Skip to main content icon/video/no-internet

Frequency Effects in Word Recognition

A word's frequency is a measure of how commonly it is used. Most often, it is expressed as a relative frequency, such as occurrences per million words in a large corpus of written or spoken language. Generally, higher frequency words are recognized more quickly and/or accurately than lower frequency words, although processing is simultaneously affected by many lexical characteristics (such as word length, spelling-sound regularity, and number of phonologically and semantically similar words). While corpus frequency provides only an estimate of the average person's experience, it correlates strongly with an individual's subjective frequency estimates and is one of the best predictors of word recognition facility. This entry briefly reviews basic phenomena of frequency effects and mechanisms proposed to account for them.

Phenomena

In a basic study of frequency, the experimenter tests performance on sets of low- and high-frequency words. The experimenter must operationalize “low” (perhaps 1 to 10 occurrences per million words) and “high” (perhaps greater than 50 or 100 occurrences per million, but specific levels vary between studies). Log frequency predicts performance better than raw frequency, such that the effect of a constant difference in raw frequency diminishes as word frequency increases (e.g., a difference between 1 and 20 per million has a stronger effect than a difference between 101 and 120 per million). One can also study frequency using a regression approach and test performance on words with a continuous distribution of frequencies. In such studies, frequency tends to account for 5% to 15% of the variance in performance when other lexical characteristics are controlled.

The basic frequency effect is easily replicated, both in the rarefied conditions of isolated word processing as well as in more ecologically valid tasks, such as reading and following spoken instructions to interact with objects in a visual display. In reading, for example, fixation time on a word is inversely proportional to its frequency. Frequency can also influence performance on processing speech sounds. For example, categorical perception of a spoken phoneme continuum such as /b/ to /p/ shifts if the continuum is between a high- and a low-frequency word, as in “best-pest” (shift toward /b/) versus “pray-bray” (shift toward /p/). Frequency also interacts with other lexical characteristics such that effects of other variables (e.g., spelling-sound regularity and neighborhood) are more pronounced in low-frequency words than in high-frequency words.

Basis

A very basic question about the basis of frequency effects is whether the causal factor is cumulative frequency of exposure (as implied by using corpus estimates) or some aspect of experience. For example, high-frequency words also tend to be words that are acquired early by children, suggesting that early age of acquisition might afford a privileged status in memory and raising the possibility that frequency effects might best be understood as effects of age of acquisition. However, behavioral studies and simulations with connectionist models have established that age of acquisition and cumulative frequency both influence lexical representations. A related question concerns the impact of long-term cumulative frequency versus that of recent changes in frequency. Although changes in recent experience can make frequency effects disappear quickly (e.g., as low- and high-frequency words in a closed or fixed set are repeated, performance on low-frequency words quickly catches up to performance on high-frequency words), such effects dissipate without continued support for short-term changes in frequency. This suggests differential weighting of long-term and recent experience.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading