Issue 8 |

Psycholinguistic experiments: From fingers and eyes to what’s in our heads

by on September 20, 2016

Picture this. You are in a crowded train station where everyone, including yourself, is dressed in black. Suddenly, you spot just one person in white. You are likely to stare at that conspicuous white spot for a while before you turn your attention away. And you do this because you have spotted something out of the ordinary. On the other hand, if you happen to be looking for a friend in this crowded station and you spot three people who look like him from the back, you are also likely to stare at the backs of this threesome, to try and make out who among them is friend.

Our natural responses of paying more or less attention to things in various circumstances are important to psycholinguists, who make it their business to study, amongst other things, how sounds, words, phrases and sentences are stored in people’s heads. For example, we might suspect that the word worm is stored as a sequence of four letters w-o-r-m. We might also suspect that it is stored as a sequence of three sounds, which linguists represent as /wɜːm/. And it might even be that the mention of worm calls up simply the mental image of a creepy crawly rather than any letter or sound sequences. Psycholinguists often like to take responses from people’s fingers and eyes as cues to how linguistic units are stored in people’s heads. In what follows, I will attempt to explain how these cues work.

Responding with our fingers

Our fingers are a cue to how we mentally store our words and sentences because we can press buttons with them. We can use our fingers to press the buttons of what psycholinguists call a “response box”, which is a box designed to track the time taken for its button to be pressed.

The principle behind how this box works is not rocket science. If by pressing a button, we signal a readiness to turn our attention from what was in front of us to something else, then the time it takes before that button is pressed shows how much attention we had paid to that thing in front of us. And if we give more attention to things that stand out (one clothed in white amongst many clothed in black), and to things that are quite alike (picking out our friend from three lookalikes), we can assume that when the time elapsed before a button press is rather long, what has been observed is either very uncommon or overly conventional.

Therefore, when measurements from a response box show people to take longer to respond to a linguistic unit, that unit is either not mentally stored in the manner presented, or is closely similar to other stored linguistic units. We may expect, for example, the sequence of sounds/letters for beeswax to be mentally stored to a greater extent than for waxbees, since people would not frequently encounter waxbees. And if we do get measurements of longer response times to waxbees than to beeswax, our expectation is somewhat confirmed. If response times to beeswax turn out, however, to be not so different than waxbees, this may not mean that our expectation of the uncommonness of waxbees to people is incorrect. Rather, it may signal that the initial sequence of letters <be> in beeswax is quite extensively found in people’s mental store of words, e.g. the <be> in beset, belittle, become and before. If so, it is unsurprising that the time taken to respond to beeswax is as long as that for waxbees, since people would have been trying to fish out beeswax from their large store of initial <be> sequences.

What I’ve said about orthographic similarities can also apply to phonological similarities. In this regard, you may want to hear about the response box-based study of Goh, Suárez, Yap and Tan (2009), who were interested in people’s recognition of words. In particular, they were interested in how this recognition was influenced both by: (i) the frequency of the words themselves; and (ii) the volume of words that sound similar to the words tested. The latter effect is often referred to by psycholinguists as ‘neighbourhood density’. In Goh et al.’s study, the words requiring recognition were played to people through headphones, following which people had to signal as quickly as they could whether what they had heard was a word or non-word by pressing the relevant buttons on a response box. The researchers discovered that people took a longer time to recognise low-frequency (or less common) words than high-frequency (or more common) words. What was also discovered was that word frequency (or commonness) had greater influence on the time taken for word recognition if the word tested was linked to a small rather than high volume of other similar-sounding words, i.e. low rather than high neighbourhood density. Beyond this study, it might be interesting to investigate whether the discoveries in Goh et al. apply equally to learners of a language as to proficient users, and what this might mean for language teaching.

Responding with our eyes

You might be wondering that if our fingers can shed light on people’s mental store of linguistic units, wouldn’t our eyes do the job as well? If we encounter waxbees and beeswax while reading, it is fairly safe to say that for most of us, our eyes are the most direct means of detecting any uncommonness or otherwise in these words. Indeed, devices known as ‘eye-trackers’ are designed to track our eye movements for this purpose. The technique associated with these devices, called ‘eye-tracking’, captures how our eyes respond to letters, words and phrases, complete sentences, and even visual images.

In the same way that people tend to take longer to respond with their fingers to linguistic units that are not quite stored in their heads in the manner presented, people may stare longer at these units. For instance, the uncommonness of waxbees may cause longer stares, or, if you prefer the in-group way of saying this, keep people’s eyes ‘fixated’ longer at this word than at beeswax. The uncommonness of waxbees may also license multiple readings of it (or ‘multiple fixations’).

However, as you might have guessed from the analogy of how staring at three lookalikes in black is as likely as staring at the one odd person in white, it is not uncommonness alone that encourages informative eye responses. These responses can also tell us when the same form of a linguistic unit is stored in our heads, but in more than one way. To explain this, it is helpful to talk about garden-path sentences, or sentences that have a high likelihood of being incorrectly parsed (or interpreted). An example of a garden-path, quoted from Fernández and Cairns (2011, p. 213), is The horse raced past the barn fell. People may shift their eyes back to reread this sentence if they initially took raced as a verb in the simple past tense; this, however, would leave fell hanging incoherently at the end. The correct way to parse this sentence is to take raced as a past participle form, as in The horse [that was] raced past the barn fell, where [that was] in [that was] raced is elided. Any evidence of an eye shift back to reread a sentence like this can then signal that the form raced is mentally stored in more than one way, perhaps with its function of the simple past being more salient than its function of the participle.

For a taste of an actual eye-tracking experiment, we can refer to the work of Jian, Chen and Ko (2013), who were keen to observe how non-specialists dealt with reading specialist physics terms in Chinese—these terms being ideographic rather than alphabetic. One discovery here was that while people did not spend a noticeably long time looking at the Chinese specialist terms on first reading, they did so in multiple readings. As the researchers suggest, it is not unusual for specialist physics terms in Chinese to comprise in themselves familiar linguistic units; an example given is the presence of the Chinese characters for electricity (電) and feel (感) in the Chinese specialist term for inductance (電感). One conclusion reached therefore was that when people first read the Chinese specialist terms, it was the familiar characters within them that were mentally evoked, and the familiarity was what explained why there were no prolonged stares at the specialist terms as a whole. It was only in multiple readings that the familiar characters were recognised as components of less common terms, which then led to prolonged stares. For the sake of fanning any sprouting curiosities arising from this work, it might be interesting to find out whether readers of Chinese who also know English may be prone to read English specialist terms by first evoking some meaningful component within those terms, e.g. induct within the word inductance, rather than the term as a whole.

There are various reasons that influence the extent to which linguistic units are recognised. Sometimes, it’s to do with how common or otherwise these units are. And sometimes, it’s to do with how similar in form they happen to be to other units that we also know. What’s important to psycholinguists are the heaps that fingers and eyes can tell us about how linguistic units are stored in our heads.


References

Jian, Y. C., Chen, M. L., & Ko, H. W. (2013). Context effects in processing of Chinese academic words: an eye-tracking investigation. Reading Research Quarterly, 48(4), 403–413.

Fernández, E. M., & Cairns, H. S. (2011). Fundamentals of psycholinguistics. Chichester: Wiley-Blackwell.

Goh, W. D., Suárez, L., Yap, M. J., & Tan, S. H. (2009). Distributional analyses in auditory lexical decision: neighborhood density and word-frequency effects. Psychonomic Bulletin & Review, 16(5), 882–887.

Leave a Comment