This experiment was originally designed to better understand how children might learn sight words for early reading. Traditional reading requires visual-verbal integration, a process by which a visual symbol is repetitively paired with a word until the symbol automatically evokes memory for that specific word. After much practice with visual symbols and their associated words, the words are eventually read in a fluent fashion, rather than requiring the reader to sound out words phonetically. A familiar word becomes associated with a visual orthographic image (Ehri, 1991). A picture or a visual symbol (e.g., dog) holds the same meaning, but the printed word doesn't look at all like a furry, domesticated pet.

This experiment investigates how learning is impacted when students attempt to memorize names of line-drawn symbols under one of two different conditions: word repetition or semantically elaborated sentences. Word repetition involves hearing each stimulus word three times. Incorporating semantically elaborated sentences involves hearing each stimulus word in a meaningful sentence that relates the word to its visual symbol. Using these two conditions, word recall with and without semantically-related sentences is assessed. To test word recall, a 15-item word list is read (one at a time) to the student accompanied by black-and-white line drawings. In this paradigm, the symbol (drawing) becomes a cue for the word. Children see the symbols and hear the words in order to verbally recall as many words as possible. The task is to determine if sentences help the learner associate the target word with its line-drawn symbol for recall. The words include five common nouns, five common verbs, and five common adjectives. Semantically elaborated sentences range from five to six words, each including a single C-unit (communication unit containing an independent clause and modifiers).

Using this paradigm, previous research has examined the impact of introducing semantically elaborated sentences (Littlefield & Klein, 2005; Littlefield, Klein, & Coates, 2012). According to Nilsen and Bourassa (2008), words that have greater semantic richness (concrete words) conjure images that make them easier to learn. Past work on word-learning paradigms (Kiraly & Furlong, 1974) found that concrete words were learned better than abstract words. To understand the meaning of words, specifically a written word, an orthographic-semantic-phonologic connection is established. In other words, the combination of written letters represents speech sounds that form a word and a meaningful concept. According to Strain, Patterson, and Seidenberg (1995), when the orthographic-phonologic connections don't work as efficiently, as is the case with learning and recalling less familiar words (i e., sword, thread), semantic input helps.


In the control condition of this experiment, participants hear 15 words, each repeated three times while viewing associated line-drawn symbols. In the experimental condition, participants hear the words three times: once in isolation prior to the sentence, once within the sentence, and then once in isolation following the sentence. The sentences chosen provide a potential connection between the word and the line-drawn symbols. Consider this symbol:

The visual symbol can be paired with the word pirate. In the control condition, participants would see the symbol while they hear "Pirate, pirate, pirate." In the experimental condition, participants would see the same symbol while hearing "Pirate. The pirate wears an eye patch. Pirate." The task involves seeing the symbol and remembering the word it represents.

In this experiment, the student is randomly assigned to either the control condition (word repetition only) or the experimental condition (semantically elaborated sentences). Memory recall scores are obtained. Do you predict that you will perform better in the control or experimental condition? In other words, what is your hypothesis?

Experimental data is collected across three memory recall trials (free recall, cued recall, and recognition). One at a time, participants see the 15 line-drawn symbols, each for 5 seconds, accompanied by the name of the symbol (with or without an associated sentence). After stimuli presentation, three recall tasks are completed and tabulated as follows:

1. Free Recall: Total number of words remembered (ranging from 0 to 15)

IImmediately after stimulus presentation, write the names of all the symbols you can remember on a pre-numbered sheet of paper. Only write down the words that you are confident about. Then, enter the number of symbol names you retained (from 0 to 15) into the computer and set aside your list. Do not refer to it during the rest of the experiment.

2. Cued Recall: Total number of correct symbol names (0-15)

INext you will see each symbol presented on the computer, and you will try to name it (type the names). Be sure to give the specific names of the symbols that you were taught.

3. Recognition Total: Total number of words correctly identified (0-15)

The recognition task involves reading a list of presented words and identifying only the words corresponding to those line-drawn symbols viewed previously. Choosing words from the same semantic class (but incorrect) are semantic errors (i e., chicken for bird). Choosing words sounding similar to the correct word (but incorrect) are phonemic errors (i e., burn for bird).

Data Format and Download

The table below shows a sample spreadsheet for a class of eight students. UserID is assigned upon completion of the experiment. ClassID, sex, age, and date of participation are self explanatory. The Condition is coded for you based on random assignment.

Sample data image from the AMT experiment

Example Data Analysis

To test the main question of whether or not learning ability varies by condition, control and experimental groups are compared for the dependent measures:

Free Recall Total
Cued Recall Total
Recognition Total
Semantic Errors (collected during Recognition trial)
Phonemic Errors (collected during Recognition trial)

Note: For columns listed with 'Total', the higher the number (up to 15) the better one's performance. For columns listed with "Errors", the lower the number (down to 0) the better one's performance.
Once individual raw data and class summary data are calculated for the Association Memory Test, different figures can be generated to depict the findings. For instance, a team of students could plot their individual performance to see if Recognition tends to be the highest score and Free Recall tends to be the lowest score (Figure 1). The impact of experimental group membership (hearing the word and a sentence) on learning can be visualized with a bar graph (Figure 2). To see how the two conditions can impact error scores, the means for the two types of recognition errors can be shown using a line graph (Figure 3).

