Introduction

When we ask young children questions, we have expectations about what constitutes an appropriate response. This is particularly so when the question requires children to identify something: for example, when we ask them about their favorite toy, or what color pencil they want, or even “which” object they are referring to (Basco et al., 2021). For children to be able to respond successfully to such requests they must employ referential communication skills: comparing all the valid referents and replying appropriately with information that identifies the target (e.g., Asher & Wigfield, 1981). In the current research, we investigated these comparison and communication processes in young children’s responses to referential questions about different object types.

In English, referential questions tend to make use of words such as “which” or “what” and, as in the examples given above, tend to refer to objects or their attributes, and require a specific response (Kearsley, 1976; Robinson & Rackstraw, 1972). Referential questions are among the earliest that young children employ (following closely after “why” and “how” questions), and by around 4 years of age are in frequent use (Searle, 1969). However, the development and understanding of children’s responses to these types of questions is less clear.

On the surface, one would expect young children to know how to respond to referential questions. Three-year-olds are good at learning the names of objects (Birch & Bloom, 2002), understand that “one” refers to a single object (e.g., Condry & Spelke, 2008; Wynn, 1992), and by 5 years of age, young children can answer questions appropriately (if the questions are logical and make sense) (Waterman, et al., 2000). Consequently, they should be able to grasp the meaning of a typical referential question that requires them to identify a desired object. For example, when asked “which one do you want?” a child must reply with enough information to communicate the identity of their chosen object. However, young children’s understanding of communication must not be overestimated (Siegal, 1997): they can have problems understanding how objects can be uniquely represented (e.g., Apperly & Robinson, 1998), can have difficulty communicating the identity of an intended referent to others (e.g., Asher & Oden, 1976), and can struggle to produce a narrative communication unless a referent is present (Carmiol et al., 2018).

Most of the referential communication tasks used in existing research (e.g., Krauss & Glucksburg, 1969; Lloyd et al., 1995; Roby & Kidd, 2008; Uzundag & Küntay, 2018) include two test phases: first, a speaker generates an informative description so that a listener can understand which one of a selection of objects is being referred to. In the second phase, a listener decides if they can identify an object following an unambiguous verbal description by a speaker, or whether they must ask the speaker for more information after an ambiguous message. The purpose of these two test phases is to assess the effectiveness of verbal communications; both generated and received (Roberts & Patterson, 1983; Roby & Kidd, 2008), to assess the development of these abilities (Lloyd et al., 1998), or to assess whether informative descriptions can be improved (Uzundag & Küntay, 2018).

To understand the development of an ability in childhood we often start by considering how adults behave in a comparable situation. So, how do adults successfully generate an informative description and indicate a referent to other people? Research suggests that a speaker must supply enough information (to distinguish between all referents) for a listener to understand which object they might be referring to (e.g., Rosenberg & Cohen, 1966). It follows that the speaker must have two abilities; first, they must be able to compare all the valid referents and understand how they are unique and different, and second, they must be able to communicate that knowledge successfully to others. An effective communication therefore appears to rely on differences being identified and a successful comparison being carried out. Comparing objects involves deep processing: it highlights any differences (and similarities) between objects that may not be apparent when those objects are examined individually (e.g., Graham et al., 2010).

One explanation of why young children can have difficulty with referential communications is that they are less likely to carry out a comparison between all referents (Asher, 1976; Asher & Parke, 1975; Asher & Wigfield, 1981; Bearison & Levey, 1977; Camaioni & Ercolani, 1988). They tend to focus on the target object, ignore or disregard other possibilities, and fail to identify the differences (Asher & Oden, 1976; Girbau & Boada, 1996). Therefore, they cannot generate an adequate description to indicate the intended object, and the referential communication fails (Glucksberg et al., 1975). However, while such research concerns referential communication, it has tended to focus on children’s ability to generate informative messages, rather than their response to referential questions.

Other research has linked comparison processing to successful referential communication in children (Camaioni & Ercolani, 1988). Five to 8-year-olds were shown eight or nine drawn images and had to match a target (presented visually or described verbally) to one of the stimuli. They then had to generate effective verbal descriptions of objects so that someone else could identify them. Performance on the comparison tasks were highly predictive of children’s communication abilities (Camaioni & Ercolani, 1988). Children who could successfully compare objects, were then able to identify what information they should to pass on to others. Children who failed to communicate effectively in this study did so because they were unable to distinguish what made the target different from those around it. It is not clear why children failed to recognize these differences, or whether it impacts on children’s ability to answer referential questions.

