1 Introduction

Demonstratives are high-frequency linguistic expressions whose referential meaning depends on the context of production (Cornish 2011; Diessel 2012). Recent studies on the processes underlying demonstratives’ use have shown that speakers select demonstratives according to a number of different perceptual, conceptual and interactional parameters such as space and action (Coventry et al. 2023); referent-intrinsic factors (Peeters et al. 2020); semantic features, (Rocca & Wallentin 2020); speaker-intrinsic factors, such as bilingualism (Lease & Shin 2025; Rubio-Fernandez 2022), mood (Kruse et al. 2024; 2025), interlocutor position (Rubio-Fernandez 2022), attention (Jara-Ettinger & Rubio-Fernandez 2024) and interaction (Rocca et al. 2019b). Collectively, we may think of these parameters as reflecting physical, social and conceptual/psychological distance.

Physical distance refers to the perceptual parameters that locate the referent in space (and time) in relation to the speaker, or the deictic centre (Bühler 1934/2011). Examples of physical distance include space and action parameters (Coventry et al. 2023), such as reachability, visibility and size in a concrete context, as seen from the speaker’s perspective, as well as the physical position of the listener during interaction (Coventry et al. 2008; 2023). Social distance, on the other hand, takes into account the social context of the communicative situation. Deictic expressions primarily serve to indicate the location of a referent to interlocutors and shift their attention onto the referent (see also Diessel 2006; Jara-Ettinger & Rubio-Fernández 2024) on a perceptual level. Social distance, thus, takes into account the listener’s attention, knowledge and collaborative function during interaction (Rocca et al. 2019b; Shin et al. 2020; 2024; Rubio-Fernández 2022) and integrates socially relevant information, such as referent ownership (Coventry et al. 2014). Conceptual/psychological distance goes beyond referencing objects in physical space and relates to the relationship between the speaker and the referent, whether it is physically present or only present in discourse. This type of distance may rely on referent-intrinsic factors (Peeters et al. 2020), such as the semantic features of the referent (Rocca & Wallentin 2020), e.g. the emotional connotations of a word (valence) in combination with the personal experiences of the speaker, e.g. bilingualism and language dominance (Vulchanova et al. 2020; Rubio-Fernández 2022; Todisco et al. 2024) and personality, e.g. mood (Kruse et al. 2024; 2025). All things considered, saying that fiancé may refer to a person who is physically distant, but it may also be someone else’s fiancé, or someone whom the speaker wishes was distant. In the current study, we focus on conceptual/psychological distance.

To date, much of the research on the use of demonstratives has focused on how salient physical and social aspects of the communicative event elicit the choice of one demonstrative form over another. When analyzing the degree of physical distance, visibility, shared attention toward the referent or ownership (Coventry et al. 2008; 2014; 2023; Peeters et al. 2020; Shin et al. 2020; 2024; Rubio-Fernandez 2022), the overall results show a strong relationship between the use of proximal demonstrative terms (i.e. this) and referents at a closer physical distance, visible, present in shared attention and owned by the speaker. The relationship is independent of the number of demonstrative terms the language might have (i.e. two terms like English this/that or three terms like Spanish este/ese/aquel). The position of the listener has also often been analyzed to assess whether a change in the position and physical distance between the listener and the referent affects the speaker’s choice of demonstrative during interaction (Jungbluth 2003; Coventry et al. 2008; 2023; Rubio-Fernandez 2022). The general assumption is that when speaker and listener share the same physical distance to the intended referent, the speaker chooses the demonstrative form depending on the distance to the referent; whereas when the hearer assumes a different position from the speaker, the speaker might use a different form depending on the distance from the hearer to the referent (cf. Diessel 1999 for more details). The effect of the position of the hearer is more evident in some three-term languages, such as European Spanish (este/ese/aquel; ‘this/that/that yonder’) or Japanese (kono/sono/ano; ‘this/that/that yonder’), where the medial terms ese or sono, respectively, are often used for referents close to the hearer and far from the speaker (Diessel 1999; Rubio-Fernández 2022). Notwithstanding, evidence of the effect of this parameter has also been found in languages with two-term demonstrative systems (Rocca et al. 2019a; 2019b; Shin & Mendieta-Rodríguez, 2024). Studies by Rocca and colleagues (2019b) and Shin and Mendieta-Rodríguez (2024) found that – in two-term systems such as English – the proximal form this is also used in functionally relevant situations to redirect listener attention toward a referent that is closer to the listener. These results seem to be in line with what Jara-Ettinger and Rubio-Fernandez (2024) found comparing demonstratives as attention tools in ten languages. More specifically, Jara-Ettinger and Rubio-Fernandez found that people used polysemous demonstrative expressions (such as this one) in two-term demonstrative systems, both to redirect the attention of the listener toward a referent and to signal when joint attention was successfully established (2024: 7).

Although physical and social parameters seem to be the most intuitive factors to drive demonstrative choice, conceptual and psychological factors have been shown to play a role in demonstrative use, especially when no interactive context is provided. As demonstrated in a series of studies using the Demonstrative Choice Task paradigm (Rocca et al. 2019a; Rocca & Wallentin 2020), saying this shoe when no shoes are present, may indicate that the shoe is experienced as conceptually close to the imaginary self/body (Rocca & Wallentin 2020; Todisco et al. 2021; 2022). By the same token, saying that tiger in the absence of a relevant context may be used to indicate that it is not the typical pet we imagine waiting for us at home after work. In accordance with Peeters et al. (2020), the above-mentioned studies found evidence that intrinsic semantic features of a referent affect the choice of one demonstrative form over another. More specifically, features conveying small manipulable, inanimate, self/body-related information about the intended referent foster the use of proximal demonstratives (i.e. this apple or this hand); while big, animate or danger-related information foster the use of medial and/or distal demonstrative forms (i.e. that tiger or that hurricane). These results are cross-linguistically consistent, independent of the number terms in the demonstrative system (see Todisco et al. 2021; 2022 for more details).¹

To sum up, the literature indicates that demonstratives act like modifiers of the intended referent, by providing additional information to a shared or contextual knowledge. When more than one option is equally grammatical, opting for one form or another gives us access to additional cues not only about the referent, but also about the perspective the speaker has on it and him/herself (Kruse et al. 2024; 2025). These findings relate to which demonstrative is used, but an additional question is where the demonstrative is used in the noun phrase (henceforth NP). Some languages allow prenominal and postnominal syntactical structuring of the demonstrative term in the NP, but how might this choice be related to the different linguistic and interactional features of discourse?

European Spanish has a three-term demonstrative system which allows two positions of the demonstrative terms in the NP. Adnominal forms can either appear in a prenominal position, as in este hombre ‘this man’, or in a postnominal position, as in el hombre este ‘this bad man’. Gómez Sánchez and Jungbluth (2015) argue that the two constructions convey different meanings. The prenominal demonstrative option (i.e. este hombre) is said to indicate a person close in perceptual or psychological space. The postnominal construction (i.e. el hombre este) is argued to add additional information at the conceptual level, by attributing a derogatory valence to the referent during conversation (see Ranson 1999; Alexander 2008; Shin & Vallejos Yopán 2023 for a review). In addition, pre- and postnominal demonstrative constructions show differences when used anaphorically. According to a series of corpus studies, the prenominal demonstrative construction is used for direct anaphora (i.e. with a referent overtly expressed in the antecedent production); whereas postnominal demonstrative constructions are used more in cases of indirect anaphora, that is when the referent is not present in the interactional scenario and the interlocutors need to recover it from extralinguistic context, shared or common knowledge (Alexander 2008; Ranson 1999; Zulaica Hernández 2017).

