Word Encoding Models

Words: 2127
Topic: Linguistics
Updated:

Introduction

Word recognition refers to the ability of an individual to recognize written words with ease. There are different models that seek to explain how human beings process words. These models are based on different principles. For example, there is the multistream model that is based on several processing channels and the hierarchical model that emphases the use of single input. This paper compares the two models in order to offer an insight into the model that best describes how human beings process words.

Introduction

Any form of reading involves the recognition of the words is read. As such, reading is an involving task and, thus, an astoundingly complex procedure (Glezer, 2009). Even though many people consider reading in the form of a singular action, research has shown that the human brain engages in different tasks at the same time whenever one is reading (Park & Han, 2011).

This explains the role that comprehension reading, fluency, phonemic awareness, phonics, as well as vocabulary plays in the reading process. According to Glezer and Riesenhuber (2015), word recognition refers to the ability of an individual to recognize written words without a lot of effort. Sometimes, it is known as the ‘isolated word recognition” since the process involves the capacity of a reader to individually recognize words that are in a list without necessarily relying on other words.

Many scholars and researchers argue that for fluent reading and word recognition, readers should embrace the word recognition process that is both rapid and effortless (Pelli, Farell & Moore, 2003). For beginners, they sound words through phonics and word analysis. However, as the words are repeatedly used, the readers are in a position to recognize them as whole units (Pelli et al., 2003).

This explains why beginners are advised to familiarize themselves with connected texts. Reading of connected texts is significant since it plays a great role in solidifying the reader’s ability in word recognition, as well as in word analysis. In addition, such a technique is known to help learners to transit from the process of merely sounding out words, to the development of word recognition abilities.

The ability of humans to recognize words in the process of reading has been of much interest to many researchers (Allen, Canfield, Kaunt, Lien, & Smith, 2009). In spite of the fact that there are many researchers interested in the knowledge of experimental psychology and the ability of human beings to recognize words, there are different postulates regarding word recognition. For instance, some researchers, such as Catell postulated that the processing of words is based on an analytical technique (Allen, Smith, Lien, Grabbe, & Murphy, 2005).

According to Dahaene and Cohen (2007), the formation of words is from component letters. Allen et al. (2009), on the other hand, believes in the holistic processing of words whereby they are seen as whole units, as opposed to component letters. Research has shown that even though different researchers are concerned with the experimental psychology of word recognition, only a few of them have adopted the visual system of word recognition (Allen et al., 2005).

As a result of the different approaches used by different researchers, several theories have come up to explain the controversy in word recognition. However, these theories have postulated that there are two primary techniques in word processing, including holistic and analytical processing. However, Allen et al. (2009), in their theory, point out that word recognition can be looked at from the viewpoint of visual system whereby spatial frequency plays a major role in visual word recognition.

Methods

In the study of word recognition, different methods are used, such as eye movement, measurement of reaction time, and the tachistoscopic presentations. However, there are other methods, such as the visual search, detection of letters, and the Stroop interference (Pelli et al., 2003). As far as the measurement of responses is concerned, the method employs naming, lexical decision, and categorization.

In the naming technique, the participants are required to name a word loudly as quickly as possible. The participants are supposed to tell whether a given letter string can be termed as a non-word, or it is a word. On the other hand, categorization requires the subjects to decide the category of a given the word. In this case, the study employed the experimental approach whereby the participants were to be subjected to 384 items under different conditions.

The 384 items included 192 non-words and 192 words. In all the items, it was intended that 24 of the items contained eight categories that were formed after crossing several categories of length with four levels of word occurrences. Then, the establishment of any given letter from a group of the words established from a necessary nonword. However, it was important to form orthographically correct nonwords.

The items were arranged in different strings, including lowercase, monochrome, mixed-hue, and mixed-case monochrome. The experiment was carried out with the aim of evaluating the primary predictions of the multistream model. First, the study was to analyze the response rate of words with the same case, mixed-case. In addition, the study investigated the response rate of hue mixed words and nonwords. Data collection was done using computers that contained non-interlaced monitors.

