Capaz and Quizás in Argentine Spanish: Epistemic Adverbs with Mood Variability

Abstract

This article explores the use of two epistemic adverbs in Argentine Spanish, quizás and capaz. While scholars have noted the existence of capaz as an epistemic adverb, it had not been systematically studied prior to this investigation. To examine the epistemic stance of the adverbs, 117 participants completed an experimental sentence completion task in which mood selection (i.e., percentage subjunctive use) was measured. The dependent variables considered were situational certainty and temporal reference, as well as adverb. Situational certainty (certain, not certain) and temporal reference (future, present, past) were communicated in written discourse scenarios that preceded the sentence to be completed, and the certain or not certain status was verified through a norming procedure. Through a mixed model analysis, findings indicated that temporal reference was the only significant dependent variable. Mood selection was significantly different in contexts with future, present, and past temporal reference. Future temporal reference was associated with the greatest subjunctive use and past with the least subjunctive use. The findings were similar for sentences with capaz and quizás, indicating a similar epistemic stance based on current data.

1. Introduction

In language, epistemic modality allows speakers to express the possibility or likelihood of events occurring (Nuyts 2001; Ojea 2005). Example (1) demonstrates one way in which a limited possibility is expressed, namely through the use of an epistemic adverb like quizás ‘maybe.’ Another way in which some languages, like Spanish, express epistemic modality is through the indicative or subjunctive mood, where comparatively the indicative expresses an increased likelihood of the event occurring (1), while the subjunctive expresses a decreased likelihood of the event occurring (2) (Butt and Benjamin 2013; Pountain and Kattán-Ibarra 2003).

(1) Quizás viene (I) mañana.

(2) Quizás venga (S) mañana.

These two linguistic tools can function independently or jointly to express modality. Some epistemic adverbs like a lo mejor almost always coincide with the use of the indicative mood, thus presenting two epistemic markers that jointly indicate a great likelihood that the event in question will occur. Other epistemic adverbs, like quizás, tal vez, or capaz may occur with verbs that express the indicative or subjunctive mood, as shown in (1) and (2). Thus, with these types of adverbs, the resulting utterances may express a greater range of epistemic modality or epistemic stance. The degree to which a speaker indicates the possibility of something is scalar (Ojea 2005), meaning that the finite points of ‘unlikely’ and ‘likely’ are not the only options but that there are a multitude of gradient stances towards the possibility of a proposition.

Considering that speakers may optionally use the subjunctive or indicative mood with certain adverbs, prior research has explored the factors that impact mood selection in these cases. From semantic, historical, and sociolinguistic perspectives, it has been found that mood selection in sentences with epistemic adverbs that allow the subjunctive or indicative is subjective, meaning that it is related to the speaker’s perspective, which includes the speaker’s certainty (Houle and Martínez Gómez 2011; Sastre Ruano 1997; Wasa 2002). Pragmatic theories related to mood selection indicate that the subjunctive is associated with low information contexts (Gregory 2001; Lunn 1989), also potentially leading to reduced speaker certainty. In addition to speaker certainty, temporal reference (e.g., past, present, future) has also been found to impact mood selection (Ávila 2010; DeMello 1995; Lunn 1995; Renaldi 1977; Schwenter 2011; Woehr 1972).

Along similar lines, prior research has also examined the relationship between epistemic adverb and mood variation as a way to enhance the understanding of the use of those adverbs. DeMello (1995), Finanger (2011), and García (2011) have taken up this question, examining mood selection in utterances with epistemic adverbs that select the indicative or subjunctive mood, such as quizá, quizás, tal vez, posiblemente, and probablemente, finding varying rates of the subjunctive use depending on the adverb.

The current study continues the investigation of the relationship between adverb and mood. The epistemic stance of an adverb that has not previously been investigated, capaz, is examined in comparison with the epistemic stance of quizás, since they are often presumed to be synonymous in varieties of Spanish where both are used. Quizás is used throughout the Spanish-speaking world and the use of capaz as an epistemic adverb is mostly limited to varieties of Argentine and Chilean Spanish (RAE).¹ While scholars have noted the use of capaz as an epistemic adverb (e.g., Aleza Izquierdo 2009, 2010), there has only been one study of the historical development of the use of capaz as an epistemic adverb (Ávila 2010) and no study of its epistemic stance in comparison to that of other epistemic adverbs. Therefore, a main contribution of this investigation is to provide insight about this understudied adverb capaz in Argentine Spanish, specifically as it compares to quizás.