The current research aimed to clarify why young children sometimes fail to recognize what differentiates one object from others around it, and to establish whether the focus on “differences” is necessary for children to answer referential questions. To do this, we had to get children to demonstrate their understanding of how objects might be similar, and how they might be different, as well as answer a referential question about an object. Therefore, in Study 1, children were presented with pairs of objects and demonstrated their comparison abilities by categorizing the objects according to their similarities or differences. Children then demonstrated their ability to communicate the identity of one of each pair of objects by responding to a referential question. In Study 2, we focussed on children who made errors in categorizing objects by similarity and difference and asked then to explain their decisions.

Study 1

Method

Participants

One hundred and fourteen children participated from a primary school serving a working-class population in the UK. There were 35 four-year-olds (Mean = 4 years and 4 months (4;4), range 3;10–4;9), 28 five-year-olds (M = 5;2, range 4;10–5;8), 29 six-year-olds (M = 6;2, range 5;9–6;8), and 22 seven-year-olds (M = 7;3, range 6;9–7;10). Half of each age group were female. Children belonged to following ethnic groups: white British (85); Asian (26); black African (1), and other (8). Teachers described all the children who took part as having a good understanding of English.

Materials

The materials consisted of four pairs of “target” objects, 20 “grouping” objects, an opaque bag, plastic containers, two plates, and a teddy. The four pairs of target objects were: a toy sheep and toy pig; a toy car and a toy train; a small red ball and a small blue ball; a small yellow block and a small green block. The opaque bag contained the target objects when they were not in use. The grouping objects consisted of toy farm animals; toy vehicles; colored balls; colored clocks (five of each). The plastic containers held the grouping objects when they were not in use. The plates were used to display the grouping objects while they were being used in the study. The teddy acted as a recipient for a target object in the communication task.

Design

The study used a mixed 4(age: 4 years vs. 5 years vs. 6 years vs. 7 years) × 2(comparison task: similarity vs. difference) design with repeated measures on the last factor. Responses to the communication part of the task were recorded. The experiment consisted of four separate trials. In Trial 1 children were presented with two diverse kinds of toy animals; Trial 2 was two distinct kinds of toy vehicles; Trial 3 was two different colored blocks; and Trial 4 was two different colored balls).

Each trial consisted of grouping the pair of objects by similarity and by difference, and a communication task that involved identifying the target object of the pair (see Table 1). The identification task always followed the comparison task, regardless of their ordering (since in referential communication tasks the comparison of possibilities must come before identification). We counterbalanced trials to avoid order effects.

Table 1 Means and standard deviation (in parentheses) for all ages of children for similarity, difference, and identity judgements in Study 1
Full size table

Procedure

The researcher tested children individually while sat at a table in a quiet part of the school. For each trial there was a familiarisation task, followed by a comparison task, followed by an identification task.

Familiarisation task. We felt it was important to be explicit about the similarities and differences between objects, so that the children (especially the younger ones) were clear about these factors. At the beginning of each of the four trials, the researcher gave children a pair of toys (the target objects) that were different in one way, and similar in another. Children were encouraged to look at and pick up the toys while the researcher stated how they differed, and how they were similar (order counterbalanced).

Comparison task: The purpose of this task was to check children’s recognition of how each pair of objects was similar and how they differed. After each familiarisation part of the trial, the researcher carried out a similarity sub-task and a difference sub-task (order counterbalanced). For each sub-task, the researcher placed ten additional toys on two plates on the table (this kept each grouping distinct and stopped toys moving around on the table). The researcher stated that all the toys on each plate “belonged together” and asked the children to put their toy toys “where you think they belong.” The researcher told children that the two toys might go together in one group (i.e., they could both be put on one plate) or they might go in separate groups (i.e., one could be put on each plate). The researcher noted on a record sheet children’s placement of the two toys. The experimenter then returned the two toys to the children and removed the grouping objects from the plates. The other sub-task followed the same procedure. Once both the comparison tasks were completed, the researcher removed all the additional objects and plates, leaving the children with the initial two toys.

