READING SKILLS Introduction ‘A fandamental concern in second language reading involves the identifica tion and nature of reading sil and whether those sills that exist are situa ed within an ordered hierarchy. The reading and leeray literature is replete with references to reading skills, higher-level skill, and lower-level skill However, the precise meanings of the terms are not always teansparent. [Much of the basic work in the literature about reading skills comes out of research and concerns in fist language reading that was fst our in the 1970s and 1980s. Some of that research addresses the processes of children learning to read, turning weitten language into oral language, and the stages cha they go chrough in order to become macure readers (Clymer 1968; Gibson and Levin 1975; Goodman 1968; Gordon, 19825 Just and Carpenter 19875 Lunzer, Waite, and Dolan 1979; Perfetti 1991; Rayner and. Pollawsek 198; Strain 1976). Because so much of the initial literature into skills has focused on the first language education of childten learning to read, itis important ro examine the implications for second and foreign language reading rather closely. As previously noted, unilaterally adopting concepes and terminology from the first language literature must be done with caution, if at all. Nevertheless, they are an important starting poine and. source of common vocabulary. This chapter examines issues in reading skills and the next chapter will be somewhat of an extension devoted specifically to issues of reading strategy acquisition, instruction, and use. As will become lear, his division is not always cleaty a dichocomyy asics not always easy to distinguish skill from strategies. Reading skills in the first language I's difficule to clearly discuss the concerns around reading skills without addressing the overlap of several terms and concepts. There isa need to rec ‘ognize how the term sills used in multiple ways. The terms ski subs, _procesing strategies, ability, and Stanovich’s (3980) term, knowledge sources,n Reading ile are terms used variably in the applied linguistics and language teaching, literature. The distinctions in usage can depend upon whether the term is used to refer toa language component orto an individual reader's ability © perform. Richards, Pact, and Weber (1985) define language skills as being: the mode or manner in which language is used. Listening, speaking, reading, and writing are generally called the four language skills... Ofeen skills are divided into subskills,soch as discriminating sounds in connect- ced speech, of understanding relations within a sentence (Richards etal. 1985: 160) Harris and Hodges define skill as ‘an acquired abiliey to perform well proficiency’ (r981: 298). Thus, the term can be used co denote a reading, behavior such as making appropriate phoneme-grapheme correspondence, ‘oF it can be used to indicate a relative level of eading ability asin the concept cof skilful reader. This ambiguity inthe use of the term sbils can be seen in the following excerpt from Proctor and Dutta (1995): First, skills acquired chrough practice or raining. A defining characteris- tic of the skills discussed in this book is tha they are not innate bue must be learned. Second, skilled behavior is goal directed. Skill develops in response to some demand imposed by the tak environment on che organism, although some learning may occur that is incidental to that demand. ‘Thied, skill i said to have been acquired when the reading bbchavior is highly integrated and well organized. Theough experience, the components of behavior are structured into coherent patterns. Finally, cognitive demands ate reduced as skills acquired, freeing limited mental resources for other activities. From these characteristics, we derive our definition of skill: Skill is goal-dicected, well organized behavior that is acquired chrough practice and performed with economy of effort, Asstated earlier in the chapter, mos skills involve the entire information- processing system, However, ata gross level, skis can be divided into those with primarily percepcual, motor, or cognitive components (Proctor and Dutta 19518) In thisexample,skllis used both as skill level (well organized behavior that is acquired through practice and performed with economy of effort) as well, as separate types of behavior (skis can be divided into those with primarily perceptual, motor, or cognitive components) Inthe present discussion, the term will be used to indicate different reading processing activities rather than as rlaive ability level Additionally, in much ofthe literature there isa mixing of the verms reading. ‘hill and reading proce. Ln part, much ofthis lst conllation depends upon. Reading ils ‘whether he authori primarily concerned wich pedagogy or with psychology: Finally, at times the discussion focuses either on reading sills comprehen- sion skills, and in some discussions comprehension skills are seen as being cataled in reading skills while at other times they ate seen as distinely different cognitive process ‘Throughout che next few sections ofthis chapter about whether skills are separable and hierarchically ordered, we will find studies that view reading skills as decontextualized abilities existing ourside of any particular literacy context. It is proposed here that this assumption is one ofthe reasons we find conflicting results from various studies. For example, ic is proposed that inferring unknown socabulary from context is a skill of some particular dif culty oF facility, then there is no acknowledgement chat some texts are ‘co-operative for inferring unknown vocabulary while other tests are not, not isitacknowledged that some lexical items may be easier to infer than others. ‘The relative salience of skills as they are operationalized in empirical stadies will affece whether they are seen as separable skills oF not, and ths relative salience will be due to the context of the lteray act in which the reader isengaged. ‘As weexamine the differing views of reading skills, we can see how under! ing assumptions abou language and research affec the different reading worlds that are being presented. Which ofthese worlds we favor wil affect what we do inthe reading classroom and the shape ofthe language teaching ‘materials we design, Ie will affece whether a clas adopts a skils-centered syllabus or focuses more closely on task- and performance-based contextual activities I will affect how the 1eading activity is discussed and the extent {o which we focus on the relationship between first and second language behaviors. Consequenty its important co take a close look at how reading skills have been identified and described in the reading erature Separability of skills Generally reading sills represented in categories representing: 1) word stack kl 2)compychension sil) fancy skill and) erie reading Skills. Word ack sl Sometimes called Decopiwe skis) tepreee the sil necessary to comet orthographic symbols nt langage, This set of ills equires ha the reader ecogne thatthe Serpe tepesens Units of language, sch 5 phonemes labs, and words, Some ofthe subskl of thts type would be coating lable pacein, converting string sound on occasions, rcogntang upper and lowercase leer an recogni ing word boundaries, Comprehenion sls represent the ality te Uc Content and knowlege to deve meaning om whats ead. Examples of compichenson ails would be grammatical competence and knowledge of morpclogy, syn, meckari, using concert gin eating wing ”we Reading ills schemata as aids, using merAcoGNITIVE knowledge, recognizing text structure, and predicting what will come next in a text. Fluency skills are directed at allowing the reader to soe larger sentences and phrases as wholes, process which assists in reading more quickly Fluency skills would involve such abilcis as sight word recognition and recognizing high-frequency lercer clusters, rapid reading, and possessing an extensive vocabulary. Finally, critical reading skills provide the reader with the skill co analyze, synchesize, and evaluate what is read, This process involves such activities as seeing the ‘cause-and-effect or comparison relationships in the text, or adopting a critical stance toward the text. There is a great deal of variety in the specific sills tha are identified by various teachers and researchers as well asa lack ‘of precision in the language used to describe the skills. In an examination of how ceading skills have been viewed in pedagogical citcles, Rosenshine (1980) examined the comprehension kill ientifie from five authoritative educational sources. The frst was from Science Research Associates (Shub etal. 1973).The second source was fom the skills rested on the 1973 National Assessment of Educational Progress: The third was from the scope and sequence chart for Scot Foresmans reading series (Aaron et. 1976). The fourth was from textbooks by Farris and Smith (1976). Te final source was from the reading program developed by the Wisconsin Research and Development Center (Cte and Askov 1974). In their examination, they found that there were common general reading skill categories specified in all five sources and that they fl into three different eypes of skills, all of which appear be associated with comprehension skills. The fstype of kill could be termed locating detail. Thisis the simples ofthe skills and involves recog- nition, paraphrase and/or matching, The second type of kill group is labeled simple inferential lands made up of uch skills 28 understanding wordsin context, recogniing the sequence of events, and recognizing cause and effect relationships. The tied group was labeled complex inferential skiland relates to working with longer texts than in che second skill ype. This group might ‘comprise such activites as recognizing the main idea, drawing conclusions, and predicting outcomes. hough Resenshine found that there were many skill in common among the five sources, he alo noted tharall ut arcs and Sih had unique skis cha did not fe wiehin the ehree types of common ski. Insome cases thishad to do with how specific the authors were in defin- ing 2 particular skill. Rosenshine notes that some teachers and researchers right believe that rather than. kil called recognizing the main ide, i might bbe morejusiable to spi the sll neo recagnzing the main idea in diferent text types such asa narzative or exposition. In general, Rosenshine finds some consensus tha reading involvesat least the followingseven subsillsacossthe three general reading skill ategories: recognizing sequence recognizing words in context Reading ils identifying the main idea ~ decoding detail = drawing inferences = recognizing cause and effect comparing and contrasting. However, he also notes that many sources break these dawn further into subskills Such subskils mighe reflect che recognition of different clues in reading, such as recognizing synonym or antonym clues, summary clues, ‘or clues provided by tone, setting, and mode, tc. In some cass lists such as these led to over 30 different skills and subskll. In addition, discussions of reading skills Frequently discuss lowertevel sills and. higher-level kil The lower-level skills tend to be related co word-attack skills, while the higher-level skills tend eo be linked to reading comprehension skills. This notion of lower-level skill being linked to word-ateack sill mayin pare bbe a historia product ofthe way eading as been taughe to children in fst language context, most of whom already havea ccively well-developed oral frst language buchaye had lel experience matching the wounds and words of the language roa princrepesentation. This concept reflects the twin process inreading of automatic text recognition versus recoding known orl language through text Just and Carpenter (987) take a reading Bueney and compre- henson development focus, and note that there ate «wo routes in text decoding ‘ivect visual recognition’ and ‘spcch recoding (327). They eport thar uene readers rey primarily on direct visual recognition while children and carly readers make peaer use of speech recoding. In early ist language reading, the accuracy and speed of word-decoding kil sa strong preicror of comprehension, a stronger predictor than listening comprehension abil. However, in later school grades, listening comprehension becomes a beter predictor of reading comprehension than worddecoding skill jse and Carpenter 1987: 328). Thus, 2s decoding sill improves, differences in reading ability are contclled more by a general comprehension facto. Stanovich (4986) posies thar any causal elaconship benween phonological scesitvty and ‘word recognition is developmentally constrained. Thats, the relationship is confined to the very eatly stages of reading acquisition (Stnovich 2000) Likewise, Doctor and Coltheart (80) found a shife rom phonological to visual encoding as age increased when the readers were attending ro meaning rather than attending to remembering the text. This indicates that as readers mmatue they have less dependence on phonological encoding and develop more reliance onthe visual code. The earch findings indicate chat children inal conver print into sound to help in comprehension, but as chey develop increased eading skill they rely more heavily on visual processing in comprehension. ‘This last Factor regarding type of processing in comprehension reveals another source of confusion in the literature about the nature of reading, a& Reading lle skills, Ie is important to be explicit about whae consticutes reading and the ‘concomitant implications for reading skills. Clymer (1968) points out that some authors define reading as responding orally to printed syrabols while others view reading as the change that takes place in one's knowledge as a result of having interated witha text. Those who take the fist view restrict their view of reading to those instances in which the outcome of reading could only take place in the narcow context ofthe reading activity, notinany ‘other form of communication. This, then, restricts the focus of reading skills to word perception. Bruck and Waters (1990) state, for example, ‘allthough reading may critically rely on che ability to recognize aword,onceithas been recognized the reader must employ general language’ comprchension mechanisms to understand the decoded items.’ (189). In this instance, reading is identified as visual decoding while reading comprcheasion is just special case ofa general comprehension ability. In one approach, the skill of interest i visual word decoding while in the othe the skill of interest is comprehension of written tex or the subskill that make it possible. For example, Gordon (1982) indicates that ‘there are cree distinct sets of ‘competencies that a student must develop in order to become an effective reader: reading skills development, reading comprehension, and reading research and study sill’ (41). He goes on co present the eype of breakdown for a reading program that spans ftst language K-8th grade in Table 4.1 opposite ‘This ehree-part breakdown conceptually corresponds to what might be termed lower-level skills (decoding), higher-level skills (comprehension), and strategies (wEtACOGNrIT1ON). It is interesting co note thac in this categorization ‘reading’ skills are separate from ‘comprehension’ skills. The reading skills are heavily focused on leter-sound correspondence. Iris clear ‘hac this collection of skills is pitched at a much earlier level of seudent than, ‘those that Rosenshine described above. ‘The componene ills pproach ro reaingis one ofthe approaches tha views skills as relections of cognitive process, The approach seeps t0 view readings the product ofs complex but decompostle information proces ing system (Care eta. 990: 9). Proponents ce this decomposable stom constng ofa specific numberof mental operations, or component sl, thae are dines fom one anther and empirically separable, Although the All ae separable, they workin concere with one another and may exert ‘more or lest influence depending upon the reading ask and rade equite- ‘ments, Levy and Hinchle (990) contend that ther likely ro be no single rganizcion ofthe muliplecomponenc sills that characterizes a good reader or poo reader Rather, good eadrs may organize ther sil ier enxly depending upon whether tey are dealing witha speeded tsk, com prehension reading ask, ora memory-based task. Ths, sill may incorporate Reading ils Reading skills development Names oflexees Vowel digraphs ‘Root words Introduce words Sileneleters, Prefixand sufi ‘Consonants Plurals Synonyms Vowels Compound words Aaronymns Sight words (Context clues Homonyms Blends Contractions ‘Malki meaning Word division Roving words Pronunciation key Consonanediagraphs Hand and soft'@ and'y’ Word definitions Reading comprehension development ‘Cateporaing Summarize Figurative language Sequencing Prediceourcomes Literary forms Follow direcions Recognize emotions Evaluate characters Read fo ets Make inferences Evaluate settings Retell tory Relibiity ofsource Factual conclusions Main idea CCompareand contrast Fat fiction, and opinion Keywords Make judgments Reading research and study sie Alphabetize Clasify books Addses, maps, graphs TTableofcontents Information from Crosr-referncing Dictionary sills varioussources Use of index Encyclopedia Useof glossary Table 4.1 Gordon (82) tree sets of eompetencies organizational flexibility in reading as opposed to some fixed model in which he diffecent component skills consistently operate in the same fishion in relation to one another. This provides some acknowledgment that literacy acts present the reader with different goals and purposes. Such 2 view also generates additional named reading skills, such as organizational flexibly. Brown and Haynes (198s) note that for component skils ‘there are no firmly established se of required subskills for reading’ (22). However, they indicate ‘hac research from a number of lines of evidence identifies sx skills that can be viewed usefully: 1 systematicty of variations in patterning of letters in the perception of ‘graphic features application of orthographic rules in phoneme-grapheme correspondence vocabulary knowledge use of semantic and syntactic context ability to hold information in shore-term memory ‘co-ordination of word