Processing Lexical Ambiguity And Visual Word Recognition -PDF Free Download

Processing lexical ambiguity and visual word recognition
06 Feb 2020 | 43 views | 0 downloads | 11 Pages | 1.32 MB

Share Pdf : Processing Lexical Ambiguity And Visual Word Recognition

Download and Preview : Processing Lexical Ambiguity And Visual Word Recognition


Report CopyRight/DMCA Form For : Processing Lexical Ambiguity And Visual Word Recognition



Transcription

14 BENTIN AND FROST, Koriat 1984 but the nature of this process has never ent phonological representations of a Hebrew consonant. been elaborated One possible assumption is that although string are related to only one visual logogen. lexical access is mediated primarily by orthographic. codes the decision is delayed until one phonological al EXPERIMENT 1. ternative is determined Thus a specific word must be. deciphered from print before the positive decision can be Experiment 1 compared lexical decision performance. generated A second possibility is that a positive decision for unvoweled ambiguous consonant strings and for their. can be based on the phonologically ambiguous homo voweled disambiguated alternatives Each ambiguous con. graphic cluster common to all alternatives and that lexi sonant string could represent either a high frequency or. cal disambiguation is a subsequent process We attempted a low frequency word The same voweled words were. to investigate the validity of each of these assumptions presented in two task conditions that were determined by. by comparing lexical decision performance on Hebrew the nature of the nonwords employed In the optional con. consonant strings in the unvoweled ambiguous form and dition the nonwords were legal but meaningless permu. those disambiguated by the vowel marks tations of the consonantal letters of real words In this. Early studies revealed that in English lexical decisions condition words and nonwords were distinguishable sim. for homographs are faster than for nonhomographs H ply on the basis of the consonantal pattern so that there. Rubenstein Garfield Millikan 1970 H Rubenstein was no logical necessity for the subject to attend to the. Lewis M A Rubenstein 1971 One common expla vowels in order to make a lexical decision In the obliga. nation for this effect of homography is that words with tory condition special nonwords were used The con. multiple meanings have multiple entries and therefore the sonantal pattern of these nonwords corresponded to real. probability of encountering during lexical search one of words in Hebrew but they were presented in an inap. many entries of a word with a large number of meanings propriately voweled manner Thus the special nonwords. is greater than the probability of encountering the single employed in the obligatory condition were in fact words. entry of a word with only one meaning Recently Jastr when unvoweled or when voweled correctly therefore. zembski 1981 elaborated on this model on the basis of in this condition the subject was forced to attend to the. Morton s 1970 1979 logogen theory assuming that vowels without which discrimination between words and. words having multiple meanings are represented by one nonwords could not have been made Recall that the. logogen for each meaning Jastrzembski assumed that difference between the high and low frequency alterna. logogens accumulate evidence in a probabilistic fashion tives of each homograph was indicated in print only by. Thus the more logogens a word has the more likely one the vowel marks Therefore the relative sizes of the fre. of them will reach threshold Accordingly a word with quency effect in the optional and in the obligatory condi. many meanings and therefore many logogens will be tion may reflect the extent to which the information. more likely to have one of its logogens reach threshold provided by the vowels was processed in each of these. sooner than a word with few logogens Note however conditions by subjects while generating the lexical de. that all the probabilistic models of lexical access that were cision. proposed to account for the effect of homography were In addition to comparing performance for ambiguous. based on responses to homographs that were also homo strings Experiment 1 also compared lexical decision per. phones In contrast it has been reported that in English formance for voweled and unvoweled unambiguous. the latency of naming homographs that have different words Unambiguous words were consonant strings that. pronunciations for different meanings e g wind is sig represented only one legal phonological derivation Thus. nificantly longer than that of naming homographs with although the phonology presented in print was not com. a single pronunciation e g fall Kroll Schweickert plete if vowel marks were missing access to these words. 1978 did not require a choice between different phonological. These data suggest that in English phonological am alternatives Therefore any difference between voweled. biguity must be resolved before the lexical decision is and unvoweled unambiguous words should reflect only. generated If the same strategy is employed in Hebrew the effect of missing phonemic information but no ef. lexical decisions about unvoweled phonologically ambig fects of ambiguity Comparison of lexical decision per. uous strings should be slower than decisions about any formance for ambiguous and unambiguous words in the. of their explicitly voweled phonologically disambiguated optional and obligatory conditions is particularly impor. alternatives On the other hand faster responses to the tant because regardless of the ambiguity of the words. ambiguous homograph than to any of its disambiguated different nonwords might differently affect performance. alternatives would suggest that phonological disambigu on low and high frequency words Duchek Neely. ation is not required for lexical decision We will claim 1984 cited by Balota Chumbley 1984 James 1975. that such a result supports the possibility that lexical de If lexical decisions for unvoweled and for voweled. cisions for phonologically ambiguous consonant strings words with ambiguous