DEMOGRAPHICS OF THIS STUDY

Following Labov's (1984) methods of sociolinguistic interviews, I interviewed 20 men and 21 women between the ages of 23 and 81 who were randomly referred to me during a six-month period. All the participants were asked to recount an incident that had an impact on their lives.

As seen in Table 1, the majority of participants were between 20 and 59 years old. The education level of the participants was slightly different between males and females (Table 2). More females had college experience than did the males. Participants in the professional category were in banking, teaching, TV reporting, nursing professions, and so forth (Table 3). The managerial category was reserved for participants who supervised people. The skilled category included printers, clerks, postal workers, and so forth. The unskilled category was used mainly for delivery workers, supermarket staff, doormen, maintenance workers, or unskilled retirees.

Distribution of participants according to age and gender

Distribution of participants according to educational level

Distribution of participants according to occupation

Participants self-reported the information appearing in Table 4. The majority of participants were in daily contact with English at home. Those who claimed to use only Spanish were still in daily contact with English, because they worked in bilingual settings. In Table 5, we find that the majority of participants had lived in the City for 16 to 45 years.

Distribution of participants according to gender and language at home

Distribution of participants according to years in New York City

THE ENVELOPE OF VARIATION

The envelope of variation refers to the set of environments within which the SPPs are said to vary or to constitute a linguistic variable. In this study, the envelope of variation was defined as a tensed verb whose [+human] subject can be either a null or expressed SPP. If the alternation could not occur, the verb was excluded from the analysis.

The examples that follow represent instances in which a speaker had the option of expressing the overt SPP or omitting it in the Spanish variety of Puerto Rico:

In (1), the speaker had the option of using all three verbs (a) quiero, (b) sepas, and (c) ibamos without the overt SPP. In this example, all three verbs were entered in the analysis because they represented a site where the variable use of SPPs existed.

In (2), the same speaker used null forms with verbs (a) acordé, (b) tenía, and (c) tuve, but used an overt 3rd person SPP in verb (d) ayudó. All four verbs were entered in the analysis because the option of expressing a null or overt form existed in each case.

There are certain environments, however, where SPPs must be obligatorily expressed. Therefore, the verbs in the following examples were not included in the analysis:

In (3), the verb llegó was entered in the analysis because it represented a site where the variation could take place. However, the verb dijo did not enter the analysis because the speaker had to use the overt form obligatorily to preserve the meaning and contrast with the competing 3rd person Él. Without the overt expression of ella, the verb dijo would be interpreted as having él as its subject, as in (3a):

Therefore, clauses like (3) with pero + NP₂ Verb did not enter the analysis.

Instances in which the use of a null SPP was the only option were also omitted from the analysis:

Here an expressed SPP produces:

In Spanish, in general, including the Spanish variety spoken in Puerto Rico, (4a) is not considered grammatical.

As previous studies have indicated (Bayley & Pease-Álvarez, 1997; Cameron, 1992; Silva-Corvalán, 1994; Otheguy & Zentella, 2000), there are specific uses of certain verbs in which pronouns are obligatorily expressed or are obligatorily absent. These particular verbs, in which the alternation could not have existed, were also excluded from the analysis:

Nonpersonal subject as:

Existential verbs haber as:

Verbs with inanimate subjects such as:

Verbs with nonpersonal pronouns such as:

Verbs with uno as their subject such as:

Verbs with impersonal se such as:

Verbs that had subjects referring to atmospheric conditions as:

Verbs in pseudo-cleft constructions in which the SPP falls to the right of the verb (D'Introno, 1989) as in:

Verbs in constructions of contrast with pero as in:

The subject of the verb pensaron must be expressed in this sentence because the referents are not the same. In this instance, the verb pensaron is excluded from the study.

The subject of siguió is the same as that of redujo. Variation can exist with siguió.

Verbs in constructions where the pronoun is expressed with mismo as in:

Subject headed relative clauses:

Set phrases:

The varbrul program, commonly used for variationist research, was also used in this study to analyze over 15,000 tokens. The dependent variable was identified as the overt or null SPP, and in the binomial application of varbrul, the probabilities were defined relative to the occurrence of the overt SPP. A probabilistic weight of above .50 was considered as favoring the presence of the overt SPP, and a weight lower than .50 was considered as disfavoring the use of the overt form. Generally speaking, .50 is considered as an absence of an effect. However, it is important to note that a value of .50 relative to a value of .42, for example, can also be interpreted as showing a stronger probability, and therefore a higher favoring of overt SPPs. Likewise, a value of .75 relative to .65 can also be interpreted as showing a stronger probability, and therefore a higher favoring of overt SPPs. In general, we can consider that the probabilistic weights also serve to indicate a degree of strength that the independent variable has on the use of the dependent variable, the overt SPP. For varbrul see Paolillo (2002).

