has led to the emergence of nasal vowels, which may be found in a swathe of dialects ‘running from northern and eastern Piedmont through western and central Lombardy and down to the south-eastern area of Emilia-Romagna’ (Sampson 1999:247), e.g. Mil. [sãː] ‘healthy’ < SANUM ~ [saː] ‘salt’ < *ˈsale, [sa] ‘he knows’ ~ [san] ‘they know’ (both verb forms revealing the analogical influence on SAPERE of the paradigms of DARE, STARE; Rohlfs 1968:285). Note that the nasal consonant reappears in morphological alternation and derivational forms: Mil. [mã] ‘hand’ < MANUM ~ [maˈnina] ‘little hand’; [bũ] < ‘good.MSG/PL’ < BONUM, -OS ~ [bɔn] ‘good.FPL’ < BONAS. However, vowel nasality is retreating under the influence of the standard language. Elsewhere vowels denasalized, resulting often in a sequence of vowel and velar nasal coda [ŋ] or [ŋ]: Cai. [dɛŋʧ] ‘tooth’ < DENTEM, Gen. [maŋ] ‘hand’ < MANUM, Ven. [viˈsiŋ] ‘neighbour’ < UICINUM. Since parts of the northwest and Emilia-Romagna also experienced nasalization in stressed syllables followed by an intervocalic single nasal, the restoration of a velar occlusion has resulted in nasal sequences such as Prm. [ˈloŋne], Bol. [ˈloŋna] ‘moon’ < LUNAM, which in the west have long been reduced to intervocalic [ŋ]: Tur./Gen. [ˈlaŋa] ‘wool’ < [ˈlaŋna] < LANAM.

Table 13.1 Palatalization of the cluster C + [l]

The dialects of northern Italy

The palatalization of the liquid in the cluster C + [l] is a development that characterizes most Italo-Romance varieties (excluding some Abruzzo dialects, Sardinian, Friulian, and Ladin) (cf. §39.4.1), but a distinctive feature of most northern Italian dialects is the further palatalization of the first consonant of some of the clusters, especially the velars, but also the labials [p], [b], and labio-dental [f] in Liguria (see Table 13.1). The affricates [ʦ] and [ʣ] that developed from Latin velar and dental consonants followed by a front vowel or yod have simplified to fricatives [s] and [z] in all but conservative areas, such as the Val Bormida: [ˈʦɛjna] ‘supper’, [u leʣ] ‘he reads’.

Some instances of palatalized velars before [a], typical of Gallo-Romance, Ladin, and Friulian, suggest the phenomenon was more widespread in the past (Tuttle 1997) and they survive in peripheral areas, e.g. in Alpine Lombard [ɡjat] ‘cat’ (Sanga 1997:259). Palatalized sibilants are characteristic of some parts of the north, especially as the first element of a consonantal cluster in Romagnol and Trentino varieties, but also in parts of Piedmont, Lombardy, Emilia, and the Veneto, e.g. [ˈmuʃka] ‘fly’ < MUSCAM. Palatalized final consonants also contribute in many areas to the marking of number (see below).

13.3 Morphology

13.3.1 Nouns and adjectives

In the north of Italy the inflectional morphology of nouns and adjectives has been drastically reduced compared to that of Latin, following the loss of case distinctions (partially retained, however, in the pronominal system), the fall of most final vowels other than [a] in some regions, and the spread of dominant patterns of variation at the expense of others. The grammatical categories of number and gender continue in most areas to be marked through inflection, although this may be conveyed by stem change due to metaphony or consonantal palatalization affecting certain groups of nouns. Invariable nouns, however, are more common than in the centre and south due to the loss of unstressed vowels, the analogical spread of feminine plural -[e] to former Latin third declension nouns, and the regression of metaphony. Diachronic developments also show the spread of inflectional syncretism, which may neutralize the gender distinction, especially in the plural.

In Liguria and southern Veneto, where final vowels were largely maintained, we find two main inflectional alternations corresponding to gender and number, e.g. Gen. [ˈfiʤu, -i] ‘son, ‑s’, [ˈfiʤa, -e] ‘daughter, -s’, Ven. [ˈnɔno, -i] ‘grandfather, -s’, [ˈnɔna, -e] ‘grandmother, -s’.

This dominant pattern, with clear gender marking, attracted members of the Latin third declension, such as Gen. [ˈnɛvu, -i], Vic. [neˈvodo, -i] ‘nephew, -s’ < NEPOTEM, -ES, Gen. [ˈnɛsa, -e], Vic. [neˈvoda, -e] ‘niece, -s’ < NEPOTEM, -ES; Noli (Liguria) and Ven. [ˈava, -e] ‘bee, -s’ < feminine APEM, -ES. (p.192) Otherwise, the rest of the class shows an [e]/-[i] number alternation in the masculine: Gen. [ˈpɛʃe, -i] , Ven. [ˈpese, -i] ‘fish’< PISCEM, -ES, but feminine nouns have typical FPL -[e] and are thus invariable: Gen./Ven. [ˈʧave] ‘key, -s’ < CLAUEM, -ES. Despite the loss of final -[e] due to apocope, e.g. Mil. [ˈdɔna ~ dɔn] ‘lady, -ies’ < DOMINAM, ‑AS, it was kept or reinstated as a feminine plural marker in eastern Lombardy and most of Piedmont, Tur. [ˈdɔna ~ ˈdɔne], and as -[i] in some dialects: Finale Emilia [ˈdɔni]. Maiden’s (1996a) comprehensive overview of the much discussed issue of whether Romance -[i] and -[e] plurals derive from the nominative or oblique case forms, concludes that the latter most likely continues the Latin accusative feminine plural -AS, as the fact that it produced no instances of palatalization of the preceding consonant suggests that at the relevant period it featured the diphthong *[aj] and not the front vowel one would expect, were it a reflex of the nominative plural -AE. Zörner (1995) also adduces persuasive evidence regarding the different development in northern Italian dialects of stem-final nasals in first and third declension feminine nouns: in reflexes of the former their evolution parallels that of word-internal nasals in contrast to the word-final pattern of nasalized vowel found in the latter, e.g. [vi'zəjn] ‘female neighbours’ < UICINAS ~ [parˈzõ] ‘prisons’ < PREHENSIONES. Invariable Friulian, Ladin, Alpine Lombard nouns in -[a] < -AS (Val Bregaglia, Ticino), e.g. ['kabra] ‘goat, -s’ < CAPRAM, -AS, as in parts of northern Tuscany (Rohlfs 1968:30) reveal a similar tenacity of the most open vowel, but total loss of -S.

In the core Piedmontese and Lombard varieties, and in northern Veneto (Feltrino, Bellunese), most masculine nouns are invariable, as are all nouns ending in a stressed vowel and feminine nouns ending in a consonant or a support vowel, e.g. Tur. [əl ɡɑt ~ i ɡɑt] ‘the cat, -s’, [əl ˈprɛvi ~ i ˈprɛvi] ‘the priest, -s’, [la fjur ~ le fjur] ‘the flower, -s’ [la ˈmɑre ~ le ˈmɑre] ‘the mother, -s’; Mil. [l œʧ ~ j œʧ] ‘the eye, -s’, [el maˈri ~ i maˈri] ‘the husband, -s’, [la reːt ~ i reːt] ‘the net, -s’. Specific phonetic changes have resulted in a number of exceptions within the masculine group, due to the palatalization of a stem-final coronal consonant prior to the loss of postulated plural -[i] inflection: Tur. [fil] ~ [fij] ‘thread, -s’ [kaˈval] ~ [kaˈvaj] ‘horse, -s’; [ɑŋ] ~ [ɑɲ] ‘year, -s’ but also modern, [ˈɑni]; Mil. [fraˈdɛl] ~ [fraˈdɛj], Ven. [fraˈdɛjo] ~ [fraˈdɛj] ‘brother, -s’. Palatalized stems may cooccur with inflexion, where this exists: Gen. [aˈmigu] ~ [aˈmiʒi] ‘friend, -s’, [ˈkasu] ~ [ˈkaʃi] ‘kick, -s’, [ˈaze] ~ [ˈaʒi] ‘donkey, -s’ (Toso 1997:53f.), whereas Pie. [aˈmiz] ‘friend’ shows a singular analogically formed from a palatalized plural. Northern Italian palatalized plurals are deemed to have derived from the Latin nominative case in -[i], while their coexistence with -[s] plurals in Friulian reflects the persistence in this area of a two-case system (nominative vs oblique), similar to that found in early Gallo-Romance texts, for longer than in the rest of Italo-Romance (for details see Benincà and Vanelli 1978, which develops the findings of Sabatini 1965a; also Maiden 2000b). The two-case system disappeared before the appearance of the first Italo-Romance texts, but according to Benincà and Vanelli the retention of two plural inflection types was favoured by the marked nature of the palatalized forms, which also guaranteed the number distinction in words ending in -[s]; elsewhere in northern Italy the presumed loss of -[s] resulted in invariable plural forms. An alternative phonetic development is proposed by Forner (2005) in a comparative study that draws on Alpine Ligurian variational data that involve Latin accusative -OS > [uj] > [i].11 He suggests that such a hypothesis, which does not have recourse to a two-case system, is not implausible for other areas, especially northern dialects.