To investigate quizás and capaz, Argentine Spanish speakers participated in an experimental sentence completion task. Mood selection (i.e., percentage subjunctive use) was measured as the independent variable. Since “the subjunctive may be triggered in a wide range of contexts” (Portner 2009) and is impacted by the speaker’s degree of certainty and temporal reference (Ávila 2010; DeMello 1995; Lunn 1995; Renaldi 1977; Schwenter 2011; Woehr 1972), both the degree of certainty and temporal reference were controlled in the experimental discourse scenarios to provide a more detailed understanding of the relationship between mood and quizás and capaz.

This methodology, which allows for the analysis of the impact of degree of certainty on mood selection, complements corpus studies of mood in which a speaker’s interpretation of context certainty is indeterminable. In sum, through the exploration of mood variation in controlled discourse contexts, this experiment investigates the epistemic stance of quizás and capaz as epistemic adverbs in Argentine Spanish and offers the first empirical analysis of capaz as an epistemic adverb.

2. Literature Review

Prior research related to the relationship between epistemic adverbs and mood has considered the diverse or similar epistemic stances of certain adverbs (DeMello 1995) and has also addressed the linguistic and extralinguistic factors that impact mood selection in the presence of epistemic adverbs (Finanger 2011; García 2011).

2.1 Types of Epistemic Adverbs Based on the Relationship to Mood

The relationship between epistemic adverbs and mood in Spanish has been investigated using questionnaires (Studerus 1995), specific examples from texts (Renaldi 1977; Wasa 2002), and corpora in order to identify trends of mood selection with different adverbs (DeMello 1995; Houle and Martínez Gomez 2011; Wasa 2002; Woehr 1972). Varying rates of mood selection have been found in combination with different adverbs, such as with quizá(s), tal vez, posiblemente, and probablemente (DeMello 1995; Finanger 2011; García 2011; Renaldi 1977), resulting in the findings that adverbs like quizás, quizá, and tal vez form an adverbial group based on their general semantic meanings and also their relationship to mood selection (DeMello 1995).

DeMello’s (1995) investigation of epistemic adverbs and mood, investigating acaso, probablemente, posiblemente, tal vez, quizá, quizás, puede/pueda ser que, and ser possible que, in various varieties of Spanish, concluded with the above grouping based on the fact that quizás, quizá, and tal vez were found to be used 39%, 30%, and 30% of the time with the subjunctive mood, respectively. This mood distribution distinguished these adverbs from others, such as acaso, which aligned with the subjunctive mood in 12% of the tokens (DeMello 1995). This implies that there is a difference in the behavior of the epistemic adverbs like quizás, quizá, and tal vez compared to the other epistemic adverbs with respect to the epistemic stance communicated. The overall rates of subjunctive use with adverbs like quizás, quizá, and tal vez found by DeMello (1995) (30–39%) reflected Renaldi’s (1977) finding of a 40% subjunctive rate with tal vez and 39% with quizá and quizás combined. Yet contrasting with Renaldi’s (1977) investigation, DeMello’s (1995) results pointed to potential differences in the use of quizás, which was used with the subjunctive 39% of the time, and quizá and tal vez which were used with the subjunctive 30% of the time. These findings draw attention to the need to further investigate the epistemic stances of adverbs like these.

2.2 Linguistic and Extralinguistic Factors that Impact Mood Selection with Epistemic Adverbs

In more recent studies, not only have researchers examined mood variation with adverbs like tal vez and quizás, but they have included more in-depth analyses of sociolinguistic variation considering distinct varieties of Spanish as well as other linguistic contextual factors such as polarity, verb lexeme, and temporal reference (Finanger 2011; García 2011; King, McLeish, Zuckerman, and Schwenter 2008). As reported in Schwenter (2011) and García (2011), King et al. (2008) investigated various epistemic adverb constructions in Argentine, Mexican, and Peninsular varieties of Spanish, analyzing 3,022 tokens from the Corpus de Referencia del Español Actual (CREA). The adverbs investigated were tal vez, quizá, quizás, probablemente, and posiblemente. Main findings indicated that Argentine Spanish highly favored the subjunctive with tal vez and quizás. Through the analysis of other linguistic variables (polarity, adjacency of the verb and adverb, temporal reference of the verb, person/number, style, and verb), it was found that temporal reference had the largest impact on mood variation in all three varieties of Spanish. Specifically, the present temporal reference favored the use of the subjunctive and past temporal reference favored the indicative in all varieties and with all adverbs. Distinguishing quizás and tal vez from other adverbs, the future temporal reference favored the subjunctive when these adverbs were present.

