Line-drawing standardized sets

Line drawings are schematic, simple and prototypical representations of concepts that have been widely used in research with adults since the publication of the pioneering study of Snodgrass and Vanderwart [3] to date. This standardized picture database consists in 260 black-and-white line drawings with norms for a series of variables, such as name agreement, image agreement, familiarity and visual complexity. Since then, this set continues to be relevant in basic and applied research and has been adapted to numerous languages, such as, for instance, Chinese [14], French [15], Dutch [16], Italian [17], Icelandic [18], Japanese [19], and Russian [20]. Likewise, it has also been adapted to different linguistic and sociocultural contexts. For instance, in Spanish-speaking countries, it has been used in Mexico [21], Cuba [22], Argentina [23, 24], and Spain [25–27].

Currently, there are also other banks of line drawings with normative data (e.g., [28, 29]), and in different languages. For instance, the International Picture-Naming Project (IPNP [30], in English, German, Italian, Bulgarian, Hungarian, Mandarin and Mexican Spanish), the Protocole Européen de Dénomination Orale d'Images (PEDOI [31, 32], in Dutch, English, German, French, Italian, Russian, Swedish and Spanish) and the Multilingual Picture databank (MultiPic [33], in Spanish, British English, German, Italian, French and Dutch).

Most of the research has been carried out with adults; however, some normative studies were carried out with children of different age ranges. To our knowledge, all of them used the original set of Snodgrass and Vanderwart [3]. It has been adapted to English ([34], with norms of 40 children from 7 to 9 years old), Chinese ([35], with norms of 66 children from 4 to 6 years old); French ([12], with norms of 1,440 children from 3 to 8 years old); and Cuban Spanish ([11], with norms of 150 children from 5 to 12 years old). In addition, we are only aware of two studies that expand the Snodgrass and Vanderwart [3] set with norms for children. One of them is the set published by Cycowicz et al. [4], which included 400 line drawings with norms of 30 English-speaking children from 5 to 6 years, as well as norms of 36 Portuguese-speaking children from 5 to 7 years old [13]. The other one is the International Picture-Naming Project [30, 36] which included norms of 250 drawings of 34 Italian-speaking children from 5 to 6 years old [37]. At present, available normative datasets of line drawings are often composed by pixelated stimuli (e.g. [3, 4, 30]) and these result are of poor quality compared to the images that we are all used to find in our daily life.

Finally, several studies have reported differences across languages and cultures in adults [30, 31], while this issue has been rarely studied with children. One study examining these differences in children was conducted by Wang et al. [35]. Their results showed significant differences in H index, familiarity and image complexity between Chinese and American children and in the expected name agreement between Chinese and French children. According to the authors, these results may reflect differences in language but they also indicate cross-cultural differences in the image properties in children from different cultures. The researchers highlighted the need for culturally specific norms when using pictorial stimuli with children.

Children and adults comparisons in line-drawing naming

Comparison research in line-drawing naming indicates that there are differences between children and adults in name agreement, modal name assigned (i.e., the most common name given to a picture in a sample), number and variability of alternative names provided, and percentage of unknown responses [4, 13, 34]. For example, in English, the seminal work offered by Cycowicz et al. [4] revealed that children provided modal names that differed from adults’ modal names for 13.5% of the pictures and produced different and a larger number of alternative names than adults did. Moreover, children reported not knowing the name or object for 9.8% of the pictures while the adults reported not knowing only 1.7% of the pictures. Previous studies also found that children were slower than adults in picture naming [37]. For example, D'Amico et al. [37] revealed that, in Italian, adults´ naming mean overall reaction time to produce the target name was 950 ms, 300 ms faster than children. The ratings of familiarity and visual complexity in previous literature were similar (though slightly lower) in children than adults [4, 13, 34].

