Home > Language, Memory > False memories in native and non-native English speakers

False memories in native and non-native English speakers

Memory – a simple word consisting of six letters. Memory – a term we frequently use to encompass a broad range of concepts. Memory – the thing that’s left after an event has long passed. But what happens when memory fails us? What happens when we fail to remember the past as accurately as we thought we would?

False memory

In cognitive research, false memories describe memories of events that did not take place or they happened quite differently from how they are remembered. The most common technique to induce false memories in a laboratory setting is a word learning paradigm called Deese-Roediger-McDermott (DRM), in which people study a list of words (such as bounce, throw, basket, bowling, and golf) that are all related to a common item (in this case, ball). When given a memory test people will often indicate that the non-presented common item (ball) was on the list with high confidence (Deese, 1959; Roediger & McDermott, 1995).

This is what researchers described as false memory: remembering something that did not happen.

There are various theories that can help us better understand how false memories happen, such as the Activation Monitoring Theory (AMT). The assumption is that when participants are studying semantically related items (items related in meaning) or phonological items (items related in sound), there is an activation of items related in meaning or sound which further activates the non-studied item in people’s minds.

An example of a semantic activation is when you are presented with a list of semantic words such as nap, tired, dream, wake, snooze, and rest then the word sleep might come to your mind because the studied items activate the word sleep.

An example of phonological activation is when you are presented with a list of phonological words such as slip, sleek, weep, heap, slot, and sleet, and then again the word sleep might come to your mind.

In addition to phonological and semantic lists, false memories can occur when participants study hybrid lists. Hybrid lists are lists of words that combine both semantic and phonological words. For example, people would study the list nap, sleek, dream, heap, rest and sleet — and guess what, the word sleep was activated in both the semantic and phonological networks, creating a hyper-additive effect (Watson et al., 2003).

A hyper-additive effect is when the activation of both semantic and phonological networks accumulate and double the false memory rates. In other words, the whole is greater than the sum of its parts. When we combine phonological and semantic items in a list, there is something special about hybrid lists that make people make a lot of mistakes and double false memory rates.

Learning about the DRM paradigm at first can be daunting because we are confronted with convincing evidence that human memory is actually not as accurate and reliable as we always thought it was. However, as the DRM showed, memory is pretty malleable and perceptive to changes and errors. We clearly remember the day in our seminar when we devastatingly looked at our professor Jen and asked her “so what’s the point of studying human memory”?

what if

After this short existential crisis, we jumped into action and came to the realization that studies on false memories not only discredit human memory and highlight its shortcomings, but also help us untangle highly complicated cognitive mechanisms. The rationale behind this is if we understand how false memories are produced then we can get important insights on how memory is organized.

All four of us (the researchers of the study) are non-native English speakers and were interested in exploring how the DRM paradigm could be used to understand language acquisition, specifically the development of semantic and phonological organizations in second language learners. Memory and language go hand in hand. A common saying states that communication is key and indeed it is. In a more and more globalized world, learning a second language is an important skill to acquire. About 93% of high schools in the United States offer a second language class (Friedman, 2015). An interesting question that arises in the context of false memories is whether second language learners are more sensitive to false memories than native speakers, or whether their semantic and phonological networks are equally developed.

Studies looking at second language acquisition (SLA) have reported that people with less language proficiency rely more on sound than meaning because they have less developed semantic networks. As proficiency increases, there is a switch from relying on phonological to relying on semantic networks (Namei, 2004). In addition, previous studies have shown that less proficient speakers were less likely to produce false memory for semantic DRM lists than more proficient speakers (Arndt & Beato, 2017). Less proficient speakers were equally likely as speakers with higher proficiency to remember the presented items, but it was harder for them to extract the gist from the studied items, producing fewer memory errors.

Down that road: The study that we conducted

The purpose of the study we conducted at Colby College was to examine how language proficiency affected the type of false memory in native and non-native English speakers. We recruited first and the second year domestic and international students from Colby. Participants studied a series of semantic, phonological, and hybrid lists, each list followed by a recall test, in which participants were asked to write down the studied items they remembered. Afterwards, they took a vocabulary test and a questionnaire to assess their proficiency level.