Following up on the initial finding of difference in the use of quizás and quizá in Argentine Spanish, where quizás was found to favor the subjunctive (factor weight = .55) and quizá to favor the indicative (factor weight = .38). King et al. (2008) and García (2011) examined quizás and quizá in six Spanish varieties: Cuban, Peruvian, Venezuelan, Argentine, Mexican, and Peninsular Spanish. Cuban, Peruvian, and Venezuelan tokens were drawn from CREA, while Argentine, Mexican, and Peninsular Spanish tokens were taken from the subset of the CREA corpus used by King et al. (2008). García’s (2011) results confirmed King et al.’s (2008) findings of a distinction between quizás and quizá, namely that while present temporal reference favors the use of the subjunctive with both adverbs and past temporal reference disfavors the use of the subjunctive with both, future temporal reference with quizás favors the use of the subjunctive, while the opposite is true for quizá. Considering geographical variation, García (2011) also found that the subjunctive was more favored with quizás in Argentine Spanish as compared to the other varieties, highlighting the use of quizás with the subjunctive in Argentine Spanish.

In a separate study of quizás in Argentine and Peninsular Spanish using CREA, Finanger (2011) also found that “non-past time reference favors Subjunctive in both dialects … while past time reference strongly disfavors Subjunctive” (97). These findings corroborate past research from both quantitative and theoretical approaches (DeMello 1995; King et al. 2008; Renaldi, 1977; Woehr, 1972). Explaining the impact of temporal reference, Butt and Benjamin (2013) state that the subjunctive is optional when an event referred to “is happening in the present or happened in the past” but that future events employ the present subjunctive or future indicative, limiting the use of present indicative (247). This may be due to the notion of possible worlds in which the speaker inherently thinks that the past must have a fixed truth value while the truth values related to the present and future are less (or not yet) readily accessible (von Fintel 2006).

In sum, there is evidence that Argentine Spanish may employ the subjunctive more often with epistemic adverbs compared to other varieties of Spanish (King et al. 2008), and Argentine Spanish may implement a greater use of the subjunctive with the adverb quizás (García 2011). Additionally, temporal reference seems to be a key factor in determining mood in contexts of epistemic adverbs (Finanger 2011; García 2011; King et al. 2008), with present temporal reference in these adverbial contexts favoring the subjunctive, past temporal reference favoring the indicative, and future temporal reference interacting more differentially depending on the adverb—this last case being found by García (2011). Thus, in the current examination of mood and epistemic adverbs, temporal reference is considered and quizás was selected as the adverb with which to compare capaz in Argentine Spanish. Furthermore, situational certainty was examined in the current study since, theoretically, a speaker’s certainty status may impact mood selection (Sastre Ruano 1997; Wasa 2002).

2.3 Capaz as an Epistemic Adverb

Though traditionally an adjective meaning ‘capable,’ capaz is increasingly found in Argentine Spanish as an epistemic adverb (Aleza Izquierdo 2009; Butt and Benjamin 2013; Di Tulio 2007; Moreno Fernández 2010; Pérez Martín 1999). However, there is a limited amount of information about its usage and relationship to mood selection. In a comprehensive volume on the use of Spanish in the Americas, the following description of capaz as an epistemic adverb was offered: “Se utiliza como adverbio de probabilidad o duda junto a un verbo en indicativo generalmente: Capaz llega tarde, o subjuntivo: Capaz llega [sic] tarde” (Aleza Izquierdo 2010: 188), providing limited information beyond a basic description of epistemic adverbs. This description indicates that capaz is used with the indicative generally, yet there are no studies to confirm this proposal. Furthermore, the absence of a uniform gloss or contextually constrained glosses for capaz (e.g. en una de esas, a lo mejor, tal vez, es posible que, posiblemente, a no dudarlo, quizás, es probable que (Butt and Benjamin 2013; Moreno Fernández 2010; Pérez Martín 1999) indicates the need to study the meaning and use of capaz. Published material with mentions of capaz typically list lexical entries and glosses for capaz with examples of its relationship to mood, but there are no empirical studies relating to this specific epistemic adverb. Therefore, capaz was selected as the focus of this study considering its common use in Argentine Spanish as well as the dearth of information about capaz and its relationship with mood selection.