Palatalized plurals, which link these varieties to Ladin and Friulian and which were once widespread in the Po valley, are today in regression but survive in conservative varieties: Bresciano [ɡat] ‘cat’ < CATTUM ~ [ɡaʧ] ‘cats’, [aˈsɛŋ] ‘donkey’ < ASINUM ~ [aˈsɛɲ] ‘donkeys’; Venetian nouns in -[oŋ]: [balˈkoŋ] ‘balcony’ ~ [balˈkoj], ‘balconies’; and in certain closed paradigms, e.g. the quantifier ‘all’, Mil. [tyt] ~ [tyʧ] ‘all’, Cai. [ˈtyʧi], but Tur. [ˈtyti]. Similarly, metaphonic plurals are becoming rarer, although still characteristic of Emilia-Romagna, Liguria, and southern Veneto (not Venice): Biella: [maŋ] ~ [mɛŋ] ‘hand, -s’; Bol.: [ba] ~ [buː] ‘ox, oxen’, [paʃ] ~ [pɛʃ] ‘fish’; Lugo [ˈojum] ~ [ˈujum] ‘elm, -s’, [ˈvedar] ~ [ˈvidar] ‘glass, -es’, [ɡat] ~ [ɡæt] ‘cat, -s’ (Pelliciardi 1977); Gen. [kaŋ] ~ [kɛŋ] ‘dog, -s’, [feˈraː] ~ [feˈræ] ‘blacksmith’; SVen. [ˈmeze] ~ [ˈmizi] ‘month, -s’, [niˈsoeo] ~[ niˈsui] ‘sheet, -s’ (Marcato and Ursini 1998:68).

Despite a wide range of dialect variation in plural formation (Rohlfs 1968:25–51) and unusual developments such as the western Lombard feminine plural in -[ˈan], e.g. [ˈtuza] ~ [tuˈzan] ‘girl, -s’, noun inflection in northern Italian dialects reveals a consistent trend towards regularizing paradigms through the extension of dominant patterns of variation. Regarding the small group of nouns that preserve the Latin neuter ending [a] for pairs or collectives (MSG -[o] / FPL -[a]), this trend dates back to medieval times (see §42.4). The plural of such nouns either acquired the regular feminine plural -[e] as in Liguria: Gen. [ˈlɛrfu] ~ [ˈlɛrfe] ‘lip.MSG, lips.FPL’, [ˈɔsu] ~ [ˈɔse] ‘bone, -s’, or zero ending in Romagna: Lugo, [l ɔs] ~ [aʎi os] ‘the bone, the bones’, [l ov] ~ [aʎi ov] ‘the egg, the eggs’, but [e braʦ] ~ [al ˈbraʦa] ‘the arm, the (p.193) arms’, (Pelliciardi 1977:53f.), thus remaining anomalous in the change of gender (although new masculine -[i] plurals emerged for some, with semantic differentiation), or the group was totally assimilated into the dominant masculine one, as in Piedmont, Lombardy, and the Veneto: Tur., Mil. [l œv] ~ [j œv] ‘the egg, -s’, Ven. [ˈlavaro, -i] ‘lip, -s’, [ˈvɔvo, -i] ‘egg,-s’. However, -[a] plurals (and feminine gender) survive in the Ravenna area, without the metaphony found in the corresponding masculine plural formations, which have metaphorical meaning: [ˈbrasa] ‘arms’ ~ [i brɛs] ‘the arms (of a cross)’ (Masotti 1999).

13.3.2 Verb morphology

13.3.2.1 Tenses

Verb systems normally include the following finite forms: indicative (present, imperfect, past, future), subjunctive (present, imperfect, past, and pluperfect), and conditional (present and past).

In modern vernaculars, perfect tenses are normally analytic even if the simple past (preterite) was robust in northern Italian varieties from the Middle Ages to the Renaissance. Among the synthetic verb forms the conditional is noteworthy for its morphological diversity, reflecting the different evolution of the form in the various regions. Like the future forms, of which it originally represented a past version (future-in-the-past), its various exponents derive from a periphrasis involving the infinitive and an inflected form of HABERE ‘have’ (in this case past tense). Some early texts, especially Lombard, retain instances of analytic forms of both future and conditional (Rohlfs 1968:334, 346), e.g. Barsegapé, a portare ‘(he) ‘has.PRS carry.INF (= he will carry)’, Bonvesin, have fa ‘(I) had.PAST do.INF (= I would do)’, but modern conditional paradigms can be divided into three main types deriving from (cf. §§27.6, 46.3.2.2): (a) infinitive + imperfect indicative of HABERE (the most common pan-Romance type); (b) infinitive + preterite of HABERE (as in Italian), including a now common subtype which shows influence of the imperfect subjunctive in the spread of an -[s] infix; (c) mixed paradigms, showing the overlap of types (a) and (b). AIS map 1019 offers first person singular conditional formatives for ‘I would eat’ as follows: type (a) [manˈʤria], [manˈʤ(r)ɛjva], corresponding broadly to Piedmont and Liguria respectively; type (b) [manʤɐˈre], broadly characterizes Emilia-Romagna, while type (c) [manʤɐˈrisa], flourishes in Lombardy. Venetan dialects, furthermore, show mixed paradigms in which the above types are distributed according to person distinctions (Zamboni 1974): Pad. 1SG magnarìa, 2SG te magnarissi, 3SG el magnarìa; Vnz. 1SG magnarave, 2SG and PL magnaréssi; Bel. 2PL magneréssi, all other persons magneràe/magnerìe.

However, a small area on the border of Piedmont and Liguria has witnessed the almost total assimilation of conditional forms to the past subjunctive paradigm (see the discussion on southern Italian dialects in §16.4.2.1), presumably due to the combined effect of their semantico-syntactic affinity (Harris 1978a:181f.) and to the phonological weakness of /r/ in this area (see Parry 1990):

(1)

The dialects of northern Italy

The fact that the very few /r/ conditionals found in the area (in Cairese just ‘be’: COND. [saˈɹɛjsa] ~ PST.SBJV [ˈfusa]) belong to frequent irregular verbs, such as ‘do’, ‘know’, suggests that these are relics of a previous state of affairs and that the current formal identity of verbs in the protasis and apodosis of Valbormidese conditional structures is not analogous to the continuation of the two Latin pluperfect subjunctives found in Swiss Raeto-Romance and southern Italo-Romance (Parry 1990).

13.3.2.2 Person endings

In many dialects (excluding Piedmontese and some Emilian and Ligurian varieties), third person singular and plural endings are syncretic.

Moreover, as noted by Rohlfs (1968:§527) and Meyer-Lübke (1895:§§131–6), in northern Italian dialects as in French the first singular ending -o is expected to disappear as a result of the general rule deleting all final vowels except -A. In fact, northern Italian dialect and French texts show that the rule had regularly applied, e.g. OFr. aim ‘I.love’, Occ. cant ‘I.sing’, OBgm. laf ‘I.wash’. The same rule affected the -o ending of second and third conjugations (OFr. dor ‘I.sleep’, vent ‘I.sell, OOcc. ven ‘I.sell’), which is still zero in French and many northern Italian and French dialects. The translations collected by Salviati in 1584–6 (Papanti 1975) have Mil. digh ‘I.say’ < DICO, intend ‘I.mean’ < INTENDO; Bol. digh, pregh ‘I.pray’ < PREGO(R). Mantuan dig, vugn ‘I.come’ < UENIO, but Bgm. has already dighi, zuri ‘I.swear’ < IURO. Where old varieties had a zero ending, more recent forms present an -i/-e ending. Papanti’s (1875) translations present Bgm. dighe, zure, Bol. degh, pregh, Mil. Disi, vegni, preghi. Most Piedmontese varieties have -o, -u, though some still display the zero ending, as do most Ladin and Romansh varieties.

