What is Lexical Access?
In order to understand what lexical access is, it is first important to briefly explain what the mental lexicon, lexical entries and lexical storage are. The lexicon refers to a systematic organization of vocabulary that is stored in the mind in the form of individual lexical entries . It has been referred to as our mental dictionary and analogies between accessing a written dictionary and accessing the mental lexicon have emerged. For further insight into the concept of the lexicon, please refer to the section of this course titled The Mental Lexicon. Lexical entries are defined as the information stored in the mind regarding a specific word. In order to recognize and comprehend words, also known as lexical items, information about its content is needed. Levelt (1989) suggested that lexical entries contain two types of information that allows individuals to recognize and understand words. This information includes content about the form and meaning of lexical items, Figure 1 . The form component of lexical entries refers to phonological and morphological information, while in contrast, the meaning component refers to the syntax and semantic information of lexical entries .
Lexical storage refers to the way in which lexical items are organized for optimal accessibility in the lexicon. It is important to mention here that there are two types of items: lexical items which carry a meaning (i.e. nouns, verbs, adjective and adverbs) and grammatical items which do not have a clear meaning but contribute to syntactic structure (i.e. conjunctions) . Only words which carry meaning are stored in the lexicon; these words are stored associatively in the mind in relation to other items . For example, when searching our lexicon for the word apple, the storage of associated items may appear as illustrated below, Figure 2.
Now that some background information has been provided on the mental lexicon, lexical entries, and lexical storage - lexical access will be discussed. Lexical access is defined as the way which individuals access words in the mental lexicon . Studies have identified several factors that can affect lexical access, such as: the frequency effect, the word/non-word effect, word superiority effect, the length effect and the imageability effect        . Some of the above mentioned factors will be discussed in greater depth below. Much research has been devoted to investigating the concept of lexical access, as understanding how lexical items are accessed is central to speech and writing comprehension.
Factors that affect Lexical Access
In visual and auditory modalities, the more frequently a lexical item is used the more quickly it is recognized . Zipf's Law (1949) developed by George Kingsley Zipf draws upon the relationship between probability of usage and frequency. Further, this law is based on the principle of least effort . The basis of the least effort principle can be outlined by the following logic: the more frequently a word is used the easier it is to process, as mentioned above. Studies have suggested that low-frequency lexical items produce longer decision times and therefore are accessed more slowly in comparison to high-frequency lexical items . As such frequency plays an important role in determining which lexical item is chosen in models of lexical access that involve competition between two items. Moreover, the frequency effect has been well attested to in studies of lexical access   . For example, a study by Balota & Chumbly (1984) revealed that high frequency words were named more rapidly than low-frequency words .
Word Concreteness and Imagery
Words such as camera and banana are easy to imagine in our mind, where as words such as justice and evil are more difficult to mentally picture. This issue which relates to the difficulty and ease of picturing some words in comparison to others refers to the concept of word concreteness and abstractness. Word concreteness is also known as imageability and as implied by its name, is the ability to visualize lexical items. Concrete words are those that describe tangible nouns where as, abstract words describe nouns which may be intangible (i.e., apple and freedom) respectively. Please see Figure 3 for further examples of concrete and abstract words. Several studies have attested to the notion that concrete lexical items such as apple are easier to imagine, while abstract words such as evil are not as easy to imagine . Moreover, studies on the word concreteness effect have revealed consistent findings that concrete lexical items have been found to be processed more accurately and quickly than abstract concepts in a variety of cognitive tasks  . These tasks include but not limited to: cognitive recall, lexical decision, word recognition and sentence comprehension  .
More specifically, Paivio (1969) found that high-imagery words were more easily recalled in an memory test than low-imagery words . This study also found that the principle of imageability interacts with the principle of frequency in word access. In short, that high-frequency high-imagery words such as student were more accurately accessed and recalled where as low-frequency low-imagery words such as excuse were least easily accessed . Bleasdale (1987) also found that in a lexical decision task, words primed other words only when both words were of the same, for example concrete-concrete, rather than concrete-abstract. From this it was concluded that the lexicon organizes concrete and abstract words separately .