2.4 The Current Study

To examine the epistemic adverbs capaz and quizás in Argentine Spanish, an experimental design involving a sentence completion task was conducted. The dependent variable was the percentage subjunctive use in the sentence completion task. Considering that prior research has highlighted the potential impact of certainty and temporal reference on mood selection, both situational certainty and temporal reference were controlled. Quizás was selected as the comparative adverb, since it has been shown to be commonly used and studied in Argentine Spanish (Finanger 2011; García 2011; King et al. 2008), plus it has been found to have a greater association with the subjunctive compared to other epistemic adverbs (DeMello 1995; García 2011).

The research question that motivates this investigation is: How does the mood selection pattern for capaz compare to that of quizás, considering both situational certainty and temporal reference? Based on prior research, it is expected that the subjunctive will be employed more often in not certain contexts compared to certain contexts. It is also expected that non-past references will provoke a greater use of the subjunctive, compared to past temporal references. Finally, considering mood selection in sentences with capaz compared to those with quizás, prior literature does not lead to a definitive hypothesis. If the two adverbs are similar in their epistemic stances, then similar mood selection patterns would be expected. Alternatively, if the epistemic adverb capaz entails some meaning of ‘capable,’ perhaps greater use of the indicative may be expected with capaz as compared to sentences with quizás. Overall, the absence of research about capaz as an epistemic adverb does not lend itself to strong hypotheses about capaz, which further indicates the need for this investigation.

3. Methodology

Mood selection in epistemic adverb constructions with capaz and quizás was investigated using an experimental approach. A sentence completion task was devised to control for certainty, temporal reference, and adverb. Participants completed each sentence with an indicative or subjunctive verb. The percentage subjunctive selection was the dependent variable.

3.1 Participants

This study attained 117 participants, all native speakers of Argentine Spanish who were between 18 and 73 years old (Mean = 29.5; SD = 11.9). There were 29 males and 88 females, and the majority of the participants were from Buenos Aires, Entre Ríos, and Santa Fe.² Seventy-three participants had a college-level degree, while 27 participants had completed high school. No participant had expert knowledge related to language or linguistics.

3.2 Questionnaire Design

The complete questionnaire consisted of 66 items (six demographic questions, 48 target questions (24 with capaz, 24 with quizás), and 12 distractors). The demographic questions requested the participants’ age, gender, city, and province where they had lived most, highest level of education, and whether the participant had studied or taught other languages or linguistics. Departing from Studerus’s (1995) forced choice test, in which “survey participants were asked to choose what seemed to them the most natural of two verb forms (indicative or subjunctive)” (95), the current questionnaire employed a sentence completion task. A sentence completion task was selected considering that forced choice tasks have been criticized for lacking evidence of naturally-occurring conversational data (García 2011). The sentence completion task did not prime the participants with a binary choice of either the indicative or subjunctive, allowing them to freely choose the most natural option. This approach also served to conceal the nature of the study as targeting mood selection. Moreover, a sentence completion task allows for controlled contexts in which to study the epistemic adverbs.

3.2.1 Target Questions

All target questions included scenarios with a preamble and then a sentence to be completed by the participant (3). For comparative purposes, the 24 scenarios (i.e., preamble plus associated sentence) were used with both capaz and quizás in the sentence completion task, yielding the 48 target questions.

(3) Preamble: Alguien ve a un grupo de empresarios y no sabe quiénes son. Su amigo tampoco sabe. ¿Quiénes son esos empresarios?

Sentence: Capaz _______ (ser) los amigos de Cecilia.

The 24 scenarios were created controlling for formality and person, as well as the independent variables of temporal reference and certainty. In light of Nieuwenhuijsen’s (2001) finding that formality may trigger the subjunctive, any use of tú, vos, or usted was avoided and a neutral stance toward formality was communicated. Also reducing the potential influence of formality, participants did not respond to any specific addressee in the task. To control for grammatical person, third-person subjects were used in each sentence that was to be completed. Regarding temporal reference, a third of the scenarios contained future temporal reference, a third contained present temporal reference, and a third contained past temporal reference.

While the temporal reference was clearly communicated through the use of verb tense, the level of certainty (i.e., certain or not certain) was expressed at the discourse level, and a norming procedure presented in section 3.2.2 verified that each scenario was rated as certain or not certain, as intended. Example 4 demonstrates a not certain scenario that expressed a low level of certainty about the event expressed in the sentence following the preamble, and Example 5 demonstrates a certain scenario that expressed a high level of certainty about the event in the final sentence. The amount of information in the scenarios at the disposal of the participants was modulated, with less information presented in the not certain scenarios and more information in the certain scenarios, similar to the approach in Czerwionka (2012). This notion of certain or not certain also aligns with pragmatic explanations of the use of the subjunctive in contexts of low information (Gregory 2001; Lunn 1989).