consonant strings in the optional. are based primarily on the abstract orthographic represen condition are based primarily on the ambiguous consonan. tation which is common to all phonological alternatives tal information then 1 the reaction times RTs for un. In other words such a result might suggest that the differ voweled ambiguous words should not be longer than RTs. LEXICAL AMBIGUITY AND WORD RECOGNITION IN HEBREW 15. for any of their voweledhigh and low frequency disam words were high frequency and 16 were low frequency accord. biguated alternatives and 2 word frequency should have ingto the same criteria as above Rating was performed by a different. a larger effect in the obligatory than in the optionalcon group of 50 judges in a similar manner Frost Katz Bentin. dition since subjects in the obligatory condition were in press The mean ratingwas 4 18 for the high frequency unam. biguousgroup and 1 71 for the low frequencyunambiguous group. forcedto processthe vowelsin order to discriminate be The nonwords in the optional condition were 32 pronounceable. tweenwordsand nonwords On the other hand if phono but meaningle lS permutations of consonants arbitrarily assigned with. logicaldisambiguation is mandatory for lexicaldecisions vowel marks We shall label these stimuli regularnonwords Note. and if positive responses for Hebrew homographs are that regular nonwords could not be read as words even if presented. based on successful access to an entry uniquely related unvoweled In contrast in the obligatory condition the nonwords. to one alternatemeaning then 1 RTs for the unvoweled were in fact words when unvoweled or when voweled correctly. the consonantal patterns corresponded to words but in this study. words should be longerthan for the voweled words since. they were inappropriately voweled These nonwords were labeled. vowels simplify the phonological disambiguation of the special nonwords The special ronwords were generatedfrom words. ambiguous consonant string and 2 word frequency of average frequency. shouldhave a similar effect on lexical decisionsfor am All stimuli were generated by a PDP ll 34 computer and dis. biguous and unambiguous words in the optional and in played at the center of a Tektronix CRT The size of each letter. the obligatory conditions was 1 2 x 1 2 em and the length of the whole word was between. 4 5 and 6 5 cm three to five letters subtending a visual angle of. approximately 4 5, Method Design Six test lists were assembled In Lists A to D all stimuli. Subjects Ninety six undergraduate students either participated were presented with the vowel marks In Lists E and F the stimuli. as part of the requirements of an introductory psychology course were presented in the regular unvoweled manner Lists A and B. or were paid for their participation They were all native speakers were presented in the optional condition List A comprised the 16. of Hebrew with normal or corrected to normal vision high frequency alternatives of the ambiguous pairs the 16 low. Stimuli The ambiguous words were 16 unvoweled consonant frequency unambiguous words and the 32 regular IlOIlwords List B. strings Each string representeda pairof different nouns with differ comprised the 16 low frequency alternatives of the ambiguous pairs. ent meanings different pronunciations and different word frequen the 16 high frequencyunambiguous words and the same 32 1100. cies Lexical decision performance for each of these unvoweled words used in List A Lists C and D were presented in the obliga. strings was compared with lexicaldecision performance for its two tory condition List C comprised the same words as List A but. voweledalternaterepresentations Figure 1 The 16pairs of nouns the set of 32 special nonwords was employed List D comprised. were selected from a pool of 50 similar pairs on the basis of fre the same words as List B and the special nonwords Lists E and. quency evaluation as follows The 100 words were printed with F were similar to Lists A and B respectively but words were. the vowel marks on a single page in random order Fifty under presented without the vowel marks Therefore Lists E and F in. graduates were asked to rate each word on a 5 point scale ranging cluded identical ambiguous words because the different alterna. from least frequent 1 to most frequent 5 Estimated frequency tives of the ambiguous words were indistinguishable without the. for each word was calculated by averaging the rating across all 50 vowel marks but different unambiguouswords Different groups. judges Words rated above 3 5 and below 2 5 were includedin the of 16 subjects each were randomly assigned to each of these. high and low frequency groups respectively The 16 word pairs lists. selected for this study were those in which one member was rated Procedure The experimenttook place in a semidarkenedsound. high and the other was rated low in frequency The mean rating treated room Subjects sat approximately 70 em from the screen. of thehigh frequencygroup was4 23 and that of the low frequency They were instructed to press one of two alternative microswitch. group was 1 69 Although in the voweled form these consonant buttons according to whether the stimulus on the screen was or. strings were unequivocal1y specified since their consonantalstruc was not an actual Hebrew word The dominant hand was always. tures were shared by different words we labeled the critical pairs used for the yes i e word responses and the other hand for. as ambiguous and differentiated between voweled and unvoweled the no i e IlOIlword R llpOIJlleS Subjeds presenIIld with Lists C. ambiguous words and D were warned about the special nature of the nonwords. The unambiguous words were 32 consonant strings each of which Following the instructions 32 practice trials 16 words and 16. represents in print only one word Sixteen of the 32 unambiguous nonwords were presented Words nonwords on the practice trials. High Frequency Low Frequency Unvoweled,Alternative Alternative Print. re presenta t ion,davar dz vz r,transcription,th i ng pest. translation, Figure 1 Example of ambiguity in Hebrew print Note tbat the two wonl i represented.