A careful examination of Friulian, Ladin, and Swiss Romansh permits us to reconstruct the development of this innovation, which seems valid for northern Italian dialects, French, and French dialects. The phenomenon involved more or less all northern Italian dialects, namely all those dialects where final unstressed vowels (except -a) (p.194) fall. The frozen developmental stages of Friulian show us that it starts in the present indicative of the first conjugation, with -a as thematic vowel; the extension to the other conjugations (occasionally to other tenses and moods) follows, but in many varieties never happened: we have -i in the first conjugation and Ø in the others, or -i in all conjugations (or, in the past, Ø in all conjugations), but we never find Ø in the first conjugation and -i in the others.

This means that the reason for the insertion of -i must be a situation produced in the first conjugation: the thematic vowel coincides with the ending in the second and third persons singular, which are never deleted, following the rule (-AS > -is, -os, -as; -A > -e, -o, -ə); -o, the first person singular ending, however, was targeted by the deletion rule, and this produced an asymmetry in the syllabic alignment of the singular paradigm:

(2)

The dialects of northern Italy

This description is valid for all varieties of northern Romance that share the same format of the deletion rule.

Analogical processes used -i / -e / -u (according to dialect) as a first singular morpheme, extending the ending to other conjugations, or to such tenses as the imperfect indicative, which displayed no asymmetry.

The nature of the epenthetic vowel is in some cases that of a neutral vowel, also used to repair disallowed consonant clusters and optimize syllabic structures (in Friulian, Ladin, or Piedmontese), in other cases it clearly represents the encliticization of the first person singular subject pronoun i, e.

For first plural present indicative, the common Romance basis is -amo, -emo, -imo, subsequently reduced to -émo, -ìmo, prevalent in northern Italian dialects. In western dialects ‑on / -um(a) (in old Bergamasque we find -om, Pie. -uma). The latter may be comparable to the French expression of first person plural on chant ‘man sings > we sing’ < HOMO CANTAT; this could mean that the indefinite form substituted the regular morphological continuation of the first person plural ending. A parallel development is found in Flo. si canta (cf. §14.4.2.2), Frl. si cjante ‘one sings, we sing’; on the other hand an indefinite subject omo ‘man’ appears in early old Italian (see Egerland 2010; cf. also §60.7). The ending -òm, -ùm(a) could be the result of enclisis, to be compared with the above-mentioned first person singular vocalic ending, and with enclisis of other subject pronouns in Lombard (cante-t ’sing=you.SG’, cantu-f ‘sing=you.PL’).

Support for this hypothesis may be found in the fact that there exist varieties of Lombard where the accent remains on the verb root: cànt-um, tròv-um, véd-om ‘we sing, find, see’ (with weakened atonic vowel also cànt-en, tròv-en, etc. ‘we sing, we find, we see’; cf. Bgm. om canta ‘man (= one) sings’ > ‘we sing’, OMil. (Bonvesin) um era (< HOMO ERAT) ‘we were’ (later èr-um, with enclisis of grammaticalized -um); see Meyer-Lübke 1895:§135; Rohlfs 1968:§530).

13.3.2.3 Root alternations

Root alternations result in cross-paradigmatic and intra-paradigmatic patterns of allomorphy and suppletion. The former are due to analogical changes among a given class of verbs, usually athematic ones. In northern Italian dialects, for instance, the velar of the root dic- [dik]- ‘say’ fac- [fak]- ‘do’ is often extended to other roots of the present indicative, starting from the first person singular as in Ven. dag-o ‘I give’, stag-o ‘I stay’. Intra-paradigmatic effects, conversely, cause the extension of the irregularity across different cells of the paradigm of the same lexeme. (3a,b) shows the distribution of this velar element in the present indicative paradigm of veder(e) ‘to see’: notice that first and second plural roots display no irregularity (see Maiden 2005; Maschi 2007):

(3)

The dialects of northern Italy

Table 13.2 Paradigm of ‘to go’ in Ladin, Venetan, and Lombard dialects

The dialects of northern Italy

Northern dialects differ with respect to the treatment of first and second person plural endings, which can either maintain the etymological root (as above), select for a different allomorph, or take the same root as the other persons. These possible outcomes are illustrated in Table 13.2, which compares the present indicative of ‘to go’ (< IRE) in three dialects. In all dialects, the root v- has replaced the etymological root in some forms of the present indicative except for first and second persons plural, which (p.195) show a different behaviour: in Dolomitic Ladin varieties, they are reflexes of the original etyma (often with a prosthetic consonant); in Venetan and central Friulian dialects they take a suppletive root (> It. andare, Fr. aller, Frl. , Ven. nare: see Prosdocimi 1993; 2001 for one proposed etymology); in Lombard dialects (Maschi 2007), the root v- has been extended to the whole paradigm, as in Milanese. Some Lombard dialects show a mixed pattern in which both suppletive and levelled forms coexist, showing that the second person plural resists paradigmatic levelling more than the first person plural.

13.3.2.4 Past participles

Table 13.3 Past participle endings of athematic verbs

The dialects of northern Italy

As previously noted, athematic verbs tend to exhibit cross-paradigmatic analogies regardless of their etyma. This is true also for the endings of past participles, which in many dialects is ‑t(o)/‑it(o) (< FACTUM ‘done’, DICTUM ‘said’) even for participles in -ATUM, -ITUM, where a voiced coronal was the expected output. The unvoiced alveolar has been extended to other roots as illustrated in Table 13.3 comparing Veronese, a dialect which has preserved the etymological distinction, with Ligurian and Piedmontese, in which past participles have been levelled.

Several dialects have developed innovative endings which arguably derive from the extension of the formative -sto of reflexes of POS(I)TUM ‘put’ and QUAES(I)TUM ‘asked’. By combining this ending with the thematic vowel -é- of the second conjugation we obtain the Venetan ending -èsto which from the fifteenth century onwards gave rise to alternations such as movesto/mosso ‘moved’, piasesto/piasso ‘liked’, coresto/corso ‘run’, tasesto/tasù ‘fallen silent’ tolesto/tólto ‘taken’. In some Alpine dialects, both in Lombardy and Veneto, the same suffix gave rise to a series of participles ending with -óst(o)/-óstu (e.g. curostu ‘run’, muvostu ‘moved’), possibly by analogy with posto ‘put’, nascosto ‘hidden’ (Rohlfs 1968:§624).

13.3.3 Word formation

Cross-dialectal variation is mainly due to differences concerning the semantic value of a given suffix. Such differences are particularly striking in the case of evaluative suffixes. Lat. ‑ACEUS, for instance, which was originally used to form adjectives from nouns, has acquired in Tuscan/Italian a pejorative meaning (e.g. amorazzo ‘flirt’), while in several northern dialects it has an augmentative value, e.g. Ven. caenaso ‘big chain’. Milanese exhibits the peculiar augmentative/pejorative suffix -atter, e.g. ongiatter ‘long nail’, whose unclear etymology (Rohlfs 1969:381f.) might be related to the Friulian derogatory -at < *-attu.

The morphophonological shape of affixes is affected by the phonological evolution of each vernacular, which may lead to overlap of the reflexes of etymologically distinct suffixes as in Piedmont, where the reflex of suffix ‑ORIA has replaced that from -tora in the formation of nomina agentis, e.g. filoira ‘spinner.F’, sartoira ‘dressmaker.F’.

As for nominalizations, the northern dialects exhibit a series of suffixes which etymologically have nothing or little to do with nominalization such as Bgm. -aja (< ‑ALIA, Salvioni 1896, e.g. marüdaja ‘maturation’ < marudà ‘to ripen’), Ven. and Frl. -one (e.g. rosegón ‘bite’ < rosegar ‘to nibble’, sburtón ‘shove’ < sburtar ‘to shove’), Pie. ‑er(o) < ‑ARIUM (e.g. semineri ‘sowing’).

13.3.4 Pronouns

Nominative pronouns occur in a double series: strong and clitic. The former derive from Latin oblique (non-nominative) forms, while the latter are reflexes of the Latin nominative series. Diachronically, the emergence of the clitic series appears parallel to the progressive specialization of oblique forms with subject function, observable since the earliest attestations (see §13.4).