Although the postnominal use of demonstratives in European Spanish is quite common in conversation, no study – to our knowledge – has investigated whether the position of the demonstrative in the NP might depend on the semantic features of the referent, in the absence of additional context. In the present work we aim at filling this gap. Using the DCT paradigm, we presented participants with a set of 150 nouns, distributed across 12 different semantic factors, to investigate whether the intrinsic semantics of the nouns influence the choice of demonstrative position in the NP. We hypothesised that nouns characterised by a negative emotional valence (i.e. terrorist) would be more likely to elicit a postnominal construction, due to the observed link to negative/derogatory attitude in use. We also hypothesised that this would happen in addition to the previously observed preponderance for using medial/distal demonstrative forms in combinations with nouns with negatively charged meaning. To better interpret our findings, we interviewed a group of European Spanish native speakers and participants to compare our results with their reported use of demonstrative forms and positions.

2 The derogatory connotation of demonstrative constructions

The choice of one demonstrative over another may be used to convey positive or negative connotations about the referent. An example is the use of iste in Latin (Todisco et al. 2022). This demonstrative was used to refer to referents at medial distance and/or proximal to the listener. It went on to convey a derogatory stance on the part of the speaker, based on this. This stems from judicial rhetoric where iste referred to the adverse party in court, i.e. iste homō tended to mean ‘this/that scoundrel’ (see Jacquesson 2015). Recent studies on Catalan and Italian, two languages characterised by a reduction from a three-term to two-term demonstrative system, have shown that they used their medial terms not only to convey medial spatial information, but also specific semantic features, such as negative moral value, or more generally negative emotional valence (Ramat 2015; cf. Todisco et al. 2021; 2022). Conveying negative valence seems to be an intermediate stage before shifting to the proximal position (iste in Latin replaced hic and developed into the Spanish este) or disappearing from the demonstrative system (the Italian medial term codesto, and the Catalan aqueix). It is an open question how the prenominal and postnominal constructions fit into this picture. The postnominal demonstrative construction is found in many languages such as European Spanish (Gómez Sánchez & Jungbluth 2015), Egyptian, Moroccan and Syrian (Brustad 2000), Irish Gaelic (Kane 2015), Russian (Grenoble & Riley 1996) and Romanian (Manoliu 1999), either as marked or unmarked structures.²

In European Spanish, the syntactic position of the demonstrative provides additional information about the intended referent. As mentioned previously, the change from prenominal to postnominal seems to be accompanied by a change in the semantic connotation of the referent, as the examples (1) and (2) show:

It should be noted that the same semantic change that European Spanish achieves with the pre- and postnominal positioning of demonstratives is also conveyed by reinforced structures in other languages, such as English or standard Italian. Standard Italian, like English, presents a reinforced structure composed of a prenominal demonstrative term and a postnominal adverb, such as shown in the following example (3):

This construction generally conveys a higher level of specificity related to the physical location of the referent.³ Da Milano (2015: 61) defines this reinforced or double structure of standard Italian as an emotive, symbolic closeness of distance formula. Its use can either be emphatic – to emphasise the identity of the intended referent – or express derogatory meaning, as in lo ha detto quello là (‘that guy said it’, using a distancing demonstrative distal form).

Studying the use of demonstratives in prenominal and postnominal constructions might contribute to a fine-grained understanding of the conceptual parameters that guide the speaker’s production. Following Zulaica Hernández (2017), we believe that postnominal demonstrative constructions can be used to express a range of subsidiary values, such as the speaker’s degree of involvement, derogatory overtone, as well as evocative information. To contribute to this debate, we investigate the demonstrative prenominal and postnominal choices – in absence of an interactive context – in two groups of European Spanish speakers using the DCT paradigm (Rocca & Wallentin 2020; Todisco et al. 2021) to see whether the range of subsidiary values attributed to demonstrative prenominal vs. postnominal position might be explained as a function of the semantic factors characterising the intended referent.

3 The present study

The present investigation focuses on the effect of semantic factors on prenominal vs. postnominal use of demonstratives in European Spanish. Here we report two studies with two different groups. Study I focuses on a sample of European Spanish monolinguals; whereas Study II focuses on a sample of European Spanish/Catalan simultaneous bilinguals.

We employed the Demonstrative Choice Task (DCT), developed by Rocca and Wallentin (2020). This is an online questionnaire in which participants are asked to match a given word (i.e. a noun) with one demonstrative form (e.g. the choice between this shoe or that shoe), based on their first intuition. The noun is presented without any additional context, meaning the participant must construct the necessary information to match the noun and demonstrative by building on their own internal context. The paradigm involves a simple choice, which has the benefit of avoiding reliance on a subjective acceptability scale (such scales are difficult to calibrate). It can be used to compare equally acceptable constructions (as in the example above) and thus be used to study subjective preference rather than experienced grammaticality. This forced-choice method, however, where all possible constructions are presented simultaneously, may cause participants to choose rare constructions more often than they would have otherwise due to demand characteristics, and thus overestimate the frequency of use. However, it should not influence the relative frequency of choices across words and semantic features. As such, the pattern of the choices may be more informative than the frequencies of the choices themselves. Previous studies on English, Danish, German, Italian, Spanish, Catalan and Tagalog Filipino (Rocca et al. 2019b; Todisco et al. 2021; 2022 Kruse et al. 2025) have found that speakers of different languages are affected by similar semantic factors when matching nouns and demonstrative forms. This suggests that, even when the context is not provided, participants always recreate a context based on personal and shared knowledge about the referent.

The aim of the present investigation is twofold. First, (a) we want to replicate the observed semantic features contributing to demonstrative choices in a shortened version of the DCT, allowing both prenominal and postnominal demonstrative constructions. Second, (b) we want to assess the effect of semantic factors influencing the postnominal use of demonstratives in the NP. In a previous study run on European Spanish using only demonstratives in the prenominal position (Todisco et al. 2021), we showed that referents with meanings related to Time, negative Valence and Loudness elicited an increased use of medial (ese, ‘that’) and distal (aquel, ‘that yonder’) demonstrative forms (i.e. aquella noche, ‘that night’; ese criminal, ‘that criminal’). We interpreted these choices as a linguistic distancing strategy that conveys information about the conceptual distance the speaker establishes with the referent (see above). These things considered, we hypothesised that if the postnominal demonstrative form conveys a derogatory meaning (Lamíquiz 1967; Hottenroth 1982; Alexander 2008; Gómez Sánchez & Jungbluth 2015; Zulaica Hernández 2017), then some intrinsic semantic factors of the noun might increase the postnominal choices. More specifically, we also hypothesised that the noun characterised by a negative Valence factor (i.e. terrorist or gunshot) might elicit more postnominal occurrences to indicate increased linguistic distancing. Among the three demonstrative forms este ‘this’, ese ‘that’ and aquel ‘that yonder’, we did not expect to find many occurrences of the distal postnominal aquel ‘that yonder’, because of its (apparent) marked status in the postnominal use of demonstrative system. As Ranson (1999: 124) highlights, negative connotations can be expressed by postponing demonstrative forms. However, all of the examples collected in her work refer to proximal and medial demonstrative forms with no mention of the distal form aquel ‘that yonder’. This information is corroborated by additional studies. According to the exhaustive analysis conducted by Zulaica Hernández (2017), postnominal demonstrative constructions can express the speaker’s degree of involvement in the utterance situation (in combination with the proximal este), derogatory overtone (in combination with medial ese) or past tense (in combination with distal aquel). Gómez-Sánchez and Jungbluth (2015: 243) emphasise that the act of distancing conveyed by postposed demonstrative forms influences the selection of the medial-distance term ese ‘that’, which, in these contexts, conveys the speaker’s feelings of disgust and reluctance. Moreover, in a more recent work by Shin and Vallejos Yopán (2023: 433), there is a reference to the negative value of non-proximal demonstrative forms, with the specific mentioning of the medial form ese in postnominal position (i.e. el hombre ese, ‘that bad man’). Once again, there is no mention of the distal demonstrative form aquel ‘that yonder’ in postnominal constructions referring to negative connotations.