Some researchers have pointed out that although the name agreement measures and naming latencies varied in children and adults, these differences are trivial, especially in older children, and the performance correlated highly in both age groups [34, 37]. However, we should be cautious with these conclusions because few studies made comparisons directly between children and adults using the same set of stimuli for both age groups, and they were carried out with small samples (< 40 children; [13, 37]). Furthermore, apart from quantitative differences, Cannard, Blaye, Scheuner, & Bonthoux [38] highlighted the importance of qualitative differences in normative research when the participants are children. One reason is that some modal names provided by the children differed from the names expected and provided by adults and, indeed, children present numerous mistakes. For instance, Pompéia et al. [13]compared qualitative differences in modal names and found that Portuguese-speaking children named 103 concepts (around 25%) differently from adults, and most of those were mistakes (e.g., cockroach for beetle). These results were in line with those provided by Cycowitz et al. [4]. The qualitative analysis of object names also makes it possible to identify the comprehension of word meanings of children from the earliest usages and it has been observed that the extensions' of children's objects name are like those of adults [39], but children are more likely than adults to omit responses because of lack of knowledge. Consequently, authors such as Cannard et al. [38] emphasize that previous literature comparing naming responses across age groups has underestimated the importance of these qualitative differences and warn about the need to be cautious when using the measures of picture naming in developmental studies, because they are dependent on modal name. In addition, the qualitative study of responses contributes to explore different cognitive processes involved in picture naming such as perceptual processing, activation of semantic information, lexical selection or name retrieval [40, 41].

Photograph standardized sets

Traditionally, research has focused on line drawings, but the standardization of photographic stimuli is progressively increasing in recent years because digital photography is today easily accessible due to technological development.

Photograph stimuli are realistic representations of objects, as they include details such as texture, shades, brightness, and volume cues. Thus, the interest on more ecological stimuli than drawings has begun to be highlighted [42, 43]. To our knowledge, the largest existing normative photograph dataset is the Bank of Standardized Stimuli (BOSS; [44, 45]). It contains 1468 photographs with norms collected from English- and French- speaking adults. Other photographic datasets with norms for adults are the set published by Viggiano et al. [43], in English and Italian; the Hatfield Image Test [46], in English; the C.A.R.E. set [47] in English; the set of Shao & Stiegert [48] in Dutch; the set of Saryazdi, Bannon, Rodrigues, Klammer, & Chambers [49], in Turkish. The ecological version of the Snodgrass and Vanderwart [3] set [42] is the only one in Spanish, and has recently been published also in Italian [50].

To the best of our knowledge, no standardized photographic set with norms for children has been published, despite its potential use as well-controlled stimuli in picture naming tasks in Primary School. In particular, colored photographs offer a rich representation of real objects, are more ecological than line drawings, and their use is increasingly widespread.

Additionally, although the number of online websites that offer free photographs in the public domain have increased (see publicdomainpictures.net for a website of public domain pictures), very few standardized banks are easily accessible with a license description and fewer are in the public domain (e.g. [47, 48], for normative picture sets under a CC0 license). If standardized picture banks were published under a public domain license, their use could be extended to assessment and educational intervention materials, given that free and unrestricted use of controlled stimuli would be allowed, even in commercial tests and textbooks.

Participants

A total of 118 Spanish-speaking children from 3rd and 4th grade of Primary School (52% female, M age = 8 years 9 months, SD = 0.6, range = 7 to 10 years) and 89 adults (79% female, M age = 23 years 10 months, SD = 8.3, range = 18 to 58 years) participated in the study. All had Spanish as a mother tongue and none of them had received a diagnosis of neurological damage, or speech or language problems.

All of the children attended public schools in different regions of Northern Spain (Cantabria, n = 80; and Burgos, n = 38). Adults were mostly undergraduate students from Northern Spain as well (Vizcaya).

Ethics statement

The ethics committee of the University of Deusto approved the procedure used in the present study (Ref: ETK-14/17-18). Written informed consent was collected from the parents or guardians of the children included in the study. Adults gave their informed consent to participate.

