Imagine that you are talking to your three-year-old who is trying to explain that he sent a letter to his buddy at the kindergarten earlier that day. To make sure that you have understood what he has just told you, you say: Så han var ikke i barnehagen? (‘So, he was not at kindergarten?’). The child replies as in (1) below:¹

In (1), the child provides the information that his friend was present at the kindergarten that day. However, the child also signals that there is an opposition between what you seem to think (that his friend was not present at the kindergarten) and the actual state of affairs (that his friend was present at the kindergarten). The child denies the truth of the proposition expressed by your utterance, and he does so by using a particular intonation pattern, the so-called Polarity Focus (PF; e.g., Fretheim, 2002), characterized by a focal accentuation of a polarity carrier (in (1), the finite verb var (‘was’)) followed by an additional accentuation later in the utterance (in (1), barnehagen (‘the kindergarten’)). What is particular about such PF utterances compared to utterances realized without PF, is that the use of PF signals that the only new information in the utterance is the truth value of the proposition expressed by the speaker. This makes PF a valuable tool for a speaker who wants to convince her interlocutor of whether a contextually given thought is a true or false description of some state of affairs.

For a speaker to be able to signal a denial of her interlocutor’s thought, it must be manifest to her that her interlocutor believes this thought, that is, she must be capable of mentally representing that her interlocutor believes this thought and accepting this as true or probably true (Sperber and Wilson, 1986/1995, p. 39; Wilson, 2012). The use of PF therefore requires the abilities to metarepresent and attribute thoughts. Also, since the production of PF arguably requires the speaker to be attentive to, evaluate, and express an attitude (of endorsement or denial) toward the truth-conditional content of an attributed thought, it is a higher-level metarepresentational ability that is required (cf. Wilson, 2012), in addition to a capacity for ‘epistemic vigilance’ toward utterance content (cf. Sperber et al., 2010). The example in (1) above (taken from the first author’s diary notes of her son) suggests that PF may occur early in children’s language production. In this study, we ask whether preschool children tend to produce this intonation pattern, and, if so, what their production of PF utterances can reveal about the development of their early pragmatic abilities.

Instead of (1), the child could also have had responded as in (2a), (2b) or (2c) below to communicate the same (or similar) content:

In addition to investigating PF production, this study explores children’s use of two particles, both orthographically expressed as jo, which often co-occur with PF and share some of its pragmatic features: they both enable the speaker to signal her attitude toward the truth value of the proposition expressed. Jo can appear either as a sentence-initial response particle (Fretheim, 2017), such as in (2ab) above, or as a sentence-internal pragmatic particle (Berthelin and Borthen, 2019), as in (2c).²

According to Fretheim (2017), the sentence-initial response particle jo is used to contradict a communicated negation by affirming the embedded positive proposition. Thus, both PF and the response particle jo require that the speaker metarepresents a contextually manifest thought and they both relate to the truth of the proposition expressed.³ The response particle jo can be used as a single word response, such as in (2a) above, or it can precede a PF utterance, such as in (2b).

As to the sentence-internal pragmatic particle jo, Berthelin and Borthen (2019) suggest that its semantic contribution is to signal that the proposition expressed should be interpreted as mutually manifest to the speaker and hearer, and to signal that the utterance can be taken as a premise for deriving a contextual implication. Sentence-internal jo therefore naturally accompanies PF in (2c), providing the hearer with an additional cue to the speaker’s intention: that the hearer should accept as true that the speaker’s friend was indeed present at the kindergarten. Because of the overlapping pragmatic features between PF and the jo particles described above, more careful analyses of these early jo productions may serve to cast light on the development of metarepresentational abilities in children.

In (1) above, PF was used to signal the speaker’s denial of the truth of the metarepresented thought. However, PF can also be used to affirm the truth of a thought, as in (3) below where A concludes after hearing that B’s fried was present at kindergarten when B had sent his letter:

The PF in B’s utterance is realized as a focal accentuation of the finite verb fikk (‘received’) followed by three additional non-focal accentuations. Using PF, B metarepresents and affirms the truth of the proposition explicitly expressed by A in (3).

PF can also be used to affirm or deny a thought that is not explicitly mentioned but attributed to someone (or to oneself). Consider the conversation in (4) below:

‘The car is charged (despite what you seem to think)’.

