Therefore, the results of online testing of children should not be directly compared to those obtained in person, the researchers emphasise.

The study’s author, Natalia Banasik-Jemielniak, PhD, told the Polish Press Agency (PAP) that the idea to compare children’s grammatical development in online and in-person settings emerged after the COVID-19 pandemic disrupted traditional research with young participants.

“It was impossible, as before, to visit preschools or family homes and conduct tests/tasks. During the pandemic, both research teams and specialists conducting language assessment and therapy for children adapted and did many things remotely,” the researcher recalls.

She notes that this shift was necessary at the time. “But when we can conduct in-person research again, conducting research online is very tempting because it saves time, money, and allows us to reach people we otherwise would not be able to reach,” she says.

However, she said an important question remained: whether remote testing produces results comparable to traditional face-to-face meetings with children.

“I decided to investigate this as part of a larger project I was leading,” she says.

Banasik-Jemielniak said she hoped both formats would prove equivalent, which would allow the project to expand its reach to Polish families abroad, including children living in countries such as New Zealand or Australia.

“If online testing yielded comparable results, it would solve the logistical and financial problems,” she explains.

The Language and Humour Research Team compared children’s performance in the so-called sentence repetition task conducted in two formats: synchronously online and face-to-face. The task is commonly used to assess children’s language development, particularly grammar.

A total of 147 children aged four to seven took part in the study, including 92 multilingual children speaking Polish and English or German, and 55 monolingual Polish-speaking children.

Researchers attempted to keep the procedure as similar as possible in both conditions. In each case, children looked at a laptop screen and repeated sentences played through speakers, using the same recording and animation featuring a teddy bear searching for treasure.

“The difference was the researcher either sitting next to the child or connecting via videoconference. So the testing procedure itself was as similar as possible,” Banasik-Jemielniak says.

The results showed that the two formats are not interchangeable. Children — both monolingual and multilingual — scored higher in face-to-face testing.

The study also revealed differences between the two groups of participants.

“I expected the format of the study to have some significance, but I did not anticipate that the groups would react so differently,” the researcher says.

Monolingual children consistently performed better in face-to-face conditions. Multilingual children, by contrast, improved their scores during the second test attempt regardless of whether it was conducted online or in person.

“It was as if they adapted more quickly to the testing situation itself. This is an interesting clue that requires further research,” she says.

One possible explanation for higher scores in face-to-face testing may be the presence of the researcher, according to Banasik-Jemielniak.

“The physical presence of the researcher likely has a supportive effect — even if the task itself involves repeating recorded sentences, not interacting with a person,” she says.

The researcher refers to the theory of social presence, which suggests that communication through a screen lacks many subtle signals.

“Communication via screen is devoid of many subtle cues: microexpressions, eye contact, closeness,” she says. “For children whose communication skills are still developing, this reduction may be significant.”

She adds that meeting a researcher in person may also help establish rapport. “Besides, meeting in person with a researcher who came specifically to see the child and sits next to them probably builds a certain rapport,” she says.

Banasik-Jemielniak recalls that some children in her research remained silent when sentences were played from a laptop.

“They understood the instructions, looked at the screen, but did not repeat the sentences. However, when I started reading the same sentences to them myself, they tried. This demonstrates something that cannot be easily captured in quantitative data: for some children, a voice from a loudspeaker is not the same as the voice of a live person sitting next to them,” she says.

She stresses that the findings have practical implications for specialists working with children.

“If the format of a test can influence the results, then specialists working with children — in diagnosis, therapy, counseling, and education — need clear interpretation guidelines,” she says.

According to the researcher, the results should not be seen as an argument against remote studies.

“Rather, it indicates that the results of online research with children should not be directly compared to the results of in-person research,” she explains.

She adds that remote research remains important because it allows researchers to reach families who might otherwise be excluded.

“Remote research is still necessary and has a strong justification. It allows us to reach families who would otherwise not participate in projects due to distance, travel costs, time constraints, sibling care, or health conditions,” she says.

Online studies can also include families living outside major urban centres and groups underrepresented in traditional research.

“In practice, this means greater sample diversity and a better chance for developmental research not to describe only a narrow group of children,” she adds.

However, Banasik-Jemielniak warns against applying norms from traditional testing directly to online results.

“If a diagnostic tool is standardised in the office, we must not automatically apply the same thresholds to results obtained via a screen. A child with a score below the norm online could perform better in an in-person setting. We risk misdiagnosis,” she says.

She gives an example of parents living in the United Kingdom who want to assess the development of their bilingual child’s Polish language skills.

“A Polish speech therapist conducts the test online, because there is no other way. They use a test with norms developed in Poland, in an in-person setting. The child performs at the borderline of the norm. What does this mean? Is their language actually developing slower? Or would they perform better in a face-to-face setting? Our research suggests the latter is very likely. And that means the risk of misdiagnosis,” she says.

The researcher argues that separate standards should be developed for different testing formats.

“If the test was standardised in the office, its norms do not automatically transfer to the online version,” she says. “Secondly, in research reports — both scientific and diagnostic ones — we should always indicate the format in which the data was collected.”

The project took several years to complete, including months spent recruiting and training the research team.

“We used several tasks in our research, and it was very important to us that the researchers were thoroughly prepared. Psychological training — essential in this type of endeavour — provided a good foundation, but proficiency in specific tools was also required. The research was conducted in Germany, among other places, and organising such trips is a separate challenge,” Banasik-Jemielniak says.

She adds that coding the data was also time-consuming.

“All the children’s utterances had to be manually transcribed. No automatic transcription system, no AI, is yet capable of handling children’s speech well. Most recordings required repeated listening,” she explains.

The research was financed by the Polish National Science Centre as part of the OPUS project.

PAP - Science in Poland, Anna Mikołajczyk-Kłębek (PAP)

wl/ bar/

tr. RL