Psycholinguistics/Neural Bases of Lexical Access

IntroductionEdit

Lexical Access and the BrainEdit

This chapter aims to explore and determine the neurological aspects of lexical access in the brain. Lexical access, or word retrieval, is a process that has undergone significant study and been the focus of debate in psychology for quite some time. Thus far, studies have identified brain regions that are accessed for semantic, syntactic, and phonological lexical tasks. In effect, there are certain neural areas that are activated for processing these tasks. Although the early models of accessing the mental lexicon were formulated in recent decades, relatively recent developments in technology have vastly improved the understanding of the actual neural circuits or basis of this process. The introduction of Event-Related Potentials (ERP) was an integral first step, paving the way for further techniques such as Magnetoencephalography (MEG), Functional Magnetic Resonance Imaging (fMRI) and Positron Emission Tomography (PET) scans to provide an unprecedented amount of spatio-temporal information pertaining to neural activation. This chapter aims to demystify the bases of lexical access in the human brain, explaining the intricacies, conditions and factors that affect lexical retrieval, using data to identify neural systems that correlate with their respective brain areas.

Fundamentals of Neuronal ProcessingEdit

On a neuronal level, it is convergence and divergence from a neuron that determines the direction of the processing that occurs. In other words, the direction of the stimuli that a neuron experiences governs feedforward and feedback responses, which are both unique in the nature of their pathways. Feedforward processing is analogous to bottom-up processing, where the basic elements of the stimulus are interpreted and used as a foundation to construct a response. Essentially, the stimulus is interpreted as “building blocks”, upon which higher processing is then based upon. On the other hand, the feedback pathway can be likened to top-down processing, where the final construct or unit is dissected into its building blocks and information is broken down from a larger concept. Since it primarily operates via the dissection of the stimuli, it is working backwards in relation to feedforward processing; it is thusly referred to as feedback processing. In the context of lexical access, a feedforward response can involve the interpretation of a visual grapheme into a phoneme, after which the phoneme is processed into a morpheme which is then interpreted semantically etc. The process continues until a word is retrieved from the mental lexicon. Conversely, the feedback pathway occurs in the opposite direction, with the word being correctly identified due its semantics being used contextually in a sentence.

Studies Exploring Word FrequencyEdit

ERP StudiesEdit

Event-related potential studies have been useful in identifying the time process of lexical access. Most notably, the N400 component of an ERP has been shown to reflect the brain activity during word retrieval. It was found that less frequent words (words found to be used less in the English language) elicited a stronger N400 response than more common and frequent nouns; this suggests that the brain was making a greater effort to perform lexical retrieval of the less frequent word.[1]

 
An example of a N400 peak from a typical ERP chart. For words with low and high frequency, the peak would generally be higher and lower accordingly. The peak can also be negative in some instances.

As stated above, it has been shown that there is a strong relation between frequency of a word in language and the time and effort taken for lexical retrieval. This general relation states that the less frequent a word is, the greater the time taken for lexical access. In terms of gauging word frequency, the most oft quoted corpus is the Kucera and Francis MRC Psycholingustic Database [2]. This relation has also been shown behaviourally, with eye tracking showing gaze duration as being longer for unfamiliar, less frequent words than their counterparts.[3] Less frequent words (closed class) such as nouns take a few hundred milliseconds longer to process than their more frequent synonyms.[4]

Language comprehension relies on a variety of codes or cues for processing, such as orthographic, thematic, phonological, semantic and syntactic cues; the decoding of each of these cues was thought to be encapsulated into specific modules of the brain.[5][6] Less frequent words also have weaker orthographic, phonological and thematic cues, all of which have been thought to play a role in lexical retrieval. These cues are interpreted by the brain and are used to facilitate lexical access; hence, weaker cues will result in slower access of words.[7]

fMRI StudiesEdit

However, a recent study explores the notion of an inter-related network being drawn upon for lexical processing. An fMRI was conducted on subjects to examine lexical and syntactic processing in reading comprehension using a lexical decision task. Results from the study suggest that a variety of lexical and syntactic factors affected activation of a wide portion of the left hemisphere, including the aforementioned perisylvian areas, as opposed to being localized in a specific area.[8] A similar purported view is that the interaction of different areas of the brain enhances the efficacy of lexical access.[9][10] From the study of the activation of brain cortical regions using fMRI, it was deemed that several areas were integral for lexical processing, and not just confined to one modality as previously thought. A point of consensus is that the left inferior frontal gyrus is indeed involved, however, its role is in question. Some researchers assert that the left IFG (also containing Broca’s area) is concerned with semantic processing of words, while others stipulate that is involved in the selection of multiple meanings.[11]