Two models have emerged in an attempt to explain the concreteness effect, one of which is the dual-coding theory and the other, the context availability theory. The latter theory supports that concrete words are activated with broader contextual verbal support and greater contextual associations in semantic memory  . The context availability theory further asserts that concrete words automatically activate more associative information in comparison to abstract words, thus resulting in faster processing. In contrast, the dual-coding theory as proposed by Paivio (1986) upholds that concrete words result in faster processing due to access to semantic meaning  . This theory support that the meaning of concrete words is facilitated easier because concrete words have verbal and visual semantic representations, while abstract words generally have a verbal semantic representation . The superiority effect of concrete words being processed more accurately and quickly than abstract words in cognitive tasks has been attributed to three reasons. First, it is believed that abstract words lack the direct sensory referents of concrete words, therefore supporting a slower processing for abstract words. Second, there is a greater availability of contextual information for concrete words than abstract words  . Lastly, the notion that concrete words have more associative meanings than abstract words is thought to attribute to this superiority findings swayed toward concrete words .
Ambiguities in spoken or written language can arise in a number of ways. For example, homophones are two words that sound the same however have two different meanings (i.e., air; heir). In light of the existence of ambiguities in language, a great deal of research has been devoted to understanding lexical ambiguity to date . The primary goal of this research is to address how readers and listeners retrieve the contextually appropriate meaning of lexical items which have multiple meanings . One key discovery that has fueled research into this topic is the understanding that words and meanings do not necessarily a one to one ratio . As most individuals have likely experienced, one meaning can be shared by numerous words and conversely, one word can share numerous meanings. To illustrate my latter example, I would like to use the phrase “tick the right box” . Without context, the reader is left wondering if right refers to the correct box or the box opposite to the left. Through examples such as the one used above it was suggested that individuals use more cognitive resources when processing ambiguous words than words with only one meaning  . Readers and listeners often use the context of a word to disambiguate its meaning . To emphasize the importance of context and its effect on disambiguating words, studies such as one by Swinney (1979) have been conducted. In this study, experimenters provided participants with the following sentence: "The man was surprised when he found spiders and other bugs in the room" . The targeted word was bug and participants were tested to see if context had an effect on which meanings were activated; the possible answers for bug were either an insect, a spy gadget or the control word sew . The results showed that if the answer choices were given in less than 200 milliseconds after the target sentence was read, both meanings of the word were activated; but if the answer choices were given after 200 milliseconds the irrelevant meaning was suppressed . These results support that if given time to reflect on the sentence, context helps individuals disambiguate the word bug to correctly mean an insect.
Exhaustive and Selective Lexical Access
There are two major theories that examine the role of context in influencing what meanings of lexical ambiguous words are activated. The first theory known as selective access which supports that context biases the interpretation of an ambiguous word, so that only the intended meaning is accessed . In essence this view considers context to sufficiently provide enough information that only the most relevant meaning of an ambiguous word is activated. In contrast, the alternative theory claims that even with contextual cues, multiple meanings for ambiguous words are still activated; this theory is known as exhaustive access . Further, some advocates of the exhaustive access theory have argued that ambiguous words do not all simultaneously get activated, but rather the more frequently used meanings and those influenced by context are activated first. Currently, the exhaustive access theory has more empirical support from studies than the selective access theory .