Results and Discussion

The primary focus of this study was on the analysis of word recognition. Does the process of word recognition involve the identification of single components for every lexical item, or does the process consider the independence of readers whereby words are identified from holistic units? The multistream model of word recognition postulates that the recognition rates of words under different conditions are different from different words and nonwords.

According to Grainger and Hannagan (2014), the response rate for same-case words is faster than the response rate for mixed-case words. The model, as well, predicts improvement of performance for mixed-case words with consideration of the monochrome presentation. According to Park and Han (2011), any items that have mixed colors and of lowercase can be recognized by the Magno-dominated stream, and so are the lowercase monochrome words and nonwords.

The interblob-dominated stream is responsible for the detection of the mixed-case monochrome items. However, under some circumstances, the ID and BD are responsible for the processing of mixed-hue and mixed-case items. Despite the fact that the processing rates for ID and BD streams are similar, these two streams show difference in the individual response rate (Park & Han, 2011).

This is the case, especially for mixed-hue and mixed-case items, whereby the response rate is depended on which of the two streams is fast than the other. For this reason, the responses of mixed-hue and mixed-case items are often high than the responses of mixed-case monochrome words.

From the analysis, it was evident that different word cases had a different effect on the ease of word recognition. For example, in a case where the letter cases were mixed, the results were that word recognition was slow relative to lowercase presentation (Colombo & Sulpizio, 2015). Such results can be attributed to the fact that lowercase words are recognizably based on the high rate of information carriage by the MD stream.

For a case where the words are mixed in terms of upper and lower cases, the slower BD and ID streams are responsible for word recognition. The results showed that performance advantage during the experiment was on the mixed-case and mixed-hue when compared to the words under the mixed-case monochrome category. As such, it is evident that holistic channels are prevented by case mixing from the word-level formation.

For this reason, it can be said that the ID and MD streams are responsible for the recognition of words that are under the mixed-case string. Such results thus are in line with the postulates of the multistream model of word recognition. However, these results differ with the findings outlined in the model postulated by Dehaene and Cohen (2007). In this model, constituent letters often do the formation of words.

An analysis of the response rate for mixed-case and lowercase stimuli showed a considerable difference where the response rate was slow for the mixed-case. In addition, the performance for the mixed case was facilitated by hue mixing, and different for both words and nonwords in lowercase.

Evidently, the mixed-case, along with the mixed-hue, had some level of performance benefit when compared to the items under the mixed-case monochrome category. As such, the MD processing streams are responsible for the lexical decisions for the lowercase stimuli. On the other hand, the ID streams account for the mixed-case stimuli. This is attributed to the fact that the ID stream analyses items using the grayscale. The BD stream, on the other hand, uses the hue contrasts.

In this study, the primary focus was on the evaluation of the process of word recognition under the human visual model. According to the behavioral and neuroscience models of word recognition, words are identified through spatial frequency, where different processing channels are in play. As pointed out by Allen et al., (2009), there is a close relationship between the ID streams and the word-level as postulated in the hybrid model.

Similarly, a rough correspondence exists between the MD streams and the letter-level channel evident in the hybrid recognition model. As such, the multistream model of word recognition is different from other models, such as the analytical model presented by Dehaene and Cohen (2007). +To begin with, the multistream model makes use of spatial frequency filters in its illustration as opposed to applying detectors in the model’s illustration. In addition, the model adopted by Allen et al. (2009), advocates the use of several streams that are used in processing various characteristics of the stimuli input. This is contrary to the LCD model of word recognition that uses only one stream, and in other cases, employs several streams, but the streams are used in processing one input differently.

Multistream model and the Hierarchical model

A critical look at the multistream theory of word recognition and the hierarchical model of word recognition shows that the multistream approach is best suited in explaining how words are processed. As evident from the study, the multistream model relates the response rate of mixed-case, mixed hue, lowercase items with the response times for MD, ID, and MD streams.