Unless otherwise indicated, the significance value of all runs in Flores-Ferrán's large-scale study was (p < .0001). The p value represents the probability that the results obtained were due to chance. The log likelihood for the best run was found to be equal to −8900.612, with a chi-square of 1.390 per cell. Chi-square values have been omitted from the tables that show comparisons between two or more studies.²

FINDINGS

To begin, Table 6 shows the frequencies and weight probabilities of overt SPP (PRO+) expression for the entire group of NYC residents who participated in the study. In general, Table 6 shows that at .44, the Puerto Rican residents of NYC, as a group, do not tend to favor the use of overt SPPs, but rather favor the null form at .56, a tendency found in most Spanish varieties. We can assume then, at this point, that a shift in the tendency favoring the use of overt SPPs would have indicated a departure from most Spanish varieties.

Frequencies and weights for the use of overt SPPs in New York City

Even though there were differences in the agendas of these two studies with regard to the number and types of factors examined,³

Comparison of New York City to San Juan speakers' use of overt and null SPPs

PERSON AND NUMBER

In Table 8, comparisons of the frequencies of overt SPP use among NYC and San Juan residents are made with the purpose of determining the differences between the two geographic areas, if any.

Overt SPPs for person and number: San Juan and NYC

A number of interesting points can be drawn from Table 8. First, the NYC speakers as a whole do not tend to deviate from the current patterns of pronominal expression when compared to speakers residing on the island. The NYC group tends to produce overt SPPs within or below the range of frequencies of these two studies, with the exception of 3rd person SPPs. Second, there are similar patterns found in the order or rank in which the frequencies appear. For instance, Ávila-Jiménez' (1995, 1996) and Flores-Ferrán's (2002) studies showed identical ordering: 2nd person nonspecific tú,⁴

There are two observations that need to be made with respect to the frequencies in Table 8, however. First, there seems to be a slight difference in the frequencies with which speakers use 2nd person nonspecific tú. Cameron's (1992) study shows 69%, whereas Flores-Ferrán (2002) shows 60%. But note that even though we find slight differences, we still see that all three studies show a higher frequency in the use of the nonspecific tú than of specific tú. Furthermore, the frequency found in the NYC residents is not higher than that found in the other two studies. Note that the frequency is not working in the direction of a contact hypothesis. Second, there also seems to be a slight difference with regard to the use of 3rd person el/ella. The NYC group uses higher frequencies than those reported in both San Juan groups. This difference requires further examination.

If we observe the probabilistic weights reported for nonspecific tú and 3rd person el/ella, reported by Cameron (1992) and Flores-Ferrán (2002), more similarities than differences can be uncovered.

The varbrul weights in Table 9 show that for the use of overt 3rd person el/ella the islanders have a probability weight of .47, disfavoring the use of overt 3rd person el/ella, whereas the New Yorkers have one of .52, which represents a stronger probability in favor of the overt expression. Relative to .47, this .52, although not a very big difference in the strengths, suggests a positive effect with regard to the use of overt SPPs in the 3rd person in the NYC group. But, at this point, we cannot distinguish if this difference is correlated to increased exposure to NYC. Therefore, I will revisit the discussion regarding 3rd person SPPs later in this article.

Comparison of New York City and San Juan probability weights of overt SPPs with person and number

In Table 9, however, we are still left with the most marked difference in the probability weights of nonspecific tú. NYC speakers have a .58 probabilty of expressing the nonspecific pronoun overtly, whereas the islanders have a much stronger probability, .72. This difference in the probability weights does not favor a contact hypothesis, but suggests that perhaps this gap exists in part because of the differences in the type of corpora analyzed when examining the use of overt SPPs with 2nd person nonspecific tú.⁵

In sum, although it is understood that varbrul weights and frequencies cannot be compared on a strictly equivalent basis, because these three studies contained different factor groups, the similarities found in the use of overt SPP for person and number are striking. Thus, and in general terms, it seems reasonable to assume that the NYC group, although separated by distance and exposed to an urban environment where English is spoken, seems to have similar patterns of overt SPP usage when compared to their counterparts on the island.

