Romance plurals

Plurals in Romance languages appear to be formed in two quite different ways:

1. By adding -s (in all languages north and west of Italy, including Portuguese, Spanish, Catalan, Occitan, French, Romansh, etc.); also in Sardinian.
2. By changing the final vowel (in all languages of Italy (except Furlan and Sardinian), as well as Romanian and related languages).

It is generally agreed that the plurals in -s derive from the Latin accusative endings -ŌS -ĀS -ĒS. However, there is some debate over the origin of the plurals of Italy and Romania, with some claiming that they derive from the Latin nominative endings -Ī -AE and others that they partly derive from the Latin accusative endings. The "nominative" theory appears more straightforward at first; however, the "accusative" theory is more common currently.

The Italian endings are -i (for nouns in -o and -e), and -e (for nouns in -a). The nominative theory suggests that the -o plural -i and the -a plural -e are derived straightforwardly from nominative -Ī and -AE, respectively (it is known that AE > e in all Romance languages), and that the -e plural -i is derived by analogy with the -o plural. (The corresponding nominative form in Latin is -ĒS. With the loss of final /s/, singular and plural would both have -e, which is problematic and was rectified by borrowing -i.)

The accusative theory starts by noting that Proto-Romance actually had both nominative and accusative endings. This is clear from languages like Old French, which have preserved both cases. However, all languages with both nominative and accusative case (Old French, Old Occitan, Old Sursilvan) agree in having forms derived from -ĀS in both nominative and accusative plurals of feminine nouns in -a, suggesting that Proto-Romance also had nominative -ĀS. Nominative -ĀS is also attested in Old Latin, suggesting that Proto-Romance may have preserved an older state of affairs that changed in urban Latin of Rome (but not in the spoken Latin of the provinces, which underlies the Romance languages). If true, there is no -AE from which Italian -e can be derived from. There is other evidence suggesting that Italian -e must derive from -ĀS:

1. In Italian, masculine amico has plural amici with /tʃ/ (the expected palatal outcome before -Ī), but feminine amica has plural amiche, with /k/ that is unexpected if e < -AE, but expected if e < -ĀS. (The change AE > e occurred long before palatalization, hence /tʃ/ is expected here too. It is unlikely that this unusual distribution is due to analogy; if so, either /tʃ/ or /k/ would be expected in both plural forms.)
2. Neopolitan and certain other minority Italian languages have unexpected alternations like gatto "(male) cat", i atti "the (male) cats" vs. gatta "(female) cat", e ggatte "the (female) cats". In these languages, loss of /ɡ/ is expected between vowels; the form with /ɡɡ/, as in e ggatte, would normally only occur if there was a lost consonant formerly preceding the /ɡ/. This suggests that Neapolitan e (standard Italian le) comes from Latin (ILL)ĀS, not *(ILL)AE.
3. The isolated Italian word dunque "thus" corresponds to Sardinian duncas. Neither word can be derived from Latin DUMQUAM, and the isolated nature of the word means that analogical change is unlikely. Sardinian duncas suggests Proto-Romance *DUNQUAS, with dunque the expected outcome (even down to the unusual qu preceding e) if -AS > e.

The "accusative" theory essentially suggests:

1. Italian plurals are in fact derived from the nominative plural.
2. However, Proto-Romance had nominative plural -ĀS, not *-AE.
3. The following sound changes took place:
   1. /as/ > /ai/, /es/ > /ei/.
   2. In unstressed syllables, /ai/ > /e/, /ei/ > /i/.

The first of these changes is almost certain, given examples like tu stai "you stand" < TŪ STĀS; Southern Italian crai "tomorrow" < CRĀS; tu sei "you are" < TŪ S(ED)ĒS; sei "six" < SEX (probably Proto-Italian *sess); Southern Italian trei "three" < TRĒS. Note also noi we < NŌS. The second sound change is cross-linguistically extremely common. Furthermore, it explains a number of otherwise unexplainable forms in Italian:

• The plural -i corresponding to Latin -ĒS
• Verbal tu dormi "you sleep" < Proto-Western-Romance /tu dɔrmes/ < TŪ DORMIS
• Verbal tu tieni "you hold" < TŪ TENĒS
• Subjunctive (che) tu ami "you love" < TŪ AMĒS

Indicative tu ami "you love" < TŪ AMĀS is unexpected; we would expect *tu ame. However, tu ame is in fact attested in Old Tuscan. Subjunctive tu dormi < TŪ DORMĀS is similarly unexpected; again, we would expect *tu dorme. In this case, it appears that -i was generalized as the universal tu ending, both indicative and subjunctive, at the expense of -e. (Note the even more striking generalization of first plural -iamo, originally only the subjunctive form of -ere and -ire verbs.)

If this theory is correct, something similar must have happened in Romanian.

References

• D'hulst, Yves, Romance plurals, Leiden University, 2005.

See also

• Classification of Romance languages
• Diachronics of plural inflection in the Padanian language
• La Spezia-Rimini Line