Identification task: The purpose of this task was to check children’s ability to recognize the identifying information about an object. Following the comparison task in each trial, the researcher placed the teddy on the table and advised the child that Teddy has just arrived to play and that they should, “give Teddy one of your toys.” The researcher then asked the test question, “Which one have you given Teddy?” Children responded and their answers were recorded as either correct (identifying information given) or incorrect (no identifying information given).

At the end of the identification task, the researcher placed the two objects in the opaque bag and the next trial began until all four trials were complete. At the end of the trials, the experimenter gave children a sticker and returned them to their classroom.

Results

Coding

For the comparison task, children received a score of 1 for each correct placement of a target object in the similarity sub-task and a score of zero for each placement that was incorrect. They also received a score of 1 for each correct placement in a difference sub-task, and a score of zero if it was incorrect. Children could receive a maximum score of 4 for similarity judgements, and a maximum score of 4 for judgements of differences. For the identification task, children received a score of 1 if they correctly communicated the aspect of the object that uniquely identified it from the other object in the pair and a score of zero if they stated some attribute which did not uniquely identify the object or if they failed to give any identifying information. For example, if a child gave Teddy one of two different colored balls, they were deemed correct if responded to the test question, “Which one have you given Teddy?” by answering, “the red one” and deemed incorrect if they said, “the ball.” Children could receive a maximum score of 4 for the identification task. The means of all scores are displayed in Table 1.

Analyses

In line with previous similar studies that gave children repeated trials we used parametric analyses to evaluate the data (e.g., Camaioni & Ercolani, 1988). Initially, a mixed analysis of variance (ANOVA) was carried out between age (4 years vs. 5 years vs. 6 years vs. 7 years) and comparison task (similarity vs. difference) with repeated measures on the latter. As expected, there was a significant main effect of age (F(3, 110) = 4.67, p = 0.004, η2p = 0.11), with pairwise comparisons showing that 7-year-olds performed significantly better than 4-year-olds (p = 0.021). There was also a significant main effect of comparison task, (F(1, 110) = 15.46, p < 0.001, η2p = 0.12), with recognition of differences (M = 3.83, SD = 0.48) significantly higher than recognition of similarities (M = 3.41, SD = 1.09). There was no significant interaction between age and comparison task.

We then wanted to establish whether there was a significant difference between the age of children correctly answering the identity question A one-way independent ANOVA was carried out with the independent variable of age (four levels: 4 years versus 5 years versus 6 years versus 7 years) and the dependent variable of correct responses. The results showed no significant difference between the different age groups (p = 0.08) (see Table 1 for means).

Next, a multiple regression analysis was carried out to determine what would best predict successful communication of the identity of an object. Age, similarity judgements, and difference judgements were the predictor variable, whilst the identity scores were the outcome variable. Correlations suggested that the age of children was positively and weakly associated with recognition of differences, r(112) = 0.31, p < 0.001, recognition of similarities r(112) = 0.24, p = 0.006, and correct identification, r(112) = 0.19, p = 0.021. Judgement of difference were also positively but weakly associated, and with correct identification, r(112) = 0.28, p = 0.001, with judgements of similarity, r(112) = 0.21, p = 0.014. There was no significant correlation between judgement of similarity and correct identification.

The relationship between the predictor variables and the outcome variable was weak (R = 0.31). The results of the regression indicated that the three predictors explained only 9.3% (adjusted R2 = 0.68) of the variance (F(3, 110) = 3.76, p = 0.013) (see Table 2). It was found that correct judgement of the differences between pairs of objects significant predicted correct identification (β = 0.45; t = 2.61, p = 0.010; 95% Cl 0.059–0.43). Neither judgements of similarities nor age of child predicted identification (lowest p = 0.20).

Table 2 Regression model of age, similarity judgements, and difference judgements predicting correct identification (study 1)
Full size table

Study 1

Discussion

The aim of Study 1 was to clarify the comparison process in 4–7-year-old’s ability to answer referential questions. Additionally, we sought to investigate the recognition of similarities and differences during the comparison process and clarify which predicted children’s ability to recognize identifying information and answer referential questions.

All the 5–7-year-olds (bar one) and most 4-year-olds answered the referential question correctly. As expected, older children found the similarity and difference tasks easier than younger children, indeed 7-year-olds performed with few errors and much better than one would expect at standard comparison or referential communication tasks (e.g., Lloyd et al., 1998). Our comparison task required the objects to be distinctly categorized by their similarities and differences, unlike previous comparison tasks (e.g., Lloyd et al., 1995). However, we had pointed out the similarities and differences to children, as part of the familiarisation procedure, so it is likely that this helped the older children to recognize the prominent features of the objects.

