From demonstrative to filler: este in Amazonian Spanish and beyond

2.1 Demonstratives

One basic discourse function of demonstratives is to encode spatial deixis. That is, demonstratives are used to indicate the distance between speakers and a given entity. However, the emphasis of this study is not on the uses of demonstratives for reference to real or imagined entities or events present in space at the moment of utterance. Instead, this study focuses on the discourse functions of demonstratives. In discourse, demonstratives are referential expressions used primarily to coordinate the communicative interaction between interlocutors, rather than to indicate spatial distance (Himmelmann 1996). They make reference to individuals, events, facts or propositions. Demonstratives can also indicate a referent’s degree of accessibility and informational status (Ariel 1988; Givón 1983; Gundel et al. 1993). Another important function of demonstratives is to establish a joint focus of attention between interlocutors (Diessel 2006).

In discourse, demonstratives can hold the following functions: discourse deixis (6), referent tracking (7), identification (8), situational use (9), and recognitional use (10) (Diessel 1999; Himmelmann 1996).

A demonstrative is employed for discourse deixis when it refers to a proposition, as in (6), where este refers back to ‘she destroys her seat.’ The referent tracking function involves contrast to another participant, a similar referent or a shift in focus of attention. That is, “demonstratives are used for tracking only if other tracking devices fail” (Himmelmann 1996: 227). Example (7) illustrates this use, where there are two participants – a worker and his boss. Ambiguity resolution between the two motivates the employment of a proximal demonstrative pronoun. The identification function illustrated in (8) is typical of copular constructions. The situational function applies when the referent is present in the utterance context. For example, if the husband of a speaker is present during an interview, and she says (9) pointing to her husband. Finally, in the recognitional use the referent is to be identified via specific shared knowledge between the interlocutors. For example, in (10), both speaker and hearer must know a specific bird, its behavior, and what a palizada is to be able to identify the referent of este. These contexts invite problems in establishing reference which is often accompanied by signs of hesitation (Himmelmann 1996). The hypothesis here is that the recognitional use helps to explain the emergence of the filler function in PAS.

Spanish demonstratives are grouped into three categories: proximal, medial, and distal. Demonstratives also carry gender and number inflection, as seen in Table 1.

In canonical Spanish, the feminine and masculine forms can function as modifiers in adnominal constructions, as seen in (1), and pronominal forms in pronominal constructions, as seen in (3). In these constructions, feminine and masculine demonstratives reproduce the gender traits of their anaphors. Neuter demonstratives differ from feminine and masculine demonstratives in some respects. Unlike the feminine and masculine forms, neuter forms do not participate in adnominal constructions. Also, neuter demonstratives can refer to diverse notions, but generally inanimate entities, especially when the speaker is not sure of their identity (RAE 2009: 1308). The entity can be concrete or abstract. Finally, neuter forms are used to make reference to entire propositions. An important point to keep in mind is that non-neuter forms can refer to a person, an animal or a thing; this may have favored the selection of a non-neuter form for the filler function (see discussion in Section 5).

However, the frequency of use of the demonstratives listed in Table 1 varies according to form, region, and language modality (oral versus written). Figure 1, from Shin and Vallejos-Yopán (2023) presents the frequency of each singular demonstrative, normalized for every million words (cpm) in the Corpus of XXI Century Spanish (CORPES).^[1] Note that the distal demonstratives aquel, aquella, aquello were combined given their overall low frequency.^[2] This finding has led some scholars to propose that the system might no longer be tripartite but bipartite with respect to spatial distance. For example, in oral data from Spain, Zulaica Hernández (2012) found that aquel represented only 4% of all the demonstratives, while este and ese represented 57% and 38%, respectively.

Figure 1:

Frequency of demonstratives in CORPES (Shin and Vallejos-Yopán 2023: 249).