To summarize, we expected to (a) replicate the pattern of choices of demonstratives, given semantic factors, independently of their position in the NP. Although the prenominal demonstrative form is the most frequently used (and thus unmarked) structure, we expected to (b) find an increased use of postnominal demonstrative forms for nouns characterised by a negative valence. Furthermore, to provide a more fine-grained assessment of the demonstrative in use, we expected that only the proximal este ‘this’ and medial ese ‘that’ demonstrative forms would be used in the postnominal constructions.

3.1 Study I

Study I focuses on the prenominal vs. postnominal use of demonstrative in a sample of European Spanish monolingual speakers.

3.1.1 Materials and Methods

Participants. Participants were recruited online. The questionnaire was distributed through word-of-mouth strategies and presented at different universities and Escuela Oficial de Idiomas (Language Schools), mainly based in the Community of Aragón (Huesca) and Andalusia (Seville). The initial sample consisted of 146 participants. Based on the information collected in the ethnolinguistic questionnaire at the end of the study, 40 participants were excluded from the study for the following reasons: 6 participants were excluded because they were not native speakers of European Spanish, 25 participants were excluded because of simultaneous bilingualism, and 9 participants were excluded because they did not finish the task. Thus, the final sample was composed of 106 adult monolingual speakers of European Spanish (26 male, 4 other, and 76 female; Mean_age 23.61; SD 14.07 and an age range from 18 to 60). All participants read and accepted the consent form for data treatment and analysis before the start of the experiment.

Materials. 150 words were selected from the DCT dataset reported by Rocca and Wallentin (2020) and Todisco et al. (2021; 2022) for English, European Spanish and Catalan. The DCT dataset incorporates the 65 semantic dimensions outlined by Binder and colleagues (2016) alongside the 11 features of the Lancaster Sensorimotor Norms (Lynott et al. 2020). Rocca and Wallentin (2020) reduced the 76 semantic features to 12 semantic factors via a subsequent factor analysis. These factors cover a variety of semantic domains thought to be relevant for human cognition. The 12 factors were named: Vision, Valence, Loudness, Human, Taste/Smell, Motion, Manipulability, Scene, Time, Torso/Legs, Arousal and Self/body.⁴ For each factor, we selected 20 nouns representing the 10 highest and 10 lowest loadings on each factor among the dataset in Rocca and Wallentin (2020). Since the hypothesised most important variance to be measured was Valence, we selected 40 nouns from this factor, thus ending with 20 nouns with high (negative) and 20 with low (positive) loadings on the Valence factor. In selecting the top 20 words per factor, the maximum value of the word was taken into account. The rationale behind this decision was to compare the positive and negative values of each factor, as well as the positive/negative of the factor Valence, in order to gain a more reliable comparison and analysis. Although nouns were chosen, based on their scores on particular semantic factors, all nouns in addition had scores (usually more average) on all the other factors. All scores from all nouns and factors went into the statistical analyses described below. The total sample of nouns included masculine (N = 73; e.g. ron, ‘rum’), feminine (N = 66; e.g. casa, ‘house’), common gender (N = 5, e.g. criminal, ‘criminal’) and forms where both genders were specified (N = 6, e.g. actor/actiz, ‘actor/actress’). All nouns were accompanied by morphologically aligned adnominal demonstratives in both prenominal and postnominal position. The grammatical features of the nouns were in accordance with the linguistic regulation of the Real Spanish Academy (Real Academia Española: https://www.rae.es). A sample of the nouns used is provided in Table 1.⁵

Table 1: A sample of words from Study One.

English version	Spanish version	Semantic factor	English version	Spanish version	Semantic factor
terrorist	terrorista	N_Valence	handshake	apretón de manos	Manipulability
torment	tormento	N_Valence	keyboard	teclado	Manipulability
criminal	criminal	N_Valence	pen	bolígrafo	Manipulability
applause	aplauso	P_Valence	hospital	hospital	Scene
gratitude	gratitud	P_Valence	farm	granja	Scene
joy	alegría	P_Valence	airport	aeropuerto	Scene
goldfish	carpa dorada	Vision	night	noche	Time
elephant	elefante	Vision	evening	tarde	Time
butterfly	mariposa	Vision	day	mañana	Time
bagpipe	gaita	Loudness	cold	frío	Torso/leg
tuba	tuba	Loudness	hygiene	higiene	Torso/leg
trumpet	trompeta	Loudness	worker	trabajador/a	Torso/leg
actor	actor/actriz	Human	rose	rosa	Arousal
politician	político/a	Human	cash	metálico	Arousal
woman	mujer	Human	flower	flor	Arousal
beer	cerveza	Taste/Smell	mouth	boca	Self/body
rum	ron	Taste/Smell	nose	nariz	Self/body
chocolate	chocolate	Taste/Smell	leg	pierna	Self/body
stampede	estampida	Motion
ricochet	rebote	Motion
train	tren	Motion

• Note. The table shows a sample of the nouns used in the questionnaire. Here we report the original English version from Rocca and Wallentin (2020), followed by its Spanish translation from Todisco et al. (2021) and semantic factor associated.

The questionnaire was distributed on the platform Qualtrics, which allowed both computer and mobile device responses (Figure 1). In both versions, the questionnaire was divided into three parts. The first part provided the participants with all the information necessary to give consent to participate and complete the task. In the second part, they were asked to match each noun that appeared on the screen with one of the six demonstrative options provided below, according to their first intuition. The 150 nouns were presented with the six demonstrative options in a randomised order – both for the nouns and for the demonstratives.

Figure 1: Experimental design providing an example of the mobile version of the DCT.

The third part of the experiment introduced a series of questions about age, gender, native language and potential foreign language knowledge. This information enabled us to limit our sample to European Spanish native speakers, excluding simultaneous bilinguals, non-European Spanish speakers, and second language learners of European Spanish. The total duration of the test was approximately 20 minutes. All procedures contributing to this work complied with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008.

3.1.2 Results

Analyses were conducted with R Studio (R Core Team, 2020). Data and analyses are publicly available at https://doi.org/10.17605/OSF.IO/97YMB. Participants chose the prenominal position for 81.2% of the trials and the postnominal position for 18.8%. The proximal demonstrative was chosen 35.1% of trials while the medial was chosen 35.7% and the distal 29.2% (Figure 2).

Figure 2: Participants made choices to couple nouns with a combination of demonstrative and pre/postnominal position. Choice distributions for the set of nouns differed from random choices (dashed line). Analysis of variance revealed a main effect of position, a main effect of demonstrative and an interaction between position and demonstrative. Follow-up t-tests indicated that prenominal position was chosen most. The main effect of demonstrative reflected that distal demonstrative was chosen less often than the proximal demonstrative. The interaction reflected that the difference in frequency for pre- and postnominal choices was larger for proximal demonstratives than for both medial and distal demonstratives. Prenominal demonstrative forms appear in the upper part of the graph.