Etymologically oblique subject pronouns spread rapidly to other contexts, replacing nominative strong forms. This led to widespread syncretism between nominative and accusative tonic forms. Third person forms are reflexes of oblique/genitive pronouns such as lui ‘he/him’, lei ‘she/her’, loro ‘they/them’ < GEN.PL ILLORUM (sometimes M lori, F lore with analogical plural). First and second person plural strong pronouns derive from reflexes of NOS/UOS compounded with ALTERI ‘others’ arguably originating from an exclusive form. Thus the strong series of the dialect spoken in Brescia (Lombardy): me ‘me’, te ‘te’, lü/le ‘he/she’, noter ‘we’, voter ‘you.PL’, lur/lure ‘they.M/F’.

In various dialects of Lombardy, third person strong pronouns can combine with deictic particles expressing proximity/distance, e.g. lüche ‘he/him, who is near the speaker’ vs lüla ‘he/him, who is far from the speaker’.

Things are more complicated with clitic pronouns, as their paradigms have been subject to noticeable changes.

(p.196) As for subject clitics, the early vernaculars (thirteenth and fourteenth centuries) do not exhibit any clear attestation of them. Cases of morphologically reduced nominative forms are in fact attested (cf. ello > el or lo ‘he’), but their status as syntactic clitics is far from straightforward until the sixteenth century (Vanelli 1984b).

Paradigms of subject clitics in present-day dialects are systematically defective (§45.2.2). Furthermore, subject clitics display various enclisis/proclisis asymmetries. Zero exponence, for instance, is more frequent in proclisis, while enclitic forms, if present, are closer to their etyma than those in proclisis.

First singular, first plural, and second plural subject clitics are often expressed by a single syncretic exponent, which usually coincides with a reflex of EGO ‘I’. In some dialects, certain subject clitics are expressed by the combination of two exponents, a ‘vocalic’ formative (possibly derived from EGO) followed by an etymological one, e.g. Frl. i ti ‘you’ (cf. §45, §47).

In Piedmontese (Parry 1993), the first person plural enclitic =ne has been extended to the first person singular and, later on, to other plural persons:

(4)

The dialects of northern Italy

Object clitics (§45.2.1) are often expressed by syncretic exponents as well. In all dialects, a single exponent expresses both the third person dative and the so-called locative clitic (on the origin of the syncretism, see Benincà 2007a).

(5)

The dialects of northern Italy

In some dialects, the exponent /ɡ/ stands for a first person plural clitic too:

(6)

The dialects of northern Italy

In other dialects, the first person plural clitic is expressed by a reflex of Lat. ME ‘me.ACC’:

(7)

The dialects of northern Italy

A special case of syncretism concerns reflexive forms; in a number of varieties the third person reflexive (< SE/SIBI) has been extended to other persons. In the following example, for instance, the reflexive form se occurs with a first person plural subject:

(8)

The dialects of northern Italy

Benincà and Poletto (2005) show that the extension of the se-type clitic to other persons follows an implicational scale, whose starting point is the first person plural pronoun and endpoint is the second person singular pronoun (see §48.3). In some Lombard dialects, first and second person reflexives are expressed by combining the first and second person clitic with the third person reflexive one, e.g. Lugano va sa ‘yourselves’.

Compound forms, i.e. clitics formed by the combination of two morphological exponents, are frequently attested to express locative and partitive clitics (see Ch. 45). In many Venetan dialects, for instance, the genitive/partitive clitic is formed by a combination of the locative clitic ghe /ge/ and the partitive element ne. Penello (2003) reports cases of composite partitive forms which may be analysed as reduplicated forms of the usual partitive ne/en, e.g. nin ‘of it/them’ in Romagnol; Cai. nun (before a consonant, Parry 2005:176).

13.3.5 Articles

Underlying extensive formal microvariation, all northern varieties have definite articles deriving from the Lat. ILLE paradigm, although reflexes of the IPSE paradigm are prevalent in the northwest with weak demonstrative value (Ascoli 1901). The preconsonantal masculine singular definite article exhibits two main types, el and lo, which in the medieval period were contextually determined (see Vanelli 1992 on their evolution). Most regions have generalized one or the other, e.g. Liguria has the lo type (the [l] has been lost but reappears before vowels): [u kaŋ] ‘the dog’, [l ˈœvu] ‘the egg’; the el type dominates elsewhere: Ver. [el ˈspozo] ‘spouse’, Bel. [al paŋ] ‘bread’, but Burano [lo ˈvɔvo] ‘the egg’; Bgm. [ol luf], ‘the wolf’ but [o] before an initial sibilant (Bernini and Sanga 1987:80), Pie. [əl kaŋ] ‘the dog’. In the (p.197) Monferrato and Val Bormida a variant [ɛɹ] alternates with [u] before initial coronal consonants: Cai. [ɛɹ kaŋ] ‘the dog’, but [u lɒjt] ‘the milk’, [u ʃpeʤ] ‘the mirror’, but this [u], unlike Lig. [u], which is the product of aphæresis, represents a vocalization of preconsonantal [l] (Parry 2005:139). Similar morphological variation is found in the feminine plural, although in this case the le type dominates (Liguria, Veneto, Turin), but the el/er type is common in southern Piedmont, and Romagna has al. Syncretic forms are found before vowels in Piedmont and Lombardy, often [j] via palatalization, while [i] for both masculine and feminine plural is typical of Lombard and northeast Piedmont (see Vanelli 1998b).

Unlike central and southern dialects, northern dialects (and northern Tuscany) have a partitive article, as does French, composed of the preposition ‘of’ and the definite article. Use of the preposition on its own for indefinite quantification is found in many Piedmontese varieties, e.g. Canavesano (9), and after a negative in Piedmontese, Lombard, and Ligurian (10):

(9)

The dialects of northern Italy

(10)

The dialects of northern Italy

As in standard Italian, the plural of the indefinite article takes the form of the partitive article, which is however not always obligatory (see Parry 2005:142f. for Cairese).

13.3.6 Demonstratives

On the whole demonstrative systems are based on a binary spatiotemporal opposition that simply relates the referent positively or negatively to the deictic centre, namely the speaker (cf. §54.1.1). Short forms derived from ISTUM exist alongside long ones from ECCU ISTUM, e.g. Ven. [sto] / [ˈkwesto] ‘this’ ~ [ˈkweo] ‘that’; Bol. [st] / [kwast] ~ [kwal], Gen. [stu] / [ˈkwestu] ~ [ˈkwelu]. In parts of Emilia-Romagna, the feminine plural of the proximal demonstrative ends in -[l], presumably by analogy with the distal demonstrative and the definite article: [stil/stal ˈdɔni] ‘these women’, but in predicative function the full form is used: [ˈkwesti/ˈkwisti] (Manzini and Savoia 2005, III:591–600).

Orientation to the addressee may still be achieved using the postnominal spatial adverbs [ki] ‘here’ and [li]/[le] ‘there’: Gen. [sta ˈkaza ki] ‘this house’, [sta ki] ‘this one’. In some dialects (Lombard, Emilian, Friulian), these postposed adverbs are the sole exponents of the spatial distinctions, especially in adnominal function, since the ECCU+ILLUM form, just like original ILLE, has lost its specific ostensive value and been generalized: Lmb. [kel ˈliber ki/li/la] ‘this/that (near you)/that book’. In Brugé’s generative analysis (1996), these adverbs occupy the base position of the demonstratives in the noun phrase, whereas the first component raises to the determiner position.

