About the project

Number morphology is a major line of continuity both between Romance languages and Latin, and across Romance languages (all these languages retain at least some morphological expression of number on the noun). Despite its fundamental place in Romance morphology, number-marking has received rather little detailed attention from a comparative-historical perspective. None of the older major comparative surveys come close to recognizing the density and richness of the data — and in many respects they could not have done, given that a good many data from lesser-known dialects have come to light only in recent years. The most comprehensive comparative overview probably remains that of Wilhelm Meyer-Lübke in his Grammatik der romanischen Sprachen, now well over a century old, but lacking in various respects where the Italo-Romance and Romanian domains are concerned.

Why should we care?

To those familiar with the four best-known Romance languages, French, Spanish, Italian and Portuguese, this lack of attention may seem unsurprising, for there does not appear to be too much to say. In Spanish it is not too much of an oversimplification to claim that plurals simply comprise the singular form + (e)s (e.g., sg gato - pl gatos 'cats'; sg casa - pl casas 'houses'; sg mes - pl meses 'months'). Much the same system (plural in -s) historically underlies many other 'western' Romance languages, such as Portuguese and French. Italian also has a simple rule whereby virtually all nouns whose singular ends in an unstressed vowel end in -i in the plural (or in -e, if they are feminine and end in -a in the singular) and this state of affairs, too, can largely be traced back to a historical final -s. In Latin, the inflectional marker of accusative plural was -s, in all non-neuter nouns; given the general loss of the neuter gender, and the tendency for case-forms other than that which originally expressed the accusative to disappear, many Romance languages emerge simply with a plural in -s.

These generalizations are, of course, in urgent need of local qualification. For example, in many varieties (e.g., Portuguese and French) -s has had a phonological effect on the root of the noun (the part that expresses lexical meaning), so that we often find ‘root allomorphy’ — one meaning expressed by more than one form, according towhether the word is singula ror plural (e.g., French singular cheval vs plural chevaux). In spoken French, loss of final -s in pronunciation has very often led to a situation of invariance between singular and plural (e.g., singular chat vs plural chats, which sounds exactly the same in the plural as in the singular, even if it is written differenyly).

Perhaps the best-known type of 'strange plural' among Romance languages appears in standard Italian, which has about a score of nouns whose plurals conserve the Latin neuter plural ending -a, are feminine and stand in opposition to masculine singular forms in -o (e.g., sg uovo - pl uova 'eggs'). This class in Italian has recently been the object of some theoretical attention (e.g., Ojeda 1995), culminating in the penetrating analysis by Acquaviva (2008, esp. ch.5), but it emerges in the latter study, precisely, as having the status of a 'special' case, comprising 'lexical plurals' lying outside the normal, productive, techniques of plural formation for Italian; in effect, these plurals are separate words in the mental lexicon from the corresponding singulars.

Our novel perspective

It may still not be clear that there is a great deal, overall, to be said about Romance plural formation beyond what has already been said. But our project inverts the usual perspective on Romance languages by placing Romanian (the fifth major standard Romance language) at the centre of enquiry. This is a language for which the relatively straightforward predictabilities of plural formation encountered in many other Romance varieties are not always available, and in which it is a by no means uncommon occurrence (in our direct personal experience) for even educated native speakers to stop in mid-sentence because they have forgotten, or simply do not know, how to form the plural of some noun. It is also a language in which the singulars of nouns are frequently not of the form expected in terms of their etymology, but quite obviously analogical reformations whose basis is the plural. It is a language in which it is even possible for the singular and the plural forms of a noun to be borrowed from different languages, with resultant erratic number morphology.

In Romanian, the plural of masculine and feminine animate nouns is (largely) predictable and rather similar in nature and origin to that of Italian: most nouns have the plural inflection -i, feminines with singulars in -a have plurals in -e (each with their attendant, and largely but not wholly predictable, patterns of root allomorphy). Case is not an important factor, since although Romanian has a rudimentary inflectional case-system, (bare) noun plurals are invariant for case; it should be noted, however, that the form that realizes plural in feminines also realizes genitive-dative singular, with some interesting exceptions (which we propose to explore) where there is an opaque semantic relationship between singular and plural. The main difficulties arise, however, in inanimate nouns. First, feminine inanimates with singular in -a may — apparently unpredictably — have plurals in -e or in -i, although in this category there are even some animates (see Table 1).

Second, there is in Romanian a vast, and productive, class of nouns with inanimate reference whose singulars are syntactically masculine but whose plurals are syntactically feminine, and characterized by the endings -e or -uri (see Table 2).

These endings can be shown to originate, respectively, in the Latin neuter plural ending -a (later replaced by feminine plural -e), and -ora (later replaced by *-ure > -uri), derived by a resegmentation of the plural of nouns such as corpus - corpora 'bodies'. This phenomenon has attracted considerable attention in the general linguistic literature on gender and agreement (e.g., Corbett 1991:288-90 and cf. Acquaviva 2008:135-40), but much less so in the specific respect of its inflectional realizations. It is fair to say that in the internationally accessible literature very little has been written on the subject (see, for example, the bare three pages dedicated to the subject by Mallinson 1986:243-46); a good deal more has been said, naturally, in Romanian by Romanian linguists, but the morphological discussion is too often concerned with the (arguably rather uninteresting) taxonomic question of whether Romanian is like Latin in having a third, 'neuter', gender, and one usually encounters mere lists of endings, occasionally accompanied by comments on semantic distinctions in cases where there exist inflectional doublets, observations on alleged trends in preference for one ending over the other, and some comments on possible phonological determinants of the selection of one ending or the other (-uri is usually found only with nouns whose singular is oxytonic, although not all oxytones select -uri). It is usually observed that -uri is a dedicated inflectional ending for nouns of this kind, and rather less often that it can also occur with nouns that have feminine singulars; -e is usually treated as essentially identical to the feminine plural inflection encountered for nouns with feminine singular nouns in -ă, but one aspect that has, to our knowledge, never been commented on, let alone explored, is the simple fact that it does not behave diachronically like the 'ordinary' feminine plural ending -e in inanimates, because it is apparently never susceptible to replacement by -i.