In general, the use of demonstratives is more frequent in oral data than in written data, with the exception of distal ones, which are more frequent in written data. This marked difference in the use of demonstratives in the modalities may be due to several factors. In communicative interaction, the range of functions of demonstratives is broader. Beyond the prototypical functions such as anaphoric and cataphoric deixis, demonstratives are frequently used to make exophoric reference to elements in the extralinguistic context, such as pointers to attract the interlocutor’s attention to a referent, and as will be discussed in Section 3, as placeholders or hesitators. The distribution of forms in Figure 1 also shows some differences between countries. Note, for example, the minimal use of distal forms in Peruvian varieties (59 examples, 0.4% of all demonstratives in oral data from Peru). Another notable detail of relevance to this paper is the frequent use of este in the oral data of Latin American countries compared to those of Spain. I return to this point in Section 4.1.

2.2 Fillers

Crosslinguistically, spoken language is often accompanied by intra-turn pauses, sound stretches, false starts, repaired utterances, repetitions, self-addressed questions, fillers, etc. The use of fillers is motivated by constraints in cognitive processes, such as difficulty in remembering an appropriate lexical item or articulating ideas during the course of utterance production (Hayashi and Yoon 2006). Given that patterns of conversations are highly conventionalized, strategies employed in the context of conversational interaction arise from shared conventions of language use (Bybee 2010; Schegloff 1979; Schegloff et al. 1977). In that sense, the use of fillers represents a conventionalized solution to the preference for progressivity in real-time human interaction (Fox 2010; Heritage 2007; Jasperson 2002). Fillers are common and frequent enough in real-time communication, and therefore they should receive analytic attention in their own. The present study is concerned with two types of fillers: placeholders and hesitators.

3.1 Local context

The Spanish variety explored here is spoken in Loreto, an Amazonian region of Peru. This region is home to the highest indigenous population in Peru (Instituto Nacional de Estadística e Informática INEI 2017) and the highest linguistic diversity (∼30 languages, 8 families) in the country. However, the speakers of all these different languages combined make up only 3% of Peru’s total population. Given the linguistic and cultural diversity of the region, place of origin is an important variable in any study of Amazonian Spanish (Vallejos 2014; Vallejos et al. 2020). Broad categories such as “rural” and “urban” are not necessarily relevant to study the Spanish spoken in the region since each community has a unique shared history and specific sociolinguistic profile (Vallejos-Yopán 2022). Thus, attempting to make generalizations about “rural” or “indigenous” villages would be unwise. In the present study, I only analyze linguistic phenomena in the Spanish of monolinguals in Kukama villages, which share a distinctive profile (see Section 3.2), and then compare these findings to what I find in the city of Iquitos. Iquitos is the biggest city in the Peruvian Amazon, which attracts migrants from different ethnic groups.

To the best of my knowledge, there are no previous studies of demonstratives in this variety of Spanish, besides my own work. Vallejos (2017) looked at the overall distribution of demonstratives in spoken PAS called Conversaciones en Loreto. An examination of a total of 1,390 demonstratives in a sample of natural conversations found zero tokens of aquel, aquella, aquello. This finding suggests that the demonstrative system in PAS might be different than canonical Spanish. As seen in Table 1, above, the canonical system of Spanish demonstratives is characterized as tripartite with respect to spatial distance (proximal, medial and distal); in PAS, however, distal forms are no longer used.

Another finding reported in Vallejos (2017) is that within the proximal category (este, esta, and esto, n = 415), there was an extremely high number of masculine tokens (n = 369/415) compared to their feminine (n = 30/415) and neuter (n = 16/415) counterparts. However, only 17% of este tokens (64/369) were used as demonstratives, 83% of them (305/369) were used as fillers, which is almost a 5:1 ratio in favor of the filler over the demonstrative uses. It was also noticed that este in the filler function seems to display phonetic and prosodic properties that sets it apart from este in the demonstrative function. Building on these previous findings, this study focuses solely in the demonstrative este.

In the next subsection, the focus shifts to the participants, the data and the coding system employed in this study.