A cool thing we found is that the DRM paradigm caused false memories across both the native and non-native groups in all three list types. This showed that the DRM paradigm has a robust effect regardless of proficiency. It also showed that the non-native speakers, just like native speakers, have strong semantic and phonological connections. Another intriguing thing we found is that participants across proficiency groups showed hyper-additive effects when studying hybrid lists. This is a novel aspect of our study because our research was the first to examine hyper-additive effects across language proficiency.

down that memory lane

Our study has shown an interesting finding, that when we compare true memory in native speakers and non-native speakers, we can see that there are no differences across groups. In other words, non-native speakers and native speakers were able to equally remember the words as they were presented in the lists. However, when looking at false memories, non-native speakers showed a tendency to have less semantic false memories compared to native speakers, indicating that they might have underdeveloped semantic networks. This could indicate that people with less language experience might not be able to draw inferences from the world as well as people with higher language experience. For example, let’s say that one of the tasks includes to read a passage and then answer some questions about it. It could be the case that native and non-native speakers will be able to recall word for word about what happened in the passage, but when asked more complex questions in regards of the gist of the passage, non-native speakers might not do as well as native speakers. Furthermore, when looking at rates of false memories obtained from hybrid lists, we see no difference in performance across proficiency group. This indicates that once non-native speakers receive cues from bottom-up and top-down processes, they work together to form a complete picture that would allow non-native speakers to draw inferences from the environment, as efficiently as native speakers. In this sense, the DRM paradigm could be a good assessment of how well we are able to draw inferences from the world that surrounds us.

Universities and colleges in the US are making an effort to recruit students from a diverse background, including international students and students that are non-native English speakers. Hence, research on the DRM paradigm across language proficiency can help us develop strategies in order to support people with less language experience in learning environments. Especially in academic settings, it is crucial to provide adequate services and learning materials to non-native speakers in order to ensure that they perform as well as native speakers. If schools offer materials that include cues that can engage both phonological and semantic networks, non-native speakers would be able to better draw inferences from information. For example, professors could provide lecture materials in auditory and visual forms so that students could have a multi-modal input of information. This would further allow students to encode, and make use of the information faster and more efficiently than with single-modal input of information.

By the time you get to the end of this blog post, you might have forgotten what we had for the opening statement (aka forgetting is real). To refresh your memory (get it???), we asked what happens when we do not remember the past as accurately as we thought we would. The answer is that although false memories are real, to have false memories is not the end of the world as long as we are aware that our memories do not reflect an absolute truth (because there is no such thing).

This research was conducted by Zhichun (Robbie) Yu, Miriam Kopp, Vianny Lugo Aracena and Adela Ramovic with the assistance of professor Jen Coane.


Arndt, J. & Beato, M.S. (2017). The role of language proficiency in producing false memories. Journal of memory and language, 95, 146-158. doi.org/10.1016/j.jml.2017.03.004.

Deese, J. (1959). On the prediction of occurrence of particular verbal intrusions in immediate recall. Journal of experimental psychology, 58, 17-22. doi:10.1037/h0046671

Friedman, A. (2015, May 10). America’s lacking language skills: Budget cuts, low enrollments, and teacher shortages mean the country is falling behind the rest of the world. The Atlantic. Retrieved from https://www.theatlantic.com/education/archive/2015/05/filling-americas-language-education-potholes/392876/

Namei, S. (2004). Bilingual lexical development: a Persian–Swedish word association study. International Journal of applied linguistics, 14(3), 363–388. doi:10.1111/j.1473-4192.2004.00070.x

Roediger, H. L., & McDermott, K. B. (1995). Creating false memories: Remembering words not presented in lists. Journal of experimental psychology: Learning, Memory and Cognition, 21, 803-814. doi:10.1037/0278- 7393.21.4.803

Watson, J. M., Balota, D. A., & Roediger, H. I. (2003). Creating false memories with hybrid lists of semantic and phonological associates: Over-additive false memories produced by converging associative networks. Journal of memory and language, 49(1), 95-118. doi:10.1016/S0749-596X(03)00019-6

  1. April 22nd, 2018 at 15:38 | #1

    This research is very interesting; I believe that we briefly touched on your research in our cognitive psychology course this semester. Reading through your post made me wonder how these results could be expanded to age, as well. Following the findings from your study, I predict that while children (who have a less developed mental lexicon) might demonstrate relatively equal true memory, they would demonstrate fewer false memories in the semantic DRM. This reminds me of the findings from the Stroop task, which suggests that children (whose reading capabilities are less automatic than those of an experienced reader) have an easier time successfully performing the external task demands of color naming. This is because given their reading capabilities are less automatic, they are better able to inhibit reading the word. I drew these connections between these two different topics from cognitive psychology because they both converge on the idea that children have a less developed (either reading or semantic) system than do most adults. I also wonder, therefore, how a non-native english speaker might perform on the Stroop task relative to a native english speaker. Perhaps, they would have slightly (or significantly) fewer errors due to the fact that reading in english might be less automatic. It is likely that if the Stroop task were conducted using less common color names (violet, rose, gold), the results would be even more significant.