Individual Differences in Speech Error Research Program

IDSERP: Until someone comes up with a better name

Summer 2006

Eventually, this line of inquiry will make it possible to help people with language production difficulties such as children with language delays or victims of stroke and traumatic brain injury.

The mainstay of speech error research to date has been the collection and analysis of large corpora of naturally-occurring speech error data. To gather such a corpus, which usually contains thousands of individual errors, a researcher (or less often a small team of researchers) hits the streets with notebooks ready and ears peeled ready to write down all the errors heard over a given time period. Thus, all the errors, whether made by butcher, baker, candlestick maker, linguist or upstairs neighbor, are grouped together. This has made it possible to see general trends within the data so that it is possible to determine what kinds of errors are commonplace and allows one language to be compared to another. For convenience, and for obvious reasons of confidentiality, identities of individuals who perpetrate speech errors are generally not recorded. Most of my best errors remain forever publically unattributed (although I still will never forget you dear souls who made them when I had a pen in hand!)

This technique is reasonable and beneficial if the goal is to gather a general picture of what happens. In any scientific endeavor, that is always one of the first steps. However, the mingling of data may, in fact conceal important findings.

Certainly, my speech production system is similar to yours. We can, afterall, understand one another. We get the same job done. But, is my speech production system a more or less exact copy of yours like your Black and Decker Handy Steamer is more or less like my Black and Decker Handy Steamer? Or do our speech production systems differ in subtle or even not so subtle ways? We can't tell unless we compare them.

Here are some things I'd like to find out:

• I believe, based on informal observation that some people are better sources of error than others. How common is this and what counts as 'a better source of error'... What would be the average number of speech errors the average person makes in a day? As part of this ongoing research program, I started recording all the errors I make. On June 12th, 2006 I recorded 16 errors. I recorded 34 errors on June 14th. How usual are these numbers? How usual is this range?
• Are there different styles of speech production that are better in some aspects and weaker in others? That is, some people who make more than their share of lexical errors or syntactic errors?
• Could we learn to identify what type of speech production a particular person has?
• Do these types correlate with any other aspect of cognition or behavior?
• How many 'types' might there be?
• If we know what type of system any particular person has, how would that information be helpful to him or her? Would it be helpful in any pedagogical or theraputic setting?
• What influence does context have on kinds of errors: do parents of small children really make as many lexical errors as they seem to (the error where you call everyone's name before getting to the child you want)?
• I have also informally observed that some people have 'favorite' erorrs: ones they make all the time. How common is that and what characterizes these particular errors?

Hypothesis to be Tested

Hypothesis: There are no measurable differences between individual speech production systems.

Method

1. Locate between 30-60 willing, good hearted folks who care to participate.
2. Train them in
   1. What a speech error is and
   2. How it's best to record them.
   3. Turn them lose for two weeks (three if not enough data are gathered) and ask them to write down all their errors 24/7. Here is an example of one one morning of my errors
   4. Tally the kinds and amounts of errors each person makes.
   5. Look for patterns that emerge.
   6. Apply statistical tests to see to what extent these differences are robust.

   And there you have it

   As of June 14, 2006, we have 23 volunteers. If you'd like to join us Email sheri Wells-Jensen.

   I am happy to allow anyone who participates seriously to

   • Assist in categorizing his or her own errors.
   • Be part of the follow-up brain storming sessions as we look over the accumulated results.
   • Assist in preparing the data for publication.
   • Co-author the resulting paper.
   • Attend the party that results!