knowledge and cextual information to elaborate ‘comprehension.a Reading sills Inany examination ofthese skills, Brown and Haynes (198) note tha there js ako a need for some other general measure of reading comprchension in addition to cess measuring each ofthese sil in order to examine the correlational structre ofthe differing component killa variables. Cate al. (i990) indicate cha focus on direc teaching ofthe sills associated with the various component skills using activities such as direct instruction and pretice of identifying gist, using context ro determine word meanings and extensive instuction in phonics, can improve reading abilcy in. fist language children ‘Thus, there are differing definitions of reading sls and dieing assump- tions about at what level of desripion they are separable. The extent to ‘which skills ae discretely defined comes in part fom whether one i aking developmental view ora more general cognitive-procesing view: Hierarchy of skills In many models of sills, ver time the focus of attention may change from lower-level skills to higher-level skills as reading abilcy is acquired, requiring Jess attention and becoming more automatic (Brown and Haynes, 1985; LaBerge and Samuels 1974; Logan 1979). Thac is, the lower level skills involving visual perception and phonic analysis become automatic with practice and require less conscious monitoring. This releases attention capacity to deal with comprehension aspects of reading. I is apparent from the discussion on reading sills chat many educators and researchers have identified reading skills at differen levels of decal. This is fone of the central issues in determining whether there are separable sills. ‘The levels of skill descriptions presented in Table 4.2 ae taken from Lancer exal.(979). Most of us would agree thar the Level r skill of ‘decoding prine’ isa skill different from the Level description ‘making sense. Hovieve, there may be quite abit of overlap inal of che Level 3 descriptions regardless of designa- tions such as decoding print, making sense, or questioning. This concern with how fine to draw a particular skill goes to the heart of whether there appear to be distinct and separable skills. As the skill becomes nartower in scope, the more dificuleic iso find distinct examples of reading acts that re ‘not comprised of moze than one skill. Furcher, as the reading act is made Lup of mote than one skill it becomes more and moze difficult to measure the skill without ic being overpowered by the existence of overlapping skills. This factor in curn males it diffcule ro determine whether the skills are cruly separate. As we begin to look atthe different approaches ro determin- ing whether reading skills represent a hierarchy of skills, this problem with identifying speci skills will become more apparent. Reading tile Level Level2 Level 3 Decoding pine Identifying levers, words, Scanning, fxating, phrases anticipating, categorizing, testing, matching, verifying Makingsense Assigning meaning to Anticipating syntactic and phrasesand sentences semantic categories, matching, verifying Questioning Noting discrepancies Retvieving material fom benween diferent statements orbetwcen hati read and what is known longeserm memory, ‘comparing inferring Table 4.2: The process of reading: thee levels of description (Adapted fom Lancer etal. (1979) In terms of any implicational hierarchy order of comprehension skills, (Clymer (1968) presented a taxonomy, developed by Barrett (N-D) in an unpublished paper. The taxonomy is divided into five ordered skill levels: (a) literal comprehension; (b) reorganization; (c) inferential comprehension; (@) evaluation; and (e) appreciation. According to Clymer, these categories are ordered to move from easy o difficult in terms of the demands of each ‘category. The Barret taxonomy is shown in Table 4.3. Itisinteresting to note that, in this eaxonomy of reading skil, the lowese level of reading skills discussed earlier such as Gordon’ (1982) names of letters, silent leters, etc, ‘or word-attack skis in general are nor listed. The reader is assumed to be beyond chat inital sage oftexe processing, Barret xonomy was leary influenced by Blooms more general processing taxonomy (Bloom 1956), which attempted to define eves of cognitive and affective procesing that represen differen level of educationsl abjecivesin areasother than reading comprehension. As with Barrett taxonomy, Blooms taxonomy assumes that the lowcr-numbere level are more basic than chose ‘vith higher numbers, Although intuitively somewhat reazonable, he presen- tation of such orderly categories may indicate more precision in categories than may psychologically exis. is nocatall lar that in Bartet’s taxonomy Inferring character traits (3.6) is more basic or essential than Judgments of fact ‘or apinion (42). There is certainly more overlap inthe categorie, and tasks devised to rflece cach category, than the ls suggests. Adiionally, he taxon- ‘omy does not reflet in any reasoned way that background knowledge will have a signifcanc effect upon the difficulty of any given category 0° how context in general will modify the proposed difficulty orease. ‘There have been a number of studies to determine whether there is empirical support forthe distinctiveness of reading skill, and for whether there is any 8Reading ils ; Reading sill 1.0 Literal Comprehension Recalling word meanings* 11 Recognition Drawing inferences about the mesning of word from context, nx Recognition of Deals tuna Recognition of Main ldese 1:13 Recognition ofa Sequence Recognition of Comparison tung Recognition of Cause and Effect Relationships 1.26 Recognition of Character Tris 12 Recall 1a Recall of Deals 1122 Recall of Main Ideas 123 Recall ofa Sequence 1124 Reel of Comparison a4 Recall of Cause and Elfet Relationships 112.6 Recall of Character Tats ‘20 Reorganization a Classifying 2.2 Ousliaing 23 Summarizing 24 Synthesiing ‘ye Inferential Comprehension 5.1 Inferring Supporting Details 3.2 Inferring Main Ideas 3.3 Inferting Sequence 5.4 Inferring Comparisons 4.5 Infersing Cause and Efect Relationships 366 Infersng Character Tees 5.7 Predicting Outcomes 538 Interpreting Figurative Language 40 Evaluation 142 Judgments of Realy or Fantasy ‘42 Judgments of Fac or Opinion 43 Tadgments of Adequacy and Validity 4-4 Jndgments of Appropriateness 445 Judgments of Wort, Desirability and Acceptability ional Response to the Content 5. Ienuicaion with Characters or Incidenes 5 Reactions othe Authors Us of Language 54 Imagery Table 4.3: Outline of the Barres taxonomy of cognisive and affective dimensions ofreading comprehension Finding answers to questions answered explicitly or merely in paraphrase” ‘Weaving rogether ideas from content Drawing inferences from the content” Recognizing writer's purpose, atitude, tone, nd mood* Ideauyinga writers technique Following the structure of passage’ identified ar unig in Davis (68) ‘Table 4.4: The eight reading comprehension kil addveued by Davis (1988, wr) hierarchical ordering of those skills identified as distincs. One of the frst language researchers most identified wich the proposition that there are separable processes or skillsinvolved in reading comprehension is EB. Davis (0944, 1968, 1972). Davis originally posited nine operational skills of com. prehension among mature readers. Thus, he was not addressing the early prerequisite skills of leter identification or phoneme-grapheme cortespon- lence. In his 1968 study, he reduced the hypothesized skills to be examined to eight. These are listed in Table 4.4. OF these eight, Davis found five variables to be unique contributors to comprehension cheough multiple regression. These were t 5,3 6, and 8, indicated with an asterisk () He noted that: comprehension in reading among mature readets is not a unitary mental skill Te is apparently a composite ofa leas ive of six underlying mental sills. Whether there is hierarchy of these skills by which one or more ‘must be available to the reader before hecan call on othersis nor shown by the data (Davis 1968: 655) ‘Through subsequent Factor analysis, Davis (1972) determined that he could, specify four unique factors or skills. These four skills were: 1 knowledge of word meanings 2 drawing inferences from the content 5 finding answers co questions answered explicitly or in paraphase in the passage and weaving together ideas inthe content 4 drawing inferences about the meaning of a word from context Davis (1972) notes hat while there are four unique skills in reading compre- hhension, his findings do not show evidence that the skills'can be arranged ina clear-cut order of cumulative agglomeration of simple skills in more complex oa8 Reading sill stalls” (672). Thus, while he does claim to identify distinct reading compre hension skill, he does nor see that they represent a hierarchically ordered set ‘ofskills Ieisnoc lear whatincerpretation to make ofthe fact thatthe skis are different in his 1968 study from those in his 1972 seudy. For example, drawing inferences about the meaning of a word from context was not unique in the first study but was unique inthe second. This most likely i