Semantically implied irregular inflection.

Author:Deibler, Scott


This study examines the effect on inflection by implied irregularity through manipulation of semantic information. Old English words and non-words are used here in an effort to increase their novelty and, thus, decrease their familiarity with the Modern English-speaking participants. It is expected that the participants would favor the regular formation of the past tense in an Old English verb when the irregular nature of the verb could not be inferred semantically because the sentence object appeared in Old English. They would then favor irregular formation of the past tense once the sentence object is revealed in Modern English and the verb could be inferred to be irregular. Participants rated their preference for three inflectional forms of a series of Old English words under conditions where the object of the sentence was in Old English or Modern English. The results favored the regular form when the object was in Old English and showed no significant difference between the regular and irregular forms after the object was presented in Modern English. The effect of semantics upon the inflectional preference of the participants appears to support the single-model theory of language, which theorizes that language is processed in a generalized manner like any other stimuli. This would cover the processing of all relevant stimuli, including semantics, which matches the findings of this study. It is hoped that the novel approach employed here will encourage further research to settle the ongoing theoretical debate within psycholinguistics between supporters of the dual-mechanism and single-system theories of linguistic cognition.

For the last quarter-century, psycholinguistics has been embroiled in a debate between two dialectically opposing theories concerning how the human brain processes language. The disagreement has concentrated on the English past tense form, which is created by the predictable addition of the suffix ed to the root verb in the majority of English verbs (e.g., acted, walked). Yet, there are as many as 180 exceptions where an irregular past tense form is created in an erratic manner, such as seek-sought. The English past tense has been the central focus of this language-processing research since it has these two distinct forms which could be determined differently by these opposing theories. (1) One should bear in mind that in English the verb can also be isolated from syntax (word order), since syntax has no effect upon tense. (2) These opposing theories differ in the cognitive processes that they believe are used to determine inflection, that is, the variation of verb form due to a change in tense, and whether semantics should influence inflection.

The dual-mechanism theory is based upon the premise that the brain is a collection of functionally specific regions and that certain areas have evolved to process language. This theory posits that a main area of the brain processes the structures found in language that follow distinct rules and patterns, such as regular past tense in English. (3) A secondary associative memory area learns the exceptions to the main rules, like irregular past tense in English, and stores them entirely. (4) When determining which past tense form to take for an English verb, the word is first compared to the known exceptions, taking the matching irrational form if found. If not, the regular past tense verb form is created by the addition of -ed. (5) Evidence of the over-regularization of verbs exists to corroborate this theory. (6)

The single-system theory posits that the brain is a homogenous general processor of information comprised of countless neural connections which interact with each other and process all stimuli in the same manner, regardless of their source. Language information is therefore processed in a single generalized routine just as other conditions from the environment. An input enters the neural network, is processed along with all other relevant data, and an output is determined. This theory is embedded in the idea that language is processed like other stimuli, so the differing outputs are contextually specific based upon the differing additional data. (7) The past tense form of an English verb would therefore be determined not by comparing the exceptions or adopting the rules, but rather by factoring in other data such as meaning, spelling, and sound while processing the verb. (8)

Both theories provide explanations for the general processing of language, but they are less able to cope with specific details in actual research. One of the main methods of research has been measuring reaction time during priming tasks, which determine whether a preceding prime word influences the response to a related target word. (9) Studies performed using morphologically similar primes (e.g., rake, raking) (10) find that the reaction times to the paired taget words are decreased. However, these results are not consistent with other studies. (11) Cognitive psychologists Lorraine K. Tyler, Emmanuel A. Stamatakis, Roy W. Jones, Peter Bright, Kadia Acres, and William D. Marsten-Wilson performed a study on four participants with semantic deficit, which found only one with irregular inflective impairment, which, they argued, would not be possible under the single-system theory. (12) Psycholinguists Marc F. Joanisse and Mark S. Seidenberg conducted a computer simulation which indicated that irregular inflective impairments could occur from deficits related to semantics (meaning) and phonology (sound), thus countering the assertion put forth by Tyler et al. (13)

The effect of semantics upon irregular inflection has been explored by psycholinguist Michael Ramscar, who proposed that a study of semantics could yield results to end this debate. (14) He posited that if semantics played an influential role in verb inflection, participants would inflect novel non-words (e.g., frink) in the way they would similar sounding English verbs (e.g., blink for regular or drink for irregular) based upon the semantics of the sentence. (15) His study placed these non-words (e.g., frink) into paragraphs, where their meaning could be interpreted and associated with either regular (e.g., blink) or irregular (e.g., drink) forms, and the participants were asked to create past tense versions of these verbs. (16) Ramscar found that most participants produced past tense versions as predicted by the semantic inference. (17) The impact of this research, however, has been downplayed by critics who argue that Ramscar did not isolate semantics and only presented participants with a choice between the two associations, such as drink or blink. (18)

There is a large quantity of cognitive research involving the English past tense. Much of it has been conducted to support preferential theories or to undermine opposing theories. (19) Methods of research that have yielded contradictory results are repeatedly used while novel approaches seem to escape inquest. Novel research needs to be stressed in order to resolve this dispute. Accordingly, this study pursues a novel approach to cognitive research involving semantic...

To continue reading