In Piedmontese and some western Ligurian varieties three terms may be found, e.g. Tur. [stu] / [kust] ~ [əs] < IPSUM ~ [kul], but the IPSE reflex is not associated specifically with the addressee, unlike in some southern varieties. It can also relate to the speaker, while in southern Piedmontese and western Ligurian varieties it has been generalized to refer to all three deictic persons, as has also the [stu] paradigm in Ventimigliese (Azaretti 1977:171f.): [sta/sa ˈdona ki/li/lɑ] ‘this/that woman’. In some dialects, e.g. Val Bormida, the reflex of IPSE represents the only real demonstrative, with binary or ternary spatiotemporal distinctions expressed by locative adverbs that are obligatory in pronominal usage, but optional with demonstrative adjectives (like Fr. ce): Cai. [ɛs ki] ‘this one’, [ɛs li] ‘that one’, [ɛs lɑ] ‘that one over there’, [ɛs kaŋ ki/li/lɑ] ‘this/that dog’. Lexicalized reflexes of ISTE survive for temporal reference, e.g. [ʃtamaˈtiŋ] ‘this morning’, while [kul] (< ECCU ILLUM) is used for cataphoric and anaphoric reference. Valbormidese demonstratives show a special evidential use in narratives: whilst anaphoric [ɛs] refers to things/events which are presented in their immediacy, [kul] marks them as distant, either temporally or psychologically (Parry 1994). Azaretti (1977) describes the Ventimigliese [stu] < ISTE / [su] < IPSE ~ [ˈkelu] < ECCU ILLUM opposition in terms of ± physical or temporal presence of the referent.

13.3.7 Possessives

Table 13.4 Prenominal possessives in Turinese

The dialects of northern Italy

Possessives normally agree in number and gender with the head noun, but often show reduced forms (with respect to their predicative forms) in attributive functions, especially in the case of the singular and the third person plural, which is normally identical with the singular (cf. §45.2.3). These reduced forms may not agree for number or for gender, as for instance in Bgm. 1SG [me], 2SG [tɔ], and 3SG/PL [sɔ]; Gen. [mɛ], [tɔ], [sɔ]; and Rmg. [mi], [tu], [su]. The permutations, however, are many: whereas Turinese masculine possessives do not vary for number (like most nouns), Cairese possessives vary just for number in the masculine, and Milanese possessives nowadays just for gender in the singular (Nicoli 1983:169) (Table 13.4).

In predicative constructions we may find fuller agreeing forms: Cai. [a ɹɛ ˈmia/ˈtua/ˈsua] ‘she’s mine/yours/his(/hers/theirs)’.

(p.198) A noteworthy difference between northern dialects and the other Italo-Romance varieties regards the order of noun and possessive adjective when used attributively (Renzi 1997:164f.): the possessive normally precedes the noun in northern dialects, as in Italian, whereas the postnominal position is characteristic of most centre-southern dialects (cf. §§14.4.1, 16.4.1) and Sardinian. Views differ on whether this difference may be attributed to noun movement or to different base-generation positions of the possessive (Manzini and Savoia 2005, III:563–9). Within the NP the possessive is generally preceded by a determiner, by default the definite article, but, as in the standard, this is usually omitted with singular names of relatives in most northern dialects, including Friulian—although this does not seem to be the case in eastern and Alpine Lombard, or Romagnol. The omission of the definite article can be restricted to a subgroup of kinship names, while in some varieties, e.g. in the Veneto, the definite article may be absent in the plural also, and some central Piedmontese dialects, such as Turinese, omit the definite article with all common nouns, except for the masculine plural (see Manzini and Savoia 2005, III for rich exemplification).

13.3.8 Agreement in the nominal group

In the case of the invariable nouns described above, their gender and number usually emerge from agreeing determiners (but see Manzini and Savoia 2005, III for microvariation), exemplifying the move towards head- rather than dependent-marking (Ledgeway 2012a:290). We have seen that reduced adnominal possessives may not show agreement, but in some dialects adjectives show agreement morphemes even though nouns do not, as with Tur. [ˈtyti] ‘all’ above, and Masi Torello (Ferrara): [ki ˈbeli/ˈbruti dɔn ki] ‘those.PL beautiful.FPL/ugly.FPL women here’ (Manzini and Savoia 2005, III:600). Aly-Belfadel’s (1933) description of Turinese grammar distinguishes between prenominal and postnominal position for some adjectives: [kaˈtivi vziŋ] ‘bad.MPL neighbours’ ~ [vziŋ kaˈtiw] ‘neighbours bad.MSG’; [ˈbravi fiœj] ‘good.MPL children’ ~ [fiœj braw] ‘children good.MSG’, but such variation seems now obsolete. Conversely, as with possessives, some dialects show no agreement marking on prenominal adjectives but it appears postnominally and in predicative position: Oviglio [kul vɛʤ ˈdɔni] ‘those.FPL old.M women.FPL’ ~ [kul ˈdɔni ˈvɛʤi] ‘those.FPL women.FPL old.FPL’ ~ [i suŋ ˈvɛʤi] ‘they.F are old.FPL’ ~ [i suŋ vɛʤ] ‘they.M are old.M’ (Manzini and Savoia 2005, III:595).

Northern dialects often show gender agreement on cardinal numbers up to three, e.g. Mil., Ven. [duː] ~ [dɔ] ‘two.M/F’, [triː ~ trɛ] ‘three M/F’; Cai. [duj ~ ˈduɛ], [træj ~ trɛ], and quantifiers in some varieties show agreement even in partitive constructions, e.g. Ven. [un ˈpoka de ˈpasta] ‘a. MSG little.FSG of pasta.FSG (= a little pasta)’, [un ˈpoki de ˈbizi] ‘a.MSG little.MPL of peas.MPL (= a few peas)’ (Cinque 1997:188). Ligurian forms for ‘too much’ and ‘little/few’ may not agree (Toso 1997:115): Gen. [g ɛ ˈtrɔpu ˈʤɛŋte] ‘there is too much.MSG people.FSG (= there are too many people)’, [g ɛ ˈpɔku ˈgɔti] ‘there is little.MSG tumblers.MPL (= there are few tumblers)’; this is presumably a development from the partitive construction described by Battye (1989:109–11).

13.4 Sentence morphosyntax

13.4.1 Subject clitic pronouns

The most evident phenomenon characterizing northern dialects is the presence of subject clitics. This feature of (p.199) northern dialects (including Florentine) and the French area has been the topic of intensive investigations and analyses since Kayne (1975). In general, as overt indexes of functional positions, subject clitics offer direct or indirect evidence for a detailed analysis of many grammatical constructions, as shown in Ch. 47. The systems that have been observed and described for northern Italy display much fine-grained variation; empirical research has shown that the different cases can be ordered in a series of implicational generalizations (see Poletto 2000 for a detailed overview which resumes and analyses decades of research).

In a broad sense northern dialects are varieties of non-pro-drop languages. The dichotomy between northern Romance and dialects of southern Italy in this respect has been recognizable since the earliest attestations of Romance; in medieval texts, northern Romance varieties show what has been called ‘asymmetric pro-drop’, a consequence of the asymmetric V2 of early Romance (see Benincà 2006 and bibliography therein): in northern dialects, together with old Florentine, the verb requires an overt subject, unless its syntactic position allows it to structurally control the subject position. In main clauses the inflected verb moves to the vacant Comp(lementizer) position, from where it governs the subject position, licensing a null subject (below indicated as Ø); in dependent clauses, verb movement to the Comp position is partially or completely inhibited (depending on the kind of subordinate clause), because in a dependent clause the Comp position is assumed not to be vacant; when the verb does not move to this position, the expression of the subject is obligatory. The following examples are from Benincà (2006):

(11)

The dialects of northern Italy

(12)

The dialects of northern Italy

The strongest inhibitors of movement are dependent interrogative structures; the few apparent counterexamples can be accounted for by recognizing that dependent interrogatives can borrow the structure of a headless relative clause (see discussion in Benincà 2006; 2010).

All Romance languages in the medieval stage (until the mid-fourteenth century) are V2 languages (cf. §§31.3.3, 62.5); only those belonging to northern Romance have an asymmetric pro-drop, which must be related to properties of verbal inflection.

At this time, subject pronouns were not clitic, but weak, pronouns, in the sense of Cardinaletti and Starke (1999). Around the fifteenth century they became clitics and formed a series, matched by a series of new tonic pronouns (see Vanelli 1987). In almost all varieties of northern Romance (including French and Florentine) the tonic pronouns are not the continuation of the subject pronouns of Latin and early Romance, but arose in the fourteenth and fifteenth centuries from stressed non-nominative pronouns (cf. §13.3.4). The issue is clear for first and second person singular: in French, as in Milanese or Florentine, the tonic pronouns are not from Latin EGO, TU, but are new formations, based on accusative or non-nominative ME, TE. Texts of the period immediately before the grammatical change show that in fact the first occurrences of non-nominative pronouns used with a function related to the subject are precisely cases in which the pronoun does not receive nominative case: even though it is related to the subject, it is in a position in which it is forced to receive a default case. This happens with subjects of non-finite verbs, subjects in postverbal position, coordination of subjects, complements of come ‘as, like’ (see, for Venetian, Stussi 1965:213, 215; Benincà 1994:170-72; for French, Foulet 1919:§§205–7). Leaving Friulian to one side, as its system is different with respect to other relevant aspects also, this evolution of strong pronouns distinguishes the varieties of northern Romance, and is not found in proper null subject Romance languages. Rohlfs (1968:§434; cf. AIS VIII, 1627) interestingly notes that the only contexts in which we find a non-nominative pronoun connected with a subject in southern dialects are precisely the same as those which in early Romance permitted a non-nominative pronoun (< ME, TE).