Here B’s utterance is realized with PF, involving a focal accentuation of the finite verb er (‘is’) followed by another (focal) accentuation of the infinite verb form ladet (‘charged’). A expresses that he cannot get to the grocery store, and by responding with a PF utterance, B communicates that she denies a (false) belief that she attributes to A: that he cannot use his (electric) car to drive to the grocery store because it is discharged.

The analysis of the use and function of PF has an affinity with the Relevance Theoretic notion of ‘echoic utterances’. Using an echoic utterance, the speaker metarepresents and attributes a thought or utterance with a similar content to someone else (or to the speaker herself), and at the same time conveys the speaker’s attitude to this thought or utterance. It is this signaling of the speaker’s attitude to the attributed thought, which is characteristic of echoic utterances, and by which they achieve relevance (Wilson, 2012, p. 249). Using an echoic utterance is one linguistic tool available to the speaker if she intends to modify her interlocutor’s epistemic state. Utterances carrying PF serve a similar communicative function in that they involve the affirmation or denial of an attributed proposition as a description of some state of affairs, and could in this way be said to involve an echoic element.

An overarching aim of this study is to gain a deeper understanding of intonation as part of a broader pragmatic competence. According to Relevance Theory (Sperber and Wilson, 1986/1995), intonation serves an important pragmatic function by contributing to an utterance’s relevance (where ‘relevance’ is understood as a trade-off between so-called cognitive effects and processing effort). On this view, intonation functions procedurally as a guide to the speaker’s intended interpretation:⁴ by signaling an utterance’s information structure it contributes to making some contextual implications more salient than others (Fretheim, 2002; Wilson and Wharton, 2006; Wilson, 2011; Scott, 2021).

The early emergence of prosodic competence in first language acquisition (for an overview, see Kehoe (2013)) and the pragmatic nature of intonation provide us with an opportunity to use intonation as a window into children’s early pragmatic competence. Children’s acquisition of intonation in the period prior to five years of age is still a quite unexplored field of research (Peppé and Wells, 2014). Furthermore, although there are studies on the role of (intonational) focus in pragmatic reasoning (e.g., Tomlinson et al., 2017; Gotzner and Spalek, 2019), there have been few attempts to combine suprasegmental phonology with cognitive pragmatic theory in the study of language acquisition (Wharton, 2012, 2020). Thus, the question of how children’s ability to master intonation as a communicative device develops remains largely unresolved. Analyses of children’s production of intonation utterances can provide us with a deeper understanding of what this ability amounts to. In addition, this study presents the first attempt of accounting for Norwegian speaking children’s use of intonation as a communicative device in language production.

The aim of our study was to investigate whether Norwegian-speaking children aged two to five years can produce intonation utterances realized with ‘Polarity Focus’ (PF), and if so, whether we would find a gradual development of their productions in contexts of increasing complexity. While we found productions of PF in all age groups tested and only in felicitous contexts, our hypothesis of a gradual development was only partially supported by our data.

Overall, our findings show that children produce PF from as early as age two. We take this to suggest that they are also (at some level) able to evaluate the truth or falsity of a proposition and attribute a contextually available proposition to their interlocutor from around this age. Furthermore, our findings show that young children can express their affirmation or denial of this truth-conditional content by intonational means. Already from the age of two and onwards, children seem to use intonation naturally and efficiently as a communicative device and in this case specifically to signal epistemic vigilance toward an attributed propositional content. This is also likely to involve an intention to modify their interlocutor’s epistemic state.

As expected, the ability to use PF to express the denial of an inferred false belief seems to arise around four years of age. However, the percentage of PF productions in this condition was overall low (7%). Since this is the most complex condition of our design and only expected to be mastered by the older children, the low percentage of PF productions was not surprising. Our findings align with previous findings in the Theory of Mind literature where the ability to linguistically express an understanding of others’ false beliefs manifests around four years of age (cf. Wellman et al., 2001).