Stimuli

PicPsy bank consists of 106 concepts each depicted both as a photograph and as a matched drawing (212 pictures total; see Fig 1 for a sample of the stimuli). The bank has also been reproduced in a colored drawings version but the norms of the present study are collected only for line drawings and photographs.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Examples of pictures of the bank in photograph, colored drawing (non-normative), and line drawings.

The photographs were retrieved from https://pixabay.com under a CC0 license. The colored drawings and line drawings were created by the authors. The PicPsy bank also includes a colored drawing version but the norms of the present study are collected only for line drawings and photographs.

https://doi.org/10.1371/journal.pone.0238976.g001

Concepts were selected considering different psycholinguistic such as lexical frequency and length, and considering subjective variables such as familiarity, imaginability, concreteness and subjective age of acquisition. Specifically, concepts of different lexical frequency were selected (M = 135.92, SD = 209.58, range = 1.14–1,580.43) according to Martínez & García [58]. Additionally, selected words had high familiarity indexes (M = 6.03, SD = 0.71, range = 3.56–7.00), high imagination indexes (M = 6.05, SD = 0.56, range = 4.29–6.85) and high concreteness indexes (M = 5.87, SD = 0.62, range = 3.74–6.97) according to Duchon, Perea, Sebastián-Gallés, Martí, & Carreiras [59]; and a subjective age of acquisition under 8 years old (M = 4.24 years, SD = 1.32, range = 1.96–7.52) according to Alonso, Fernández, & Díez [60]. Most of the concepts used in PicPsy (75%) were presented in the study of semantic categories by Goikoetxea [61] for 3rd and 4th grade students, and corresponded to the following 30 semantic categories: animals, atmospheric phenomena, birds, buildings, clothing, feelings, flowers, fruits, furniture, geographical accidents and natural land formations, insects, kitchen utensils, light sources, mammals, media, parts of a house, parts of the human body, reading material, tools, trees, types of boats, types of building material, types of drinks, types of fabrics, types of food, types of plants, types of professions, types of relatives, types of toys, and types of vehicles.

In order to depict concepts, public domain photographs under a Creative Commons CC0 license were downloaded from https://pixabay.com, https://commons.wikimedia.org, and https://www.publicdomainpictures.net. Similar to editing steps employed in Brodeur et al. [44], the image background of the photographs were removed, words or brands were blurred, and images were resized to fit in a frame of 500x500 pixels using CorelDraw (Corel Corp., Ottawa, Canada) and Removebg. In order to improve the representation of certain concepts, arrows were added or multiple photographs were combined (e.g., a boy and a girl could be shown together with the arrow pointing to the girl).

The criteria for the arrangement of the photographs were similar to those of Snodgrass and Vanderwart [3]. The photographs of animals or parts of the body were oriented to the right and to the left in a balanced way. Fine and elongated objects (e.g. a pencil) were inclined with an orientation of 45° and the functional part was placed downwards.

The drawings were created directly from the photographs, once edited. Each of the drawings was done delineating the photographs by hand by one of the authors. Each drawing was then digitized and transformed into a vector image, in order to provide definition in any size. All the line drawings had the same visual characteristics of scale, shape and orientation as the corresponding photographs.

Procedure

Each participant completed the picture naming tasks in one of the two formats, line drawing or photograph. Seventy-three children (61%) and 48 adults (53%) took part in the assessment in line-drawing format, and the remaining participants in colored photograph format. Participants were evaluated from November 2017 to March 2018.

Most of the participants were tested in class groups by one of the examiners in a quiet room of the school or the university, except for 30 adults, who performed the same taskautonomously through the Internet. In order to ensure that children performed the task by themselves, unaffected by the responses of other children, a school teacher accompanied the experimenter in the session and the classroom desks were separated to leave a space of at least one meter between participants. In addition, the children were asked to perform the task in silence and to raise their hands whenever they had questions.