(4) Preamble: Alguien discute con su amigo sobre la política y propone una opinión con poco conocimiento de la actualidad.

Sentence: Capaz la presidenta _______ (creer) en la política actual.

(5) Preamble: Alguien comenta a su hermano sobre la vereda muy limpia de los vecinos, algo un poco inusual en ese lugar.

Sentence: Capaz a los vecinos les ______ (importar) mucho su vereda.

Summarizing the 24 scenarios, there were 12 certain and 12 not certain scenarios. Within each of those groups, there were four with future temporal reference, four with present temporal reference, and four with past temporal reference. Each scenario was used with capaz and quizás, resulting in 48 target items (Table 1).

Table 1.

Contexts considering temporal reference and certainty

3.2.2 Norming Procedure

Prior to the norming procedure, all items were examined by a native speaker of Argentine Spanish to verify their acceptability for the participant population. Then thirteen native Spanish speakers, different from the main participant group, participated in the norming procedure. The norming procedure was conducted to verify that the certain and not certain contexts were evaluated as such, preambles that provoked people to be more or less certain about the events expressed. The second purpose of the norming procedure was to test six preambles in each of the six blocks (i.e., certain-future, certain-present, certain-past, not certain–future, not certain–present, not certain–past), to identify the four preambles for each that provoked the most certainty or uncertainty about the events.

Six preambles were created for each of the six blocks (N = 36). The norming participants read each preamble, a related question, and responded to that question to indicate how certain they felt about the events in preamble. Example 6 demonstrates a certain-present preamble, the norming question that followed, and the response options where 1 (no sé) indicated the greatest uncertainty and 4 (sí) indicated the greatest certainty about the events.

(6) Preamble: Alguien nota que una persona que lleva bastantes bolsos puede necesitar ayuda, aunque parece hábil.

Norming question: ¿Esta persona necesita ayuda?

Response options: 1 = No sé 2 = Posiblemente 3 = Probablemente 4 = Sí

While there were four response options, the survey was conducted in Qualtrics, a common online survey tool, using a sliding scale so that participants could provide scalar numerical responses. This option allowed participants to more precisely choose the level of certainty.³ Participants viewed one question at a time, which avoided previous items affecting subsequent items, and the questions were randomized to avoid participant detection of patterns of certainty or temporal reference.

Based on the norming results (i.e., participants’ numerical ratings for all 36 preambles), the two preambles in each of the six blocks that least aligned with the certain or not certain intention were eliminated to reduce the overall length of the final questionnaire and to more accurately provide certain and not certain contexts. For example, within the certain-present block, the average ratings for the individual preambles ranged from 1.5 to 3.24 on the certainty scale, where ratings closer to 4 (sí) represented more certainty perceived by the participants. The two preambles with the lowest scores (i.e., least certain) were eliminated. This procedure was followed for all blocks.

Responding to the main purpose of the norming procedure, the numerical ratings of the remaining four preambles in each of the six blocks were analyzed with a univariate ANOVA, primarily to test whether the certain and not certain preambles were evaluated as significantly different from each other. The results indicated a significant main effect considering the certain and not certain contexts, F (1, 314) = 75.48, p < .001, meaning that participants were more certain about the events that occurred in the certain preambles (Mean = 3.0, SD = .90) and less certain about those in the not certain preambles (Mean = 2.10, SD = .95). Also, as expected, there was not a significant main effect when considering the effect of temporal reference on the certainty ratings of the preambles, F (2, 314) = .31, p = .73, (Future Mean = 2.54, SD = 1.04, Present Mean = 2.55, SD = 1.07, Past Mean = 2.63, SD = .97).

Qualifying the main effect, there was a significant interaction between the certain and not certain contexts and temporal reference, F (2, 314) = 4.58, p = .011. The interaction indicates that the certain and not certain contexts were more distinctly evaluated when they included present or future temporal reference than when they included past temporal reference. Yet, results from a posthoc pairwise comparison with a bonferroni adjustment, setting the alpha at .017, indicated that evaluations of the certain and not certain items were significantly different from one another with future, present, and past temporal reference, F (1, 84) = 49.54, p < .001; F(1, 82) = 7.69, p = .007; F (1, 81) = 41.215, p < .001, respectively (present-certain Mean = 3.1, present–not certain Mean = 2.0; future-certain Mean = 3.1, future–not certain Mean = 2.0; past–certain Mean = 2.8, past–not certain Mean = 2.4) (see Figure 1). As expected, the certain preambles were evaluated by participants as provoking more certainty and the not certain preambles were evaluated by participants as provoking less certainty, within all three temporal

Figure 1.

