Skip to main content icon/video/no-internet
icon backReturn to Entries

Cohort Model of Auditory Word Recognition

The core idea at the heart of the cohort model is that human speech comprehension is achieved by processing incoming speech continuously as it is heard. At all times, the system computes the best interpretation of currently available input combining information in the speech signal with prior semantic and syntactic context. Originally proposed in 1980 by William Marslen-Wilson and Lorraine Tyler, the cohort account has been subject to ongoing refinement in response to new empirical data and neural network simulations. Predictions of the model for neural responses to speech are currently being tested.

Origins of the Cohort Model (1973 to 1985)

During the 1970s, response time data collected by Marslen-Wilson and others demonstrated the speed and accuracy of speech perception and comprehension. Native speakers can shadow (i.e., repeat aloud) heard sentences with minimal delays between perception and production while correcting for mispronunciations of key words. Detection tasks similarly show rapid word identification in sentences with participants responding within 300ms of the start of a target word—substantially before all of the relevant speech has been heard. Early identification is achieved by comparing incoming speech with known lexical items (for example, the word trespass is distinct from all other words once the segment /p/ has been heard) in combination with contextual cues provided by the sentential context.

The cohort model proposed by Marslen-Wilson and Tyler in 1980 thus suggests that word identification begins with initial activation of a set of candidates that match the start of a spoken word (the word-initial cohort that for trespass would include words such as tread, treasure, treble, etc.). Activated candidates are rejected as incompatible speech segments are heard; word recognition occurs when information in the speech signal uniquely matches one single word (the uniqueness point). The importance of the sequential structure of spoken words in predicting word recognition has been confirmed in a range of response time tasks. For example, people can decide that speech does not match a real word (i.e., make lexical decisions about spoken pseudowords) as soon as they hear a segment that deviates from all spoken words. Hence, decision responses occur with a constant delay when measured from the /s/ of the tromsone or the /p/ of trombope; it is at these positions that participants can determine that these pseudowords are not the familiar word trombone. These findings uniquely support accounts of speech perception in which word recognition can occur when sufficient information has been perceived in the speech signal and provide strong support for the first instantiation of the cohort model.

Computational Instantiations (1986 to 1999)

The predictions of the cohort model for the timing of word recognition depend on assessing phonetic information that is shared by or distinguishes between different spoken words. These can be estimated from computerized pronunciation dictionaries revealing the proportion of items (like doll embedded in dolphin) that challenge early identification. A further methodological development was the use of cross-modal priming to test the time course of word recognition. Consistent with cohort assumptions, Pienie Zwitserlood showed that speech that matches multiple words (e.g., the start of captain or captive) activates multiple meanings to a degree that is modulated by prior sentence context. Word frequency also affects activation of word candidates, so captain would be more active than capstan because of its higher frequency of occurrence. These findings motivated a revision to the cohort theory in which word activation and identification are graded processes that combine the speech input, lexical information, and contextual cues.

...

  • 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

Read next

More like this

Sage Recommends

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

close

Have you created a personal profile? Login or create a profile so that you can save clips, playlists and searches

close
close
close Signed In to profile You are signed in as:
close
Logged in Institution Institution