A mixed-effects analysis was conducted on aggregated response frequencies for each word per participant with position and demonstrative as predictors and participants as random effects. Prenominal position and proximal demonstrative were used as reference levels. There was a significant main effect of position (F (1,585) = 791.4, p < 0.001), there was also a main effect of demonstrative (F (2,585) = 8.51, p < 0.001) and an interaction between position and demonstrative (F (2,585) = 11.54, p < 0.001). Follow-up t-tests revealed that the postnominal position was chosen less often (coef.: –0.24, s.e.: 0.01, t (585) = –19.6, p < 0.001). No difference in choice frequency was found between the proximal and medial demonstrative (coef.: –0.017, s.e.: 0.012, t (585) = –1.43, p = n.s.) while the distal was chosen significantly less than the proximal demonstrative (coef.: –0.071, s.e.: 0.012, t (585) = –5.97, p < 0.001). The significant interaction effect was due to the difference in choice frequency between pre- and postnominal being significantly smaller for medial (coef.: 0.041, s.e.: 0.017, t(585) = 2.34, p < 0.05) and distal demonstratives (coef.: 0.083, s.e.: 0.017, t (585) = 4.81, p < 0.001) than for proximal demonstratives, meaning that participants were less likely to choose the proximal demonstrative in the postnominal position and less likely to choose the distal demonstrative in the prenominal position (Figure 2).

To test the effects of semantics on choice of demonstratives, three logistic regression analyses were performed, one for each demonstrative relative to the two other outcomes combined (e.g. este vs not este). All semantic factors, except Vision had a significant impact on the choice of demonstratives (Table 2, Figure 3). Negative Valence, Loudness, Human/Social roles, Motion and Arousal yielded significantly fewer proximal demonstratives, while Manipulability, Torso/Legs and Self/body yielded significantly more proximal demonstratives. These results were largely mirrored in the choice of distal demonstratives. Negative Valence, Loudness, Motion, Scene, Time and Arousal yielded more choices of distal demonstratives while Taste/Smell, Manipulability and Self/body significantly decreased the likelihood of distal demonstratives. The choice of medial demonstrative was significantly more frequent for Loudness, Human/Social roles and Motion and less frequent for Scene, Time and Torso/Legs. It is important to note that these results are not mutually independent as there are only two effective degrees of freedom in the choices.

Table 2: Semantic factors affecting demonstrative choice.

	Proximal (‘este/a’)				Medial (‘ese/a’)				Distal (‘aquel/la’)
	coef.	s.e.	z-value	p (corrected)	coef.	s.e.	z-value	p (corrected)	coef.	s.e.	z-value	p (corrected)
Vision	0.030	0.018	1.654	n.s.	–0.040	0.018	–2.273	n.s.	–0.004	0.020	–0.216	n.s.
Valence	–0.163	0.016	–10.457	<0.0001	–0.009	0.014	–0.632	n.s.	0.167	0.015	11.348	<0.0001
Loudness	–0.187	0.016	–11.674	<0.0001	0.094	0.015	6.083	<0.0001	0.095	0.017	5.596	<0.0001
HumanRole	–0.092	0.020	–4.624	<0.0001	0.059	0.019	3.154	<0.05	0.028	0.020	1.392	n.s.
Taste/Smell	0.057	0.022	2.614	n.s.	0.020	0.022	0.883	n.s.	–0.082	0.025	–3.238	<0.05
Motion	–0.153	0.017	–8.806	<0.0001	0.063	0.017	3.832	<0.05	0.069	0.018	3.868	<0.05
Manipulability	0.151	0.016	9.231	<0.0001	–0.025	0.016	–1.537	n.s.	–0.152	0.019	–8.188	<0.0001
Scene	–0.039	0.018	–2.185	n.s.	–0.064	0.018	–3.615	<0.05	0.116	0.019	6.246	<0.0001
Time	0.028	0.015	1.834	n.s.	–0.148	0.016	–9.239	<0.0001	0.117	0.016	7.411	<0.0001
Torso/Legs	0.064	0.016	3.944	<0.001	–0.053	0.016	–3.252	<0.05	–0.010	0.018	–0.554	n.s.
Arousal	–0.111	0.016	–6.822	<0.0001	0.001	0.016	0.046	n.s.	0.121	0.018	6.683	<0.0001
Self/Body	0.181	0.016	11.149	<0.0001	–0.019	0.016	–1.184	n.s.	–0.222	0.020	–11.158	<0.0001

• Note. The table reports the results of regression analyses for demonstrative choice, regardless of position. Results reflect the choice of a particular demonstrative relative to the two others. A positive coefficient reflects an increased choice while a negative reflects a decreased choice. P-values are Bonferroni-corrected.

Figure 3: Three logistic regression analyses, treating each demonstrative as the target variable relative to the two others largely replicated previous findings, suggesting that the choice of proximal demonstrative is affected by the investigated semantic factors. Manipulability and Self/body-relatedness make proximal demonstratives more likely, while negative Valence, Loudness, Human roles, Motion, and Arousal are associated with fewer proximal demonstratives. This pattern is largely mirrored in the choice of distal demonstratives. Medial demonstratives seem to be negatively associated with Time in the current set of words. * p < 0.05; ** p < 0.001; *** p < 0.0001, Bonferroni corrected for multiple comparisons.

A mixed-effects logistic regression was applied to test the effects of the 12 semantic factors on the choice of demonstrative position. Participants were entered as random effects. Four semantic factors remained significant after Bonferroni correction for multiple comparisons (Table 3, Figure 4): Vision, Loudness, Manipulability, Time and Self/body. A negative effect (Vision, Loudness, Manipulability) reflects that prenominal position was chosen relatively more often and a positive effect (Time and Self/body) reflects that a prenominal was chosen relatively less often. No effect of Valence was observed after correction for multiple comparisons, but an uncorrected effect was found, indicating that Negative Valence may be associated with a weak tendency to use the prenominal position more often (coef.: –0.04, s.e.: 0.017, z = –2.2, p(uncorrected) < 0.05) (Figure 4), contrary to our hypothesis.

Table 3: Semantic factors affecting positional choic.

	Pre/postnominal position
	coef.	s.e.	z-value	p (corrected)
Vision	–0.080	0.023	–3.480	<0.05
Valence	–0.040	0.018	–2.250	n.s.
Loudness	–0.063	0.019	–3.220	<0.05
HumanRole	–0.049	0.023	–2.105	n.s.
Taste/Smell	0.022	0.029	0.781	n.s.
Motion	–0.010	0.021	–0.492	n.s.
Manipulability	–0.062	0.021	–3.010	<0.05
Scene	–0.037	0.022	–1.682	n.s.
Time	0.066	0.020	3.302	<0.05
Torso/Legs	–0.056	0.020	–2.751	n.s.
Arousal	–0.020	0.021	–0.939	n.s.
Self/Body	0.139	0.022	6.386	<0.0001

• Note. The table reports Results of regression analyses for positional choice. P-values are Bonferroni-corrected. A high score on the Time and Self/body factors yielded a greater likelihood for choosing a postnominal construction.

Figure 4: Logistic regression revealed that the choice between pre- and postnominal position for demonstratives was primarily associated with nouns denoting semantics related to the self/body in the current selection of words. * p < 0.05; ** p < 0.001; *** p < 0.0001, Bonferroni corrected for multiple comparisons.

To summarise, our results confirm that demonstrative choice is affected by semantic features of the referents. Manipulability and Self/body relatedness make proximal demonstratives more likely to be chosen, whereas Negative Valence, Loudness, Human roles, Motion and Arousal are associated with fewer proximal demonstratives (este, ‘this’). This pattern is largely reflected in the choice of distal demonstratives (aquel, ‘that yonder’), whereas medial demonstratives (ese, ‘that’) seem to be negatively associated with Time. The prenominal construction – as an unmarked construction – is the preferred choice of the sample, whereas the postnominal is the less frequent choice, confirming its marked status. Contrary to our expectations, European Spanish speakers do not seem to choose between prenominal and postnominal constructions on the basis of the positive or negative Valence of the intended referent. Instead, we saw effects of Time and Self/body factors, which were more likely to elicit postnominal choices.