In a separate study, it was found that even experienced computer typists were not spared from the frequency effect; low frequency words were typed slower, further establishing the inhibitory effect of low frequency on lexical access.[12]

MEG StudiesEdit

Magentoencephalography has shown to be particularly useful in correlating brain activity to lexical access. In particular, there are three MEG response elements that pose the most significant relation to lexical brain activation: the M170, M250 and M350 components. The M170, being the earliest component, is suggested to reflect the earliest occipito-temporal activity in visually processing the word form, chiefly by processing the letter strings that are present. The M250 is at a later latency, and thought to deal with the phonotactic processing, or interpretation of the phonological, orthographic or thematic components as a sort of medial level decoding, as mentioned above. However, is the M350 component that has been subjected to the most scrutiny, as it is believed to coincide with the actual tapping of the mental lexicon in word retrieval.[13] Moreover, studies have shown that M350 can be likened to the N400 component of an ERP, in terms of their representation of the effect of lexical frequency on word retrieval. The M350 latencies, along with their amplitudes, were found to be particularly sensitive to word frequency.[14] Furthermore, it was found that the neighbourhood density, meaning the similarity of certain parts of a word with each other (such as Have and Shave), are thought to elicit greater responses of the M350. As such, it is suggested that the brain indulges a much higher effort in discriminating between words that are in the neighbourhood density, as multiple phonemic variations of the word are accessed, but only one can be selected as the correct form.[15] This strengthens the relation between phonological cues and lexical access, as the similarity or differences between cues and the strength of the cue itself can facilitate or impede lexical retrieval to a certain extent.

With regards to word frequency, the less frequent a word, the higher the amplitude of the M350 response. This is similar to the behaviour of the N400 component. In addition, word repetition, which entails that a word be repeated several times throughout the course of an experiment or sentence, has shown to create a smaller M350 response; this suggests that less mental effort is required to process repeated words. To summarize, neighbourhood density of words is linked to an inhibitory effect on the M350, while word frequency and repetition are suggested to facilitate it.[16]

In another MEG study conducted by Embick and colleagues, a lexical decision task used six frequency word categories, each defined by a one-step decrease in the log frequency; the latencies of the M350 component were analyzed to solidify its relation to word frequency.[17] The word frequencies for this experiment were selected from the Cobuild corpus. The main difference for this experiment was that a single word system was used, as opposed to a sentence structure to present the stimuli. The results however, were similar to the experiments noted above. Three primary peaks were found at 150 ms, 250 ms and 350 ms. The response times for the M350 component ranged from 357 – 593 ms for the most frequent words, while the least frequent words elicited responses ranging from 392 – 732 ms. A point suggested by the researchers was that repetition of words that were presented had a facilitating effect on the M350 latency; semantic priming, achieved by presenting the subject with a single word with a similar meaning to the subsequent word, also had the same effect.[18]

Orthographic/Visual Lexical Retrieval StudiesEdit

Initially, word retrieval for vocal or silent verbalization was thought to be in the left cerebral hemisphere, prominently featured in Broca’s area and Wernicke’s area, otherwise known as the classic language areas. It has indeed been shown that Perisylvian structures play a major role in the phonemic representation of word forms, also known as the grapheme to phoneme conversion.[19]

A PET study performed by researchers on patients with brain lesions in their telencephalon regions and normal control individuals identified areas responsible for word retrieval.[20] More specifically, areas of retrieval for unique and non-unique entities were identified. Participants were given a visual naming experiment, with three categories of pictures being presented: well-known persons (unique entity), animals (unique and non-unique) and tools (non-unique).

Regional cerebral blood flow was measured using a PET scan; the experiment yielded strong results indicating the left temporal pole (TP) and left interotemporal (IT) sectors were involved in unique and non-unique items respectively. It was found that the left TP was stimulated in naming persons, while the left IT sector was particularly activated by animal names. However, it is important to note that these regions are not thought to be primarily responsible for storing words; rather, they have an intermediary role in lexical retrieval. It is far more likely that these regions interpret the immediate information being perceived, such as the phonological or orthographic nature of the stimulus.