Our finding that PF was produced by two-year-olds in both the Pos-Aff, the Pos-Den and the Neg-Den conditions support previous research that show that children are able to both reject false and accept true statements before their second birthday (Lyon et al., 2013), and that two-year-olds spontaneously correct assertions they believe to be false (Pea, 1982). In fact, it was the Neg-Den condition that had by far the highest percentage of PF productions in all age groups. This finding is surprising since the literature suggests that an increasing number of negations increases the complexity of the test conditions (cf. Just and Carpenter, 1971). Although this may be the case at a general level, we see no evidence of this in our data: Even two-year-olds master the more complex context where they need to contradict a prior negative belief. This result is in line with Pea (1982), who shows that the ability to correct false statements appears prior to their expressing agreement with true statements. Intonation, and in this case PF, seems to offer young children an easily accessible linguistic strategy for communicating their attitude toward a propositional content, enabling even children as young as two years to express this higher-level metarepresentational content without having to verbalize it.

This high percentage of PF productions in the Neg-Den condition suggests that it was the most natural context for PF production in our task. A growing body of research shows that from very early on children monitor the reliability of the information communicated (Gliga and Csibra, 2009; Koenig and Woodward, 2010; Sperber et al., 2010). It is possible that signaling a denial or an opposing opinion might be more socially important than signaling an endorsement. It could also be that PF is more frequently used by adults in contexts like the Neg-Den condition, and therefore possibly more familiar to children. We know from the study by Turco et al. (2014) that adults produced Verum Focus in 70% of the ‘polarity correction’ contexts where the verbal stimuli used was similar to our Neg-Den condition, involving a mismatch in form of a false negative statement about what was depicted in the visual stimuli. Future research should investigate experimentally how Norwegian-speaking adults use PF, focusing on the different contexts for eliciting PF and in what ways they differ.

We further observed that the Neg-Den condition also provided a natural context for responding with the sentence-initial response particle jo, often in combination with PF (such as in (11) above). While according to Noveck et al. (2021) the accurate but surprisingly fast si response by the four-year-olds in their study suggested that these children did not aim at modifying the epistemic state of their interlocutor when responding with si, our results suggest otherwise. First, they indicate that the ability to produce pragmatically felicitous responses using the response particle jo could be present as early as two years of age. Given what we know from the developmental literature of children’s ability to evaluate the truth value of propositional content at such early age (Pea, 1982; Lyon et al., 2013), together with some level of perspective taking (O’Madagain and Tomasello, 2021), it seems likely that, if two-and three-year-olds produce jo accurately, they have by the age of four developed a pragmatic maturity that goes beyond relying merely on the minimal semantic representation of the particle. Second, children younger than four years showed mastery of PF production and especially in the Neg-Den condition, a context highly similar to the Negative-Si condition used by Noveck et al. (2021) to elicit the response particle. Furthermore, in our study, children younger than four years spontaneously and felicitously produced the combination of the sentence-initial response particle jo and PF. Together, this mastery of both PF and the response particle jo at such early age, we believe speak against a limited pragmatic competence involved in the use of this particle at four years of age. To gain a deeper understanding of how the response particle jo and PF are related and what the use of them separately and in combination can reveal about children’s developing pragmatic abilities, future research should address this relationship directly, using different approaches and methodologies and a broader set of context types to elicit the two phenomena.

Our data also included participant responses that, although realized without PF, seemed to have a similar pragmatic effect. In the example in Figure 8 above, we discussed how this effect could be due to the presence of a sentence-internal jo, which often involves some sort of oppositional feature. Berthelin and Borthen (2019) argue that the procedural meaning encoded by jo involves an instruction to the hearer to interpret the proposition expressed as mutually manifest to speaker and hearer, and to use the proposition expressed as a premise for deriving and supporting a contextual implication. As they (2019, p. 25) point out: “jo is a useful tool when speakers suspect that the hearer will not accept the information they are communicating.” This is also the case for PF. Just like sentence-internal jo, PF can be used when a speaker needs to convince her interlocutor of the epistemic status of the proposition expressed and of the conclusions that can be drawn from it. Future research should also investigate the relationship between the sentence-internal particle jo and PF in more detail. If the two phenomena are closely related, in what ways do they differ, and what triggers the use of them in combination?