3.2 Participants

To explore the research questions stated above, the speech of eight monolingual PAS speakers was examined. The data comes from my corpus of Amazonian Spanish, which at present consists of about 40 h of unscripted interviews collected from a total of 52 speakers. The interviews selected for this study were conducted between 2011 and 2015. The first topic of the conversation was introduced by the interviewer (e.g., main events taking place, their high school/college experience, preferred activities, etc.), after which the participants took the lead for the most part. The total duration of the selected interviews ranges from 50 to 65 min. The interviews took place in rural villages and the city of Iquitos (see more details, below). The sound was collected using a solid field recorder (Marantz PMD660) and a condenser omnidirectional lavalier microphone (AT803). Interviews were transcribed into intonation units using the Transcriber software.

For the present study, I have constructed a balanced sample in terms of place of origin, gender, age, and occupation of the interviewees.^[3] The demographic background of the speakers is shown in Table 2.

The sample is balanced in terms of place (village = 4, city = 4), and gender (female = 4, male = 4). The age of the speakers ranges from 21 to 43.^[4] The participants have various occupations, and in terms of formal education, all the speakers have had access to more than 10 years of schooling.

The first group of speakers are from three Kukama villages located along the Samiria and Marañón River, around 20 h away from Iquitos traveling by boat. In these villages, community members shifted to Spanish about five decades ago. Some participants in this study know Kukama words and phrases; nonetheless, they do not self-identity as bilingual. The Kukama villages in this study have access to radio, but very limited access to television or the internet. The villages offer primary education (K–6) and secondary education (7–11). The Kukamas travel regularly from villages located along navigable rivers to other cities in the vicinity.

The second group of speakers are from Iquitos. Iquitos is the biggest city in the Amazon of Peru with a total population of 160,497 (INEI 2017). The city is surrounded by suburban districts such as Belen, Punchana, and San Juan.^[5] In Iquitos, 85% of the population declared to have learned to speak in Spanish (INEI 2017), and although Iquitos attracts immigrants from rural villages, only 0.2% declared knowing another local language.^[6] Usually, young people from rural villages come to work and study in Iquitos. Data from Kukama villages will henceforth be referred to as Village PAS and Iquitos as City PAS.

3.3 Data and coding

All instances of este were manually extracted from the corpus of unscripted interviews which resulted in a total of 219 tokens. The tokens were coded by the constructions in which they appear and their position within the intonation units. A phonetic analysis of each token was also performed. The details of the coding are explained below. The data for this study is available in Open Science Framework (OSF), including values and z-scores at https://doi.org/10.17605/OSF.IO/DVAQX.

4.1 What is the distribution of este in terms of its main morphosyntactic functions?

The main finding regarding the functions of este is that this form is mostly used as filler. As shown in Figure 2, only 23% (50/219) of the tokens are used as proximal demonstratives, 28% (62/219) as placeholders, and 49% (107/219) as hesitators. In other words, the placeholder and hesitator functions account for 77% of the tokens. Hence, the primary function of este in PAS is not that of a demonstrative. A closer look at the demonstrative category reveals that most of the forms are in prenominal position, none of them occur as postnominal, and there are only 2 tokens in pronominal function.

Figure 2:

Distribution of este by function and place.

4.2 Do the different functions of este exhibit phonetic correlates?

To answer this question, I implemented three measurements: word length, consonant cluster length, and vowel duration. Recall that the values have been normalized, the box plots below represent z-scores. For the sake of completeness, I also provide the mean values in milliseconds after the z-scores.

The results for word length are given in Figure 3. Demonstratives are significantly shorter (z-score −0.66, 232 ms) than placeholders (z-score 0.26, 445 ms).^[13] However, hesitators are not significantly different from placeholders (z-score 0.16, 420 ms).^[14] The results for consonant clusters, given in Figure 4, point in the same direction. The /st/ cluster in demonstratives is significantly shorter (z-score −0.50, 124 ms) than in placeholders (z-score 0.21, 171 ms),^[15] but there are not differences between /st/ in placeholders and hesitators (z-score 0.10, 167 ms).^[16]

Figure 3:

Word length by function.

Figure 4:

/st/ length by function.