Figure 1 - Learning Performance by Case
Figure 1 - Learning Performance by Case

Figure 1- Learning Performance by Group
Figure 2 - Learning Performance by Group

Figure 3 - Recognition Errors by Group
Figure 3 - Recognition Errors by Group

Applications and Extensions

Findings from this experiment apply to cognitive theory and may inform teaching practice. As in the case across cognitive psychology learning experiments, recognition is anticipated to be easier than cued recall, and cued recall is anticipated to be easier than free recall. Through this experiment, the three scores can be compared for individual and group data.

Additional cues may not always enhance recall of information. Sometimes, providing more details during the learning process can actually impede encoding and later recall, especially with one-trial exposures. In this experiment, semantically elaborated sentences were given to enhance recall, but for some, the added information may actually cause recall accuracy to decrease. When individuals are susceptible to semantic interference, limiting the amount of information to be encoded can lead to more successful learning. Were you susceptible to interference? Take out the sheet of paper you completed during the free recall phase of the experiment. Review the learning trial again to find out how many words you recalled correctly. Compare findings with your peers. Students who were randomly assigned to the experimental condition (semantically elaborated sentences) may discover a higher rate of errors. Learning can be more successful when extraneous information is limited.

Sentence stimuli from the Association Memory (AM) Test (Klein & Littlefield, 2000, unpublished manuscript) were designed to provide meaningful connections between symbols and words to enhance recall. However, Littlefield, Klein, and Coates (2012) found that students who were not exposed to the sentences (control group) achieved higher recall scores on average than the experimental group (who heard semantically elaborated sentences). Further, Littlefield and Klein (2005) revealed that students diagnosed with dyslexia recalled fewer AM Test words than typically-reading students when both groups heard the semantically elaborated sentences.

Why might reading ability relate to AM Test performance? Two potential theoretical explanations are offered here. One explanation is that working memory may function differently in typically-achieving readers and poor readers. Working memory describes a thought process wherein information is temporarily held and manipulated. Integration of differentially-coded information (visual with verbal) is an important part of new learning that occurs during working memory processing. Research shows that people diagnosed with dyslexia tend to have lower scores than age- and intelligence-matched peers on working memory measures (Pickering, 2006; Wang & Gathercole, 2013). Another explanation for why reading skill may influence AM Test performance is that visual-verbal paired associate learning is related to reading ability (Warmington & Hulme, 2012; Windfuhr & Snowling, 2001). In paired associate learning, pairs of items are presented together so that one eventually becomes a memory cue for the other. Paired associate tasks requiring verbal output are most closely related to reading ability (Litt, de Jong, van Bergen, & Nation; 2013). Since dyslexia is a language-based learning disability, the verbal demands of the AM Test may, in part, have influenced the group differences that were previously observed.


Ehri, L. (1991). Development of the ability to read words. In R. Barr, M. Kamil, P. 
Mosenthal,  PHandbook of reading research, 2,  383-417.  New York: Longman. 
Kiraly, J., & Furlong, A. (1974). Teaching words to kindergarten children with picture, 
configuration, and initial sound cues in a prompting procedure. Journal of Educational Research,
Klein, E.R., & Littlefield, L.M. (2000). Association Memory Test.  Unpublished manuscript. 
Department of Psychology, Drexel University, Philadelphia, PA. 
Litt, R. A., de Jong, P. F., van Bergen, E., & Nation, K. (2013). Dissociating crossmodal and 
verbal demands in paired associate learning (PAL): What drives the PAL-reading relationship?  
Journal of Experimental Child Psychology, 115,  137-149. doi: 10.1016/j.jecp.2012.11.012
Littlefield, L.M., Klein, E.R. (2005). Examining visual-verbal association in children with and without
reading disorder. Reading Psychology,26,  363-385. doi:10.1080/02702710500285706 
Littlefield, L.M., Klein, E.R., & Coates, S. (2012). An experimental evaluation of the effects of 
using training sentences to aide young children's word recall. 
 Effective Education (First online article), 1-15. doi:10.1080/19415532.2012.712840

Pickering, S. J. (2006). Working memory in dyslexia (Ch. 2). In T. P. Alloway and S. E. Gathercole's
Working Memory and Neurodevelopmental Disorders.  New York: Psychology Press. 
Nilsen, E., & Bourassa, D. (2008). Word-learning performance in beginning readers.
Canadian Journal of Experimental Psychology, 62,  110-116. 
Strain, E., Patterson, K., & Seidenberg, M.S. (1995). Semantic effects in single-word naming.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 21, 1140-1154. 
Wang, S. & Gathercole, S. E. (2013; Epub ahead of print). Working memory deficits in children
with reading difficulties: Memory span and dual task coordination. Journal of Experimental Child Psychology. doi: 10.1016/j.jecp.2012.11.015
Warmington, M. & Hulme, C. (2012). Phoneme Awareness, Visual-Verbal Paired-Associate
Learning, and Rapid Automatized Naming as Predictors of Individual Differences in Reading Ability. Scientific Studies of Reading, 16, 45-62. doi: 10.1080/10888438.2010.534832
Windfuhr, K.L., & Snowling, M.J. (2001). The relationship between paired associate
learning and phonological skills in normally developing readers. Journal of Experimental Child Psychology, 80, 160-173.