Results of norming evaluations considering certain and not certain items with three temporal references (*p < .017, ***p < .0003).

references. Considering the results of the norming study, 24 preambles were shown to accurately reflect the intended certain or not certain contexts and, therefore, to be appropriate for the investigation. Appendix A contains additional samples of the scenarios.

3.2.3 Item Distribution

Because the 24 scenarios served both for the capaz and quizás items (N = 48 target items) and because no participant was to see the same scenario more than once, two questionnaires were created. Each contained 24 items, half with quizás and half with capaz, with no repetition of scenario. The items in each had an even distribution considering certainty and temporal reference. Each questionnaire also contained 12 distractor items not only to help conceal the nature of the study but also to check whether a participant demonstrated prevailing mood selection patterns (e.g., optatives: quiero que 1 S and assertive statements: sé que 1 I) (see Appendix A for samples). To ensure equal response rate for each questionnaire, Qualtrics was set to alternate between each questionnaire. The items in the assigned questionnaire were randomized within Qualtrics for each participant.

3.3 Procedure

The first author emailed Argentine contacts and asked them to send an email with a description of the study and a link to the Qualtrics online questionnaire to Argentine people with little or no metalinguistic knowledge. Emails with links to online questionnaires are the most recommended types of electronic surveys, and snowball sampling, as in this case, has been shown to find similar participants eligible for the study (Dörnyei and Taguchi 2010).

Each participant responded to the demographic questions and one of the two questionnaires that contained target items and distractor questions in a randomized order. For the target and distractor questions, participants read the preamble and then completed the following sentence by supplying a conjugated verb of the infinitive that was indicated in parentheses. Participants were allowed to type the response and, in the instructions, they were encouraged to write their first idea. The survey took 20–30 minutes, following the survey length suggestions by Dörnyei and Taguchi (2010).

3.4 Approach to Questionnaire Analysis

Responses were coded as 1 (subjunctive) or 0 (indicative). Some responses required specific considerations. When responses contained a different verb than the one requested, additional information, or more than one response, they were not included in the analysis. Responses containing typos, such as in the use of comenza for comienza or an error in subject-verb agreement, were included, since the error would not likely impact mood selection. Finally, since the distractors had rigid mood selection parameters (e.g., optatives: quiero que 1 S and assertive statements: sé que 1 I), if a participant did not select the standard mood choice for a given distractor, that participant’s data was removed from analysis. Of the 2,808 responses, 236 (8.4%) were discarded from the analysis due to no response or for the reasons outlined above.

Once all data were coded, the percentage subjunctive use was calculated for each participant within each of the blocks varying in certainty, temporal reference, and adverb. For example, if a participant selected the subjunctive in three of the four certain-future sentences, the mood selection for this participant in this block was 75%. This measure was used to create a continuous measure of mood selection. These percentages for each participant were analyzed as the dependent variable in a mixed model analysis employing least square means, using Statistical Analysis Software (SAS). Certainty, temporal reference, and adverb were included as fixed effects within the model. Participant, hometown, sex, age, and education were included as random effects, since all potentially impact the linguistic variation under examination, yet the fixed effects were the focus of the study.

4. Results

The descriptive statistics related to mood selection considering certainty, temporal reference, and epistemic adverb are presented first. Then, the results of the mixed model are presented, in order to examine the impact of the fixed effects on mood selection within a more complex model.

4.1 Descriptive Statistics: Certainty, Temporal Reference, and Epistemic Adverb

Comparing the not certain and certain contexts, a greater use of the subjunctive with not certain contexts compared to certain contexts was expected. However, the subjunctive was selected only slightly more often in not certain contexts (Mean = 41.8%, SD = 35.0%) compared to certain contexts (Mean = 39.4%, SD = 36.4%)

As to temporal reference, greater subjunctive use was expected with non-past references. The current descriptive results demonstrate this expected pattern. The subjunctive was selected on average in 62.5% of the contexts with future temporal reference (SD = 34.8%) and 41.7% of the contexts with present temporal reference (SD = 31.7%). Highlighting the difference in mood selection in contexts with past temporal reference, the subjunctive was selected on average in 17.6% of the contexts with past reference (SD = 24.6%).

When analyzing the descriptive results comparing the contexts with capaz and those with quizás, the subjunctive was selected at a similar rate. In contexts with capaz, participants selected the subjunctive an average of 40.3% of the time (SD = 34.8%), and in contexts with quizás participants selected the subjunctive an average of 40.9% of the time (SD = 36.7%).