These phenomena regarding verb agreement include Florentine, yet not Italian. On the other hand, these phenomena, which involve many syntactic constructions, link all northern dialects specifically with French and French dialects.

13.4.2 Wh-movement constructions

The more salient aspects of the reflexes of clitic pronoun syntax are described in Ch. 47. We highlight some points here, in particular concerning those varieties, such as (p.200) Venetan dialects, where a wh-pronoun is incompatible with a doubling clitic. The following examples are from Paduan:

(13)

The dialects of northern Italy

(14)

The dialects of northern Italy

In Venetan varieties, this property allows us to distinguish restrictive relative clauses from other kinds of relative clauses, such as appositive (non-restrictive) and kind-defining relatives (see Benincà and Cinque 2014). From Paduan:

(15)

The dialects of northern Italy

This pattern is by no means shared by all varieties; in particular, subject clitics very often double a ‘wh’ subject, in relative as in interrogative clauses (see Benincà 1986; Poletto 2000; Parry 2007b). Subject clitics invert with the verb in main interrogatives, but this property, which was widespread in northern dialects, has slowly receded since the seventeenth century. The loss of verb movement to the vacant Comp(lementizer) position is revealed by the order subject clitic–verb, and in some dialects (except most Ligurian and Emilian dialects) is accompanied by the insertion of a complementizer [ke] ‘that’ to follow the wh-word (see Parry 2003, in particular for Piedmontese). Furthermore, some Lombard and Venetan dialects apparently do not move the wh-pronoun to the clause left periphery, but leave it in a postverbal position (see Munaro 1999; Munaro et al. 2001).

An interesting generalization emerges from the observation of the strategies of interrogative clause formation in northern dialects: no dialect shows the insertion of the complementizer [ke] ‘that’ together with subject clitic inversion, and no dialect shows subject clitic inversion in dependent interrogatives. The few exceptions can be accounted for by recognizing that interrogatives can adopt the structure of headless relative clauses (see Benincà 2012).12

13.4.3 Negation

Northern dialects display all the different stages of Jespersen’s Cycle (see §51.2.1) and the distribution of negators in the sentence can throw light on the diachronic process, suggesting specific hypotheses on the internal composition of negation words and the functional projections involved in negation. Like other phenomena above, this characteristic concerns Piedmont, Lombardy, and Emilia Romagna, but not the Veneto and Liguria.

The analysis of microvariation of western northern dialects (and French) strongly suggests that each feature has a location in a functional position in the structure; northern Italian dialect negative particles perform the functions subsumed by the single negator in Latin, Italian, Venetan, Ligurian, Tuscan, and central and southern dialects. Zanuttini (1997) distinguishes two kinds of preverbal negator (a clitic and an independent negative head; see also Cinque 1999:21) and three positions for postverbal negators. The origins of the negative polarity items that become postverbal negators are mainly either lexemes indicating a minimal quantity, e.g. mi(n)ga ‘crumb’, or grammaticalized expressions containing a negation.13 Milanese, for example, (p.201) has two postverbal elements: a minimal quantity, mica/minga ‘crumb’, and a negative , which adds a presupposition.14 They have different positions in the sentence: minga precedes, and no follows, the past participle:

(16)

The dialects of northern Italy

Turinese has two postverbal negations: nen, the simple negation, and pa, the presuppositional negation (Zanuttini 1997), as in the following examples:15

(17)

The dialects of northern Italy

Both negators seem to occupy the same position with respect to the past participle; actually, their position is, respectively, before (pa) or after (nen) the adverb già ‘already’:

(18)

The dialects of northern Italy

(19)

The dialects of northern Italy

Most varieties have only one lexeme as negator, which appears in a different position when it assumes a specific presuppositional value.16

13.4.3.1 Negators

The predominant Italo-Romance negation strategy, namely preverbal negation alone (Neg 1), continuing Latin morphology and syntax, is found in the northeast and northwest: Venetan, Friulian, and some Ladin varieties, as well as Ligurian. In the central Po area, Piedmontese, and Lombard have mainly Neg3, in which a postverbal element, originally used to ‘reinforce’ the preverbal negative, has taken over the preverbal negator’s functions; it can be a noun with general reference merged with a negative, Pie. nen(t) ‘nothing’ < NE GENTEM ‘no people’ (Rohlfs 1968:218; Iliescu 2011); or non-negative minimizers such as Lmb. mi(n)ga (< MICA(M) ‘crumb’), Eml. brisa (< *ˈbrisa ‘crumb’), which became negative through frequent collocation with the preverbal negative in an intermediate discontinuous strategy, n…nen/minga/brisa (Neg 2), or less frequently, a resumptive holophrastic negator added clause-finally, Lmb. . The transitional Neg2 type is still found in Emilian as well as in some Alpine Lombard (Ticinese) and border Piedmontese–Ligurian varieties, while Venetian seems to be currently developing a form of Neg2 (Garzonio and Poletto 2009). AIS map 653 shows the three types:

(20)

The dialects of northern Italy

Table 13.5 Piedmontese negation

The dialects of northern Italy

These structural types do not correlate with homogeneous geographical areas: particularly in transitional areas, microvariation involves two or even all three strategies, as in the Val Bormida (Parry 1997b), where Neg2 prevails, but Neg1 is found in irrealis clauses and in structures such as n…âtr ‘only’, with Neg3 favouring two (p.202) main contexts: (a) with the verbs BE and HAVE (frequent as perfective auxiliaries) and (b) in the presence of preverbal complement clitics, particularly nasal ones.17 This contemporaneity of variants may be captured by adopting van der Auwera’s (2010) five-stage evolutionary model to illustrate the evolution of Piedmontese negation (variation is governed by text type, register, and linguistic context) (Table 13.5).

All early Italo-Romance varieties have strategies that reinforce the preverbal negative marker in particular pragmatic contexts, but the northern Gallo-Italian group present the same continuation of Jespersen’s Cycle as found in Gallo-Romance. All these varieties were prone not only to syllabic and vocalic weakening (NON > no, ne, n, also found in some southern varieties, e.g. Abruzzese and Campanian), which often resulted in near or total homophony with two common preverbal clitics (the first person plural complement clitic derived from NOS ‘we; us’ and the partitive from INDE ‘thence’), but also to the development of subject clitics and interrogative verb-subject clitic inversion. Thus, in addition to pragmatic influences that obtained elsewhere too, these factors arguably contributed to the grammaticalization of new expressions emerging from the cyclical renewal of negative structures by placing structural constraints on the preverbal space. Diachronic Piedmontese data, together with the current microvariation of Val Bormida dialects, show that if grammaticalized discontinuous structures exist, the preverbal negative is most likely to be dropped when other proclitics occur, thus avoiding the more costly operation of substitution:

(21)

The dialects of northern Italy

Furthermore, the tendency to drop the preverbal negative in interrogative inversion structures (see Martineau and Vinet 2005 for French) is confirmed by Piedmontese and contemporary Venetan data (Parry 2013b).18 A feature of the development of negation in northern Italian dialects which has stimulated much theoretical discussion is the way in which the preverbal negative marker came to interact with other preverbal clitic elements. Up to and including the sixteenth century the former always follows subject clitics and precedes complement clitics in all Italo-Romance texts, as in modern French. However, from the seventeenth century some northern texts show variations: in Paduan, the preverbal negative now precedes all subject clitics, except for the vocalic clitic a that is not marked for person or gender agreement, while in early seventeenth-century Milanese the preverbal negative follows the second person singular, but precedes the third person singular subject clitics occurring with auxiliary verbs, and by the end of the century negation precedes the former also (Vai 1995:161–3). By the twentieth century the Ligurian second person singular subject clitic is similarly affected, although variability is still found in Genoese (Toso 1997:229).