We have suggested that utterances carrying PF have an affinity with echoic use in the relevance-theoretic sense (Wilson, 2012): PF utterances involve both an attribution of the proposition expressed, and they enable the speaker to convey her attitude to this proposition in the form of a denial or an affirmation. Although this is a rather simple form of echoic use, the attitude being explicitly conveyed, our study has shown that children master such echoic uses from a very early age. This has potential implications for theoretical accounts of the development of other, more complex forms of echoic use such as verbal irony, which is characterized by the speaker tacitly echoing and expressing a dismissive attitude to an attributed thought (Wilson, 2012). These uses have been shown to have a protracted development (Falkum and Köder, 2020) with some comprehension abilities emerging around the age of six years. Our results show that the ability to express an endorsing or dismissive attitude to an attributed thought (expressed explicitly in the context in the simplest cases) emerges much earlier. In this way, intonational competence, and more specifically the ability to use PF appropriately, could be seen as a precursor to ironical uses.

Finally, we would like to mention some caveats. First, we have claimed that PF production involves attribution of the thought that is affirmed or denied by the use of PF. However, the design of the verbal stimuli in our first three PF conditions (Pos-Aff, Pos-Den, Neg-Den) makes it difficult to tease apart this ability from the ability to metarepresent a thought (without having to attribute it), since the handpuppet explicitly expresses his prior beliefs. There is solid evidence that four-and five-year-olds can attribute thoughts. One possibility then is that the two-and three-year-olds do not attribute the thought they are affirming or denying, but are merely metarepresenting a contextually available thought. However, this analysis leaves open the question why a speaker would produce PF in the first place: The informative intention of a speaker who uses PF is to convey her affirmation or denial of a metarepresented thought. Why would she convey this if not to modify her interlocutor’s epistemic state (which does involve thought attribution)? The production of PF does not make sense in a context where the speaker merely metarepresents the thought without attributing it to someone. If two-and three-year-olds did not have this ability (and as a consequence they are not aiming at modifying their interlocutor’s epistemic state), we should expect them to produce less PF overall than the older children, simply because they would not experience relevant situations for the production of PF. However, we find no significant differences in PF productions between the age groups in our study. It seems likely, therefore, that the ability to attribute thoughts is also involved in the felicitous production of PF, and that this ability is present already from the age of two years.

Second, our experimental setting posed some challenges especially for the youngest participants. Although the experimenter made an effort to make the conversation as natural as possible, some of the two-year-olds had trouble adapting to the experimental setting or did not feel familiar enough with the experimenter (even though all participants who wanted to bring a familiar caretaker were given the opportunity to do so), and refused to speak. This could have masked their intonational competence. Finally, the use of production data as a source of evidence for pragmatic competence requires an interpretation of children’s communicative intention, which is inevitably speculative (Zufferey, 2020). Moreover, production data are often thought to underestimate children’s performance compared to comprehension data. However, since the conditions in our structured elicitation task are specifically designed to elicit PF, less is left to speculation compared to spontaneous productions in unstructured contexts. We also believe that for the study of early pragmatic development, production data provide a valuable source of insight, especially because controlled comprehension experiments may not be feasible with children in the youngest age groups. However, our conclusions in this paper inevitably rest on our interpretation of the production data.

Our study provides the first experimental evidence that Norwegian-speaking children are able to produce intonation utterances realized with ‘Polarity Focus’ from an early age. We suggest that the mastery of the production of PF, as well as their ability to produce the jo particles in appropriate contexts, can be seen as an early linguistic manifestation of the cognitive abilities for the attribution of thoughts and epistemic vigilance toward propositional content.

At a more general level, our study provides insight into the role of intonation as part of a broader pragmatic competence. An overarching aim was to start filling a gap in the literature by combining suprasegmental phonology and cognitive pragmatic theory. We provide experimental evidence for the pragmatic functions of intonation, which in the case of PF allows speakers to communicate a positive or negative attitude toward a metarepresented proposition. Our exploratory analyses of the jo particles also contribute some insight into children’s developing metarepresentational abilities.

We believe to have shown that studying intonational production can be useful as a window into children’s pragmatic competence. Although our results did not fully support the developmental hypothesis, they provide evidence of the intonational productions of children aged two to five years and a piece of information about their developing pragmatic competence which is currently missing in the literature. We hope to see many more studies of children’s intonational competence during this crucial developmental period in the coming years.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by NSD - Norwegian Center for Research Data. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

LH: conceptualization, experimental design, data collection, coding of the data, writing–original draft preparation, writing–review and editing, visualization, data analysis, and project administration. IF: conceptualization, experimental design, writing–review and editing, data analysis, and supervision. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.