4.2 Mixed Model and Related Descriptive Statistics

The mixed model considered the fixed factors of certainty, temporal reference, and adverb in addition to the random factors listed in section 3.4. Considering the main effects, certainty was found not to be a significant factor, indicating that there were not significant differences between the not certain and certain contexts in terms of mood selection, F (1, 431) = 0.89, p = 0.18, r = 0.05. Regarding temporal reference, there were significant differences in mood selection among the three temporal references, F (2, 431) = 222.15, p < .0001. Based on a post-hoc analysis with Bonferroni adjustment the mood selection in contexts with future reference were significantly different from those with present and past reference, t (431) = 8.92, p < .0001, r = 0.39, t (431) = 21.04, p < .0001, r = 0.71 respectively. The mood selection in contexts with present reference was also statistically different from that with past reference t (431) =—11.08, p < .0001, r = 0.47. The effect sizes of these differences were large or very large, indicating their impact on mood selection. Finally, with respect to epistemic adverb, no significant differences in mood selection in sentences with capaz and quizás were found, F (1, 431) = 0.31, p = 0.72, r = 0.03. No interactions between fixed effects were significant.

Figure 2, in conjunction with Table 2, depicts the similar mood selection trends for capaz and quizás in not certain and certain contexts considering future, present, and past temporal references. The most impactful explanation for mood selection in sentences with epistemic adverbs capaz and quizás is the difference in temporal reference. The subjunctive mood is used significantly more often when the context refers to a future event, compared to contexts that refer to present or past events. Furthermore, the subjunctive mood is selected significantly more often in contexts with present reference than in those with past reference. While results for capaz and quizás were similar, the subjunctive was quite common in sentences with capaz when referring to future and to a lesser degree present events, challenging the notion that the epistemic adverb capaz is generally used with the indicative, as stated by Aleza Izquierdo (2010: 188). While no other results demonstrated a significant difference, Figure 2 showed a slight trend related to the interaction between adverb and certainty. Specifically, in not certain contexts, the subjunctive was selected more often in sentences with capaz than quizás. Conversely, in certain contexts, the subjunctive was selected more often in sentences with quizás.

Figure 2.

Percentage subjunctive use by adverb considering certainty and temporal reference.

Table 2.

Percentage subjunctive use by adverb considering certainty and temporal reference

It is worth noting that main effects indicated that mood selection was significantly different in contexts with future and present reference, future and past reference, and present and future reference (see Figure 2).

5. Discussion

In response to the research question “How does the mood selection pattern for capaz compare to that of quizás, considering both situational certainty and temporal reference?” we discuss the findings considering certainty, temporal reference, and then adverb. The hypothesis for certainty was that not certain contexts were expected to provoke the subjunctive mood more than certain contexts. This hypothesis was based on prior research on subjunctive selection (Lunn 1995; Sastre Ruano 1997; Wasa 2002). However, comparing certain and not certain contexts, the subjunctive was selected at almost the same rate, and consequently the certain and not certain contexts did not have a significant effect on mood selection. This unexpected result, in comparison to theoretical understandings of factors that impact mood, does not suggest that situational certainty affects mood selection in clauses with epistemic adverbs. Future research should continue to consider this question, addressing possible task effects and ways of investigating contextual certainty in naturally occurring conversations. Another consideration regarding this result is that a speaker’s certainty, and thus its potential effect on mood selection in epistemic adverbial structures, may be more entwined in another variable, such as temporal reference.

The hypothesis for temporal reference was that it would have a significant effect, and that past temporal reference would exhibit lower subjunctive selection rates than non-past temporal reference based on prior investigations (Finanger 2011; García 2011; King et al. 2008). As expected, this study supports the findings of those previous studies that temporal reference has a significant effect on mood selection. Furthermore, this study examined three temporal references (past, present, and future), where future temporal reference aligned with the greatest use of the subjunctive (Mean = 62%), present temporal reference was associated with less use (Mean = 42%), and past temporal reference was associated with the least use of the subjunctive (Mean = 18%). These findings align with past studies such as Finanger’s (2011) results for quizá(s), which found a 34% subjunctive selection with past temporal reference and 62% selection with non-past temporal reference and García’s (2011) results related to quizás that recorded multivariate analysis factor weights of 0.30, 0.60, and 0.64 for past, present, and future temporal references, respectively. In comparison with García’s (2011) findings, the current data show a greater distinction considering contexts with present and future temporal references. Thus, instead of distinguishing the effect of past and non-past contexts, this study confirms significant differences in the subjunctive use in contexts involving past, present, and future temporal references.