3.1.3 Qualitative analyses. Interviewing European Spanish speakers

Follow-up interviews were conducted to strengthen our findings and interpretations. A total of 9 native European Spanish speakers from Seville, Spain (5 female; Mean_age 35.8) were interviewed about their use of demonstrative forms in their everyday life. Three of the interviewees participated in the DCT study as well. They were asked to provide answers to the following two questions:

1. How would you explain the use of demonstrative to a person learning European Spanish?

2. European Spanish also uses postnominal forms, when? Can you provide some examples?

With regard to the first question – about the use of demonstratives in the everyday life – the general answers emphasised that demonstratives convey spatial information. All respondents reported that the proximal form este ‘this’ is used to indicate something close to the speaker; the medial form ese ‘that’ to indicate something at a middle distance and the distal form aquel ‘that yonder’ to indicate something at a wider distance from the speaker. Informants also stressed the distal form to be less used with respect to its proximal and medial counterparts (supported by the present study). Only two of the interviewees mentioned that the medial term ese ‘that’ can also indicate a referent close to the hearer and that all demonstratives can also express a distance in time.

When asked to explain the use of postnominal constructions, 8 interviewees emphasised that the construction conveys a derogatory meaning about the referent. Surprisingly, when asked to provide examples of the postnominal constructions, all interviewees used referents that had no negative valence per se. The most common examples were with human referents such as niño/a ‘baby boy/girl’ or tío/a ‘man/woman’ like in el chico este, ‘this bad guy’, mira al tío este ‘look at that bad guy’, accompanied by a derogatory tone of voice and palm-up gesture. Moreover, they reported that – given its negative connotation – they tended to avoid the structure in formal contexts, restricting its use to informal and friendly conversation. When asked to indicate which of the three demonstrative forms would be more suitable for this friendly interactive context, they all agreed that the unmarked form would be the proximal postnominal (i.e. el tío este, ‘this bad guy’, contrary to the findings in this study – see Figure 2) and that both the medial and distal forms would be used less to achieve the same purpose. More naturalistic data, including the nouns mentioned by the informants should be collected to investigate why this inconsistency between the quantitative and qualitative data arises.

3.1.4 Discussion of Study I

In Study I we focused on the interaction between semantic features of the referents and the choice of prenominal vs. postnominal demonstrative forms in European Spanish monolinguals. Our goal was twofold: a) to replicate the results of the previous DCT study on European Spanish using this shortened version of the paradigm; and b) to assess whether semantic features of the referents influence not only the choice of specific demonstrative forms (i.e. proximal vs. medial/distal), but also the position of the demonstrative in the NP (i.e. prenominal vs. postnominal). Our main hypothesis was that referents characterised by a negative valence, which in previous versions of the DCT selected a prenominal medial or distal demonstrative, would foster the selection of the postnominal option, which is known to convey derogatory meaning. In line with Gómez Sanchez and Jungbluth (2015) and Zulaica Hernández (2017), we expected to find an increased use of proximal postnominal and medial postnominal structures corresponding to referents with negative valence (i.e. el criminal este/el criminal ese). We did not expect to find many occurrences of the distal postnominal in relation to negative valence (i.e. el criminal aquel), because of its assumed marked status in the post-nominal counterpart of the European Spanish demonstrative system (see also Ransom 1999).

The results only partially met our expectations. Regarding (a), the reliability of the reduced form of the DCT, we replicated the main results found in Todisco et al. (2021): manipulable and self/body related referents elicit proximal demonstratives, regardless of their position in the NPs, whereas negative Valence, Loudness, Human roles, Motion, and Arousal are associated with an increased use of distal demonstrative forms. In the current set of nouns, the medial demonstrative ese ‘that’ seems to be positively associated with Loudness and negatively associated with Time. The latter is a replication of previous findings by Todisco et al. (2021), while the effect of Loudness has not been observed before. We did not replicate a relationship between the factor Vision and the medial demonstrative. Regarding (b), we found an interaction effect between the prenominal and proximal demonstrative vs. postnominal and non-proximal demonstratives. Moreover, the postnominal option was chosen less often than its prenominal counterpart, regardless of the semantic factor analyzed. The only two factors to yield a relative increase in choice of postnominal constructions were Time and Self/body. This was contrary to our initial intuition that the negative Valence of the referent would increase the choice of a post-nominal demonstrative form.

In order to corroborate our results and interpretations, we compared the DCT data with a series of follow-up interviews with native European Spanish speakers (i.e. some of whom also participated in the task). We asked them (1) to explain how they use demonstratives in everyday life and (2) how they would implement the postnominal options. Informants agreed that the primary use of demonstratives was to convey spatial information (1), which highlighted that they were not aware of the effect of the semantic features of referents in their previous or everyday choices. When asked to explain the use of postnominal forms (2), they essentially agreed that the postnominal structure conveys derogatory meaning and that it is used in informal contexts. Those informants that participated in the DCT also explained that they did not use postnominal structures very much in the experiment because – given the formal context – they did not feel legitimized to do so. This information is consistent with the low frequencies we found in the DCT data.

To replicate our findings and strengthen our interpretations, we ran an additional DCT study on a new dataset and with a different group of participants, which is presented in the following sections.

3.2 Study II

Study II focuses on the prenominal vs. postnominal use of demonstratives in a sample of European Spanish/Catalan simultaneous bilinguals. Participants were tested in European Spanish.

3.2.1 Material and methods

Participants. Participants were recruited at the Faculty of Spanish, Modern and Classic Philology at the University of the Balearic Islands in Mallorca, Spain. The questionnaire was promoted during class, and participation was voluntary. The initial sample consisted of 86 participants. Based on the information collected in the demographic questionnaire at the end of the study, 26 participants were excluded from the study for the following reasons: 4 participants were excluded because they were not simultaneous bilinguals of European Spanish/Catalan, 15 participants were excluded because they did not finish the task, and 4 participants were excluded because they reported to have learned Catalan or European Spanish after three years of age.⁶ Thus, the final sample was composed by 61 adult, simultaneous bilingual speakers of European Spanish/Majorcan Catalan (6 male and 55 female; Mean_age 22.72; SD 5.96; age range: 18–48 years). All participants read and accepted the consent form for data treatment and analysis before the start of the experiment.

Materials. Study II replicated Study I in all its aspects except for two main points. First, to corroborate our previous findings, we used another set of word stimuli. A new set of 150 words were selected from the DCT dataset reported by Rocca and Wallentin (2020) and Todisco et al. (2021; 2022). After excluding the words used in Study I, word selection followed the same criteria used for Study I. We selected 20 nouns representing the highest and lowest loadings on each factor among the remaining words in Binder and colleagues (2016) and 40 nouns per Positive/Negative Valence. All scores from all nouns and factors went into the statistical analyses described below. The total sample of nouns included masculine (N = 77; e.g. joven, ‘teenager’), feminine (N = 61; e.g. excusa, ‘excuse’), common gender (N = 2, e.g. abrazo, ‘hug’) and forms where both genders were specified (N = 10, e.g. empresario/empresaria, ‘businessman’). All nouns were accompanied by morphologically aligned adnominal demonstratives in both prenominal and postnominal position. The grammatical features of the nouns were in accordance with the linguistic regulation of the Real Spanish Academy (Real Academia Española: https://www.rae.es).

The questionnaire was distributed on the platform Labvanced.com, which allowed both computer and tablet device responses (Figure 5).

Figure 5: Experimental design providing an example of the laptop version of the DCT run in Labvanced.com.

The 150 nouns were presented with the six demonstrative options presented simultaneously, in a randomised order – both for the nouns and for the demonstratives. A sample of the words used is presented in Table 4.

