Everyone knows that in the sentence “Jane is washing her”, the pronoun “her” cannot refer back to Jane. Over the last four decades, researchers have established that adults reject the interpretation of Jane and her being the same person. But until they are about six-years-old, English-speaking children sometimes accept that interpretation. They accept that “Jane is washing her” can mean the same as “Jane is washing herself.”
In other words, English children are often not accurate with the interpretation of pronouns, even though they actively use them. You can pay attention to the children around you to see whether you can detect this error.
These kinds of mistakes are also made by adults learning English and new research we’re currently working on explains how this happens and how it could be used to help language teachers stamp these errors out.
Kids get it wrong
An example from a famous 1990 study shows how children get the distinction wrong. In the picture below, we see Mama bear touching her own shoulder. The experimenter says: “This is Mama Bear. This is Goldilocks. Is Mama Bear touching her?” The reasoning is that, if the children have the same grammatical knowledge as adults, they should know that “her” must refer to Goldilocks, so they should answer “no”. If they answer “yes,” it means they allow “her” to refer to Mama Bear.
The researchers, Yu-Chin Chien and Kenneth Wexler, found that five to six-year-old children get it wrong about half (51%) of the time. It’s as good as chance, or as another researcher, the late language expert Tanya Reinhart surmised: they are guessing.
Even more counter-intuitively, children are actually better at understanding questions such as “is every bear touching her”, where there is more than one bear – as in the picture below. When presented with this picture, children got it wrong only 16% of the time.
Too much going on at once
It’s truly puzzling why children are better at understanding more complex sentences than simpler ones. The answer lies in linguistic theory. Reinhart proposed that children have as good a chance of getting it right than wrong, just because it is a very complex task.
In evaluating the meaning of pronouns, children make use of something called “accidental co-reference”, typically used by adults too. To understand this, imagine the following scene from the film Side Effects. When explaining an accident suffered by one of her patients a psychiatrist, played by Catherine Zeta-Jones, says: “The patient blamed me. The patient’s husband blamed me. Even I blamed me.”
Adults understand that, under these very special conditions, “me” can refer back to the subject of the sentence: “I”. But children consider this very limited interpretation as a regular one, and juggle it together with the usual way of interpreting pronouns, to see which one fits the situation.
Evaluating two linguistic derivations at the same time is very taxing for children: it exhausts their limited processing resources. So they metaphorically throw up their hands in the air and resort to guessing.
Are reduced pronouns easier?
A 2012 study by researchers from the Massachusetts Institute of Technology, Jeremy Hartman, Yasutada Sudo, and Ken Wexler, went further and evaluated children’s understanding of full pronouns (“Cow labeled him”) versus reduced English pronouns (as in “Cow labeled ’m”). Their prediction was that children would be much more accurate with the reduced pronouns than the full ones because reduced pronouns do not allow the accidental coreference interpretation.
They found that the children they surveyed, who had an average age of five, produced 52% adult-like responses when presented with the full pronouns, but 80% adult-like responses with the reduced pronouns. In other words, the reduced pronouns were easier to understand.
Extending this to language learners
In our own forthcoming research, my colleague Lydia White and I have extended this to second language learners – French and Spanish speakers who are learning English. Our study involved two identical experiments, in which a story was presented to the participants on a computer screen in written and aural form. They heard a test sentence such as “Harry sprayed’m” in experiment one and “Harry sprayed him” in experiment two.
Participants had to judge whether a test sentence – such as Harry sprayed’m – was true or false with respect to the context. We also tested whether the number of people referenced in the sentence made a difference: they were asked if every boy sprayed him, or just one person (Harry) did, as in Chien and Wexler’s experiment above. We predicted that the complex referent (every boy) would actually turn out to be easier, because it doesn’t allow them to make accidental co-reference.
As the graph above illustrates, those learners with an intermediate level of English were more accurate when it came to the more complex test sentences referring to more than one person (the green line), compared to simpler ones with just one (the purple line). Our language learners were also more accurate with reduced pronouns versus full pronouns.
It gets easier
This means that non-advanced learners of English may have trouble interpreting English pronouns, even though they have similar pronouns in their native language. This is because processing difficulties, similar to those experienced by English children, are implicated in their interpretation of pronouns. But the good news is that they can be overcome with increased language proficiency, as the advanced learners’ behaviour indicates – the right hand points on the graph.
There is an important implication for language teaching here: teachers should be aware that pronouns are difficult words to interpret, and should give their students ample time and practice with them.