The results show that there is a correlation between the rates of recognition for mixed-case monochrome items relative to the ID stream, as well as the mixed-hue and mixed-case items relative to the ID and BD. When the results from the study are compared with the results from the model postulated by Dehaene and Cohen (2007), it is evident that there are a number of differences. The difference in these results shows that a lot of evidence is needed to understand the process of forming words and the process of word recognition.

However, an analysis of the multistream model shows that words are formed in a separate manner from letters. In addition, features are not used in the formation of letters, which explains that it is unlikely that constituent letters play a role in word-formation. This is the opposite of the analytical hierarchical model postulated by Dehaene and Cohen (2007).

According to Glezer (2009), if words were formed by the use of a single orthographic input as postulated in the hierarchical model, mixed-case, and lowercase words should exhibit the same hue effect. However, the multistream model has proved that there is a considerable difference in the impact of color for words of mixed-case and lowercase.

Conclusion

From the foregoing, it suffices that there are different models used to explain the process of processing words. The concern of this analysis was on the difference between the multistream and the LCD viewpoints of word recognition. An overview of the two models shows that the multistream model differs from the hierarchical model in two ways.

First, the multistream model makes use of spatial frequency in its processing stream, while the hierarchical model uses visual features that are analyzed by the visual features detectors. Secondly, the multistream model makes use of a variety of processing channels that respond differently to diverse characteristics of stimuli. For the case of the hierarchical model, the focus is on a single input (Dehaene et al., 2005).

As such, the findings from the analysis indicate that there is a correlation between the holistic channel (MD) and the recognition of visual words. In addition, it was evident that some circumstances can lead to the use of both the analytic channel, the BD and ID, whenever the MD stream detects unfamiliar stimulus information. The consistency of the findings shows that visual recognition of objects makes use of multiple neural streams. Thus, it suffices that words are formed directly from holistic units based on visual features independent of letters (Allen et al., 2009).

References

Allen, P., Smith, A., Lien, M., Grabbe, J., & Murphy, M. (2005). Evidence for an Activation Locus of the Word-Frequency Effect in Lexical Decision. Journal Of Experimental Psychology: Human Perception And Performance, 31 (4), 713-721. doi:10.1037/0096-1523.31.4.713

Allen, P., Smith, A., Lien, M., Kaut, K., & Canfield, A. (2009). A multistream model of visual word recognition. Attention, Perception & Psychophysics, 71 (2), 281-296. doi:10.3758/app.71.2.281

Colombo, L., & Sulpizio, S. (2015). When orthography is not enough: The effect of lexical stress in lexical decision. Memory & Cognition, 43 (5), 811-824. doi:10.3758/s13421-015-0506-6

Dehaene, S., & Cohen, L. (2007). Response to Carreiras et al: The role of visual similarity, feed forward, feedback and lateral pathways in reading. Trends In Cognitive Sciences, 11 (11), 456-457. doi:10.1016/j.tics.2007.08.009

Dehaene, S., Cohen, L., Sigman, M., & Vinckier, F. (2005). The neural code for written words: a proposal. Trends in Cognitive Sciences, 9 (7), 335-341. doi:10.1016/j.tics.2005.05.004

Glezer, L., (2009). Evidence for Highly Selective Neuronal Tuning to Whole Words in the Visual Word Form Area. Neuron, 62 (2), 199-204. doi:10.1016/j.neuron.2009.03.017

Glezer, L., & Riesenhuber, M. (2015). Adding Words to the Brain’s Visual Dictionary: Novel Word Learning Selectively Sharpens Orthographic Representations in the VWFA. Journal of Neuroscience, 35 (12), 4965-4972. doi:10.1523/jneurosci.4031-14.2015

Grainger, J., & Hannagan, T. (2014). What is special about orthographic processing?. Written Language & Literacy, 17 (2), 225-252. doi:10.1075/wll.17.2.03gra

Park, S., & Han, J. (2011). The effect of holistic versus analytic processing on gender difference in memory. Journal of Vision, 11 (11), 847-847. doi:10.1167/11.11.847

Pelli, D., Farell, B., & Moore, D. (2003). The remarkable inefficiency of word recognition. Nature, 423 (6941), 752-756. doi:10.1038/nature01516