Introduction to Language Processing and Acquisition
Senior 6 · Psychology (Languages)
Introduction to Language Processing and Acquisition
Lesson 2 of 4
25 min
In 1966 Roger Brown and David McNeill at Harvard University read students the dictionary definitions of rare words, such as “a small boat used in the harbours and rivers of China and Japan”, and asked them to name the word. Often a student was sure they knew it but could not quite say it. Instead of the answer, sampan, they came up with words that sounded similar (Saipan, Siam, sarong) or meant something similar (barge, houseboat, junk). Even without the word, they could often guess its first letter and how many syllables it had.
The tip-of-the-tongue state shows that knowing a word is not a single piece of knowledge. You can have its meaning, and even part of its sound, without the whole sound form. Speaking means retrieving and assembling these parts in the right order, several times every second.
Definition
Language production: the process of turning an intention or idea into spoken, written or signed language.
Tip-of-the-tongue (TOT) state: a feeling of knowing a word while being temporarily unable to retrieve its full sound form.
Speech error (slip of the tongue): an unintended deviation from what the speaker meant to say.
Disfluency: an interruption in the flow of speech, such as a pause, “um” or “uh”, a repetition or a self-correction.
Willem Levelt (1989) proposed that speaking passes through three main stages, with a monitor that checks the result.
Figure 1. Levelt’s model of speech production (simplified).
| Stage | What happens | What goes wrong if it fails |
|---|---|---|
| Conceptualisation | Choosing what to say and how, for this listener | Saying something unclear or inappropriate |
| Grammatical encoding | Retrieving words’ meanings and grammar, building the sentence frame | Word exchanges |
| Phonological encoding | Retrieving and ordering the sounds | Sound errors and tip-of-the-tongue states |
| Articulation | Moving the speech organs | Mispronunciation, slurring |
| Monitoring | Checking inner and outer speech | Uncorrected errors |
Ordinary speakers produce roughly six disfluencies per hundred words (Fox Tree, 1995). These are not simply mistakes: Herbert Clark and Jean Fox Tree (2002) argued that “uh” warns listeners of a short delay and “um” of a longer one. Pauses tend to occur before content words that are hard to retrieve and at the start of new ideas, where planning is heaviest.
TED-Ed: “This is what happens in your brain when you can’t recall a word”, Cella Wright. Watch for how the animation explains the gap between a word’s meaning and its sound form.
Important
Evaluate the evidence. Speech-error collections are natural evidence, but they depend on what listeners notice and write down; dramatic errors get recorded, ordinary ones are missed. Laboratory-induced errors are better controlled but artificial. Levelt’s model is supported by error patterns, tip-of-the-tongue states and reaction-time studies, but critics argue that the stages overlap and interact more than a strict sequence suggests: for example, sound similarity sometimes influences which word is chosen, which a strictly one-way model would not predict.
Warning
Don’t confuse a word exchange with a sound exchange. In a word exchange whole words swap places (“a mother to my letter”); in a sound exchange only the first sounds swap (“a well-boiled icicle”). They point to different stages: grammatical versus phonological encoding.
Tip
Memory hook: Levelt’s stages run “Think – Build – Sound – Say”, with the monitor listening throughout.
Practice
Sources and credits
Written by Ganzaa: the text, the table, Figure 1 and the exercises.