Role of Ventral Extrastriate Cortex in Visual Word ProcessingEdit

In terms of interpreting visual data, the ventral extrastriate cortex has been found to play a role in early processing of visual word form, giving rise to subsequent phonological or semantic processing. [21] Studies on orthographic forms of lexical access have specified a number of regions activated by visual lexical decision tasks, including the occipital pole, lateral and basal occipito-temporal lobes, the superior and middle temporal gyri and the inferior parietal lobule.[22]

In particular, the lateral extrastriate cortex was attributed to actively processing the visual information; the left medial extrastriate cortex was found to be most activated by the actual word, and word-like stimuli. It is interesting to note that activation in these regions was seen bilaterally as well; this challenges the notion that language stems solely from the left hemisphere. It can be surmised the right hemisphere plays a role in visual word-form lexical access.[23]

 
An fMRI scan showing brain activity in several brain regions, including Broca's area, from a lexical decision task with variable word frequency. The activity in the visual extrastriate region is noteworthy as well.

Further expanding upon lexical retrieval from a visual source, it has been shown that brain activity brought about by an orthographical code is variable as well. It has been suggested that hand-writing does indeed inhibit or slow down lexical access to a certain degree, insinuating the brain does indeed discriminate between typeface and hand-writing.[24] The study found faster response times in visual naming tasks for typeface as opposed to cursive hand-writing. This delay can be attributed to the longer time taken for the grapheme to phoneme conversion that occurs during visual naming, as described above. This effect was made even more apparent for words with low frequency.

In terms of the direction of processing, the results indicated that participants were less sensitive to phonological effects when making word/pseudo-word judgements, as they were thought to rely on the familiarity of the word; this entails that they relied on top-down processing to distinguish between the stark differences of words and pseudo-words.[25] However, forward processing was still utilized in discriminating between words in which the differences were less obvious, such as pull and lull. As mentioned previously, an additional factor to be considered is the semantic top-down processing which was most notable when a low frequency word was written in cursive hand-writing. As such, the researchers suggested this was due to the weak orthographic and phonological cues which could be gleaned from the complexity of the cursive hand-writing; the brain then utilized semantic processing for lexical access. In this case, the lateral extrastriate cortex was activated bilaterally, similar to the visual tests done in the aforementioned experiments. A similar study conducted by Hellige and Adamson found that the right hemisphere is indeed more active than usual (since the left hemisphere is typically far more dominant) for reading handwriting than print.[26]

Lexical AmbiguityEdit

A study conducted by Fiebach, Sandra H. Vos and Angela Frederici went further to determine the neural correlates of sentence comprehension (in terms of lexical ambiguity) for readers with high and low working memory capacities. The participants were subjected to a visual sentence comprehension task, where sentences were shown with and without ambiguous parts.[27] An fMRI was done on the participants, with the aim of identifying the neural bases of deciphering syntactic ambiguity. It was found that, as suggested by earlier studies, that the left-dominant network in the perisylvian region was indeed involved in sentence processing of ambiguous words. However, they pointed out that it was the superior region of Broca’s area (BA 44) and the parietal region which showed a proportional activation increase to the length of the ambiguous component of the sentence. Thus, they concluded that the BA44 activation increases as the demands for sentence comprehension increase.

As had been suggested by Just and Carpenter, working memory capacity did indeed affect brain activation; low span readers demonstrated greater activity for the same brain regions as their high span counterparts.[28] It is suggested that these readers place a larger effort on their cognitive mechanisms to process sentences, especially the ambiguous ones. This could be attributed to the lexical accessing of multiple meanings in these regions, with time taken to select the correct word depending upon the context of the sentence. In terms of identifying all the brain regions, Fiebach et al. found that processing for lexically ambiguous words occurred in the left posterior temporal sulcus, the middle temporal gyrus, left anterior superior temporal gyrus and the left superior temporal sulcus. These results are fairly consistent with other studies done investigating lexical ambiguity. Moreover, it was noted that the posterior inferior frontal gyrus (Broca’s area) also underwent greater activation depending on the complexity of the ambiguous sentence. However, since this was a visual experiment, activity was also seen in the right hemisphere in the posterior superior temporal sulcus, posterior third of the inferior frontal gyrus and in the anterior insula, supporting evidence from earlier studies.