Dominant and Subordinate Meanings
As mentioned above, several studies have revealed that lexical access of ambiguous words can be dependent on context. Ambiguous words can also be subcategorized into dominant and subordinate meanings as often two meanings of a word not equally used . Thus the dominance of a meaning refers to the relative frequency each meaning of an ambiguous word is used. For example, if we were to review the word "cram", a university student would be more likely activate the dominant meaning of studying, rather than a subordinate meaning of try and squeeze something into an insufficient space. To view some examples of dominant and subordinate meanings, please see Figure 4. Furthermore, dominant and subordinate meanings can be divided into polarized and balanced ambiguous words. Polarized words are those with meanings that have a predominant meaning which is most frequently used in relation to the word. In comparison, balanced words are ambiguous words which do not have one dominant interpretation for the word (i.e. right can mean either correct or a direction).
Words and Word Recognition
Lexical Access and Semantic Priming
Semantics is the study of word meaning and the ways in which words are related to one another in our mental lexicon. Semantic priming is the unintentional increase in speed or accuracy when responding to a stimulus such as a word or a picture that has been previously primed. For example, if an individual sees a target word of doctor and then subsequently is shown the word nurse, their response time would be expected to decrease. To visualize semantic priming in action, please see Figure 5 which depicts Collins & Loftus's (1975) Spreading Activation Model . As predicted, studies have shown that when participants are shown a word or picture that relates to a target word, their response time decreases as a result of semantic priming . Please refer to Psycholinguistics/Semantics_in_the_Brain for more information on semantics and how it relates to lexical access.
Models of Lexical Access
How language users recognize a lexical item’s meaning is an important concept. Thus the models of lexical access attempt to explain how individuals access words and their related meanings in our minds. There are two major classes of models that detail how lexical entries are retrieved during reading and listening tasks. The first type of model are known as serial search models, where as the second type are parallel access models .
Serial search models believe that when we encounter a word, we look through all lexical entries to determine whether the item is a word or not, and then retrieve the necessary information about a word (i.e., its semantics or orthography). Serial search means propose that lexical access occurs by sequentially scanning one lexical entry at a time . An example of a serial search model is Forster’s (1976) autonomous search model. In contrast, the parallel access models propose that perceptual input about a word activate lexical items directly, and that multiple entries can be activated at once. That is, a number of potential candidates are activated simultaneously and the lexical item which shares the most features with the targeted stimulus is the one that is chosen . Examples of the parallel search model are Marslen-Wilson’s (1987) cohort model, McClelland & Seidenberg's (1989) connectionist model and Morton’s (1969) logogen model . The factors that influence word access and lexical organization are addressed in both the serial and parallel processing models. At the present time there is a greater acceptance toward the parallel access models than the serial search models when explaining lexical access .
The Autonomous Search Model
The autonomous search model was developed by Kenneth Forster and views the word recognition process as being divided into several parts . More specifically this model upholds that lexical access is carried out in a two-stage process. Forster’s (1976) model of lexical access is best illustrated by comparing the lexicon to a library. A word similar to a book, can be in only one place in the lexicon and library. However, several catalogs (i.e., title, author, year) can be used to determine where the book or lexical item are located in both locations . Forster proposed three major types of access files which included: orthographic, phonological and semantic/syntactic . The first type of access file mentioned (orthographic) means that words are accessed based on their visual features; words accessed through the phonological access file are done so through how they sound; and lastly, words retrieved using the syntactic/semantic file are done so according to their meaning. Moreover, input from any modality (visual, auditory) can only access these files one at a time. Lastly, orthographic and phonological access files contain information about the beginning parts of words (i.e., the first few letters of their spelling or first few sounds they begin with) . When a word is presented either visually or phonologically, a complete perceptual representation of the word is constructed and subsequently activates in the access file based on its initial spelling or sound . Once a word's location has been established based on its access files, a search for the word entry in the lexicon must still be carried out . Relating this model back to the analogy of a library, lexical access is thought to occur by first locating the file in which the information is (i.e., searching for which section a book is in) and then the lexicon is searched for the actual location of the word (retrieving the book of the shelf). It should be clarified that all information of the lexical entry (i.e. its spelling, semantics, pronunciation, etc) are contained in the lexicon and not the individual access files .