function ofthe ‘wo differen statistical types of analysis used in the ewo differen analyses. ‘Two subsequent studies re-examined Davis’ original daca using different statistical procedures. Spearritt (1972) concluded that the statistical approach taken by Davis in his fictor-analytic study did not employ the ‘most comprehensive procedures, and he reanalyeed Davis’ data using a different statistical approach. The four skillshe identified were: 1 Recalling word meanings 2. Drawing inferences from the content 3, Recognizing a writes purpose, attitude, tone, and mood. 44 Following the structure ofa passage. He notes however, that vocabulary is the best differentiated ofthe four iden- tified skills. The other three skill were highly intercorrelated and the tests ‘measuring them may actually measuse one basic ability termed ‘reasoning in reading’, However, thismay bea product ofthe tests rather than bringing the skills themselves in question Lunzer etal. (1979) set out to examine che status of eight item types repre- senting comprehension skills. Their eight types are represented in Table 4.5 ‘They developed items for each type of skill across four diferent passages, Each test consisted of an extended passage, followed by approximately 30 ‘questions. The tests were administered to 257 pupils in thee last yearn one ‘of four British primary schools. Reliabiltes foreach test were reasonably acceptable, ranging from .83 to 85. Their factor analysis produced six Factors for cach ofthe four tests. However, the item ¢ypes on the different tess were ‘Word meaning, ‘Words in contest Literal comprehension Drawing inferences fiom single strings Deawinginferences from multiple stings Interpretation of metaphor Finding alien or main ideas Formingjdgments. Table .5: Comprebesion skill tem pes (Lancer et al. 1979) Reading sills associated with different factors and were inconsistent with one another. ‘Thus, the structure ofthe items is apparently est specific. When they put all the items of each type into the analysis, regardless of passage, they cobtaned a single factor that accounted for 71 per cent of the total variance. ‘They concluded chat the individual differences in reading comprehension across readers is nor the result of a mulrplicigy of specialized aptitudes, and they cerainly did not find suppore For any hierarchy of difficulty in che skills they examined. In short, an apparenc unitary construct of reading emerged from their study. “There have been numerous other aremprsin the firs language trate co identify sls and hei ordering. primarily wing fctr-analytic techniques to identify reading lls and any posible hierarchy of chose sl. Lennon (1962) examined 1a such studies He generalized arose the studies and identified fou basic reading kill general verbal factor: b) comprehesion of explicit informacion; ¢) comprehension of implicc information and 4) appreciation. However, in face half ofthe suis identified only a single factor andthe other studies varie inthe umber and aype of skills that were idencifed. In pare, te discrepancies tha have been found, and which oud the sue of whether cere are distinct reading sil, ae often methodologi- cal in nature First, many dffereneexising tts ranging from comprehen- sion to reading speed toler recognition have en used inthe diferent test and litle attention given tothe elas of he ress, Thorndike {0973 being the exception Additionally, in terms of question difficulty on a reading tes, the particular text may be the largest source of difficulty for the reader, nor the itera type. A scientific text may require avery diffcuelteral-inerpretation item while an item on a less difficule ext may ask for an easy inference. While there may be some general order for item type within passages, any taxonomy wil have to contain the caveat all other things being equal...’ However, it is reality of| literacy acts chat ‘all other things being equal’ rarely if ever occurs. For one thing, background knowledge will differ and this will affect the difficulty of answering a word-in-context eype item. Likewise, in addition to the text being the source of difficulty, the particular tak cha is required of the reader will affect diffculry. Whether an examinee is asked to select « mulkiple- choice item or to write our the answer will affect the relative dificult. Further, the interpretation of factors in factor analysis as much an arc a it isascience, Pethaps a different approach is needed, one that does not equate correlation of skills with absorption of sil. Thats, from the fictor-analyric approach, when tes icems identified with different skills correlate highly, i¢ isassumed that they ae no longer separate skills. The view ofthe componen- tial model offers some alternate approaches to interpreting highly correlated skills, In this view, the correlations ae reflections ofthe different seength of %~ Reading ills sclationships, and simply because ovo skills, or subsills, such as phonologi- cal awareness and word recognition, are highly correlated, does not mean thae they are pare of the same skill, To some exten, this approach requires thac research adopt an ¢ prioriconfitmatory type approach co Future research, “Through such approaches as structural equation modeling, rather chan exploratory factor analysis, skills and theie separability can perhaps be more adequately and eeliably identified Howevet, tha docs help us much right now: Whether a he present time ‘we accept that there are separable skills oF not will be duein large par (o Our particular academic orientation. There are researchers who attempt co rake a Skill and decompose i into is smallest intuitively meaningful parts (There ate several distinct skills in reading’). Thete are other researchers who are looking to combine whatever information is available in order co come up ‘with the most parsimonious and least complex explanation (There i only ‘one reading factor). Consequently, we should expect much disagreement about whether cher are skills. However, there appears to be scant evidence that chose skils that have been identified are hierarchically ordered in any ‘way other than the basic leter-decoding skills ae essential for application ‘ofany of the other reading kills. Rather, we should conceptualize reading as 4 multifceted information-processing activity requiring any number of subcomponents. Any one of these subcomponents can lead co reading differences across individuals. Ici lear that ease and facity of orthographic and phonological processing, s well as word-recognition ease, ae important contributors to efficient reading On skills and their hierarchical nature in the second language literature ‘There have been several attemps to develop sil lists and hierarchies in the second language reading ieratare. Much of the interest in ths work is designed to recognize that second language learners are generally older when they begin co read the second language than the normal age of beginning reading instruction in the fre language, are alecady literate in thei fist language, and have more specific needs than ealy fist language learners ‘The research neo sil in second language has taken one of two primary oi- emtations. The fst examines the ole ofthe traditionally termed lower-level ‘word processing skill while the second examines the role of comprehension skils in second language reading. The research into lower-level procesing Skil is reflected in studies by Brown and Haynes (1985) and several stds by Koda (1987, 1992, 1997. 2008). The research into the epantilty and ordering of second language reading comprehension skills is reflected in ‘work by Munby (1978), research on the guidelines ofthe American Council Reading ills fon the Teaching of Foreign Languages (acT#L) (Higgs 1984), and work by Alderson and Lakmani (1989), Alderson (1990), Hudson (1993), and the ‘Common European Framework (North 2000). Lower-level processing in second language Brown and Haynes (1985) examined differences in visual and orthographic coding sil of Japanese, Arabic, and Spanish speaking learners of English. “The sletionofchese language groups was mocvated bythe differen script types represented by the thee languages. Spanish largely uses the same alphabet as English, Arabic has a diferent sript, but one thats alphabetic in principle, while Japanese uses a sytem that isa combination of phono- logical, syllabic, and logographic symbols, The research was designed to examine differential visual and orthographic processing skills due to the differing nature ofthe ist anguage writing systems “To the surprise of Brown and Haynes (198s) ina task requiring the matching oforthographicstripsas sume ot different although the Spanish readers were {aster and made fewer errors than the Arabic readers, che Japanese were the fastest and most accurate a this orthographic recognition task even with ‘words up to ten leters. The Japanese were also fastest in matching same and differene numeric strings, while the Spanish and Arabic speakers were basically equal. However, on a read-aloud task, the Japanese speakers were slower and less accurate than the other ewo groups. Further, the Japanese showed