The current findings may be explained by relying on the theory that past temporal reference refers to events in worlds accessible to speakers, while present and future reference refer to events in worlds that are less accessible to speakers (Butt and Benjamin 2013). The indicative is more common in contexts with past temporal reference because the past worlds are more accessible, allowing speakers to be more certain about the events and both understand and communicate the great likelihood that that they occurred in the past. Future temporal reference allows for less accessibility and less certainty, resulting in a greater use of the subjunctive indicating the reduced likelihood of occurrence and reduced certainty associated with the event. With respect to the present tense, future analysis of the individual contexts will help to determine if some present tense contexts are more or less accessible to speakers, thereby impacting their mood selection accordingly. While not all aspects of mood selection are related to temporal reference and certainty, the findings provide new empirical information that mood selection is distinct in epistemic adverbial phrases with past, present, and future temporal references.

Finally, regarding the comparison between sentences with capaz and quizás, it was thought that different mood selection patterns may emerge. However, speakers selected the subjunctive in sentences with capaz at almost the same rate as in sentences with quizás (capaz Mean = 40%, SD = 35% versus quizás Mean = 41%, SD = 37%), showing no statistically significant main effect. Thus, the subjunctive selection rates for capaz and quizás, and thus their epistemic stances, seem to be comparable. This provides empirical evidence in support of prior literature that provides quizás as a synonym for capaz (Aleza Izquierdo 2009; Moreno Fernández 2010). This finding in isolation lends support to those who argue that capaz selects the indicative more than the subjunctive (Aleza Izquierdo 2010), but mood selection patterns with capaz vary considering temporal reference as described above. Thus, it would be inaccurate to describe capaz as selecting the indicative more than the subjunctive, since this is true according to the current data only in contexts with past and to a lesser degree with present temporal reference.

Considering all three dependent variables (adverb, certainty, and temporal reference) (see Figure 2), it is clear that temporal reference functions similarly with respect to mood selection in sentences with capaz and quizás. A slight interaction between adverb and certainty is observable, yet it was not statistically significant. Figure 2 shows that sentences with quizás provoke greater subjunctive use in not certain contexts and that sentences with capaz tend to be associated with greater subjunctive use in certain contexts. This interaction may indicate a slight difference between capaz and quizás, but additional data would be required to provide further evidence of this trend. Despite this trend, the main findings regarding the comparison of the subjunctive selection of capaz and quizás indicate that the two adverbs provoke very similar mood selections, indicating similar epistemic stances.

6. Conclusion

The main goal of this investigation of mood selection in epistemic adverbial constructions was to examine the use of capaz as an epistemic adverb as it compares to quizás. The dependent variables of certainty, temporal reference, and epistemic adverb were examined in a controlled experiment using a sentence completion task. While questionnaires are sometimes criticized for not generating authentic data through spontaneous speech, a potential limitation of this study, this study achieved what spontaneous contexts have trouble realizing, namely, controlled linguistic environments in which to test speakers’ natural linguistic tendencies. Specifically, studying certainty in naturally occurring contexts is difficult, if not impossible, as it is a cognitive variable (Czerwionka 2012), and so the normed contexts of certainty created for this experiment represent a solution that allowed for the examination of certainty. Furthermore, the current investigation improved upon previous questionnaire tasks that provided a forced choice between indicative and subjunctive forms (e.g., Studerus 1995), by using a sentence completion task that allowed participants increased freedom in their linguistic responses.

A main contribution of the current investigation was the empirical mood selection information associated with the epistemic adverb capaz, which was lacking in prior literature. The current data indicated that capaz was used with the subjunctive more often than was reported in previous literature. The detailed analysis of three dependent variables indicated that temporal reference was a significant factor impacting mood selection in sentences with capaz and quizás. Moreover, this study’s results found that subjunctive usage increases from past, to present, to future contexts, finding that speakers view the past as accessible, the present as less accessible, and the future as the least accessible, evidenced in mood selection patterns in sentences with epistemic adverbs capaz and quizás.

As the first study to compare mood selection associated with capaz to that of other epistemic adverbs in Argentine Spanish, capaz was found to function largely like quizás with respect to mood selection. The data in this study indicate that capaz and quizás likely share similar epistemic stances. Future research may consider additional data types to investigate the use of capaz, the comparison of capaz to other epistemic adverbs, and the use of capaz in other varieties of Spanish. Consideration of other social or linguistic contexts may also provide insight into the use of capaz as an epistemic adverb.