Before turning to switch reference, I would like to revisit the observation made regarding the use of 3rd person SPPs, because the large study further examined the use of overt and null SPPs with regard to 3rd person in more detail. Because 3rd person pronouns el and ella use the same morphological endings in their verbs, I hypothesized that potential interference may influence speakers use of overt 3rd person SPPs in their narratives. Therefore, a factor group named Competition or Potential Interference between previously overtly mentioned 3rd person singular subjects was constructed in Flores-Ferrán (2002). The purpose of this factor was to determine whether contexts with two or more 3rd person singular subjects in previously mentioned verbs tend to produce more appearances of overt SPPs in the verb under study. The rationale behind this factor was that the two or more 3rd person subjects in the same narrative would create an environment of potential referential ambiguity. Contrary to my hypothesis, an initial varbrul run in a step-down analysis asked for this factor group's removal. That is, the results showed that there was no connection between the use of multiple 3rd persons subjects in a narrative and the high or low numbers of overt 3rd person SPPs.

SWITCH REFERENCE

A factor that has been shown to condition the use of overt SPPs in a significant and systematic manner is that of switch reference, found across Spanish dialects (Cameron, 1992, 1995; Ávila-Jiménez, 1996; Bayley & Pease-Álvarez, 1997; Flores-Ferrán, 2002; Morales, 1986; Silva-Corvalán, 1982, 1994) and also other languages (Eid, 1983; Enç, 1986; Lira, 1982).

In search for evidence of contact showing that the NYC residents studied here have altered their use of pronominal expression because of their exposure to the City, I questioned whether or not a switch in reference influenced speakers' use of overt SPPs, and I also questioned if the patterns of pronominal expression were similar to those documented on the island. I hypothesized that if differences were found, they could be attributed to the degree of exposure to NYC.

DEFINING SWITCH REFERENCE AND HOW THE DATA WERE ANALYZED

Cameron's (1992, 1995) study conducted in Puerto Rico constitutes a valuable contribution to any study related to the variable use of SPPs. That study showed that a switch in reference has a robust effect on the use of overt SPPs. Therefore, Cameron's (1992) terminology and formula were used to analyze the data in this present study with the purpose of providing a point of comparison for the NYC data:

The relationship of switch and same reference for this present study was defined between two NPs where the second NP is the [+human] subject of a tensed verb that occurs after and nearest to another subject NP of a tensed verb. Following Cameron's (1992) terminology, NP (1) is known as the trigger and NP (2) as the target. The target is the subject NP that is marked as either switch or same with respect to the trigger. For the purposes of comparison, I strictly adhered to Cameron's (1992, 1995) formula:

The NYC narratives for switch reference analysis were coded using the following:

Examples of how a switch referent was analyzed and coded:

In example (17), when we compare the target NP of verb (b) contesta to the NP of the trigger (a) estaba, we find that there has been a switch in reference from mami to yo, and therefore contesta was coded with factor 2 (switch in reference). However, when we compare the target NP in verb (c) dice, to the NP trigger (b) contesta, we find that both NPs refer to the same subject, ella and mami. Therefore, (c) dice was coded as same in reference (factor 1). When we compare the NP target in (d) estás to its trigger NP found in (c) dice, we find that the NP for (d) estás is the null 2nd person specific tú, whereas the NP for (c) dice, is the null form 3rd person, ella. Therefore, the verb (d) estás was coded with factor 2, as a switch reference.

Table 10 shows higher frequency and probability weights for the use of overt SPPs in a switch reference environment. In instances where there is no switch, speakers expressed only 38% of their SPPs overtly. The strength of this frequency is represented by .43 varbrul weight. Said differently, when there is not a switch reference, speakers tend to favor null SPPs. However, when there is a switch reference, there is a stronger probability (.57) that speakers will use an overt SPP. That is, when a switch occurs, speakers favor the use of an overt form. This pattern was also reported by scholars in the United States, the Caribbean, Mexico, Spain, and Latin America, which suggests that a switch reference conditions the use of overt SPPs. In sum, a switch reference in NYC does show a robust effect on pronominal expression.

Null and overt SPP expression in switch reference environments

We now turn to a comparison of switch reference findings in NYC and San Juan. In Table 11, we see that the frequencies of overt SPP expression resemble those documented in San Juan. First, and in general, both groups show higher rates of overt SPPs in a switch reference, and lower rates in a nonswitch environment. Second, we find that in a switch, the San Juan speakers expressed 57% of their SPPs overtly, whereas the NYC group expressed 54%, a close proximity. A similarity also exists in the use of overt SPPs in a nonswitch environment. The NYC group expressed overt SPPs in a nonswitch environment in 38% of their verbs, while the San Juan residents did so for 31%. Although the patterns are similar, the NYC group tends to express more overt SPPs in a nonswitch environment (38%) than the islanders (31%). This observation may require further examination.