a greater sensitivity o whether the words were familiar or unfamiliar, indicating chat they relied more on sight word knowledge than phoneme _grapheme cues, The Japanese readers appeared to have more diffculy in ‘connecting visual symbols to spoken units. Brown and Haynes speculace thar these results may be due co both he literacy and educational background ofthe learners. First, the Japanese appeared to process the leer combinations visualy rather than phonologically, as they would in their firs language. Spanish and Arabic speakers, on the other hand, were accustomed. toalphabetic writing systems that have a generally clear and consistent sign— sound correspondence. Second, the Japanese educational system has gener~ ally tended co value reading over speaking in second language instruction. “The study appears to indicat chat fist language skills are applied to second language contexts, a least for skis assaciated with orthographic processing and letter recognition. However, thete are problems with basing too much fom he study. The study does not report data about the different language groups in terms oftheir proficiency levels. Nor dos it eeport descriptive oF inferential seatistcs in adequate detail. However, it does provide inivial support indicating different word-atack skills among the languages and indicates how those skills may be transferred from the first language co the second language context. aa Reading sil, Koda (987, 1990) provides additional suppor forthe finding tha Japanese readers transfer the vsual-processng skills from thee frst language tothe second language when reading English, chough this process appears to be 4 tendency rather than an eitherfor dichotomy. She investigated the relaioaship benween orthography and cognition in reading . second language. She notes chat script systms can be logographic, in which each symbol represents an entre syllable, or orthographic, in which cach symbol represents single phoneme. The different representations of the cole of pric in che differen writing systems, a wells proceses of perception and Cognition, would be expete to differ for user ofeach orthographic system Koda (1087) summarizes pas esearch showing that orchographics that are sound based iavelve more phonological coding thaa those tha re meaning based. However, some research has alo found a phonological tace in Jogographic language processing, indicating the need for readers to supply phonological information to eetreved logographic meaning, Koda posi thac it is ‘reasonable to conjecture thar phonological recoding of lexical items is vital in eading sentences and txts, whether written in sound based sexi or meaning based sexpe (13) Koda’ (1987) study investigated the phonological inaccessibility assumed for logographic readers reading a sound-based second language. Twenty-six Japanese students were divided into evo groups. One group ws prescoted with pronounceable names (fr example, deft) associated with pictures of fish. The second group was presented with the sime pictures bu associated ‘with unpronounceable names for example, fi). Alter learning the names, subjects read a pasage describing the fish, using either the pronouneeable or "unpronouneeable forms che subjects had lesned. Koda found hat, in con- «east to English-speaking readers, the Japanese readesin er study read the "unpronounceable passage in significantly shorter amoune oF time than che pronounceable passage. The results suggest cha the Japanese tanseered their second language reading skill ro the second language reading in that phonological recoding i not a common strategy among Japanese readers reading Japanese In ewo subsequent studies, Koda (1989, 1990) investigated first language orthographic infuence on processing involved in second language readin with readers from language backgrounds with contrasting orthographics (Arabic, Japanese, Spanish, and English). Two pastages were employed. The first pasage described the characteristics of ive cits fish, and the second described the characcersticsof five imaginary cocktails In the experimental condition, Sansri symbols were used in place of the names ofthe fish and ‘cocktails. Inthe control condition, pronounceable English nonscnse words ‘were substituted forthe Sanskrit symbols. Half of each experimental and control group read one ofthe passages while the other half read the second passage. After reading the passages, the subjects were given a recall test in which they were asked to match the names ofthe ishicocktals to theit Reading ils characteristic descriptions inthe passage. The results showed chat subjects spent considerably more time on the Sanskrit passages than on the English nonsense versions across the groups. This indicated that phonological inaccessibility significantly impeded reading regardless of language background. However, it was found chat reading speed was significancly slower in the readers with the alphabetic phonological languages (Arabic, English, and Spanish) than the morphographic readers Japanese), The results appear to indicate thac the different phonological coding strategies are used ‘among subjects with contrasting orthographic systems, and that some phonological processing skill cansfers from first language to second Janguage. Thatis, the Japanese readers transferred their reading sill character recognition and the alphabetic readers transfered their phonologcally related skills Haynes and Car (1990) agin investigated lower evel frst language reading skilson second language reading. They examine skilled readers whose fist language was Chinese while reading English as a second language. In Chinese, characters map primaily on to meaningful syllabic units of specch ‘while English maps primarily on to phonemic units. Haynes and Carr nore that eadets of English often show evidence of phonological recoding during the reading proces uc that Chinese characters do not appea oelice much evidence ofthis when the sentences aresmple and consequent do no place high memory demands on comprehension. The study involved a two-stage component sil approach. The subjets were 28 university Fieshmen (low frou), 32 seniors (high group) ina university in Taiwan, and x5 American undergraduate voluntees fiom a general psychology clas ata large mid- ‘western univers: The frst sage of che study compared visul-processing ficiency of English alphabetic stimuli by Rest language Chinese readers at the wo levels of English proficiency (low and high) with performance of rate readers of English. The subjects were engaged in matching same or ditfeen pairs of orthographic stings that wee: :)orhographicall regular susings; 2) onhographicaly regular non-words; and 3) real English words. “The subjects were abo asked ro match stings of digits inorder to determine ‘whether the procesing of matching was language specific or a more general process. The second stage of the tesearch involved comparing individual Aifferences a the vsual-orthographic level reading outcomes such as: 1) ‘comprehension; 3) speed of reading; and) ability o learn new lexical tems fiom context. The stdy alo included measures of vocabulary knowledge, English grammar, English istening comprehension, and Chinese reading comprehension, Relevant findings from the study were: 1 The comparison of words, pscudo-words, and leer strings was used £0 indicate whether the presence of orthographic structure improved the level of recognition. The finding was that the English readers benefited from the existence ofthe orthographic cues toa significantly greater extent than did the Chinese readers in erms of speed and accuracy. 98Reading sil 2 The more advanced Chinese leaner of English didnot difer from the lower-level Chinese learners in erm of speed of reading or reading com- prehension, However, the advanced-level learners were significantly better in terms of the vocabulary-leaning measures. Haynes and Care take this san indcacion thatthe longer exposure to English orthography provided an advantage for his aspect of reading comprehension 3, Cortltions beeween all he measties indicate relative laf elaionship berween the measures of comprehension and those of reading speed Furthes, Chinese reading comprehension scores by the Chinese native speakers correlated with English reading comprehension scores but ‘not with meanites of English reading specd. However, che measure of orthographic same-diffeent recognition was related to reading compre henson. In general, the findings indicate that speed and comprehension ae different and that the factors that constrain speed and those chat constrain accuracy ae diferent. 