Given that misnaming by familiar individuals is a widespread phenomenon, the goal of Study 2 was to understand the circumstances under which misnamings occur (e.g., when it occurs and who misnames whom). To that end, we gave undergraduates a longer survey about particular misnaming episodes, with more specific questions that were derived from data about misnaming incidents from a pilot study with Duke students.
One-hundred and nineteen Duke University undergraduates (56.3% female) who had reported being misnamed in Study 1 were recruited to complete an online survey. The mean age of the participants was 18.97 years (SD = .91; range: 18–22).
Participants were directed to an online Qualtrics survey, where they consented to participate. After reading the instructions, all participants completed the misnaming and demographics portion of the survey.
Participants were asked to report up to 10 incidents of misnaming. For each instance of misnaming, participants were first asked to (1) identify the person who had misnamed them, (2) indicate whether they had been called one name or a string of names, and (3) indicate, if called multiple names, whether the incorrect names were always said in the same order. Participants were then asked what name(s) they had been called, as well as their relationship to the individuals with whom their names were switched. Participants also rated the misnaming episode on a number of different dimensions shown in Table 1. One dimension asked how often this type of misnaming occurred (rarely, yearly, every few months, monthly, and daily or more often). We categorized these responses into two bins: rarely, yearly, or every few months (Bin 1); and monthly, daily, or more often than daily (Bin 2).
For each misnaming incident, we also coded the phonetic similarity of the participant’s name and the name that he or she was wrongly called in two ways. First, phonetic similarity was independently rated by two coders on a scale from 0 to 2 (0 = no similarity, 1 = somewhat similar, 2 = similar), and the ratings showed high reliability, Κ = .77. Disagreements were scored by a third independent coder. The third coder’s ratings were retained in all cases of disagreement. Names were rated as similar if they had identical beginning and ending sounds (e.g., Michael and Mitchell) or if they had multiple letter similarity at the beginning (e.g., Phillip and Phyllis). Names were rated as somewhat similar if they shared a beginning sound (e.g., Abigail and Agatha; Felicity and Phyllis) or if they shared a common ending sound (e.g., Joey and Mikey). Names with none of these features in common were rated as not similar.
Second, we analyzed the number of overlapping phonemes between the names. Each name was translated into the International Phonetic Alphabet (IPA) by using an online dictionary based upon the Carnegie Mellon University Pronouncing Dictionary. Next, a trained linguist checked the translations for correctness as well as translated names that were not processed by the dictionary. A proportion of phonetic overlap between the used name and the correct name was then calculated by counting the number of phonemes in the used name that were also present in the correct name, and dividing this number by the number of phonemes in the correct name.
Questions included the participants’ age, gender, race/ethnicity, and languages spoken. In addition to information about the participants, the demographic section also asked participants to describe their family members, pets, and anyone who was mentioned in the previous sections. For each person mentioned, they were asked to list their relationship, perceived physical similarity (1 = not at all similar to 7 = very similar) between the person and themselves, and whether the person lived with them or not. For each pet mentioned, they were asked the pet’s type, gender, and whether the pet was currently owned. Participants were also prompted to include information for all of the individuals that had been mentioned previously when describing misnaming incidents. Later, the names and demographics of these individuals were matched to the information pertaining to each misnaming incident in which they or their names were involved.
Results and discussion
Participants provided a total of 292 cases of misnaming. Note that here and in the subsequent studies, because of missing data, some frequencies do not sum to the total N, and some percentages may not sum to 100%. Here and throughout the paper, we report all of our results, but mainly discuss those that address our hypotheses. (See Table 1 for the relevant data on the demographics of the misnamer and misnamed, as well as characteristics of the misnaming incidents.) To summarize, misnamers were more often female than male, were almost always older than the misnamed, and saw or spoke to the misnamed regularly. Each reported misnaming was typically experienced every few months, yearly, or rarely, and the misnamed tended to be called a single name only during a misnaming episode. Perceived negative mood did not have an effect on misnaming.
We also analyzed the qualities of the named individuals whose names were incorrectly used to refer to the misnamed participant. About half of the time, the named individual was present when the misnaming occurred. On average, the named individuals were only somewhat physically similar to the misnamed participants, suggesting that perceived physical similarity between people was not driving misnaming.
Besides the qualitative characteristics of the misnaming event and the individuals involved with misnaming, the relationships between the misnamer, the misnamed, and the named were also examined. Specifically, we were interested in whether misnaming occurred because the misnamed and the named were members of the same group or semantic category (e.g., the same family), and thus their names shared some semantic meaning. The relationships between the misnamer and the misnamed, and the named and the misnamed, were categorized into one of four categories: family members, friends, other humans, and pets. For all following analyses, only the first name reported was analyzed because of the internal similarity of naming strings skewing the results. The relevant data are reported in Table 2; each incident of misnaming was categorized by the relationship between the misnamed and the misnamer (rows) and the relationship between the named and the participant (here, the misnamed; columns). The relationships between the misnamed and the misnamer and the misnamed and the named largely overlapped. In other words, family members tended to misname other family members using a third family member’s name, friends tended to misname other friends, and others tended to misname others. When analyzed using chi-square, which assumes independence between the relationship of the misnamer and the misnamed and the relationship between the misnamer and named, the results were significant across all categories. The large value of φ (.93) indicates a large effect of semantic relatedness. In this sample, only 13 pets were mentioned, and their names were used exclusively by family members of the misnamed.
The aforementioned results suggest that misnamings occur within semantic categories; however, the analysis did not exclude the possibility that the misnamings also occurred because of phonetic similarity between names. For each measure of phonetic similarity (rated similarity and proportion of phoneme overlap), the phonetic similarity of the incorrect name used by the misnamer to all of the correct names (legal name and nicknames) were averaged. Of all the cases of misnaming (n = 292), the name used and the correct name were, on average, low on phonetic similarity as rated by research assistants (M = .33, SD = .45), as well as when the proportion of overlapping phonemes between the correct name and the used name was calculated (M = .34, SD = .22). We ran a Monte Carlo procedure to determine a baseline of phonetic similarity by pairing names of all misnamed individuals in the study at random with the incorrect names used across the misnaming incidents and rating these pairs for phonetic similarity, as above (K = .85), as well as calculated the overlapping phonemes between the two names. Here too, the rated phonetic similarity and proportion of phoneme overlap between pairs was low (M = .18, SD = .34; M = .29, SD = .21, respectively). The rated phonetic similarity of the original name pairs was statistically higher than the randomly generated word pairs, t(582) = 4.61, p < .001, η2 = .04, as was the proportion of phoneme overlap, t(582) = 2.68, p = .008, η2 = .01.
The similarity of the used name to the correct name was also compared to the results of the Monte Carlo randomization using chi-square analysis (see Table 2). The Monte Carlo ratings were used as expected values for the chi-square calculation. This allows for the comparison of the effect size φ between semantic relatedness of the misnamed and the name and the rated phonetic similarity of the correct name and the name used. There was a significant difference between the rated phonetic similarity of the observed misnamings and the expected Monte Carlo-generated values, with the observed misname–name pairs rated as more similar than the Monte Carlo pairs. However, the effect size (φ = .49) is smaller than the effect of semantic relationships on misnaming (φ = .93), Fisher’s Z = 13.47, p < .001.
Consistent with Study 1, these results suggest that instances of misnaming are common occurrences that participants can remember and report on; however, the retrospective nature of the survey may influence these findings. Additionally, misnaming between familiar individuals is not random. Rather, the wrong name used tends to be within the same semantic category and is affected by the phonetic similarity between the correct name and the name used.