Comparison of switch reference environments in San Juan (Cameron, 1992, 1995) and New York City (Flores-Ferrán, 2002)

But before further examining the nonswitch environment of the NYC group, a comparison of singular and plural verbs in a switch reference environment is in order. Table 12 also shows striking similarities in the frequencies with which overt SPPs are expressed in a singular and plural context. Notice, however, that in nonswitch instances, the NYC group shows higher frequencies of singular and plural overt SPPs than the islanders. Further discussion is also needed with regard to this observation.

Comparison of overt singular and plural SPPs in switch reference: New York City (Flores-Ferrán, 2002) and San Juan (Cameron, 1992)

Before continuing, I need to address the questions asked earlier: Does a switch in reference exert an influence on the use of overt SPPs in the NYC group? The response is positive; a switch in reference does exert an influence on the use of overt SPPs in NYC. Furthermore, the patterns of pronominal expression are similar to those documented in San Juan (Cameron, 1992). When I address the second question regarding the relative frequencies of overt SPP expression and whether or not the NYC group exceeds the frequencies found in San Juan, I find that the only distinction that can be made is that the NYC group tends to use more overt SPPs in a nonswitch reference environment. At this point, I cannot find evidence of a change with regard to this factor, unless I address how the three separate groups of residents use overt SPPs in this environment.

To further examine this finding, we now look at how the three distinct groups, the recent arrivals, the established residents, and the NYC native-born, use SPPs in a switch reference environment. In this analysis some gaps begin to emerge in the data, in particular, with the NYC native-born group.

EXPOSURE TO NEW YORK CITY AND SPP USE: IS THERE EVIDENCE OF A CHANGE IN PROGRESS?

Recall that I used Cameron's (1992) formula for defining a switch reference for analyzing the corpora. I hypothesized that residents of NYC would show similar patterns of overt pronominal expression to those in San Juan, providing that the NYC group had not been affected by their exposure to NYC, which I can indirectly associate to English contact. In Tables 7, 8, 9, 10, 11, and 12, when comparisons between the NYC group and the islanders were drawn, similar patterns of SPP expression were found, as well as similar frequencies and probability weights with regard to person and number, switch reference, and the use of overt singular and plural SPPs in switch reference. But earlier in the article, it was also mentioned that if heightened use of overt SPPs were detected and those frequencies and/or probabilities exceeded those already documented in San Juan, then this would be considered a sign of divergence, pointing towards a contact hypothesis. I discovered that for nonswitch reference, the realization of the overt SPP in the New Yorkers was seen at a frequency of 38%, whereas the islanders showed only a 31% frequency. The question I need to address now is whether or not this difference detected is sufficient to claim that degree of exposure to NYC can be correlated to the use of overt SPPs.

When constructing the factor group that examined the indirect exposure that a NYC resident might have to English, I had to take into account several subfactors that were self-reported. Still, these factors indirectly address the matter of English contact. The factor of exposure to NYC was coded as: Degree of exposure to NYC, wherein absences away from the City were not subtracted from the total years:

• N = speaker has lived in NYC since birth and had any 5 of the first 8 years of education in the United States (native-born New Yorker)
• E = speaker has lived in NYC more than 16 years (established resident)
• R = speaker has lived in NYC between 0 and 5 years (recent arrival)

Total years in NYC, wherein absences from the City were subtracted from the total years:

• 1 = less than 15 years
• 2 = between 16 and 30 years
• 3 = between 31and 45 years
• 4 = more than 45 years

Age of arrival to NYC was coded with:

• B = born in NYC
• C = child until 12 years
• T = teen from 13 to 19 years
• A = adult more than 20 years

Language at home was coded with the following schema:

• E = English
• S = Spanish
• B = Both

The only factor group to survive the step-up/step-down analysis of the varbrul program was that of exposure to NYC, which did not take into account the time that a participant may have left NYC to go to Puerto Rico on vacation or left NYC for a period of time to live elsewhere. One reason why the distinction was made regarding the factor groups of degree of exposure and total years in NYC was that older participants could not identify the exact years they had spent away from the City. The nonsignificance of age of arrival, language spoken at home, and whether one arrived as a child or at an older age shows that these three categories are immaterial for describing an account of indirect English contact. However, some observations can be made if we examine the only surviving factor in this category.

Table 13 shows that only the NYC native-born group (at .55) has a probability of using more overt SPPs with their verbs than the other two groups. That is, the recent arrivals and the established residents do not show a probability weight that favors the dependent variable, the overt SPP (.39, .45). Furthermore, there is an increase in the progression of the frequencies and probability weights as degrees of exposure increase. This finding with regard to native NYC-born Puerto Ricans and their tendency to use more overt SPPs needs further discussion, for it appears to favor a contact hypothesis. But first, I will compare research documented in San Juan, Boston, and New York and then discuss in more detail how these comparisons may relate to the results found with the NYC native-born group reported in Flores-Ferrán (2002).