Nevertheless, children found it harder to recognize the similarities between pairs of the objects than recognize the differences. While it is only strictly necessary to recognize the differences between objects when looking for a unique identifier, we wondered why children were not as proficient at recognizing similarities. Even 7-year-olds, who had no difficulty with the other tasks, occasionally faltered with the similarity task. The aim of Study 2 was to investigate this further by asking children to verbalize or explain their reasons for “grouping” objects as they did.

Study 2

Method

Participants

Thirty-six children (21 girls) participated from a primary school serving a working-class population in the UK (a different school to Study 1). There were 18 four-year-olds (M = 4; 9), range 4; 3–5; 2), and 18 five-year-olds (M = 5; 8, range 5;4–6; 1) (we kept to a younger cohort due to the near ceiling performance of 6- and 7-year-olds children in Study 1). Children belonged to following ethnic groups: white British (34), black Caribbean (1), and other (1). Teachers described all the children who took part as having a good understanding of English.

Materials

We used the materials from Study 1 (except Teddy, as no identity/communication task was included). A digital voice recorder documented children’s verbal response to the explanation question.

Procedure

The familiarization procedure followed the same format as Study 1. There was no identity task. The comparison task was as used in Study 1, apart from one adjustment: When children had grouped each object (either for similarity or difference) they were asked why they had put the objects in those places. For example, children were asked, “Can you put your two toys where you think they belong” followed by “Why do they belong there?” or “Why does that belong there?”.

Design

The study used a mixed 2(age: 4 years vs. 5 years) × 2(comparison task: similarity vs. difference) design with repeated measures on the last factor. Each child received four trials. The independent variables were age and comparison task. The dependent variables were the children’s responses to the comparison tests (if they grouped the objects correctly according to similarity and difference). We also asked children’s explanations of their choices and noted their responses. We counterbalanced the ordering of the trials and the comparison tasks, to avoid order effects.

Results

Coding

As in Study 1, children received a score of 1 if they placed objects in the appropriate groupings for each of the similarity and difference tasks, and a score of 0 if they were incorrect. This meant that children could potentially gain a maximum score of four for each of the tasks.

Analyses

We carried out a 2(age: 4 years vs. 5 years) × 2(comparison task: similarity vs. difference) mixed model analysis of variance (ANOVA) with repeated measures on the last factor. We found a main effect of comparison task, F(1, 34) = 5.36, p = 0.027, η2p = 0.14, where judgements of difference (M = 3.78, SD = 0.089) were significantly better than judgements of similarity (M = 3.31, SD = 0.18) (see Table 3). There was no effect of age and no interaction.

Table 3 Means and standard deviations (in parentheses) for correct judgements of similarities and differences across age groups (study 2)
Full size table

As we had found no difference between the two age groups, we collapsed the data across these factors to examine the overall number of errors made. There were 9 errors made in the difference task and 26 errors made in the similarity task. We then examined the type of explanations children offered. Children’s error explanations fell under three broad topic areas, (a) ambiguous (i.e., they did not offer a justification, or their response was vague) (b) color error (i.e., their explanation was related to an incorrect color matching of objects), and (c) thematic explanations (i.e., they suggested an alternative relationship or story involving the objects) (see Table 4 for examples). Of the error explanations in the difference task, 4 (45%) related to color, 5 (55%) were ambiguous, and none were thematic. In the similarity task, however, 17 (65%) error explanations related to color, 9 (35%) related to thematic explanations, and none were ambiguous.

Table 4  Explanations produced for grouping errors in study 2
Full size table

Study 2

Discussion

There were more errors with the similarity task than the difference task. Children’s explanations of their errors were classified as either a) ambiguous, b) color, or c) thematic. Ambiguous errors may have been due to difficulties in verbalizing the reasoning process. Color errors could have been due to the focus on perceptual similarity that young children tend to have in categorization activities (Namy & Clepper, 2010). Or a “preference” for color might be dominant to the extent that it overrides or interferes with other types of categorizations (Catherwood et al., 1989).