Patients with lesions in Broca’s area (pars opercularis and pars triangulis), and in some cases temporal and parietal areas have exhibited weaker ability to discern thematic codes from a lexically ambiguous word in a sentence.[29][30]

Moving deeper into to the realm of lexical ambiguity, several MEG studies have been conducted to determine the effect of an ambiguous word on brain activity and MEG component latency. A study performed by Ihara et al. utilized MEG and fMRI imaging to identify the neural mechanisms of lexical access of ambiguous words.[31] Subjects were presented with words in the Japanese script, both in sentences and in a single fashion. It was found that the left posterior temporal and inferior parietal area showed significant activation, along with the left anterior middle/inferior temporal area, for sentences containing words which were ambiguous (having more than one meaning). However, the context of the sentence was shown to have been a significant factor in brain activation of these areas, due to greater activation for sentences than single words. In addition, the inferior frontal cortex was activity was influenced by ambiguities, as was seen from fMRIs of ambiguous and clear meaning words. Specifically, the left posterior inferior frontal cortex showed clear context effect for the unambiguous words, but not for ambiguous ones; the activation of the pIFC was larger for related ambiguous words, signifying that multiple meanings of the word had to be accessed.

In addition, the MEG component also demonstrated that response times were delayed for ambiguous words, as opposed to unambiguous ones. This can be explained via a study by Pylkkanen and colleagues which determined that the presentation one homonym (word with same pronunciation and spelling, but different meaning) delayed the M350; this delay can be attributed to the inhibition of the competitor word.[32] Overall, the MEG results did demonstrate differences in latencies (and hence brain activation) for ambiguous and unambiguous words. The 200-300 ms components were larger for ambiguous words compared with unambiguous; the early occurrence of these components suggest that lexical ambiguity is deciphered rather quickly using contextual information found in sentences by the brain. Furthermore, the MEG also showed activity till the 700 ms component, encompassing a variety of peaks, most notably of which occurred at 400 ms. This component was thought to be related to the N400 component.[33] Also explored in the experiment was the effect of semantic priming: the amplitude of MEG components and brain activation seen from the fMRI scans showed less activity for words that were semantically primed. fMRI imaging was correlated with the MEG component to show a timeline of brain activation: the left posterior frontal cortex was activated at roughly 250 ms, followed by the left anterior inferior frontal cortex at 350 ms, which was then succeeded by the anterior medial temporal lobe at 400 ms.

In a related study, it was shown that the distinction between the functions of the anterior and posterior portions of the left inferior frontal cortex were that the former dealt with semantic processing, while the latter was concerned with phonological processing.[34] A further point of note is the role of the right hemisphere, which has shown to also play a important part in the processing of lexically ambiguous words as determined by lesion studies.[35]

Summary Table for Lexical RetrievalEdit

A very brief summary of this chapter is provided in the table below. Please keep in mind that this is information is highly abridged; for a deeper understanding of the underlying concepts and experiments, please refer to the chapter above, or to the actual sources listed in the references section. Due to the high level of research being done in this area, this table may warrant further addition to its content as more theories are being developed; please feel free to edit the table and the chapter or add to it constructively in light of new or updated information.

Type of Study Distinguishing Feature Explanation of Feature
Event-Related Potentials N400 Less frequent words elicit a higher N400 response, while a smaller N400 is indicative of more frequent words. The size of the N400 is representative of the mental effort involved in word retrieval.
Magnetoencephalography M170 Indicative of earliest level of processing, using visual cues of words in their most basic form, such as letter strings.
M250 Represents second stage of processing, whereby phonemes and more complex levels of a word are accessed.
M350 Corresponds to retrieval of the word from the mental lexicon. Exhibits behaviour similar to the N400 with respect to word frequency; less frequent words evoke a smaller and delayed M350. Neighbourhood density of words (in phonemic terms) can also has an inhibitory effect on lexical retrieval.
Functional Magnetic Resonance Imaging Left Inferior Frontal Gyrus Activated in almost all lexical retrieval tasks. Thought to play a role in semantic processing. Posterior Region show to be accessed for multiple meanings.
Perisylvian Structures This region includes Broca's and Wernicke's areas, and so play a role in the phonemic interpretation of words
Lateral Extrastriate Cortices Involved in the initial processing of visual word stimuli.
Broca's Area (BA 44) Suggested to be involved in lexical ambiguity in sentence comprehension as well.
Positron Emission Tomography Left Temporal Pole Shown to be involved in retrieval of unique words, such as proper nouns.
Interotemporal Pole Shown to be involved in retrieval of non-unique words such as common nouns.