Table 4: A sample of words from Study Two.

English version	Spanish version	Semantic Factor	English version	Spanish version	Semantic Factor
alligator	caimán	Vision	finger	dedo	Manipulability
turtle	tortuga	Vision	ball	bola	Manipulability
ivy	hiedra	Vision	pencil	lápiz	Manipulability
sin	pecado	N_Valence	store	almacén	Scene
curse	maldición	N_Valence	embassy	embajada	Scene
insult	insulto	N_Valence	forest	bosque	Scene
rumour	rumor	P_Valence	summer	verano	Time
scream	grito	P_Valence	winter	invierno	Time
envy	envidia	P_Valence	old	viejo/a	Time
drum	tambor	Loudness	muscle	músculo	Torso/Leg
megaphone	magáfono	Loudness	driver	conductor/a	Torso/Leg
piano	piano	Loudness	beach	playa	Torso/Leg
student	estudiante	Human	bonfire	hoguera	Arousal
businessman	empresario/a	Human	fireworks	fuegos artificiales	Arousal
journalist	periodista	Human	lightning	relámpago	Arousal
barbecue	barbacoa	Taste/Smell	family	familia	Self/Body
tea	té	Taste/Smell	hair	pelo	Self/Body
mustard	mostaza	Taste/Smell	reality	realidad	Self/Body
subway	metro	Motion
cab	taxi	Motion
bus	autobús	Motion

• Note. The table shows a sample of the nouns used in the questionnaire. Here we report the original English version from Rocca and Wallentin (2020), followed by its Spanish translation from Todisco et al. (2021) and semantic factor associated.

3.2.2 Results

Data and analyses are publicly available at https://doi.org/10.17605/OSF.IO/97YMB. Participants chose the prenominal position for 87.7% of the trials and the postnominal position for 12.3%. The proximal demonstrative was chosen in 25.8% of the trials while the medial was chosen 33.6% and the distal 40.6% (Figure 6).

Figure 6: Study II. Participants again made choices to couple nouns with a combination of demonstrative and pre/postnominal position. Choice distributions for the set of nouns differed from random choices (dashed line). Analysis of variance revealed the same pattern of choices as in experiment 1, i.e. a main effect of position, a main effect of demonstrative and an interaction between position and demonstrative. We replicated that prenominal position was chosen most. We also found that both medial and distal demonstrative were chosen less often than the proximal demonstrative. The interaction replicated the findings from Study I that the difference in frequency for pre- and postnominal choices was bigger for proximal demonstratives than for both medial and distal demonstratives. Prenominal demonstrative forms appear in the upper part of the graph.

Analyses were identical to those of Study I. There was a significant main effect of position (F (1,323) = 512.4, p < 0.001), there was also a main effect of demonstrative (F (2,323) = 15.1, p < 0.001) and an interaction between position and demonstrative (F (2,323) = 19.0, p < 0.001). Follow-up t-tests revealed that the postnominal position was chosen less often (coef.: –0.33, s.e.: 0.01, t (323) = –17.9, p < 0.001). Both medial (coef.: –0.09, s.e.: 0.02, t(323) = –4.96, p < 0.001) and distal (coef.: –0.15, s.e.: 0.02, t (323) = –8.63, p < 0.001) demonstratives were chosen less often than proximal. The significant interaction effect was due to the difference in choice frequency between pre- and postnominal, it was significantly smaller for medial (coef.: 0.11, s.e.: 0.02, t(323) = 4.06, p < 0.001) and distal demonstratives (coef.: 0.11, s.e.: 0.03, t (323) = 6.05, p < 0.001) than for proximal demonstratives, meaning that participants were less likely to choose the proximal demonstrative in the postnominal position and less likely to choose the distal demonstrative in the prenominal position (Figure 6).

The three logistic regression analyses that tested for semantic effect on choice of demonstratives found that fewer semantic factors played a role than for Study I, but still 6 out of 12 semantic factors were found to influence the choice of demonstrative (Table 5, Figure 7).

Table 5: Semantic factors affecting demonstrative choice.

	Proximal (‘este/a’)				Medial (‘ese/a’)				Distal (‘aquel/la’)
	coef.	s.e.	z-value	p (corrected)	coef.	s.e.	z-value	p (corrected)	coef.	s.e.	z-value	p (corrected)
Vision	0.056	0.024	2.276	n.s.	0.005	0.025	0.191	n.s.	–0.077	0.028	–2.775	n.s.
Valence	–0.081	0.026	–3.069	<0.05	0.033	0.026	1.263	n.s.	0.064	0.029	2.221	n.s.
Loudness	–0.077	0.022	–3.461	<0.05	0.029	0.023	1.254	n.s.	0.068	0.025	2.718	n.s.
HumanRole	–0.101	0.027	–3.680	<0.05	0.001	0.027	0.030	n.s.	0.116	0.029	4.045	<0.001
Taste/Smell	0.026	0.026	1.003	n.s.	0.036	0.026	1.383	n.s.	–0.079	0.031	–2.574	n.s.
Motion	–0.034	0.027	–1.300	n.s.	0.030	0.027	1.126	n.s.	0.003	0.030	0.106	n.s.
Manipulability	0.172	0.026	6.611	<0.0001	–0.079	0.027	–2.935	<0.05	–0.138	0.030	–4.558	<0.0001
Scene	–0.060	0.024	–2.535	n.s.	0.017	0.024	0.693	n.s.	0.056	0.026	2.121	n.s.
Time	0.005	0.023	0.208	n.s.	–0.072	0.023	–3.052	<0.05	0.075	0.025	3.036	<0.05
Torso/Legs	0.016	0.024	0.652	n.s.	0.005	0.025	0.189	n.s.	–0.018	0.027	–0.661	n.s.
Arousal	0.044	0.024	1.847	n.s.	–0.023	0.024	–0.941	n.s.	–0.028	0.027	–1.041	n.s.
Self/Body	0.215	0.026	8.334	<0.0001	–0.089	0.027	–3.292	<0.05	–0.184	0.031	–5.880	<0.0001

• Note. The table reports the semantic factors affecting demonstrative choice. Results reflect the choice of a particular demonstrative relative to the two others. A positive coefficient reflects an increased choice while a negative reflects a decreased choice. P-values are Bonferroni-corrected.

Figure 7: Experiment 2. Three logistic regression analyses, treating each demonstrative as the target variable relative to the two others. The pattern is similar to that observed in Study I; however with fewer significant effects, maybe due to the smaller sample size. Manipulability and Self/body-relatedness are again found to increase the likelihood of proximal demonstratives, while negative Valence, Loudness, Human roles, are associated with fewer proximal demonstratives. This pattern is largely mirrored in the choice of distal demonstratives. Time is again found to influence the choice of medial demonstratives, again in a negative direction. * p < 0.05; ** p < 0.001; *** p < 0.0001, Bonferroni corrected for multiple comparisons.

Negative Valence, Loudness and Human roles again yielded significantly fewer proximal demonstratives, while Manipulability and Self/body yielded a significant increase of proximal demonstratives. Human/Social roles and Time yielded more choices of distal demonstratives while Manipulability and Self/body yielded significantly fewer distal demonstratives. No semantic factors were found to significantly increase choices of medial demonstratives, while Manipulability, Time and Self/body were associated with decreased use of the medial term. As for Study I, we note that these results are not mutually independent.