The master lexicon is assumed to be organized into bins with the most frequent entries stored on the top of the bins. This belief further explores why high-frequency words are accessed more quickly than low-frequency words. Entries in this model are said to be searched one by one until an exact match to the perceptual representation is found; Figure 6 depicts how this process takes place. The process of lexical access proposed by Forster's model occurs in more of a step by step process (serial search) rather than a simultaneous process (parallel access) . When the relevant lexical entry is found it is then cross referenced against the targeted input to ensure accuracy . If the selection is deemed correct, the search is terminated. However, if the selection is deemed incorrect a more exhaustive search is continued until the correct lexical entry is retrieved from the lexicon. Incorrect selections can comprise of either non-words which do not adhere to the rules of English grammar such as zdkj, or a non-word that resembles a real word such as shure. Several studies have revealed that individuals require more time to reject the non-words which represent a real word in comparison to words which clearly do not  . In short, lexical access of a word is terminated only once the correct lexical entry is located, which are scanned one at a time.
The Logogen Model
Morton (1969) proposed that words are not accessed by determining their locations in the lexicon but by being activated by a certain threshold . An analogy between Morton's model and a light bulb can be made in that a word similar to a light bulb is activated when enough energy is being delivered to the source. Thus in relation to the logogen model, words are activated when its threshold has received enough energy to access the lexical entry .
Morton (1969) claimed that each lexical entry had its own logogen which tracked of the number of features a lexical entry had in common with a targeted stimulus . Words are said to be at a resting level and have a zero-feature count when they are not being activated as a potential candidate of a the targeted word . Morton proposed that each logogen had an individual threshold which required a particular amount of input/energy for the lexicon to access a particular lexical entry . Input can be received in the form of orthographic, phonological or semantic information as individuals read or listen to language. Once all the input is received for the lexical entry candidates, the number of features for each logogen is summed up and the certain logogens reach their predetermined thresholds . Among the logogens which do reach their threshold, the lexical entry which has the highest feature count thus the most similarities with the targeted word is chosen. Once lexical access has been completed, all logogens return to their zero-feature count resting level . Logogens that have reached their threshold take longer to return to their resting level than those that have not . As depicted in Figure 7, contrary to Forster's (1976) autonomous search model, Morton’s logogen model provides no separate access routes in which the lexicon is searched for lexical items. In contrast, individuals use all available input (i.e., orthographic, phonological) in order to activate logogens with similar features to the targeted word. In addition, Morton’s model allows for simultaneous parallel searches of input from multiple modalities, where as Forster’s model only allows one access file to be used at a time .
Similar to Forster’s model, high-frequency words are accessed more quickly than low-frequency words in the logogen model. This model asserts that the frequency effects are the result of the lower activation threshold for frequently used word. That is, it takes less activation to fire a high frequency word than a low frequency word. For example, the word bear which is used more frequently would require less input to reach its threshold than a word like anteater, see Figure 8. Priming on the other hand, is accomplished by a quick and temporal lowering of the threshold of logogens related to a prime. .
Morton’s (1969) logogen model has been one of the most influential of the parallel word access models and served as the basis for all the parallel models that followed . As with any model, however, modifications were made from the original 1969 model exactly a decade later (1979). The newest version of the model asserts that separate input paths and logogens exist for words presented by reading (visually) versus listening (auditory) .