Although the picture naming tasks are usually performed orally and individually, some previous studies have been carried out with children of similar ages to our sample, who perform the studies in written form and in groups (e.g. [34]). This procedure allows for the quick testing of large groups of children, without altering the class’ and school’s routine. If the tasks are properly organized, group testing can offer an efficient method to collect data at minimal cost and time. For a proper written naming performance, children are required to have minimal spelling skills in order to write words fluently. Therefore, a sample of 3rd and 4th graders was selected since by this age children have developed enough fine motor skills to write fluently and it has been shown that Spanish speakers achieve a high level of spelling proficiency by the end of the second grade [62]. In any case, misspellings were not considered when scoring and analyzing the data.

In the classroom-based form, pictures were projected sequentially on the center of a white screen from a computer. Each stimulus was presented preceded by a fixing point (+) for 500 msec and remained on the screen for about 15,000 msec, long enough for participants to respond. The entire session lasted approximately 45 minutes. The order of presentation of the stimuli was randomly generated, but the order of the presentation was fixed.

All participants performed three tasks. These were a picture naming task, a familiarity rating task and a visual complexity rating task. These task were employed in previous picture naming literature with children [4, 12, 13, 34, 37].

For the picture naming task, participants were instructed to name each picture with the first word that came into their mind by asking “What is this picture?” They were also told that they should not worry about spelling. If they failed to recognize the picture or remember the name for a concept, they were told to specify either “Don’t Know the Name” (DKN), “Don’t Know the Object” (DKO), or “Tip-Of-the-Tongue” (TOT) by writing down the initials if they were unable to retrieve a word from memory. As in the previous literature conducted with children and because these tasks conducted collectively, not individually, no information was collected on whether the TOT responses were TOT negative (i.e., when the intended target word does not correspond to with the word that the participant was thinking of) or TOT positive (when the target word does correspond). For the familiarity task, participants were asked to rate how often they daily interact, hear or think about the concept on a scale from 1(a few times) to 5 (many times). For the visual complexity task, participants were asked to rate how difficult the picture was according to the amount of details of the picture on a scale from 1 (few details) to 5 (many details).

Participants were instructed to use an answer sheet created for this purpose, similar to previous written naming studies (e.g., [29]). The sheet contained 106 numbered lines, one for each stimulus, where participants were asked to enter the name of the concept depicted on each picture. To the right of each line, two scales from 1 to 5 were presented for the assessments of familiarity and visual complexity. Here, the participants had to indicate with an X the chosen value. The scale was adapted in a similar way to other studies with children [4, 11]. Specifically, children were instructed to use 1, 3 and 5 point ratings and these values were accompanied by squares of different sizes (1x1cm, 1x3cm and 1x5 cm, respectively) and by oral and written labels. The same scale was presented to children and adults.

In order to reduce the potential difficulties of children performing the experiment, the tasks were selected and adapted according to the following considerations: First, the selected tasks did not require judging any mental images because of the difficulty of children to understand them (see [35]), excluding variables such as image agreement or image variability. Second, the scale was adapted to facilitate its use by children [11] and participants were asked to use points 1, 3 and 5 for their ratings because children could have difficulties to use the full range of numerical values as adults do [4]. Third, a brief practice with three stimuli that were not present in the bank was presented to each participant before starting the study. At this point, questions were also clarified in order to ensure that participants had understood the instructions and the use of the scales. In addition, children were praised with positive feedback on the effort they made during the session, and with a diploma that was awarded at the end of the session. Finally, the duration and progress of the session was scheduled to prevent fatigue (see [37] for a study that finds no fatigue effect in name agreement with a session of similar duration to the one proposed here).

On the online form, stimuli were presented using Google Forms (Google LLC, CA, USA), with one stimulus per page in order to facilitate presentation until participants responded. Each stimulus was presented followed by an online answer sheet, presented as on the classroom-based form.

Scoring

Previous to the analysis, basic variants of the same name were combined when they were singular and plural forms, and when the name was written with and without spelling mistakes. For each item, the following measures were analyzed.