Again, we applied a mixed-effects logistic regression to test the effects of the 12 semantic factors on the choice of demonstrative position. Participants were entered as random effects. Three semantic factors remained significant after Bonferroni correction for multiple comparisons (Table 6, Figure 8): Torso/Leg, Arousal and Self/body. A negative effect (Torso/legs) reflects that prenominal was chosen more often and a positive effect (Arousal and Self/body) reflects that a prenominal was chosen less often. No effect of Valence was observed, even without correction for multiple comparisons (coef.: –0.03, s.e.: 0.04, z = –0.8, p(uncorrected) = 0.44) (Figure 8).

Table 6: Study II. Semantic factors affecting positional choice.

	Pre/postnominal position
	coef.	s.e.	z-value	p (corrected)
Vision	–0.040	0.037	–1.077	n.s.
Valence	–0.030	0.038	–0.771	n.s.
Loudness	–0.036	0.033	–1.100	n.s.
HumanRole	0.009	0.040	0.215	n.s.
Taste/Smell	–0.027	0.038	–0.696	n.s.
Motion	–0.009	0.039	–0.233	n.s.
Manipulability	0.096	0.040	2.413	n.s.
Scene	0.060	0.036	1.667	n.s.
Time	0.032	0.034	0.946	n.s.
Torso/Legs	–0.106	0.036	–2.981	<0.05
Arousal	0.186	0.036	5.131	<0.0001
Self/Body	0.186	0.043	4.311	<0.001

• Note. The table reports results of regression analyses for positional choice. P-values are Bonferroni-corrected. A positive coefficient for the Arousal and Self/body factors indicated a greater likelihood for choosing a postnominal construction. The effect for Self/body was a replication of the findings from Study I.

Figure 8: Logistic regression revealed that the choice between pre- and postnominal position for demonstratives was primarily associated with nouns denoting semantics related to Arousal and the Self/body in the current selection of words. The finding for Self/body was a replication of Study I. * p < 0.05; ** p < 0.001; *** p < 0.0001, Bonferroni corrected for multiple comparisons.

To summarise, our results replicated that demonstrative choice and position is affected by semantic features of the referents. Manipulability and Self/body relatedness make proximal demonstratives more likely to be chosen, whereas Negative Valence, Loudness, and Human roles are associated with fewer proximal demonstratives (este, ‘this’) across both studies. This pattern is largely reflected in the choice of distal demonstratives (aquel, ‘that yonder’), but not to the same degree as seen in Study I. Medial demonstratives (ese, ‘that’) were again found to be selectively associated with Time, as a negative effect. The prenominal construction is the preferred choice. As in Study I, we failed to find an effect of Valence on the choice between prenominal and postnominal constructions. On the other hand, we replicated an effect of the Self/body factor as linked to the elicitation of the postnominal construction.

3.2.3 Qualitative analyses. Interviewing European Spanish speakers

To bolster our ability to interpret the findings, all participants were presented with additional questions at the end of the noun/demonstrative matching task. They were asked whether they used postnominal demonstrative forms in everyday life (Question 1) and, if so, in which context (formal or informal). We also asked which meanings the demonstrative was actually used for: neutral, negative, or for a distant referent (Question 2). All 61 participants answered Question 1: Do you use postnominal demonstratives in your everyday speech (e.g. la manzana esta)? Only 26% reported that they do not use any postnominal constructions, 49% reported to use postnominal constructions exclusively in informal contexts, and the 25% reported to use them in both formal and informal contexts. Regarding Question 2: If you use/hear these postnominal demonstratives, how do you interpret their meaning, 16% reported using/hearing postnominal constructions with a neutral meaning, the 31% with faraway referents, the 39% with a derogative meaning, and 13% preferred not to answer.

3.2.4 Discussion of Study II

In Study II, we examined the interaction between the semantic features of referents and the choice between prenominal and postnominal demonstrative forms in Majorcan Catalan/European Spanish bilingual speakers via the DCT. This was done to corroborate and strengthen the results obtained in Study I. To this end, in Study II we used the same design as in Study I. However, we changed the dataset of words and involved another sample of participants.

As in Study I, our results confirmed a) the reliability of the reduced form of the DCT, stressing that the choice of demonstratives was not random and that the prenominal position was generally preferred to the postnominal option. Furthermore, we replicated the significant effect of demonstrative position (pre- or postnominal) and the interaction between position and demonstrative, as found in Study I, and confirmed that the prenominal position was chosen most frequently, with both the medial and distal demonstratives being chosen less frequently than the proximal demonstrative. Once again, the interaction replicated the findings from Study I, showing that the difference in frequency between pre- and postnominal choices was greater for proximal demonstratives than for medial and distal demonstratives.

We also confirmed that b) there is an association between demonstrative choices, positions and semantic factors. We replicated that the postnominal option was chosen less often than its prenominal counterpart, and that this effect was enhanced for proximal demonstratives and diminished in combination with distal demonstratives. Proximal demonstratives were used significantly less frequently with nouns related to Negative Valence, Loudness and Human roles, while they were used significantly more frequently with nouns related to Manipulability and Self/body. This pattern was largely mirrored in the choice of distal demonstratives. Although we did not find any semantic factor that significantly increased the choice of medial demonstratives, their use was found to decrease in the case of specific semantic factors, such as Manipulability, Time and Self/body. In line with Study I, we did not observe any effect of Valence in the choice of demonstrative position, while nouns related to Torso/Leg, Arousal and Self/body did. The choice of postnominal construction was significantly higher for nouns denoting semantics related to Arousal and Self/body for the current selection of words.

To better interpret our results, we asked all participants a series of follow-up questions. When asked (1) whether they use postnominal demonstratives in everyday life, and in which contexts, half of the participants agreed that the primary use of postnominal demonstratives was to convey information in informal contexts, while an additional 25% reported using these constructions both in formal and informal contexts. If we consider the experiment a formal setting, this information is consistent with the low frequencies of postnominal choices found in the DCT data. When asked (2) how they would interpret the postnominal options to explain the interpretation they give of the postnominal demonstrative construction, a large proportion (39%) of participants highlighted its derogatory meaning.

Overall, the main results of Study II replicated the results from Study I. Furthermore, they opened new and interesting avenues for future research.

4 General discussion

In the present study, we examined the interaction between the semantic features of referents and the choice between prenominal and postnominal demonstrative forms via two studies in European Spanish.

European Spanish uses two structures to convey deictic information. The first shows the demonstrative form in prenominal position, as in este artículo ‘this paper’. The second shows the demonstrative in postnominal position, such as in el artículo este ‘this damn paper’. According to several studies on European Spanish demonstratives, the postnominal structure can convey a derogatory connotation of the referent (e.g. Lamíquiz 1967; Hottenroth 1982; Ranson 1999; Gómez-Sánchez & Jungbluth 2015; Shin & Vallejos Yopán 2023). To our knowledge, no study has been conducted to determine whether intrinsic semantic factors of the referent, which have been shown to affect the choice of demonstrative form, can also influence the choice of the demonstrative position. To this end, we conducted two shortened version of the DCT expecting that the more the negative valence of the referent the higher the use of postnominal demonstrative structures.

The results only partially met our expectations. We confirmed the reliability of the DCT and the effect of semantics on the choice and positioning of demonstratives within noun phrases (NPs). Both prenominal and postnominal constructions were used by our participants as unmarked and marked choices, respectively, but not as a function of inherent Valence.