by the same COllllOnant string are not homophones,16 DENTIN AND FROST. were prepared in congruence with those on the test list that fol 1000. lowed The 64 test trials were presented next at a rate of one stimulus. UNAMBIGUOUS WORDS, every 2 5 sec The subject s response terminated the stimulus ex. posure If no response was given within 2 sec the stimulus was. removed and an error was marked The subject started the test trials. by pressing a ready button RTs were measured in milliseconds Unvoweled Voweled Voweled. and errors were marked, The RTs for correct responses were averaged for each 700. word over the 16 subjects who were exposed to it and 0. for each subject over the 16 words in each frequency. group RTs that were above or below two standard devi. ations from a subject s or a word s mean were excluded. and the mean was recalculated Less than 1 5 of the RTs. were outliers 500, We will describe first the comparison between the am High Low High Low High Low. biguous consonant strings in the unvoweled and the two Freq Freq F r eq Freq F r eq Freq. voweled presentations The RTs to the high frequency OPTIONAL OBLIGATORY. words in List A to the low frequency words in List B CONDITION CONDITION. and to the unvoweled ambiguous words in List E2 were. compared by a one way ANOVA 3 and the Tukey A Figure 3 Mean reaction times and standard errors to unambigu. post hoc procedure RTs to unvoweled consonant strings ous words in the lexical decision task In the obligatory condition. were faster than to any of the two voweled alternatives nonwords would become real words if presented without the vowel. and RTs to the high frequency alternatives were faster. than to the low frequency alternatives Figure 2 The. ANOVAs revealed that all differences were significant The effect of the special nonwords on lexical decisions. for the stimulusanalysis F 2 30 22 09 MSe 4 521 for high and low frequency voweled ambiguous words. p 0001 for the subject analysis F 2 45 5 15 MSe were assessed by comparing the RTs to ambiguous words. 14 700 p 01 and minF 2 63 4 18 p 05 in the optional condition List A and List B and in the. obligatory condition List C and List D in a two way. AMBIGUOUS WORDS,condition X frequency ANOVA RTs to ambiguous.
words were slower in the obligatory condition than in the. optional condition minF l 24 10 96 p 01, 1000 Across conditions RTs to high frequency words were. faster than to low frequency words minF l 49 9 97,p 01 However the most important result of this. 900 comparison was that the frequency effect was twice as. large in the obligatory condition as in the optional condi. tion Figure 2 This interaction was significant both for. a 800 the stimulus analysis F l 15 11 91 MSe 3 346. Vl P 004 and for the subject analysis F l 6O 4 25,E MSe 92 036 p 05 rninF l 73 3 13. The effect of the vowel marks on high and low, frequency unambiguous words was assessed by compar. ing the voweled high frequency unambiguous words in. 600 List B and the voweled low frequency unambiguous. words in List A with their unvoweled presentations in. Lists E and F respectively Figure 3 A two way vowel. 500 condition X frequency ANOVA revealed a significant. Without High Low High Low,Vowels Freq Freq Freq F req.
