Word recognition

2.2.1. The role of phonological mediation in word recognition

Th e focus of this section is the semantic aspect of word recognition and its role in reading. Several conceptualisations of phonological mediation in relation to L1 and FL/L2 reading are discussed.

Th e question as to whether recognition of the printed word is mediated by some version of its spoken equivalent (often named “inner speech” or “subvocalisation”) has given rise to a number of hypotheses and stimulated various studies. In the phonological recoding hypothesis, fi rst off ered by Rubenstein, Lewis, and Ru-benstein (1971), phonemes are viewed as abstract, hypothetical entities; therefore, it denies the existence of any form of speech or subvocalisation. Th e hypothesis claims that word recognition is preceded by the conversion of letters into phonemes and a search in the mental lexicon for an entry which matches this phonological form. Th e hypothesis has inspired numerous studies; however, their fi ndings fail to support its predictions concerning the presence of phonological mediation in the process of word recognition. For example, clinical studies of brain-damaged patients (Saff ran and Marin 1977) indicated that patients who had lost an ability

to recognize the spelling-sound correspondences (presumed by the phonological recoding hypothesis) and therefore unable to decode pseudowords, can still cor-rectly recognize many words. Th e fi ndings undoubtedly show that there must be a diff erent route to the mental lexicon than phonological receding, reconfi rming the assumption that there must be a dual access to the internal lexicon: one direct and the other involving phonological receding (Coltheart et al. 1979; McCusker, Hillinger and Bias 1981; Meyer, Schvaneveldt and Ruddy 1974). Th e two routes are assumed to operate in parallel, with word frequency being an important fac-tor infl uencing the speed of the direct route. Th is approach seems to explain why even skilled readers phonologically recede unfamiliar words; it also accounts for the absence of phonological mediation in reading familiar words.

Let us now focus on the presence of phonological mediation in silent read-ing, often named as subvocalization or “silent speech.” Research (e.g., Dooley and George 1988; Edfeldt 1960; Kleiman 1975; Swanson 1984) implies that an amount of phonological recoding depends on the level of the reader’s profi ciency and the dif-fi culty of texts. Good readers are less likely to subvocalize and easy texts “provoke”

less “silent speech.” Subvocalisation improves the comprehension and retention of complex material, which requires a heavier load on memory, e.g., in the case of comprehension questions.

Interesting implications concerning sounding out words in reading have been suggested by cognitive psychologists. In their experimental studies of memory, Bad-deley, Eldridge and Lewis (1981) showed that suppressing subvocalisation does not aff ect the speed of reading and understanding the gist of the text. However, it makes readers less sensitive to errors in text, such as wrong word order. Th e researchers (Baddeley, Eldridge and Lewis 1981: 1, abstract) concluded that “subvocalisation allows the creation of a supplementary articulatory code which is produced and utilised in parallel with other aspects of reading. Such a code seems particularly suitable for monitoring order information.” In further research, Baddeley (1999) focused on what he calls the articulatory or phonological loop, i.e., a process of rehearsal, usually via subvocal speech and its main function – maintaining the memory trace.

Th e exploration of the conception of a phonological loop confi rmed the as-sumption put forward by earlier studies that subvocalisation can facilitate reading comprehension. Baddley (1999: 53) says:

You probably use it [subvocalisation] when reading diffi cult prose – a legal document, for example – where accurate understanding is essential, but I suspect that you do not subvocalise very much when reading a novel. You might well argue that, although you do not subvocalise, you still think you hear a voice when you read; I suspect that this

“voice” is based on another system, an auditory imagery system, related to but diff erent from the articulatory loop.

Th e role of “an auditory imagery system” was further discussed by Smith (1994: 160), who explains that sounding out specifi c words, for example in read-ing poetry, does not so much contribute to a literal comprehension as “estab-lish a diff erent – a complementary or alternative-kind of mood or meaning.” Th is

assumption is shared by Eysenck and Keane (1995: 315), who conclude that apart from reducing the memory load in comprehension, inner speech “may provide the prosodic structure (e.g., rhythm, intonation, stress) that is lacking in written text but present in spoken language.” Similarly, Ridgway (2009) assumes that subvocalisa-tion can be important in monitoring more aff ective and interpersonal aspects of language, especially where the text resembles speech, e.g., in reading poetry and drama.

Subvocalisation has also been the subject of discussions in relation to reading in a foreign/second language. It is assumed that FL/L2 readers are less familiar with the phonology or prosody of the language than native language readers. In fact, many FL/L2 readers, if interviewed about this phenomenon, would be aware of some form of subvocalising in their FL/L2 reading (as demonstrated in the study described in this book; see Part Th ree). More laborious subvocalisation taking place in learners’

minds in reading in the FL than in their native language was observed by Ridgway (2009). Th e main reason can be the weaker linguistic competence of FL/L2 read-ers, which makes reading in FL/L2 more demanding for short-term memory (Birch 2002). Automatic word recognition in FL/L2 reading is considered important in FL/

L2 reading (e.g., Eskey 1993). Th us, developing a phonological store of words that can be automatically recognized in a written text is a factor that will undoubtedly contribute to more fl uent reading.

2.2.2. Word recognition in the rauding theory

Th e role of word recognition in reading was also explored by Carver (1977) in his rauding theory. Th is model of reading is interesting because of the theoretical assumptions and practical implications it off ers for both L1 and FL/L2 reading instruction. A detailed discussion of the rauding theory is off ered in Kusiak (2011).

In this book, only the most important aspects of this model are presented. Special attention is given to advanced FL/L2 readers.

In his investigation, Carver (1977) focuses on the similarity between reading comprehension and listening comprehension (the word “rauding” is a combination of two words: “reading” and “auding”). An ability to recognize words quickly and effi ciently is considered to play the key role in the process of reading. Reading is conceptualised as a competence that encompasses fi ve reading “gears:” scanning, skimming, general reading (called by Carver rauding), study reading and memo-rizing. At various levels of comprehension information is retained to a diff erent extent. Rauding reading, also called normal reading, involves “comprehending each consecutively encountered complete thought in a passage, … comprehend-ing about 64% or more of the thoughts in a passage” (Carver 2000: 405). Carver (2000) hypothesizes that an advanced reader would be at the raudamaticity point, which means that he/she has raudamatized all of his/her audamatized words. In other words, an experienced reader is able to decode automatically and under-stand quickly words that he/she knows when spoken and which he/she is able to pronounce when presented in print. Drawing on his theory, Carver designed

a model containing precise mathematical equations, which has been applied in both L1 (e.g., Rupley 1996) and FL/L2 reading research (e.g., Asano and Sudo 2006;

Hirai 1999; Lee 2006). All the studies confi rmed the assumptions of the Carver model, indicating that similar processes underpin reading and listening in both L1 and FL/L2.

As regards teaching implications in both L1 and FL/L2 reading, the rauding theory suggests creating opportunities which would help learners to recognise and develop their individual rauding rate. Raudamaticity training concerning vo-cabulary is recommended, also by means of computer technology. Readers are advised to develop their sight vocabulary (i.e., words recognized in print) by ex-tensive reading for pleasure. Manuals and textbooks promoting quick improve-ment in reading rate are considered as “dangerous” materials that can frustrate readers and be detrimental to the development of their comprehension skills (McLay 2007).