Overt SPP use according to degree of exposure to New York City

I use Table 14 as a point of reference to enable us to see at a glance if the NYC group's frequencies diverge from other studies. In Table 14, the NYC group as a whole shows a similar frequency of overt SPP expression to those documented in Boston and San Juan. Now I would like to further address why the NYC native-born Puerto Ricans may be exceeding in the frequencies. Recall that in Table 13 the NYC group was shown to use 53% of its SPPs overtly. By cross-tabulating the factor group of degree of exposure to NYC and the factors of switch reference and the use of singular and plural subjects, we find that, on the surface, some data may support the contact hypothesis.

Comparison of overt SPPs according to regional dialects

I ran a cross-tabulation of the factors of switch reference and degree of exposure to NYC. In Table 15, one sees that the NYC native-born group does express more overt SPPs in a nonswitch environment than the established residents and the recent arrivals. This frequency also exceeds the frequency reported by Cameron (1992), which was 31%. Therefore, although the established and recent arrivals remain within or close to the boundaries of previously reported research, the native NYC-born Puerto Ricans do not. But a closer look shows that there is an increased progression in the frequencies of pronominal expression for the nonswitch environment, but not in instances of switch. This is the first contradiction that emerges, in that the recent arrivals and the established residents show very similar frequencies (49% and 48%, respectively).

Cross-tabulation of switch reference and degree of exposure to New York City

Table 16 once again shows that the NYC native-born group leads in the use of overt singular SPPs at 57%. We also see an increased progression in the frequencies that match the degree of exposure for the use of singular SPPs. However, we do not find that for plurals the same pattern holds. That is, the native-born New Yorkers and the recent arrivals show identical frequencies in the use of overt plural SPPs (25%), a concern that does not point toward a contact hypothesis. It should be noted that previous studies (Barrenechea & Alonso, 1977; Bentivoglio, 1980; Cameron, 1992; Enríquez, 1984; Morales, 1986) have reported that plural subjects show a much lower rate of SPP expression than do singular subjects; therefore, in that sense, the NYC native-born group seems to follow patterns documented in other studies.

Cross-tabulation of singular and plural overt SPPs and degree of exposure

In sum, the explanation that states that the increased progression in the frequencies of overt SPP expression parallels years of exposure can no longer be used to support a contact hypothesis, because this is the second instance in which this inconsistency appears. If English were to be indirectly influencing Puerto Ricans who are NYC native-born, then we should be seeing increased uses of overt SPPs in plurals also because it is expected that the almost categorical use of overt plural SPPs in English may also influence NYC speakers to use more overt plural SPPs in Spanish, regardless of number.

If we go back to Table 15, one can also see that the NYC-born residents tend to express more overt SPPs in almost half of their verbs (46%), even when a switch in reference has not occurred. This observation lends itself to the question of whether the NYC native-born group is arriving at the stage of expressing overt SPPs with and without a switch. This parallel use of the overt form may suggest that perhaps the strategy for signaling a switch may be what is being altered as a consequence of indirect exposure to English, an observation that is outside the scope of this study and would need further investigation.

Because we now know that the NYC native-born group shows a stronger tendency to use overt SPP in both switch and nonswitch references, we should then compare the NYC native-born group with the San Juan speakers. I draw on these comparisons to see if there are any similarities with respect to switch reference, as the participants in this study are exposed to NYC English.

In Table 17, note that both groups have a tendency to express higher rates of SPPs in a switch reference, and both decrease frequencies in a nonswitch environment. However, one can see that the San Juan group makes a clear distinction of expressing overt SPPs in a switch versus a nonswitch, as there is a 26 percentage point gap. The NYC native-born group, on the other hand, has only an 18 percentage point gap difference between switch and nonswitch instances. One may consider this gap of 18% to be a signal indicating that the NYC native-born group does not clearly make the distinction of using overt SPPs in switch as the San Juan group does; however, the 26% to 18% difference is not very robust. It should be noted that, with regard to Madrid Spanish, Cameron's (1992) study also found a 19% difference in the use of overt SPPs between a switch and nonswitch environment. Therefore, the 18% difference on the surface does not necessarily point toward a change in a switch reference strategy.

Comparison of use of overt SPPs in switch reference: New York City native-born speakers (Flores-Ferrán, 2002) and speakers from San Juan (Cameron, 1992)