Interestingly, the thematic explanations only occurred when toy animals or vehicles were used, and not the balls and blocks. Toy animals can trigger the activation of mental representations of experiences of real-life animals (Ware et al., 2006) so it seems likely that toy vehicles could activate mental representation of experiences of real-life vehicles. These mental representations could have become dominant in children’s thoughts and overshadowed the required judgements of similarity. Or, children may have made more qualitative judgements of the features of the objects, rather than focusing on general similarity (Diesendruck et al., 2003; Hammer & Diesendruck, 2005; Medin et al., 1993) and this led to more unorthodox categorizations.

Another possibility is that children who created thematic explanations to justify their (incorrect) categorizations, simply had better skills in communication and creativity: they were able to invent a scenario to explain their choice and verbalise it. However, this seems improbable: whilst the creation of a “narrative” is one of the more complex types of generative communication (Carmiol et al., 2018), the capacity to talk about imaginative thoughts predicts Theory of Mind (ToM) ability in young children (Peterson & Slaughter, 2006). ToM, the ability to understand someone else’s mental states (Wellman, 2017). is linked to more advanced abilities in many fields, so while children with higher levels of ToM could certainly be more creative and communicative, it seems less likely that they would be making such errors of categorization in the first instance. Nevertheless, this may be an interesting route for further investigation.

General Discussion

The aim of the current research was to clarify whether children compared the similarities and differences between pairs of objects before they successfully answered referential questions about those objects. In Study 1, we found that children who were able to correctly categorise objects according to their differences also seemed to be able to answer the referential question about one of those objects. However, there was no link between categorizing according to similarity and answering the same question. Other studies have found that children of this age can demonstrate how pairs of objects differ but struggle to respond to questions referring to that difference (Waters & Beck, 2012). Our findings suggested that it was only necessary for children to recognize how the pairs of objects differed, for them to be able to communicate the identity of the chosen referent.

In Study 2, we confirmed our previous findings that children found it easier to categorise pairs of objects according to differences, than to similarities. Interestingly, children’s verbal explanations of their errors, when categorizing objects, fell under three broad topic areas, (a) ambiguous, (b) color error, and (c) thematic explanations. Children’s errors were mostly related to the color of the objects in questions: children ignored obvious similarities and differences and instead focused on the color of the objects to try to explain their categorization errors (though often their color explanations were also incorrect). The next most frequent error seemed to be based on a qualitative, experiential factor: children created a mini story or narrative to explain their incorrect categorization of objects.

In both our studies, children found it harder to categorise pairs of objects according to their similarities, than to categorise according to differences. In one sense, this was to be expected, as one only needs to recognize differences between objects to be able to identify unique referents. Nevertheless, the recognition of commonalities (or similarities) between objects, is thought to be the most important stage of the comparison process for children (Gentner & Namy, 1999, 2006; Hammer et al., 2009), especially when the objects have been clearly labelled (as in our case) (Namy & Gentner, 2002). Yet, we found no inflated preference for seeking similarities, as suggested by previous research (e.g., Hammer et al., 2009). One possibility is that when children need to answer a referential question in a more naturalistic environment (such as our studies where they chose, manipulated, and named toys); recognizing the differences between things is simply sufficient. Or it might be that only children with more advanced cognitive functioning can attend to both similarities and differences between objects (Basco & Nilsen, 2017). There is also the chance that without the support and confirmation of another person’s facial cues, children ignore similarities between referents (Basco et al., 2021). Further research may be needed to clarify why preference for seeking similarities is not always found.

In conclusion, comparison processing is important for communicating about unique referents: it highlights any similarities and differences between objects that may not be apparent when those objects are examined individually (Graham et al., 2010). The results of the current research showed that children who successfully compared objects, by recognizing how different they were, went on to answer correctly referential questions about those objects. Young children’s understanding of how they can gain certain aspects of knowledge can also depend upon successful recognition of differences (Waters & Beck, 2012, 2015). In addition, children’s success at referential communication is linked to higher perspective taking abilities (Roberts & Patterson, 1983), and higher ToM levels (Resches & Pérez Pereira, 2007). Our findings offer support for the link between recognizing the subtleties of how objects differ, and many other cognitive abilities that develop throughout young childhood. Nonetheless, there is still uncertainty about the link between the creation of elaborate narratives, higher ToM abilities, and better categorization abilities. Further investigation is clearly needed to help determine the complexities of young children’s understanding of identification of unique referents, and the role of comparison in developing responses to referential questions.