Mechanisms of cross-situational word learning

Cross-situational word learning is the notion that people can learn a word’s meaning based on the objects/ concepts it consistently co-occurs with. It is called cross-situational learning because co-occurrence between words and objects is used across situations (i.e., no single exposure is decisive). There are now a lot of data showing that adults and children can track this information can use it to learn words.

I am interested in better understanding how people use cross-situational co-occurrence of words and and objects to learn words. For example, are people able to maintain more than one meaning for a word? What kind of individual differences between people impact cross-situational word learning?


Relevant report(s):

Roembke, T. C., & Wiggs, K. K., & McMurray, B. (accepted). Symbolic flexibility still present in first graders. Journal of Experimental Child Psychology.

Roembke, T. C., & McMurray, B (2016). Observational word learning: Beyond propose-but-verify and associative bean counting. Journal of Memory and Language, 87, 105-127.


Pruning/ eliminating incorrect information

As two items (e.g., a word and an object) co-occur, associations between the two are built. However, do we also get rid of incorrect associations that might have been built by accident? If so, how quickly does the pruning of incorrect associations happen and does this process differ depending on the kind of information that is learned?

These are questions that we have explored in a model of word learning in pigeons. In addition, my dissertation explores under what circumstances incorrect associations are lost during human word learning.


Relevant report(s):

Roembke, T. C., Wasserman, E. A., & McMurray, B. (2016). Learning in rich networks involves both positive and negative associations. Journal of Experimental Psychology: General, 145(8), 1062–74.


Overlap and variability in vowel acquisition

Vowels and particularly digraph vowels (e.g., EA, AI) are hard to learn to read in English. One of the reasons why this may be the case is that in digraph vowels a letter maps to a different sound based on the configuration it is presented in (e.g., A in EA = /ɪ/, whereas A in AI = /eɪ/).

Previous data from our lab (Apfelbaum et al., 2013) have shown that variability in irrelevant consonants can help vowel acquisition in first graders: Presenting the vowel AI in variable words such as main, raid, and sail makes it easier to isolate the correct vowel sound. However, follow-up work suggests that there are also other factors that impact the beneficial effect of variability in consonants.

In addition, we have used a motor-analog task to model vowel acquisition in adults, thus allowing us to investigate more quickly how factors such as overlap between vowels can impact acquisition and generalization.


Relevant report(s):

McMurray, B., Roembke, T.C., & Hazeltine, E. (in revision, invited submission). How can field tests of principles of learning inform reading pedagogy? Overlap and variability jointly influence sound/letter learning. Journal of Cognition and Development.

Roembke, T. C., Freedberg, M., Hazeltine, E., & McMurray, B. (in preparation). Overlap among vowels during training augments learning and retention of grapheme phoneme correspondence regularities.


Automaticity in written word recognition

Literate adults cannot help reading a word: When you see the written word green, it is impossible for you not to activate its meaning. As a result, you might find it hard to name the color the word was printed in (i.e., red; this is also part of the famous Stroop test to measure interference/ inhibition). Clearly, activating words automatically is a big part of why we can read as quickly and effortlessly as we do.

I am interested in better understanding how automaticity in written word recognition contributes to our ability to read. Is automaticity more important for some aspects of reading than others?


Relevant report(s):

Roembke, T. C., Hazeltine, E., Reed, D., & McMurray, B. (in press). Automaticity of word recognition is a unique predictor of reading fluency in middle-school students. Journal of Educational Psychology.