Intrigued by our findings and keen to corroborate our interpretations, we compared the DCT data with a series of follow-up interviews and questions for native speakers and participants from both samples. Informants from Study I, were asked (1.I) to explain how they use demonstratives in everyday life and (2.I) how they would implement the postnominal options. They agreed that the primary use of demonstratives was to convey spatial information (1.I), which revealed their lack of awareness of the effect of the semantic features of referents in their previous or everyday choices. When explaining the use of postnominal forms (2.II), our informants stated that the postnominal structure conveys derogatory meaning and that it is used in informal contexts. Those who participated in the DCT also explained that they did not use it very much in the experiment because – given the formal context – they did not feel licensed to do so. This information is consistent with the low frequencies we found in the DCT data. As previously mentioned, one of the main rationales behind the DCT is that participants are asked to match a noun with a set of demonstratives, without providing any further context. In this way, participants are asked to rely on their first intuition to match nouns and demonstratives, and this intuition is based on their own internal context or knowledge. It is therefore understandable if some participants felt less comfortable using a postnominal option, as the informal and interactive context that would have constructed and supported their choice was missing. In addition, when asked to provide examples of postnominal constructions, all informants used nouns that do not have a negative valence per se, such as niño/a ‘baby boy/girl’, tío/a ‘man/woman’, which only refer to human beings. Combining the DCT data with the interview findings, we conclude that when the noun had a negative valence per se, speakers did not consider the postnominal form as the primary choice to convey distancing language. They tended to rely on the unmarked prenominal construction using medial and distal demonstrative forms to signal conceptual/psychological distance, as confirmed by the DCT data.

Building on Study I, we added two follow-up questions also to Study II to reinforce our findings. All participants answered two questions: (1.II) regarding their use of postnominal demonstrative forms and (2.II) regarding their interpretation postnominal constructions they had used or heard. In line with our informants from Study I, participants in Study II again emphasized that postnominal construction are more appropriate in an informal context. In addition, when asked to pinpoint the interpretation of these specific constructions in their daily life, the results were not as striking as we expected. Although almost one half of the respondents opted for the derogatory connotation of postnominal demonstrative forms, the other half stressed that postnominal constructions could also be used for faraway referents.

Using aggregated response frequencies across participants as outcome measures, we found across both experiments that the combination of proximal demonstrative and postnominal construction is relatively dispreferred, while the combination of distal demonstrative and prenominal construction is also relatively less common in the DCT context. If both distal demonstratives and postnominal position signal conceptual/psychological distancing (derogatory distancing in the case of postnominals), this may not be so surprising, since a combination of distal demonstrative and postnominal position would then provide a contradiction in the distancing signal in the absence of a physical context to govern the demonstrative. Whether this interaction remains in interactive contexts needs to be studied, together with the role played by the medial form.

As regards the combination of position and referent semantics, we failed to find the predicted relationship with Valence across both experiments. Only the Self/body factor was consistently associated with the postnominal construction across the two studies. Our qualitative data, however, corroborated the assumption that derogation is one of the main effects of the postnominal construction. How can we reconcile these seemingly contradicting observations? When looking at the examples of postnominal constructions provided by our participants, these did not include nouns with negative connotations, rather all participants included nouns referring to humans, such as niño/a ‘baby boy/girl’, tío/a ‘man/woman’, that do not have a negative valence per se. This helped us to interpret our results in a new light. The use of a postnominal construction may thus be used to add something extra to words that do not already have a negative meaning, e.g. in relation to the speaker’s own Self/body. The negative valence of a noun (i.e. terrorist, criminal, gunshot, …), combined with the choice of medial and distal demonstrative forms, may already have provided a sufficient amount of distancing language. Thus, the choice of a postnominal demonstrative would be redundant for nouns with a negative valence semantics, if negative valence were the only information to be conveyed about the referent.

Although redundancy is abundant in communication (Wit & Gillette 1999; Bazzanella, 2011; Trudgill 2011; Leufkens 2020), it is often thought to be avoided, due to communicative economy, e.g. following Grice’s principle of quantity (Grice 1975). If the listener already knows something, one can leave it out. In fact, demonstrative pronouns are often used as such shorthand placeholders that allow the speaker to navigate and pinpoint complex discourse elements very efficiently. Studies have shown that speakers, including children as young as five years, use this kind of audience modelling when speaking (Nadig & Sedivy 2002; Brown-Schmidt et al. 2008; 2015; Heller & Chambers 2014).

Additionally, we speculate that the relatively larger proportion of postnominal constructions used for words related to the Self/body, may also be informed by the qualitative responses. If the primary use of the postnominal construction happens in informal conversation, when in the DCT may this be established? One possibility is inner speech (Fernyhough & Borghi 2023; Wallentin & Nedergaard 2023). When establishing a context for the demonstrative choice in the DCT, participants are likely to enter into an inner dialogue about their relationship with a particular referent. When the referent is an element of the participant’s own physical or personal makeup, such as mouth, which was one of the words in Study I (see Table 1), the participant may automatically enter a more informal discussion with him/herself about her relationship with this particular referent and thus be more likely to use the postnominal position. More studies are needed to explore this hypothesis.

Overall, our results have opened up a number of future lines of research, branching out into two main areas. The first is to compare the results for postnominal demonstrative choice with those of other languages that also allow for the double position of demonstrative forms. This would allow us to replicate the results found for European Spanish. The second is to further study the prenominal and postnominal use of demonstratives in naturalistic conversation (e.g. analysing demonstrative production in existing corpora) to assess whether interaction legitimises postnominal forms over prenominal ones, as our interviewees suggested. This will allow us to zoom into the noun related to the semantics of the Human Role, which – according to our informants – elicits more postnominal constructions. Furthermore, this will help us establish a more solid way to investigate how the redundancy hypothesis and the communicative economy interact, which is something we could only briefly touch on in this paper. These new insights will improve the interpretability of our research and help us better understand the cognitive mechanisms underlying demonstrative choice and usage.

5 Conclusions

Here we investigated the information conveyed by the prenominal and postnominal position of the demonstrative in the NP of European Spanish. In this language, NPs such as este hombre and el hombre este convey different meanings, with the latter (potentially) revealing derogatory information about the intended referent. Drawing on cognitive studies on the perceptual and conceptual parameters that guide demonstratives’ choice, we hypothesised that semantic features determine not only which demonstratives are used, but also where they are used in the NP: before (prenominal) or after the noun (postnominal). We replicated the main effect of semantic features on proximal and medial/distal demonstrative choices. We failed to find support for the hypothesis that intrinsic negative valence was associated with enhanced elicitation of postpositional demonstratives. Instead, we found an unexpected effect of the Self/body factor across both studies. Qualitative interviews suggested that the use of postnominal demonstratives to express derogatory connotations primarily happens in informal contexts. From both the DCT data and the answers provided after the DCT, we inferred that when the noun is already characterised by a negative valence per se, a medial/distal prenominal, unless legitimated by the interactional context, is sufficient to convey distancing language, making the postnominal demonstrative option redundant, when only one semantic dimension is stressed. Words referring to the self/body may, we speculate, elicit a more informal inner dialogue in participants, making them more likely to use the postnominal form as a kind of self-derogatory speech. Future studies are needed to test this hypothesis.

Abbreviations

adj = adjective

adv = adverb

dct = Demonstrative Choice Task

demo = demonstrative

det = determiner

Prox. = proximal

Sing. = singular

Data availability

Data and analyses’ coding are publicly available at https://doi.org/10.17605/OSF.IO/97YMB.

Ethics and consent

All experimental procedures have been approved by the Ethical Board at the University of Balearic Islands, Exp. Number: 236CER21.

Acknowledgements

We extend profound gratitude to Professor Joaquin Comesaña-Rincón of the University of Seville for the insightful and stimulating intellectual conversations during the initial phases of this research. The authors gratefully acknowledge the crucial support and cooperation received from the Escuela Oficial de Idiomas in Huesca and the University of Seville for the first round of data collection. Furthermore, deep appreciation goes to Andrea Rodríguez-González for checking the Spanish stimuli, to the participants, and to the University of the Balearic Islands for their invaluable support in facilitating the second round of data collection for this study.

Competing interests

The authors have no competing interests to declare.