4 Despite the disocaton of eading comprehension and reading speed in general, both these factors were associated with new word learning. For laynes and Cart, cis indicates that both speed and accuracy may be factors in lescal acquisition. This is consistent with the limited-capacty models of ineractive reading which sce accuracy and speed a fictors in integration oF word clues ‘The results generally indicate that che low-level skills of orthographic processing are both differen in kind from comprehension skills and tha the particular eype of orthographic-processing. practices in general first language literacy aces are transferred ¢o some extent into second language orthographic processing. Koda (1992) also examined the relationship berween lower-level processing skills and second language reading comprehension. Her focus was on the processes involved in extracting visal informacion from print, such as word orlerter identification. She examined the concepe chat deficits in lower-level processing skills puta strain on short-term memory and hence hinder ext integration. Thslimited-capacty model, similar to that discussed in Haynes and Carr (1990), posts that when a reader is preoccupied with lower-level processing, less cognitive capaciey is available for the higher-level text processing skill, Koda notes that research ‘suggests thac the reading process Ishewily dependent upon visual information contained in the ext, and that successful performances determined largely by efficiency in sampling visual information details (503). She predicts, then, that because of dilferences becween Japanese and English scrip, second language learners of Japanese vl suffer in reading comprehension to the extent that they have problems processing Japanese script. Fiffy-cight American college students enrolled in first-year Japanese took a batcery of tests. For reading comprehension, they rook clove test, paragraph, Reading stile comprehension test and a sentence comprehension est. Verbal processing was ‘measured using wo types of speed recognition tasks: word recognition in both Kanji and Hiragana, as well as leer recognicon in Hiragana text. Koda predicted that est performance onthe comprehension measures would largely be determined by word recognition of both Kanji and Hiragana. The second, prediction was that, because grammatical morphemes such as case markings and verb tense are written in Hiragana leteridentification ability would predict loze test performance. Since the sentence-level comprehension test id not require IsTERSENTENTIAL PROCESSING, it was predicted that readers ‘would have more capacity available for processing and thus verbal processing would not affect the sentence comprehension performance as much as it ‘would the cloze and paragraph comprehension tests. ‘The results of the study indicated chat the three verbal processing cests correlated highly with the text comprchension scores, but less well with the sentence comprehension measures. Results indicated that at early stages of| ability, Kanji recognition was the single significant predictor of comprehen- sion while as ability increased, Hliragana word recognitions entered as a significant factor. The reason fo this, Kada speculates, is that because Kanji are linked directly to a word meaning while Hirigana ar linked to sounds thac are added together to ercatea word, it may be easier to make asymbol- to-word connection than symbol-t-sound connections. Thus, initial retrieval of Kanji may be more salient when full control over the sound. system has not yet been formed, The analysis Further found chat lerter identification was the only significane predictor for he cloze tests, as pre~ dicted. Finally, none of the verbal processing measures significancly predicted che sentence comprehension test scores, lending support to the Timsited-capacity argument about early reading processing, However, the multiple-regression approach taken here requires large samples, larger chan those used inthis study. Again, descriptive statistics and reliabilities are not reported soi is difficult co extrapolate from the study t0 ‘other contexts. Without descriptive statistics it is not possible for us to see juse how low or high in ability the subjects were. Further clouding the implications of the study isthe face cha alot of the interpretation rests on ‘which particular Kanji and Hiragana were selected forthe test instruments Ina review ofthe literature on fis language orthographic processing, Koda (0997) again explores the effet of orthographic processing on word identif- cation. She again argues for the importance of visul-information concrol in the text asa key factor in reading. Here she aemps to clarify the nature of orthographic knowledge and it Function in second language contexts ‘noting that curren vocabulary sade show reading proficiency to be core- laced with the ability to infer meaning of unknown words from context {Chern 1993) Failure wo use context may be due vo word misientfcation. She notes that many ideniicacon errors resule fom insufficient information %6 Reading skills being derived during orthographic processing. First language research has indicated that poor readers are inefficient in cheir lexical retrieval From print (Perfetti 1991; Stanovich 1991). Thus, reading is dependent co a large degree on visual information, She further reiterates from her previous studies chat linguistic feacutes essential co sentence comprehension and production vary from language to language. In many instances, cros-linguistic variations in processing are consistent with what would be predicted based on the morphosyntactic features of the particular frslanguage For Koda, the face that Japanese employs three major orthographic systems (logographic, sylabary, and alphabetic) isan essential factor co take into account when examining second language reading processes by Japanese learners earning English or English fest language speakers earning Japanese. Again, she argues that, since lower-level processing plays a erucial roe in reading comprehension, since different processing i equited inthe proces- ing of differenc languages, and since previous research has indicated that sore frst language processing is transferred to second language then oxtho- sraphic knowledge playsa more central ol in scond language reading than ‘many have previously assumed, In essence, then, Koda sees these low-level skillsasparcof a hierarchical ordering of reading comprehension skills. They are necessary but not sufficient skill in successful reading, ‘The research by Brown and Haynes (1985), Koda (1987, 1989, 1990, 1992, 1997), and Haynes and Carr (1990) consistently indicate that lower-level processing in second language i affected by visual and orthographic coding. Farther, this coding in many cases is affected by the orthographic conventions of the learners’ first language. Efficient visual processing is essential fr lexical access, and consequentyis essential for comprehension, [As long asthe learner is a¢ a low level of processing al will be impeded. Higher-level skills in second language For potentially higher-order skills, Munby (1978) presents taxonomy across the second language sills of listening, speaking, reading, and writing within 4 communicative framework. Some example skills chat are most identified ‘with reading ae presented in Table 4.7. The aurnbers tothe let of each sill represent the original designation of skills 1-54 provided by Munby (1978), lity, comprehension Munby (1978) developed his taxonomy of skills in order to provide teachers and materials writers with a basis from which to select skills appropriate for seudenes with particular goals and needs, Its not clear the extent to which the skills in che taxonomy are viewed in a strict hierarchy of difculy. Munby states, in relation co skills in groups 20-3 that, ‘the ability to perform, skis 20, 21,22. and 23.1 appears to involve operating ata lower (or grosser) Readingstils 97 47, Recognizing the script of language 171 discriminating the graphemes 172 following grapheme sequences (Spelling system) 173 understanding punctuation 1g. Deducing the meanings and use of unfamiliar lexical tems, chough 19.1 understanding word formation: 19.4 stesfroots 19:12 afixation 19.13 detivation 19:04 compounding 19.2 contextual clues 30. Undersandingexplicly stated information 32, Understanding information in the tex, not explicidy stated, ehrough 21 making inferences 122 understanding figurative language 32. Understanding relations berween pars ofa text cohesion devices of 32.1 refeence (anaphoric and cataphorie) Se comparison | 329 substiution 32.4 ellipsis 3205 time and place relaess 526 logical connecters ugh grammatical 4p, Distinguishing the main idea from supporting dels by differentiating 3.1 primary tom secondary significance $92 the whole fom its parts, 393 aprocess fom its stages og category from exponent 395 saement fom example 396 fice from opinion 59:7 apropesition from an angument [Fe Basie veference dlls understanding and we of «44 graphic presentation, viz, Headings, rub-headings, numbering, indentation, bold print footnotes 442 tableof contents and index 43 crosereferencing ‘dacard caalogue 1445 phonetic ranscripcion! diaciis continued98 Reading ails 4. Skimming to obsain 451 the gt ofthe ext 152 agenerl impression ofthe text, $4. Relaying information 541 directly (commentary/ description concurrent with action) 542 indvety (eporting) Table 4.7: The taxonomy ofreading sills (Adapted from Munby (i978) level of delicacy chan that for tlate skills suchas those at 32 (19)- More generally the taxonomy gos fom skill, ‘discriminating soundsin solace ‘word fox, co 54, ‘telaying