Examples of even more unusual preverbal ordering of clitics and negation are discussed in Parry (1997b): in the Val Bormida, on the Piedmontese–Ligurian border, the preverbal negative follows not only all subject clitics but also all first and second person complement clitics, singular and plural, as well as all reflexive clitics. Occasionally two preverbal negative markers may be heard, confirming the availability of two preverbal structural positions:

(22)

The dialects of northern Italy

Zanuttini (1997) argues from such data that there are two types of preverbal negative markers: a ‘strong’ one that negates the clause on its own, represents an independent syntactic head, located higher than the agreement projection, that may interact with subject clitics; the other a ‘weak’ one, which needs an adverbial reinforcer, raises with the verb from a structurally low position, and may therefore interact with complement clitics. Although this is generally the case, there exist dialects with Neg2 that show reordering involving subject clitics (NEG-SCL), e.g. in Emilia (Zörner 1994:89), and dialects with Neg1 where the negative follows some complement clitics (see Parry 1997b; 2013b, and for more examples, together with an alternative theoretical interpretation, Manzini and Savoia 2005, III:286f.).

(p.203) Another distinctive northern feature pertains to second person singular negative imperatives, which in Piedmontese and Lombard are formed simply by negating the positive imperative (as in French), unlike the dominant Italo-Romance strategy of negating the infinitive:19

(23)

The dialects of northern Italy

Zanuttini (1997:105–54) attributes this variation to structural differences involving two different types of negative marker: (a) a preverbal type (non, nu, etc.) that can negate the clause on its own but is incompatible with ‘true’ imperatives of lexical verbs, as it requires some form of syntactic expression of mood (morphologically or via an auxiliary, which may be phonetically null); and (b) a postverbal negative adverb (nen, minga, etc.) that does not interfere in the movement of the true imperative to the vacant Complementizer position. Given that early northern dialects show cases of the non type with true imperatives and some survive in Romagnol ([nu ˈkaska] NEG fall.IMP ‘don’t fall’, AIS map 1621 pt. 499), as well as in Friulian and Ladin (Vai 1998), Parry (2010b) offers an alternative interpretation, arguing that the current syntactic rule could be the result of pragmatic responses to homophony (in Piedmontese) and analogical influences (in Lombard) rather than syntactic incompatibility.

Negative concord, ‘the co-occurrence of more than one negative element in the same clause with the interpretation of a single instance of negation’ (Zanuttini 1997:9), is a well-known feature of Italian and Italo-Romance varieties, which tend to exhibit the asymmetric variety, whereby a postverbal negative indefinite must always occur in the scope of a sentential negative or other negative element, but a preverbal one cannot occur with the sentential negator, e.g. It. non viene nessuno ‘NEG comes nobody (= nobody comes)’ vs nessuno (**non) viene. Many medieval varieties allowed the latter combination (symmetric negative concord): it appears consistently in early Piedmontese and still obtains in modern Venetian (Marcato and Ursini 1998:188), which retains the preverbal negator: nissun no vien ‘nobody comes’. Although Gallo-Italian varieties have lost the preverbal sentence negator and their n-words can convey negation independently, the cycle of negative strengthening to convey more robust denial continues with the emergence of postverbal negative concord, confirming the tendency to mark negation at both sentence and constituent level (i.e. head- and dependent-marking). In this case all elements appear postverbally, but in many dialects there are structural constraints on cooccurrence that produce the following typical hierarchy: modal/aspectual periphrases > compound tenses > prepositional phrases, with the contexts on the left of the hierarchy most favourable to negative concord between the postverbal negator and the negative indefinite (Parry 1996:247):

(24)

The dialects of northern Italy

(25)

The dialects of northern Italy

13.4.4 Auxiliaries

As noted above, the preterite has disappeared in practically all northern dialects and has been replaced by analytic forms. In northern dialects BE is used as an auxiliary in combination with unaccusative verbs, and HAVE with transitives and unergative verbs; unlike other Italo-Romance varieties, indefinite se ‘self=’ does not automatically trigger the auxiliary BE, but in some varieties maintains the auxiliary required by the verb when used in non-indefinite sentences.20 In Venetan dialects we have:

(26)

The dialects of northern Italy

(27)

The dialects of northern Italy

(p.204) In Piedmontese the auxiliary is BE, not only with indefinite se, but with all types of se constructions (even though ‘have’ is found in a few varieties):

(28)

The dialects of northern Italy

(29)

The dialects of northern Italy

Parry (1998) provides a detailed description and analysis of all Piedmontese constructions involving se. Note the enclitic position of the clitic to the past participle, a specific characteristic of Piedmontese to which we return below.

Milanese as well uses the auxiliary BE with indefinite se:

(30)

The dialects of northern Italy

With reflexive se (both with argument and inherent reflexive verbs) the majority of dialects can use indifferently HAVE or BE. Some dialects, in particular of the central Veneto, display a sort of mixed paradigm: Paduan, for example, uses the auxiliary BE with first and second singular and second plural reflexives, and auxiliary HAVE with third singular and plural; second plural can more freely choose one or the other auxiliary:21

(31)

The dialects of northern Italy

In western central Veneto, in the dialect of Fossalta di Piave, only first person singular has the auxiliary BE, the other persons select auxiliary HAVE. The choice of the other auxiliary, in any case, does not produce strong ungrammaticality but just an effect of unnaturalness.

In Milanese (from Nicoli 1983, and Massimo Vai, p.c.) we find:

(32)

The dialects of northern Italy

It is possible that the preference for one or other auxiliary, at least in some varieties, also depends on the semantics of the verb (see Manzini and Savoia 2005, III:§§5.5-6; Cennamo and Sorace 2007; cf. also §49.3.4).

In the Lombard dialect of Mendrisio in southern Switzerland (Lurà 1987), the two auxiliaries are in some cases indistinguishable, due to the fact that s- is ambiguously interpretable as the clitic se or as the root of the auxiliary BE; but first person singular has clearly the auxiliary BE and third person singular and plural the auxiliary HAVE, replaceable by the auxiliary BE (Lurà 1987:169):

The ambiguity of some forms of HAVE and BE paradigms (due not to accidental phonological evolutions but to grammaticalization processes) are found in many dialectal areas, all over Italy; see data and discussion in Manzini and Savoia (2005, III:§5.9), Bertocci and Damonte (2007).

13.4.4.1 Double compound forms

In eastern varieties compound verbal forms are found with two auxiliaries; the auxiliaries can be three in the varieties that accept these forms in the passive (cf. also §58.3.4). The most common case displays a (generally grammaticalized) (p.205) past participle of the verb HAVE inserted between the auxiliary and past participle of the lexical verb; the first auxiliary can be inflected in any tense and mood (apart from the preterite, which is incompatible with the aspectual value of the form, as attested in the very few villages that have double compound past tenses and that still had the preterite until a few decades ago). The form can appear in a main clause, in which case the interpretation is that of a ‘semelfactive’ aspect: the event is exceptional, or happened just once (33); it can appear in a dependent temporal clause, introduced by ‘when, after that’, and in this case the aspect is ‘completive’ perfect (34). We give examples from the Venetan dialect of Cereda (Teresa Vigolo, p.c.):

(33)

The dialects of northern Italy

(34)

The dialects of northern Italy

Double auxiliaries with similar characteristics have been described for Friulian and Veneto (Benincà 1989:577f.; Marcato 1986), where the construction is stylistically very marked, but widely used, and for old and modern French (Foulet 1925; Schlieben-Lange 1971), where it is documented much earlier than in northern Italian dialects (the first record for northern Italy is in the Istrian variety represented in Salviati’s 1584–6 version of Boccaccio’s novella, in Papanti 1875:24).22

13.4.6 Clitic areas

The variability of the positions of clitics in sentence structure in different dialectal areas and in diachrony inspires hypotheses concerning the functional architecture of the clause. The syntax of complement clitics in medieval Romance, as well as the syntax of subject clitics in modern varieties, suggests a functional area in the left periphery available for clitics.

