Prior studies of variable subject pronoun expression (mostly on Spanish) have consistently shown significant conditioning by person and number but less social conditioning (Torres Cacoullos & Travis, 2018). In this paper we compare variable subject pronouns in Spanish and Persian, focusing on the use of local person referents in dialogues (1sg and 2sg) and on the use of informal vs. polite forms of address (tú vs. usted in European Spanish, to vs. shomâ in Persian).

We recorded spontaneous speech dialogues with 54 native Spanish speakers in Barcelona and with 98 native Persian speakers in Tehran, using the same interview protocols in both locations. Speakers were between 18 and 48 years, approximately gender-balanced, with at least a high school degree. The envelope of variation consists of 4,872 finite pronominal clauses for the Spanish and 7,594 clauses for the Persian data.

The average pronoun rates are similar: 0.16 for the Spanish variety of Catalonia and 0.17 for the Tehrani variety of Persian. Both Spanish and Persian show the highest rates with local person referents. However, the effect of person/number is very different: Spanish has 0.17 for 1sg yo and 0.26 for 2sg tú, while Persian shows 0.28 for 1sg man and (only) 0.13 for 2sg to. We argue that the different uses of polite forms explain the difference between 2sg in Spanish and Persian.

Persian polite 2sg shomâ is used three times more frequently than the informal 2sg to, whereas the Spanish polite form usted is rarely used. The pronoun rate for the Persian polite 2sg shomâ is 0.38, compared to 0.13 for the informal to. Persian also has an ‘intermediate’ politeness expression by combining shomâ with the (mismatching) 2sg verb ending (see Nanbaksh, 2011). In these cases, the rate is 0.35. Overt 2sg pronoun to is reserved to much more restricted situational settings in Persian than in Spanish. It mostly occurs among friends and relatives without salient differences in age and social status.

In a second step, we tested whether gender, age, level of education, and number of siblings (as a family network-related variable) have an effect on rate of pronoun usage. At α = β = 0.1, an ANOVA showed no effect of gender and level of education. However, age (Pearson-R = 0.232, p < 0.024), and number of siblings (F=3.070, p < 0.051) correlate with pronoun usage.
A follow-up analysis revealed that these effects are restricted to local persons, and most visibly to 1sg and the polite forms of 2sg address. We argue that the age effect is age-grading rather than change in progress (see also Angermeyer & Singler, 2003 on historic stability of politeness variation). Furthermore, it is not unusual in Iran that siblings make differential use of 2sg in their interaction, based on their respective age relation. Those with more siblings are more familiar with this practice. We discuss our findings in the light of Helmbrecht’s (2011) competing motivation approach.