Learning ExerciseEdit

The following learning task is available to be viewed at your convenience, and is intended to facilitate your understanding of the basic concepts discussed in this chapter. Even though some of the content discussed above can be overwhelming, the learning exercise will help you digest the information in a simple, straightforward manner.

ReferencesEdit

  1. Van Petten C, Kutas M Interactions between sentence context and word frequency in event-related brain potentials. Mem Cognit. 1990 Jul;18(4):380-93.
  2. Kucera F, Francis WN (1967) Computational analysis of present-day American English. Providence, RI: Brown University Press.
  3. Rayner, K., Sereno, S. C., & Raney, G. E. Eye movement control in reading: a comparison of two types of models. Journal of Experimental Psychology. Human Perception and Performance (1996) 22, 5, 1188-200.
  4. Raney, G., & Rayner, K. Event-Related Brain Potentials, Eye Movements, and Reading. Psychological Science (1993) 4, 5, 283-286.
  5. Fodor M. A method for obtaining measures of local parsing complexity throughout sentences. J Verbal Learn Behav. 1983 22:203-218.
  6. Frazier L (1987) Theories of Sentence processing. In: Modularity in knowledge representation natural-language understanding (Garfield JL, ed.) pp.291-307. Cambridge, MA:MIT Press.
  7. Bradley DC, Garrett MF, Zurif EB (1980) Syntactic deficits in Broca’s aphasia. In: Biological studies of mental processes (Caplan D, ed.), pp 269-286. Cambridge, MA : MIT Press.
  8. Keller TA, Carpenter PA, Just MA. The Neural Bases of Sentence Comprehension: a fMRI Examination of Syntactic and Lexical Processing. Cereb. Cortex. 2001; 11(3): 223-237 doi:10.1093/cercor/11.3.223.
  9. McClelland JL. (1987) The case for interactionism in language and processing. In: Attention and Performance. XII. The psychology of reading (Coltheart M, ed.) pp 3-36, London: Lawrence Erlbaum.
  10. Just, M. A., & Carpenter, P. A. A Theory of Reading: From Eye Fixations to Comprehension. Psychological Review. 1980; 87(4): 329-54.
  11. Thompson-Schill, S.L., D'Esposito, M., Kan, I.P. Effects of repetition and competition on activity in left prefrontal cortex during word generation. Neuron. 1999; 23: 513–522.
  12. Inhoff, A. W. Word frequency during copytyping. Journal of Experimental Psychology: Human Perception and Performance. 1991; 17: 478–487.
  13. Stockall, L., Stringfellow, A., and Marantz, A. The precise time course of lexical activation: MEG measurements of the effects of frequency, probability, and density in lexical decision. Brain and Language. 2004; 90: 88–94. DOI:10.1016/S0093-934X(03)00422-X.
  14. Pylkkanen, L., Marantz, A. Tracking the time course of word recognition with MEG. Trends Cogn. Sci.. 2003; 7: 187–189.
  15. Stockall, L., Stringfellow, A., and Marantz, A. The precise time course of lexical activation: MEG measurements of the effects of frequency, probability, and density in lexical decision. Brain and Language. 2004; 90: 88–94. DOI:10.1016/S0093-934X(03)00422-X.
  16. Stockall, L., Stringfellow, A., and Marantz, A. The precise time course of lexical activation: MEG measurements of the effects of frequency, probability, and density in lexical decision. Brain and Language. 2004; 90: 88–94. DOI:10.1016/S0093-934X(03)00422-X.
  17. Embick, D., Hackl, M., Schaeffer, J., Kelepir, M., and Marantz, A. A magnetoencephalographic component whose latency reflects lexical frequency.Cognitive Brain Research. 2001; 10: 345–348. DOI:10.1016/S0926-6410(00)00053-7.