informatio’. However, kill, ‘distinguishing the main idea fom supporting detail’, does not appear in any logical way to be a less demanding endeavor chan skill 44, “basic reference skis understanding, and use of” and 44.2 table of contents and index. Sil, che taxonomy does present skill 54, telaying information’ ina way that is consistent with a reading to-write or litracy-act basis. With more specificity of the tasks that indicate actual literacy aets, this provides a form of contextualzation ofthe reading sil cis importane co note that Munby developed his taxonomy outside of an empirical framework, much like the Barrett and Bloom taxonomies discussed earlier. He is identified ase of communicative needs, He then reflected on ‘what skill would be necessary for the realization of diferent communicative activites and extrapolated those skills to relationships within the taxonomy. ‘Ths, the relative order and separability of the skills preseneed are noc based on any ofthe statistical procedures discussed in the preceding section Mecarey (1998) investigated whether performance on reading skills was uniform across earners and ceading texts, and whether the skills Formed a hierarchy for non-native eaders of Spanish, In this study, 38 fourth-semester scudents of Spanish and 30 in the fifth semester of study in a Midwestern vs university were chosen to contrast the skills in two ability levels. The subjects read two text selections, each four pages long. The skills isolated were: 1) locating details—recognition and paraphrase (skill); 2) simple inferential skillsunderstanding words in concext and recognizing cause/effec skill- 2); 3) complex inferential skills—recognizing main ideas and drawing con- ‘clusions (kills). The three-way repeated measures Analysis of Variance (aova) indicated a three-way interaction among level, pasage, and skill ‘There was also a main effec for skill and interactions between skill and level and skill and passage. Post hoc. tests showed no significant difference between skillt and skill-2, but skll3 was significanely different from skill-: and skill. However, of primary importance in the study isthe indication that the variables acribuced o pasage and skill are constantly in interaction Reading stile with one another. Thus, while skll-t and skll2 do noc appear to be diferent (co the extent that they have similar means), skil- is differen from them, nd chere isan overriding indication that they operate in consort throughout the teading process. Mecarey indicates thatthe findings show hierarchy to the extent cha skis was more difficult chan the other skills regardless ofthe ‘combination of passage or level, However, what isnot clear From studies such as this is how homogencous the test items are as representative samples of a skill, The study included only ewo items per passage of each sill type In essence, a study is needed that generates a lage number of all cree skill types and chen examines the extent to which there is overlap of che skill distribution, not just the mean of each skill group. The degree of overlap across skill designations isa central concesn in deciding the extent to which we feel comfortable asserting any hierarchical nature of skills. This is particularly relevant for the Mecartty stady given thac the paper does not report descriptive statistics or reliability estimates fr the cst. Alderson and Lakmani (198) investigated whether there existed ‘identi able separate levels of comprehension (260) inthe context of an English ‘szaminaton for students neding to show evidence of cheir English reading abil. They examined the issue feom to directions. Fist, nine teachers at the Instat for English Language Education atthe Univesity of Lancaster ‘were given the task of examining test and determining what each item was testing, and then co specify whether they Flt item was measuring lower idle, or higher-level ablts. They were chen given an envelope with a lise of reading sills and asked to indicate which of those skills were tested on the examination they eviewed. The results indicated very litle agreement among the ceachers on the levels being rested by the diferente items ‘Addivonally dhe judges were unable to consistently specify which reading shill the particule items represented. In fact, thee determinations were often as divergent as one judge’ indicating that a item was measuring explicitly seated information, ge, with other judges indicating that che itera measured lnowledgeof literary style; distinguish related items or ‘understand Coherence; pick out main ides. Alderson and Lukrmani found overall that ‘forsomething es than half of the ems involved, is possible to say that some identification of level scems tobe posible judgmenclly (265) In order to examine the issue further, Alderson and Lukmani carried out an empirical phase of the study involving an examination ofthe performance of| 100 students at Bombay University, India, onthe examination under inves tigation. The analysis involved the 1 items that had achieved agreement by. the judges in the previous stage ofthe study: According to the « prior class- fication, this 4-item test contained five lower-level items, four mid-level items, and five higher-level icems. The results showed a sight tendency for the higher-level items to be more difficult. However, they found this to be only a slight tendency. For example, chey found the percentage of correct ”Reading sills scores for lower-level questions to range from 42 percent to 89 percent, while ‘the middle-level skills ranged from 52 percent t0 73 pr cent and the higher- level skills ranged from 42 percent to 64 pet cent. One of the lower-level items and one ofthe higher-level items tid asthe most difficule items. Apparently, item difficulty does not closely align with proposed level of comprehension. ‘Their conclusions from the «vo phases of the study are that chere are no clearly defined isolated skills or definite implicationally ordered levels of skill hierarchy such that an examinee must answer tes items for the lower- level skills in order to answer items asessng the putative upper-level skills Indeed, they found tha the higher-order sil items had a much lower item discrimination chan the lower-order skill items, indicating that many of the lower-level students were answering higher-level skill tems wich moresbility than would be expected ina hierarchy. Ic very well may be tha these higher- level skill items are indeed measuring cognitive abilities rather than reading specific abilities. In that case, there is no reason to believe, as Alderson and Lukmani point out, that lower reading ability will necessarily inverfere consistently withthe application of cognitive ability. In an extension ofthe Alderson and Lukmani (989) study, Alderson (1990) focused on two diferent tet, tests for which there is more published data available, The tests are the Test of English for Educational Purposes (180) and the English Language Testing Service (su est. Boh tests were based fon Manby’s taxonomy, and assume that Munby’s taxonomy i ordered such that lower-level skills are a preequisie forthe higher-level sil. Fis, Alderson selected.8 experienced teachers oF to complete thre tasksbased onthe re) ateach skill ostensibly ested on the T= EP asto whether cach ‘wasa higher-or lower-level kill: 2) inspect the THEP reading test and match cach item tothe sil ostensibly rested: and 3) judge whether each cs item is esting higher- or lower lev! sll. ‘The uges showed substantial disagreement as to wich skills were ot lower-order skills. Only ‘understanding explicidly stated ideas’ received ‘unanimity in che judgments, Nine ofthe 14 skis had disagreements among the judges of jo per cent or more. In their evaluation of which reading skill each item was testing, che judges showed similar disagreement. On one item, although the judges differed from one another markedly, chey alo all slect- ed skill different ftom that designated by the test designer. The two items with the most consistency of judgment had discrepancies of 24 per cent and 4 per cent respectively: Finally, when the judges marched each item with \whecherit was higher- or lower-level tem, chere was disagreement approach- ing chaos, and some ofthe judges changed their rating of whether a particu lar skill was lower or higher based on the item content. Thus, the study reinforces che findings from the Alderson and Lukmani (198) study using different test with different judges. Reading sil In the empirical phase of his study, Aldetson examined he test esults from previously administered tess. For the Tee, he found chat the items that ‘were eargeted as higher-order tems bythe test writer had a mean facility oF 165 whereas che items that were supposed tobe lower-level items had a mean icem facility of 3. That is, the proposed lower-level items were more diff- cule forthe examinees than the supposedly higher-level items. A further analysis focuses upon those items in which there was relative agreement among the judges who rated the items previously. In this instance, as well, there is ltde relationship beeween ranked dificulty and examinee peeform- ance. In order to address this in more deal with a larger sample of items, ‘Alderson took the results for seven different TEE? tess. Again, chere were zo systematic results supporting assigned level of reading difculty and performance across he seven tes. Final inorder co further est the findings, Alder examined resuls on seven forms of the ELS tes, a different ype of test from those used in the stady just discused and in the Alderson and Lakmani (3989) study. Again, the results showed no systematiciy inthe rela tionship berween reading skill designation as represented by test tems and score. For thee est, the higher-level sls were more dificult in unee tess there was no difference beswecn the higher-and lower-level skills, and in one test the higher-level kil items weee easier than the lower-level kil items In 1993 1 investigated the role of reading skills and skill levels using item- response theory to help identify and delimi the role that particular sills phi: In this examination, he hypothesized tha: Asal or subs will havea narrow bandwidth of diffculy. 