03 March 2017

Speaking Rate and Information

Welcome back. Isn’t it time I returned to the subject of linguistics? If you don’t count Speaking in Whistles, it’s approaching five years since I voiced my concern about the use of “mom” instead of “mother” and of “no problem” instead of “you’re welcome” in Linguistic Longings.

Today, I’ll address speed: (1) Do you talk fast, slow or somewhere-in-between? (2) Do fast talkers convey more information than slow talkers in the same amount of time?

Let’s take the first, first. The answer might depend on where you’re from.

Rate of Speech
In case you missed it, about a year ago, various news media reported the results of an analysis by a mobile advertising analytics company, Marchex. Applying its “Call DNA” software to over four million consumer phone calls to businesses between 2013 and 2015, Marchex generated state-by-state rankings of the callers’ rate of speech, density of speech (wordiness) and hold times before hanging up.

While the source of the data and apparent lack of differentiation by age or gender might lead one to question their accuracy, the results are at least fun trivia.


States with fastest and slowest talkers from
www.marchex.com/2016/02/02/talkative/
Faster talkers tend to reside in the North (fastest: Oregon, Minnesota, Massachusetts), slower talkers in the South (slowest: Mississippi, Louisiana, South Carolina).

Going beyond rate of speech, Marchex found lower density (less wordy) speakers tended to reside in the Central U.S. (Oklahoma, Kansas, Wisconsin, Minnesota, Iowa), while higher density speakers were more scattered in the Northeast, Mid-Atlantic, West and Southwest (New York, California, New Jersey, Nevada, Maryland). As an example, Marchex noted that a New Yorker used 62% more words than an Iowan in the same conversation with a business.

Callers who hung up the quickest when put on hold tended to be in the Northeast, mid-Atlantic and upper Midwest (Kentucky, Ohio, North Carolina, New York, West Virginia). Callers in the Midwest and South (Louisiana, Colorado, Florida, Illinois, Minnesota) were most patient or had nothing more pressing to do.

OK, but do those fast-talking Oregonians convey more information than those slow-talking Mississippians?

Information Content of Speech
A recently published study by a Brown University researcher found that, regardless of how fast people speak, they convey about the same amount of information in a given period of time.

To reach this conclusion, the researcher analyzed two collections of conversations. He focused first on the Switchboard Corpus, about 2400 two-sided telephone conversations among 543 speakers (302 male, 241 female) from all areas of the United States. He then replicated the results using the Buckeye Corpus, 40 speakers (20 old, 20 young, 20 male, 20 female) from Columbus, Ohio, conversing freely with an interviewer.
 

Pawing through the
lexical information.
The researcher estimated speech rate based on the actual and expected durations of words, calculated from mathematical measures of a word together with the previous and following words. He estimated information rate from calculations of two linguistic criteria, lexical (dictionary definition) and structural (syntax).


Introduction to Linguistics video:
www.youtube.com/watch?v=Q9LpTZkQeZs
Linking Speed to Content
The study found that, compared to slow speakers, fast speakers were likely to produce lower lexical information (i.e., use less informative words) as well as lower structural information. For example, fast speakers were more likely to use active voice (e.g., I will do it) than passive voice (e.g., It will be done by me).

The findings suggest that people converse within relatively narrow bounds of communication data to avoid providing too much or too little information in a given period of time. On average, speakers either speak quickly or provide high information content, but not both.

Wrap Up
One finding of interest is male speakers provided more information than did female speakers for a given rate of speech. Although untested in this study, the researcher judged that such gender differences might follow from sociolinguistic factors. Could it be that males are more concerned with speaking than with being understood?

Thanks for stopping by.

P.S.
Marchex analysis of consumer phone calls: www.marchex.com/2016/02/02/talkative/
Article on Marchex analysis on The Atlantic website: www.theatlantic.com/entertainment/archive/2016/02/speaking-fast-and-slow/459393/
Brown University study in Cognition journal:
www.sciencedirect.com/science/article/pii/S0010027716302888

Article on Brown University study on Science Daily website:
www.sciencedaily.com/releases/2017/01/170117140005.htm
Switchboard Corpus: catalog.ldc.upenn.edu/ldc97s62
Buckeye Corpus: buckeyecorpus.osu.edu/