Romance linguistics does not possess a description and explanation of the morphological manifestations of Romanian plural formation which places all of it in a wider, comparative, context. Worse, we do not even possess an account of it which places it in a proper Romanian linguistic context beyond standard Romanian. It is true that we possess an excellent overview of the history Romanian noun plural formation in Diaconescu (1970:§3.2), but this remains essentially descriptive and taxonomic, and largely limited to literary standard Romanian. Otherwise there has emerged over the twentieth century (and especially with the appearance of the Romanian regional linguistic atlases — still partly unpublished — in the last third of the century) a vast, purely descriptive, literature on the morphology of the dialects of Romania proper, and the three major varieties (Aromanian, Meglenoromanian and Istro-Romanian) spoken beyond the Danube. This material pullulates with information on noun plural formation but no attempt has ever been made at a descriptive and interpetative synthesis of the morphology of plurals (beyond some isolated studies of particular lexemes or limited geographical areas and studies of occasional 'oddities' — such as the occurrence of what appear to be special forms of plurals used after numerals). All this means that we do not know, overall, the incidence and nature of variation between feminine plural -e and -i in inanimates; the extent and determinants of penetration of -i into feminine animates; the extent and determiants of -e and -uri as plural endings; the extent and significance of inflectional doublets in plural formation; the extent of analogical reformation of singulars on the basis of plurals. Moreover, we do not have an integrated comparative account of certain innovatory sources of plural inflection.

Many of the forms and types of variation observable for Romanian and Romanian dialects actually have (and had) a much wider geographical distribution; from a historical and comparative perspective the Romanian facts cannot be discussed in isolation from Italo-Romance (generally called ‘Italian’) dialects. All Italo-Romance dialects seem originally to have had forms cognate with Romanian -e and -uri (usually -a and -ora). The two endings co-occur in most central and southern Italian dialects, sometimes as inflectional doublets for the same lexemes, and almost always without explanation of what (if any) the difference between the variant forms might be: thus Veroli in Lazio aˈnɛllu - pl aˈnɛlla or aˈnɛlləra ‘rings, curls’ (or regular masculine aˈneʎi), ˈprato - pl ˈprata ‘meadows’, ˈorto - pl ˈɔrtəra ‘orchards’. In some cases we know that the -ora ending has extended to human referents. But despite the existence of numerous descriptions we possess, again, no detailed comparative description of such forms, let alone a principled account of their distribution. The same is true of variation between -e and -i in (inanimate) feminine plurals. With regard to the -a and -ora type plurals, the Italo-Romance and Romanian facts are different in two major respects. First, we have a much greater observable historical depth for the Italo-Romance varieties; second, the -a and -ora types are recessive to the point of extinction in northern and central Italy, whilst they are robustly productive throughout Romanian. The situation observable in modern standard Italian, in which a handful of remnants of the -a type (and none of the -ora type) persist, has been the focus of considerable scholarly attention, culminating in Acquaviva's compelling analysis of these as having formal, syntactic and semantic characteristics which qualify them as distinct lexical items which are inherently plural, having special semantic characteristics, rather than 'irregular', inflectional, plurals with corresponding inflectional singulars. But Italian presents the peripheral disjecta membra of a once far more extensive system. Some important preliminary descriptive work has been done on the history of these plurals in Italo-Romance generally, but no work, to our knowledge, seeks to collocate the Italian facts in the wider Romance comparative perspective, to explore the wider implications of the kinds of microvariation observable in many dialects, or to see, for example, whether the sharp distinction between 'lexical plurals' in Acquaviva's sense, and ordinary plurals, can be discerned in modern or historical central Italian dialects in which the -a and -ora types are much more widespread, yet have been retreating.

In short, even at the descriptive level we need a comprehensive overview of the inflectional morphology of number in the Romance languages, and we possess both the means and the expertise to produce one. But we also need a description which seeks to identify as far as possible the principles (possibly phonological or semantic) underlying the distribution of the word-forms which express plural.

What we expect to find

The morphology of the singular is also important. To anticipate what we think will be one of our principal findings, we believe that we shall be able to show that in 'eastern' Romance the form of noun plurals is often truly unpredictable, and that there is massive lexical storage of plural forms alongside their inflectionally corresponding singulars, not only in 'classic' cases of semantically 'unmarked' plurals (see Tiersma 1982), but in general. This property seems to be so pervasive in Romanian that there is repeated evidence for speakers using plurals as the basis for the morphological re-structuring of singulars. Byck and Graur’s call in the 1930s (1967[=1933]:92) for a wider systematic survey of singular as well as plural morphology in Romanian regional varieties has never, to our knowledge, been systematically answered, and the vast growth of available data since the Second World War means that the time is ripe for such an undertaking. Many Italo-Romance varieties have the potential for parallel developments, but creation of the ‘wrong’ singular on the basis of inflectional ambiguity in the plural is — apparently — rare; the Italo-Romance domain also deserves careful exploration in this regard.