The duration of the vowels shows an interesting pattern, these are given in Figure 5. There are clear differences between demonstratives, on the one hand, and placeholders and hesitators, on the other hand. In the demonstrative category, the first vowel (z-score −0.29, 75 ms) is longer than the second vowel (z-score −0.74, 40 ms). In contrast, in the placeholder category, the first vowel (z-score 0.08, 83 ms) is shorter than the second vowel (z-score 0.19, 139 ms). The same configuration emerges in the hesitator category, the first vowel (z-score 0.09, 87 ms) is shorter than the second vowel (z-score 0.07, 143 ms). In addition, when we compare demonstratives with placeholders, the only significant difference between them is the second vowel. In other words, there is a significant difference between demonstratives and placeholders with respect to the second vowel,^[17] but not with respect to the first vowel.^[18] Within the filler category, there are not significant phonetic differences between placeholders and hesitators, neither with respect to the first vowel^[19] or the second vowel.^[20] In sum, the duration of the consonant cluster and the second vowel are the informative elements to distinguish demonstratives from fillers.

Figure 5:

Vowel duration by function.

A question to consider is whether the duration of the vowels in the demonstrative este is comparable with those in any other two-syllable word with penultimate stress in PAS. The mean values of the vowel in the demonstrative este display a similar pattern to those of two-syllable words in PAS (Koops and Vallejos 2018). Figure 6 highlights the similarities between PAS words, in the first two columns, and the demonstrative este, as well as the differences when compared with placeholders and hesitators.

Figure 6:

‘este’ compared to other two-syllable words in PAS.

4.3 What is the distribution of este within intonation units?

There are some differences regarding the distribution of este within intonation units. The demonstrative este occurs primarily in medial position (74%, 37/50), and only rarely in final position (6%, 3/50). In contrast, the majority of the placeholders (52%, 32/62) and hesitators (54%, 59/109) appear in final position. This is summarized in Figure 7.

Figure 7:

Position of este in intonation units.

4.4 Summary of findings

A corpus study shows that este has adopted two main roles in PAS: demonstrative and filler. Within the demonstrative category, este appears mostly as an adnominal modifier in prenominal position, and rarely as pronominal. Within the filler category, este is used as both placeholder and hesitator. Quantification of the distribution of usage of este reveals that its primary function is that of a filler. In fact, both placeholders and hesitators comprise 77% of the total number of tokens. A closer examination at their syntactic distribution and phonetic shape provides a fuller picture of the multiple roles of este in today’s PAS. These features are summarized in Table 4. For duration, I provide the z-scores followed by the mean value in milliseconds.

Regarding their syntactic distribution, demonstratives and placeholders follow certain constraints. Demonstratives occupy an adnominal modifier position within a noun phrase or the slot of a noun or a noun phrase within a clause. Placeholders parallel the distribution of demonstratives in that they also occupy the slot of a noun while the speaker engages in word searches. In contrast, hesitators have distributional freedom. Likewise, demonstratives and placeholders are referential, they point to particular entities the speaker has in mind; hesitators are not referential. Demonstratives, placeholders and hesitators differ regarding the kinds of jobs they do, demonstratives are used mainly for anaphora, but also for cataphora. Placeholders are used to buy time while one comes up with a word, hesitators are used to deal with difficulties progressing in real-time interaction. As for their position within intonation units, demonstratives appear typically in medial position (74%), while placeholders and hesitators tend to occur in final position. While I did not code for pauses, the propensity for placeholders and hesitators to occur towards the end of intonation units implies they are frequently, but not always, followed by pauses.

Demonstratives also differ from fillers (placeholders and hesitators) in terms of their phonetic shape. Demonstratives are overall shorter than fillers, mainly due to the duration of the consonant cluster and the second vowel. The phonetic analysis shows that the filler este is characterized by a clear shape when compared to the demonstrative este, from which it evolved. In the demonstrative este, the first vowel of the word, the stressed one, is longer than the second vowel, the unstressed one. In contrast, the highly frequent filler is not only longer in duration overall but shows the opposite pattern with respect to vowel duration than its predecessor: the first vowel is shorter than the second vowel. Given that vowel length is a cue of stress in Spanish, this finding may have an effect in the perception of stress, or lack thereof, in filler este.