The Cohort Model
Marslen-Wilson et al (1987) proposed that when individuals hear a word, its phonological neighbours also get activated. To simplify this concept refer back to Figure 5: The Spreading Activation Model. Rather than words of similar semantics being primed, the cohort model proposes that words with similar sounds are primed. The cohort model is comprised of three main stages. During stage one, also known as the access stage, the first few sounds of the target word activate all words with a similar sound. For example, in the sentence “Renee went to go buy a toy from the st-…,” stand, store, stranger as well as other similar phonological words would accessed during the first stage. The set of words which become activated are known as the "cohort" . The cohort model bares similarities to Morton's (1969) logogen model in that multiple words can be activated, and the system continues searching through all activated words until it settles on a single choice. The second stage of Marslen-Wilson's (1987) model is known as the selection stage, during which all activated words are progressively eliminated thus narrowing the cohort . An activated lexical item in the cohort can be eliminated either based on inappropriate context or if a better candidate is activated. All lexical items in the cohort continue to be eliminated until a single lexical item remains, known as the integration stage. Figure 9 depicts the three stages of lexical access and elimination as described above in the cohort model.
The original cohort model asserted that an exact match between a lexical item and its phonological properties was required . However subsequent studies revealed that individuals are still able to access a correct lexical item, even if words are mispronounced or left out (i.e., if an individual yawned part way though a word) . In light of this information, the cohort model was revised and currently maintains that an exact match between a lexical item and its phonology are not necessary for lexical access. The cohort model also accounts for frequency and non-word effects similar to Morton's logogen model . Both theories assume that context and primed words narrow the original set of activated lexical items, thus leading to a quicker recognition of targeted stimulus .
The Connectionist Model
McClelland and Seidenberg founded the connectionist model for lexical entry recognition in 1989. Please refer to Psycholinguistics/Connectionist_Models #McClelland_and_Seidenberg_Model for further insight into the connectionist model of lexical access.
Ambiguous : a lexical item or sentence that has more than one meaning (i.e., right - correct; direction).
Dominant Meaning : the view that ambiguous words are most activated a dominant meaning.
Exhaustive Access : the view that multiple meanings of ambiguous words are activated regardless of contextual cues.
Homograph : different words that are spelled the same but may have different pronunciations (i.e., bear-animal; bear-carry).
Homonym : a word with two meanings.
Homophone : two words that sound the same (i.e., there; their).
Imageability : the degree to which a concept can be visualized. Highly imageable words are more easily accessed than low imageability words.
Lemma : a level of representation of a word between its semantic and phonological representations it is syntactically specified but not contain sound-level information.
Lexical Access : the process of accessing and retrieving a lexical entry from the lexicon.
Lexicon : the system of vocabulary which is stored in the mind in the form of lexical entries for each item; commonly referred to as our mental dictionary.
Parallel Search Model : models of lexical item access that claim each item in the lexicon are activated simultaneously in an attempt to find to correct lexical item.
Selective Access : the view that context provides sufficient cues to activate one interpretation of an ambiguous word which is most relevant.
Semantics: the study of word meaning and the ways in which words are related to one another in our mental lexicon.
Serial Search Model : models of lexical item access that claim each item item in the lexicon is activated one at a time until the correct item is found.
Subordinate Meaning : the subordinate meanings for ambiguous words are usually those which are not the first to be activated upon reading or hearing the ambiguous word.
- Field, J. (2004). Psycholinguistics: the Key Concepts. United Kingdom, London: Routledge.
- Levelt, W. J. M. (1989). Speaking (p. 188). Cambridge, MA: MIT.
- Field, J. (2003). Psycholinguistics: a Resource Book for Students. United Kingdom, London: Routledge.
- Chumbley, J. I., & Balota, D. A. (1984). A word's meaning affects the decision in lexical decision. Memory & Cognition, 12(6), 590-606.
- Swinney, D. (1979). Lexical access during sentence comprehension: (Re)-consideration of Context Effects’. Journal of Verbal Learning and Verbal Behaviour, 18: 21-39.
- Simpson, G. B. (1984). Lexical ambiguity and its role in models of word recognition. Psychological Bulletin, 96(2), 316-340. doi:10.1037/0033-2909.96.2.316.
- Simpson, G. B. (1994). Context and the processing of ambiguous words. In M. Gernsbacher, M. Gernsbacher (Eds.), Handbook of psycholinguistics (pp. 359-374). San Diego, CA US: Academic Press. Retrieved from EBSCOhost.