2 Skills or subskills may beat different levels depending on whether their ‘general dispersion is aca more difficult or less dificult location on the dliffculry continuum. 3, Ackil orsubskil will have very discriminating items, chacis, high slope values (Fludion 1993: 65) Subjects fr the study were Spanish-speaking chemical engineeting students studying English for specific purposes at a university in Mexico. Fach of tice forms of a reading test comprising a gramomar test, a test of reading ‘comprehension, and a multiple-choice cloze est measuring general reading ability was administered co approximately 200 students. Items on the cest ‘were categorized according to the task or item type they represented. The reading skills associated with items for each subtest were characterized a5, indicated in Table 48. ‘Three judges rated the items according to che category they represented, Items were placed in the category identified by at least ewo of the three judges. The results show that ofall the skills, only specific direct informationReading ils ‘Subtest and reading skill Reading comprehension rbret 1 Specific direct information: fae contained in che passage that can be answered by single term or number 2 Specific restatement-a phrase or clause taken directly from the text 3 Gist a question cha can be answered by comprehending a section of the ‘excand selecting the appropriate answer that ia paraphrase ofthe text 4 Graphichart: a question tha directed the students oa cha illustration, orgraph toanswerthe question Gammarvubiet 1 Simple verb cense: tenses including imple present, copula and simple pase 2 Connectors: logical connectors such as cau, thus, cheer, and eomequently 3 Pasive voice 4 Modal verbs 5 Perfectverenses 6 Rela pronouns pronouns in ave ass, inlined relative forms 7. Progressive: present or past “aliphiecce 1 Reading eomprebesiongrimmar undesanding prepositions information sneer! bs tem cmp yt over vocabulary 2 Reading comprehension/ocabuly: the procesingslngange eu interclausal), but a lexical choice is required ro solve it * >. Gramnmar/eading comprehen: he ar oft dif invahes ‘late shor-ange gamma consul afew wodson Sites ofthe lnk ovina meta pre case 4 ecabultyleating comprehen: the inary pea i tpery i voctbulary aloha aol eading comprehension in atthe ‘der mus understand the information pst hin cae bout Table 4.8: Reading sill by est type (Adapeed from Hudon (1995)) appears to have a consistent order in relation to the others, Tes mean score indicated that ic was an easier tsk eype than the others, Likewise, specific restaremene items had a high discrimination and a relatively identifiable bandwidth for difficulty: However, each ofthe skills generally represented a ‘wide band of difficulties, indicating a violation of the initial hypothesis described above, That is, the items associated with the different skills did not tend to have homogeneous difficulty vals, Reading sills “The esuls fom che study indicate chat the curtene analogy ofeadingas an interaction of top-down and botom-up proceses and skills isa restcting rmecaphor. The processes and sil are definitely interactive. However, the inceraction appears to be other than top-down interacting with bottom-up, at least for readers at all but the lowest ability. Rather, iis importa to cmphasize che overlapping parallel interactions beeween language skills, inceractionsthacar independene of il ves designated as higher or lower, top-down or bottom-up. Ics kel that theres moe variation within sil levels than bervicen levels of reading skill. Consequently, we need co re-examine models using levels and discrete categories. Most likly, the sls and proceses that are the bss for reading are much more numerous and ‘complex than existing taxonomies an accommodate Different ski, knowl- edge, and vocabulary are differentially related depending upon the context, purpose, text length, time, ete. associated with individual readers Summary and conclusions In general, both the first language and second language literature related to reading skills argue against the existence of strictly hierarchically ordered read ing skills. Further, chere appear to be broad categories of skills such 25 word- attack skills, comprehension skis, fluency skills, and critical reading skills rather than a detailed lis of numerous discrece reading skills. The applica- tion of identified skill, farther, appears to be mediated by such factors as ‘ext, purpose, and content, While the detailed lists of skills may be helpful in curriculum development and scope and sequence charts associated with textbooks and series, the actual operationalization of simple unitary skills is problematic. Reading acts and literacy events are sufficiently complex that they involve muleiple skills and skills char are nor unitary in thei structu Discussion and study questions 1 From the review ofthe literature how many eading sil do you identify as mporane? What ae they? Isic problem that sills appear o overlap? 2 To what exten do you think reading comprehension tsa reading sil as opposed toa general language ably? How will your answer to this ‘question affect the Kinds of activites you would include in a seadng 5 Aleson found dfcaly in determining sis or lvls of dicley sociated with sills. Whac does this imply for leaning to read and for testing reading? In is study, he indicated than pedagogical circle, sks tay be higher ina bicrarchy of values attached to the skills rather than acruallyimplcaionally ordered. Do you consider this corer, and fo, how is eimporane 103| seq Reading sills 14, Hudson assers three hypotheses regarding skills. What are they? To what STRATEGIES AND cextent do you believe tha these are correct hypotheses? Were they shown, tcexisein the present study? METACOGNITIVE SKILLS 5 What factots can affect the stability of reading skis and their difficuly? Inlooking across the various studies, what i your view of the tabiliy, and ‘our abiligy to iden reading skill? 6 Lookeatone of the extended lists of reading ills such as that by Barrett or Munby. Selec one of the skills and try co write atest item that would allow you to rest the skill, What ate some ofthe problems you have with ‘generating more than one item for the skill that would measure roughly the same abilcy? Introduction While the lst chapter focused on the reading skills utilized by reader, this chapreradresss the rol of reading strategies and metacognitive skill The relevant ates of concern in the fst language literature ate fst examined, and then the findings and implications from a second language perspective aa addressed. It would be useful at this time to clearly define the characteristics that distinguish skill, strategies, and metacognitive skill, However, previous literature does not help in this regard. As noted in the last chapter, some literature estricts reading sills o visual and auditory processes of decoding, and attributes the rest of the text-mediated meaning-making endeavor to cognitive or comprehension skis. Other researchers restrict reading skills to automatic processes and contrast thore to the view that strategies are con- scious and deliberace repair strategies. This question of whether of not to restrict strategies to ‘consciousness isa central issue in considerations of strategy imporeance. Wellman (i988) stares chat the term strategy more narrowly denotes some routine or procedure delid- cerately employed to achieve some end ... ln my usage, therefore strategies include only deliberate or intentional attempts to help oneself, for example, tohelp oneself remember... To bea strategy, the means must beemployed deliberately, with some awareness, in order to produce or influence the goal (Wellman 1988: 5: emphasis in original) However, Richards, Plat, and Weber (985) define strategy as procedures use i learning thinking, ee which erveas away ofteaching 4 goal. In language learning, leaning strategies... and communication strategies. ae those conscious or unconscious proceses which language leaenets make use of in leatning and using language. (Richards eal 1985274)