Cohort model

The cohort model in psycholinguistics and neurolinguistics is a model of lexical retrieval first proposed by William Marslen-Wilson in the late 1980s. It attempts to describe how visual or auditory input (i.e., hearing or reading a word) is mapped onto a word in a hearer's lexicon. According to the model, when a person hears speech segments real-time, each speech segment "activates" every word in the lexicon that begins with that segment, and as more segments are added, more words are ruled out, until only one word is left that still matches the input.

The cohort model relies on a number of concepts in the theory of lexical retrieval. The lexicon is the store of words in a person's mind.; it contains a person's vocabulary and is similar to a mental dictionary. A lexical entry is all the information about a word and the lexical storage is the way the items are stored for peak retrieval. Lexical access is the way that an individual accesses the information in the mental lexicon. A word's cohort is composed of all the lexical items that share an initial sequence of phonemes., and is the set of words activated by the initial phonemes of the word.

The cohort model is based on the concept that auditory or visual input to the brain stimulates neurons as it enters the brain, rather than at the end of a word. This fact was demonstrated in the 1980s through experiments with speech shadowing, in which subjects listened to recordings and were instructed to repeat aloud exactly what they heard, as quickly as possible; Marslen-Wilson found that the subjects often started to repeat a word before it had actually finished playing, which suggested that the word in the hearer's lexicon was activated before the entire word had been heard. Findings such as these led Marslen-Wilson to propose the cohort model in 1987.

The cohort model consists of three stages: access, selection, and integration. Under this model, auditory lexical retrieval begins with the first one or two speech segments, or phonemes, reach the hearer's ear, at which time the mental lexicon activates every possible word that begins with that speech segment. This occurs during the "access stage" and all of the possible words are known as the cohort. The words that are activated by the speech signal but are not the intended word are often called "competitors." Identification of the target word is more difficult with more competitors. As more speech segments enter the ear and stimulate more neurons, causing the competitors that no longer match the input to be "kicked out" or to decrease in activation. The processes by which words are activated and competitors rejected in the cohort model are frequently called "activation and selection" or "recognition and competition." These processes continue until an instant, called the recognition point, at which only one word remains activated and all competitors have been kicked out. The recognition point process is initiated within the first 200 to 250 milliseconds of the onset of the given word. This is also known as the uniqueness point and it is the point where the most processing occurs. Moreover, there is a difference in the way a word is processed before it reaches its recognition point and afterwards. One can look at the process prior to reaching the recognition point as bottom-up, where the phonemes are used to access the lexicon. The post recognition point process is top-down, because the information concerning the chosen word is tested against the word that is presented. The selection stage occurs when only one word is left from the set. Finally, in the integration stage, the semantic and syntactic properties of activated words are incorporated into the high-level utterance representation.

...
Wikipedia