All proclitics to the inflected verb are currently assumed to be in an area within the inflectional domain (§31.2.2); by contrast, northern Italian dialects provide evidence for a low clitic placement site, immediately outside the verb phrase (see Benincà and Tortora 2009; Tortora 2010 for detailed references).23

When a complement clitic is associated with a non-finite verb, we find clitic pronouns enclitic on the past participle, but only in Piedmontese (except Canavesano dialects) do they fail to climb to the finite auxiliary, and in some cases they appear twice (proclitic on the auxiliary and enclitic on the past participle). The following examples are from Mondovì and Cairo Montenotte (ASIt data; see also Parry 1995; 2005); (35a,b) show enclisis to the past participle, whereas in (36) the clitic is both in proclisis to the auxiliary and in enclisis to the past participle (the latter scheme is attested in Biondelli’s 1853 and Papanti’s 1875 texts):

(35)

The dialects of northern Italy

(36)

The dialects of northern Italy

With respect to the idea that Romance languages display a diachronic drift progressively reducing leftward (upward) syntactic movement, Piedmontese varieties are particularly interesting, with evidence coming from different areas of syntactic structure (Tortora 2014b).

Notes:

(1) Although this chapter is the result of close collaboration between the authors, main responsibility for the work is divided as follows: Benincà §§13.1, 13.4, 13.3.2.2; Parry §§13.2, 13.3, 13.4.3.1; and Pescarini §§13.2.1, 13.3.2.3, 13.3.2.4, 13.3.4.

(2) Even though Friulian is also dealt with separately (Ch. 10), we will refer to it when necessary, because it exhibits most northern Italian phenomena, and in a form that often provides us with a key to understanding those traces found in other dialects of northern Italy.

(3) Dante, DVE I, xv–xix; Fernow (1808:259f. and passim); Biondelli (1853), Ascoli (1873:esp. 450-53), to quote just the key references (see Benincà 1996b).

(4) In a certain tradition of philological studies, only Piedmontese, Lombard, and Emilian-Romagnol are considered areas of Celtic population. In a more flexible perspective, such as that adopted by Pellegrini in his studies on toponymy (cf. Pellegrini 1981), all northern Italy shows a greater or lesser influence of Celtic languages. Even those areas, such as southern Veneto, that apparently differ and have a Venetic substratum were celticized, in particular the rural population, just before (or during) the process of Romanization.

(5) The point is crucial for his theory of vulgare latium, i.e. Italian, the vernacular of Italy: since vulgare semilatium, the vernacular of the northern half of Italy, exists (he says) together with many local varieties, we can assert that vulgare latium, Italian, also exists, and, unlike the many inadequate local varieties, can provide a common vernacular substitute for Latin (cf. Ewert 1940).

(6) Trumper (1977) uses Emilia-Romagna to illustrate a typical micro-diglossic region, where dialects are disappearing more rapidly than e.g. in the Veneto. In an interesting analysis, Matarrese (1990) shows that in the 15th c. the language used in Romagna, in particular in the court and chancery of Ferrara, was not aligned with Florentine, nor showed local features, but instead was influenced by the Venetian high level language. This means that a High local variety had not arisen.

(7) At an even higher level, between northern and central-southern Italian speakers, more radical operations are required. At this level, as pointed out by Vincent (2007c; 2012), a sort of unconscious syntactic comparison is at work, an operation that requires a pre-existing syntactic similarity—or comparability—between the languages involved. Vincent underlines the relevance of this aspect for the formation of an Italian national koiné, and draws attention to the ‘Cremona papers’ (published by Baglioni 2010); there, well before national unification, a sort of unitary Italian is attested, in northern Africa and in the Mediterranean area, including in documents of an official nature; it had indisputable non-local characteristics, not at all based on literary Italian—a koiné produced spontaneously through contact.

(8) Vincent (2006b) points out the similarity with the prototypical Greek koiné, analysed by Morpurgo Davies (1987). see also Ferguson (2003; 2007) on Venetian as a sort of koiné. On colonial Venetian, developed on the Adriatic coasts, see Folena (1968–70).

(9) Some apparent counterexamples, such as past participles of strong verbs, are in fact clear cases of analogical extension.

(10) Support for this theory comes from the fact that words with original non-high final vowels often show no diphthongs, e.g. Pie. [ˈskora] < SCHŎLAM, ‘school’, [ˈnora] < *ˈnɔra ‘daughter-in-law’, [om] < HŎMO ‘man’ as opposed to [kør] ‘hearts’ < *ˈkɔri, [bø] ‘oxen’ < *ˈbɔvi (Rohlfs 1966:140f.), where the earlier diphthong *[ɥo] has remonophthongized; analogical levelling would account for the spread of such diphthongs to the singular. An alternative theory proposed by Pellegrini (1982:17) and Zamboni (1993:467) is that the greater number of syllables in proparoxytones and words retaining final vowels led to a compensatory shortening of stressed vowels that hindered diphthongization. see also Sánchez Miret (1998) and Ch. 38.

(11) Regarding the origin of the -[i] inflection, whereas a phonetic development via palatalization of [s] and vowel raising plausibly accounts for reflexes of Latin third declension undifferentiated nominative/oblique nouns and adjectives ending in -ES (> [ej] > [i]), the possibility of [i] from second declension oblique -OS has been rejected in the main (see Maiden 1996a).

(12) This hypothesis may explain the interesting fact that in some Ligurian, Piedmontese, and Friulian dialects interrogative pronouns derive from demonstrative + complementizer. In Ligurian varieties we have kelu ke ‘who?, which?’ (lit. ‘that (pronominal) that (complementizer)’: Cuneo 1997), in Friulian là che ‘where (lit. ‘there that’). These forms derive from headless relative pronouns, which are characterized by the phonological realization of the complementizer (see Munaro 2001; Benincà 2012); some dialects and languages (e.g. Italian) use a demonstrative pronoun with non-specific interpretation followed by the complementizer to introduce a headless (or ‘light-headed’) relative clause (see Citko 2004). The complex of pronoun and complementizer can grammaticalize, producing an interrogative pronoun. This phenomenon further supports the close relationship between headless relatives and interrogatives.

(13) Cinque (1976) has shown that the negative polarity item mica involves the presupposition that the negated event was expected to happen. The presupposition is valid for Vnz. miga as well; evidence provided by Cinque is that this postverbal element cannot appear in constructions that involve a presupposition of their own, such as restrictive relative clauses or the protasis of the conditional sentence. Testing these properties in an area extending from Verona (western Veneto) to Lombardy (see Pescarini 2005 for a first account), we found that the set of incompatible constructions reduced progressively as one moved further westwards, towards the area where the negation is postverbal. This result, if confirmed, suggests that the presupposition conveyed by the postverbal negative polarity element is more complex than it seems, and should be decomposed into smaller units, which are deactivated as the dialect advances along Jespersen’s Cycle.

(14) see Vai (1996) for the diachronic development of negation in Milanese.

(15) As has been widely recognized, all negative sentences involve a presupposition; we are referring here to more explicit and stronger presupposition concerning the expectations that the positive sentence be true.

(16) On the other hand, if a dialect has more than one postverbal negator, they have different positions and values; however, one Emilian dialect has two postverbal lexical negators, both belonging to the ‘minimal quantity’ class (briza and minga ‘crumb’), and no possible interpretative or syntactic difference has been found; the reason is apparently connected to sociolinguistic factors (see Colombini 2007; Benincà and Tortora 2011).

(17) see Zeli (1968) for Ticinese varieties.

(18) Benincà and Poletto (2005) draw an interesting comparison between the diachronic development of negation and that of interrogative and personal pronouns; in all three areas we have the emergence of clitic forms and doubling structures.

(19) The infinitive is [kaŋ'tɛ].

(20) In order to get an indefinite se it is necessary to choose verbs and contexts that render less plausible an interpretation which involves the speaker, in which case a 1st person plural form of the verb is preferred. This aspect, which has also been pointed out for Milanese by Massimo Vai (p. c.), seems the opposite of that found in Friulian or Florentine, where impersonal se with the 3rd person singular verb is the normal way of rendering a 1st person plural form.

(21) A similar pattern is found in Abruzzese dialects (Giammarco 1979, and data stored in the ASIt database: <http://asit.maldura.unipd.it>).

(22) Double auxiliaries are also found in southern Italian dialects, with characteristics that differ in an interesting way from those of northern dialects (see Ledgeway 2009a:596–600, D’Alessandro and Ledgeway 2010b, and references therein).

(23) Northern dialects are also particularly rich in sentential particles, which mark semantic or pragmatic nuances, and with their ordering restrictions localize the related functional projections: see Munaro and Poletto (2005), Poletto and Zanuttini (2010), and Ch. 53.