TB5 Lecture 4; Word production and errors

• Common errors in speech give clues to the underlying mechanisms and units of a language.
• A good model of a psychological process as well as being able to explain regular circumstances, also needs to have a breakdown for more difficult circumstances such as speech errors. 
  • (See mindmap of speech errors for detailed info)

Early reports of such errors, such as William Spooner's famous 'spoonerisms' (above) arent ideal for psychological analysis as they are based on third hand reports, which have a high risk of observer bias as well as no evidence of objective recorded data; not great for detailed analysis! 

Take for example; A weekend for maniacs [A maniac for weekends]

Stress = Stress 1, Stress 2

It is clear that 'weekend' and 'maniac' have swapped, a word exchange error, however;

• 1) The stress pattern (stress 1 and 2) remains the same
  • This indicates that stress patterns are not tied to particular words.
  • There is an independence of stress assignment.
• 2) The plural marker ('s) is left behind, this is called being 'stranded'. 
  
  • Similarly, the form of the plural marker agrees with the replaced, and not the original word.
    • This suggests that critical units of organisation are morphemes, not words.
    • The realisation of the form of the plural therefore is a late process.

It is asked whether spoonerisms happen at random, or if the nature of the resultant error is important. For example some speech errors are hard to determine etiologically (freudian slips).

To resolve this, slips can be induced experimentally, called SLIPs which stands for Spoonerisms of a Laboratory Induced Predisposition (Baars et al (1975)).

• Used prime sequences intended to make phoneme substitutions more likely, such as a task where you say the words silently as quickly as you can and then vocalise them when you hear an alarm.
  • It was found that SLIPs were more likely when the resulted in words over non-words.
  • E.g. dart board --> bart doard X, darn bore --> barn door = YES!
    • This suggests that perhaps we self-moniter prior to articulation and we are more likely to spot non-word errors.
• Research into taboo slips (Motley et al (1982) found a similar result. In cases where a spoonerism would result in a taboo word, (hit shed --> ...), there were weaker effects. In these cases, galvanic skin responses were higher, and in recent studys similar effects have also been shown through EEG.
  • This evidence also supports the idea that speakers (unconciously) self moniter.

Research into Freudian SLIPs (Motley et al (1979)) examined whether contextual effects may influence tendency to produce Freudian spoonerisms.

• Had 3 context conditions with heterosexual AMAB participants.
  • Electricity; Expected to get shocked
  • Sex;  Viewed provocative images
  • Neutral; No electricity or provocative stimuli
• Used the same words in each condition (the standard SLIP method)
  • Half of the potential resulting errors were provocative, the other half were related to electricity.
    • It was found that context effects biased the potential error rates.
    • Perhaps when the error is topical, self-monitering fails.

Lexicalisation; The process of converting thoughts, to sounds, to words. This is typically thought to occur in 2 stages from conceptual representation (the thought). These are;

• 1) Lemma; First, the correct lemma is selected. A lemma is an abstract lexical unit, specified both syntactically and semantically.
• 2) Lexeme; Next, the form of the word, the lexeme is selected.

For example, "beer"...a noun...an alcoholic beverage..."/bir/"

But why are there two seperate stages? This brings us back to speech errors, there are two types of whole word error, 1) semantic substitutions (wrong lemma selected) and 2) phonological substitutions (wrong lexeme selected)

• e.g/ Semantic substitution; Fingers --> toes
• e.g/ Phonolgical substitution; Equivalent --> equivocal

The TOT state is a noticeable, extended state of uncertainty about the correct form of a word, yet they 'know' it! It often involves a partial phonological recall and can be stressful and frustrating.

Brown & McNeill (1966); Tested TOT phenomenon. Gave pps definitions of low frequency words, such as 'sextant', which pps had to give the lexeme for.

• It was found that participants produced 'interlopers', secant, secton rather than the correct lexeme.
  • This can be explained as access to the lexeme combined with a failure to access the lexeme. (Partial activation)
    • Once the lexeme is selected, the word needs to be articulated (Levelt, 2001)
    • This requires a process of syllabification, a 'mental syllabary' which contains articulation plans for all the syllables we know.
    • Therefore TOT is a failure in lexeme selection.