Anthony J. Naro, Universidade Federal do Rio de Janeiro
Marta Scherre, Universidade Federal do Espírito Santo and Universidade de Brasília

Three general processing-oriented principles underlie variation in Brazilian Portuguese third-person subject/verb and noun-phrase plural concord:

1. phonic salience: plural forms more highly distinct from the corresponding singular form are more likely than plural forms more similar to the singular. Thus the marked plural is more likely in nós fomos/nós foi ‘we went’ than in nós moramos ‘we live’ or ‘lived’/nós mora ‘we live’;
2. positional salience: agreeing forms are more likely when constituents are closer and preceding within the phrase. Thus plural marking is more likely in eles dizem ‘they say’ than in os filho dela pelo menos tá passando bem `At least her children are well´;
3. parallelism: forms tend to cluster in marked or unmarked chains on the phrase and discourse levels. Thus, marked verbs are more likely in a discourse segment such as uns camarada chegaram do lado deles assim… começaram a implicar com eles… pisaram no tênis deles ‘some guys came up to them…they began to bother them…they stepped on their sneakers’.

The salience effect is the strongest and results in a distribution of marked forms with heavy bias toward the preterit because the desinence is often stressed in this tense. First-person plural concord also obeys these principles, but the phonic salience effect is becoming functional through reanalysis of surface forms so that tense takes precedence over salience in determination of marking. Originally agreeing preterit forms highly distinct from the present were favored. In some population groups, the spoken language shows favoring of the agreeing form in the preterit even for low salience oppositions such as nós moramos /nós mora ‘we lived/live’. Thus, tense takes precedence over salience in determination of marking.

Since the present/preterit distinction does not exist in the standard paradigm of regular verbs for first-person plural this development takes on a highly functional role, with preterit nós moramos ‘we lived’ contrasting with present nós mora ‘we live’. Furthermore, more non-agreeing first-person plural forms are created by avoidance of antepenultimate stress in the imperfect, as in phonologically reduced nós morava from nós morávamos ‘we used to live’. Functional and phonological effects create a variable system that preserves meaning and penultimate stress but clashes with grammatical tradition.

Avoidance of sociolinguistic conflict is achieved by creation of a new semantically first-person subject pronoun a gente, that is diachronically derived from the noun with the same form and remains grammatically third-person singular. It is used mainly with present and imperfect verbs, setting up contrast between a gente mora ‘we live’/a gente morava ’we used to live’ and nós moramos ‘we lived’. This is fully in accord with recent trends toward restoration of concordance in urban Brazilian Portuguese.