- Mason, R. A., & Just, M. (2007). Lexical ambiguity in sentence comprehension. Brain Research, 1146115-127. doi:10.1016/j.brainres.2007.02.076.
- Tabossi, P., & Zardon, F. (1993). Processing ambiguous words in context. Journal of Memory and Language, 32(3), 359-372. doi:10.1006/jmla.1993.1019.
- Vakoch, D. A., & Wurm, L. H. (1997). Emotional connotation in speech perception: Semantic associations in the general lexicon. Cognition and Emotion, 11(4), 337-349. doi:10.1080/026999397379827.
- Whitney, P. (1998). The psychology of language. Boston: Houghton Mifflin.
- Zipf, G.K. (1949). Human behaviour and the principle of least effort. Massachusetts, Reading: Addison-Wesley.
- Balota, D. A., & Chumbley, J. I. (1984). Are lexical decisions a good measure of lexical access? The role of word frequency in the neglected decision stage. Journal of Experimental Psychology: Human Perception and Performance, 10(3), 340-357. doi:10.1037/0096-1518.104.22.1680.
- Gleason, J. B., & Bernstein, N. R. (1998). Psycholinguistics. Toronto: Harcourt Brace College Publishers.
- Paivio, A. (1991). Dual coding theory: Retrospect and current status. Canadian Journal of Psychology, 45, 255–287.
- Jessen, F., Heun, R. R., Erb, M. M., Granath, D. O., Klose, U. U., Papassotiropoulos, A. A., & Grodd, W. W. (2000). The concreteness effect: Evidence for dual coding and context availability. Brain and Language, 74(1), 103-112. doi:10.1006/brln.2000.2340.
- Paivio, A. (1969). Mental imagery in associative learning and memory. Psychological Review, 76, 241-263.
- Bleasdale, F. A. (1987). Concreteness-dependent associative priming: Separate lexical organization for concrete and abstract words. Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 582-594.
- Schwanenflugel, P., & Shoben, E. (1983). Differential context effects in the comprehension of abstract and concrete verbal materials. Journal of Experimental Psychology: Learning, Memory and Cognition, 9, 82–102.
- MacDonald, M. C., Pearlmutter, N. J., & Seidenberg, M. S. (1994). The lexical nature of syntactic ambiguity resolution. Psychological Review, 101(4), 676-703. doi:10.1037/0033-295X.101.4.676.
- Barsalou W. L. (1992). Cognitive Psychology: an Overview for Cognitive Scientists. (p. 221-244). New Jersey, Hillsdale: Lawrence Erlbaum Associates Publishers.
- Collins, A. M., & Loftus, E. F. (1975). A spreading-activation theory of semantic processing. Psychological Review, 82(6), 407-428. doi:10.1037/0033-295X.82.6.407.
- Scott, J., & Pavlenko, A. (2008) Crosslinguistic Influence in Language and Cognition (p. 72-88). New York, New York: Routledge.
- Forster, K. I (1976). Accessing the mental lexicon. In F. Wales & E. Walker (Eds). New approaches to language mechanisms (p. 257-287). Amsterdam: North Holland.
- Aitchison, J. (1994). Words in the mind: An introduction to the mental lexicon, 2nd ed. Oxford: Basil Blackwell.
- Morton. J., & Patterson, K. (1998). A new attempt at an interpretation or an attempt at new interpretation. In M. Coltheart, K. Patterson & J. Marshall (Eds.) Deep Dsylexia (p. 91-118) London: Routledge.
- Marslen-Wilson, W.D. (1987). Functional parallelism in spoken word- recognition. Cognition, 25, 71-102.
- Seidenberg, M. S., & McClelland, L. J. (1989). A distributed developmental model of word recognition and naming. Psychological Review, 96(4):527. doi:10.1037/0033-295X.96.4.523.