  18. Pylkkanen, L., Stringfellow, A., Flagg, E., & Marantz, A., (2000). A neural response sensitive to repetition and phonotactic probability: MEG investigations of lexical access. Proceedings of Biomag 2000, Helsinki, Finland.
  19. Damasio, H., Grabowski, T.J., Tranel, D., Hichwa, R.D., and Damasio, A.R. A neural basis for lexical retrieval. Nature. 1996; 380: 499-505.
  20. Damasio, H., Grabowski, T.J., Tranel, D., Hichwa, R.D., and Damasio, A.R. A neural basis for lexical retrieval. Nature. 1996; 380: 499-505.
  21. Keller, T.A., Carpenter, P.A., Just, M.A. The Neural Bases of Sentence Comprehension: a fMRI Examination of Syntactic and Lexical Processing. Cerebral Cortex. 2001; 11(3): 223-237. DOI:10.1093/cercor/11.3.223.
  22. Hart, J.Jr., Kraut, M.A., Kremen, S., Soher, B., and Gordon, B. Neural Substrates of Orthographic Lexical Access As Demonstrated by Functional Brain Imaging. Neuropsychiatry, Neuropsychology, and Behavioral Neurology. 2000; 13(1): 1-7.
  23. Hart, J.Jr., Kraut, M.A., Kremen, S., Soher, B., and Gordon, B. Neural Substrates of Orthographic Lexical Access As Demonstrated by Functional Brain Imaging. Neuropsychiatry, Neuropsychology, and Behavioral Neurology. 2000; 13(1): 1-7.
  24. Barnhart, A.S., and Goldinger, S.D. Interpreting Chicken-Scratch: Lexical Access for Handwritten Words. Journal of Experimental Psychology. 2010; 36(4): 906-923. DOI:10.1037/a0019258.
  25. Barnhart, A.S., and Goldinger, S.D. Interpreting Chicken-Scratch: Lexical Access for Handwritten Words. Journal of Experimental Psychology. 2010; 36(4): 906-923. DOI:10.1037/a0019258.
  26. Hellige, J. B., & Adamson, M. M. Hemispheric Differences in Processing Handwritten Cursive. Brain and Language, 2007; 102(3): 215-227.
  27. Fiebach, C.J., Vos, S.H., and Friederici, A.D. Neural Correlates of Syntactic Ambiguity in Sentence Comprehension for Low and High Span Readers. Journal of Cognitive Neuroscience. 2004; 16(9): 1562–1575.
  28. Just, M. A., & Carpenter, P. A. A capacity theory of comprehension: Individual differences in working memory. Psychological Review. 1992; 99: 122–149.
  29. Dronkers NF, Wilkins, DP, Van Valin, RD, Redfern BB, Jaeger JJ. A reconsideration of the brain areas involved in disruption of morphosyntactic comprehension. Brain Lang. 1996; 47:461-463.
  30. Caplan D, Hildebrandt N, & Makris N. Location of lesions in stroke patients with deficits in syntactic processing in sentence comprehension. Brain. 1996; 119:933-949.
  31. Ihara, A., Hayakawa, T., Wei, Q., Munetsuna, S., and Fujimakia, N. Lexical access and selection of contextually appropriate meaning for ambiguous words. NeuroImage. 2007; 38: 576-588. DOI:10.1016/j.neuroimage.2007.07.047.
  32. Pylkkanen, L., Llinas, R., Murphy, G.L. The representation of polysemy: MEG evidence. J. Cogn. Neurosci. 2006; 18: 97–109.
  33. Pylkkanen, L., Llinas, R., Murphy, G.L. The representation of polysemy: MEG evidence. J. Cogn. Neurosci. 2006; 18: 97–109.
  34. Stoeckel, C., Gough, P. M., Watkins, K. E., & Devlin, J. T. Supramarginal gyrus involvement in visual word recognition. Cortex; a Journal Devoted to the Study of the Nervous System and Behavior (2009) 45, 9, 1091-6.
  35. Rodd, J.M., Davis, M.H., Johnsrude, I.S. The neural mechanisms of speech comprehension: fMRI studies of semantic ambiguity. Cereb. Cortex. 2005; 15: 1261–1269.