frequency effect F I 6O 7 44 MSe 12 334, p 01 but no effect of the vowel marks and no inter. OPTIONAL OBLIGATORY action, CONDITION CONDITION The effect of the special nonwords on lexical decisions. for voweled unambiguous words was assessed by com, Figure 2 Mean reaction times and standard errors to ambiguous. paring the RTs to voweled high and low frequency un. words in the lexical decision task In the obligatory condition non. words would become real words if presented without the vowel ambiguous words in the optional condition List B and. marks List A and in the obligatory condition List D and. LEXICAL AMBIGUITY AND WORD RECOGNITION IN HEBREW 17. List C A condition x frequency two way ANOVA re However we have obtained this result with consonant. vealed that lexical decisions were slower in the obliga strings that represented in print not only different mean. tory than in the optional condition F 1 60 9 48 MSe ings but also different words i e different phonologi. 18 471 p 005 and slower for low frequency than cal representations Therefore our results are in contrast. for high frequency words F I 60 3 84 MSe to results obtained in similar studies about English Kroll. 18 471 P 05 but in contrast to ambiguous words Schweickert 1978. there was no interaction between the two factors F 1 60 There are several ways to explain our data The first. 0 31 The different effect of nonword condition on explanation is based on the hypothesis that a lexical deci. ambiguous and unambiguous words was verified by an sion for an ambiguous unvoweled consonant string is. ambiguity ambiguous unambiguous x condition op based on accessing at least one of the words it represents. tional obligatory x frequency high low ANOVA re The existence of several possible entries increases the. vealing a marginally significant three way interaction probability that one of them will be accessed This ex. F 1 120 3 57 MSe 20 067 P 06 planation is unlikely for two reasons 1 in English this. The nonwords in Lists A and B were identical as were effect was found only for homographs that represented. the nonwords in Lists C and D and those in Lists E and different meanings but only one pronunciation 2 the. F Therefore the RTs to nonwords in the optional condi probabilistic explanation is based on the assumption that. tion Lists A and B were compared with the RTs to non the distribution of the RTs to the different meanings. words in the obligatory condition Lists C and D and with represented by the homograph partially overlap Although. RTs in unvoweled nonwords Lists E and F by a one this is probably true in our study as well we tried to. way ANOVA and Tukey A post hoc procedure RTs to diminish this overlap by selecting only homographs that. voweled nonwords in the optional condition and to un represented both very high frequency and very low. voweled nonwords were not significantly different frequency words. 764 msec and 739 msec respectively and both were The second explanation of our data is based on the as. faster than the RTs to nonwords in the obligatory condi sumption that Hebrew readers are more familiar with un. tion 1 055 msec F 2 93 36 16 MSe 27 255 voweled than with voweled words This explanation. p 0001 however is contradicted by the absence of an effect of. The distribution of errors between the different stimu vowel marks on lexical decisions for unambiguous words. lus groups and conditions is presented in Table 1 and for nonwords. Although the uneven distribution of errors within cells The third explanation is congruent with our hypothe. precluded a significant statistical analysis the overall pat sis If lexical decisions for ambiguous consonant strings. tern does not suggest a speed accuracy trade off between do not require phonological disambiguation addition of. conditions vowel marks is unnecessary Moreover vowel marks add. information that probably cannot be totally ignored and. Discussion thus increases the word processing time Therefore we. The comparison of responses to voweled and unvoweled suggest that these data support the hypothesis that lexical. ambiguous words revealed that explicit presentation of the decisions for unvoweled Hebrew homographs may be. vowel marks that disambiguated the consonant strings did based on information that is common to all lexical alter. not facilitate lexical decisions relative to decisions about natives that are represented by one consonant string De. the ambiguous unvoweled strings In fact an opposite ef cisions based on the common information should be at. fect was found Apparently this result simply replicates least as fast as those based on accessing specific entries. previous findings that suggested that lexical decisions for for the additional reason that the frequency of the com. homographs are faster than those for nonhomographs mon consonant string is higher than the frequency of each. Mean Percent of Errors and Standard Error of the Mean SEM in Lexical Decision for Voweled and. Unvoweled Ambiguous and Unambiguous Words and Nonwords in the Optional and Obligatory Conditions. Ambiguous Words Unambiguous Words,Voweled Voweled Unvoweled Nonwords. Hi Freq Lo Freq Unvoweled Hi Freq Lo Freq Hi Freq Lo Freq Voweled Unvoweled. Optional Condition, Percent Errors 3 19 13 67 1 17 2 73 6 25 3 13 11 33 3 12 6 25.
SEM 1 22 1 76 061 1 10 1 56 1 10 1 85 1 26 1 42,Obligatory Condition. Percent Errors 9 66 14 50 2 34 8 20 14 35,SEM 1 84 2 90 0 87 2 39 2 05. 18 BENTIN AND FROST, of its individual lexical realizations This suggestion is EXPERIMENT 2. supported by previous results which revealed that lexi. cal decisions for low frequency ambiguous consonant In contrast to making lexical decision naming a pho. strings are faster than for low frequency unambiguous nemically ambiguous string of consonants necessarily re. consonant strings Bentin et al 1984 quires the selection of only one of the alternative phono. Word frequency affected lexical decision performance logical representations The main purpose of. in both the optional and the obligatory conditions sug Experiment 2 was to investigate the nature of this selec. gesting that subjects did not ignore the differences between tion in an attempt to enhance our understanding of the. the words even if such a detailed analysis was not abso process of disambiguation of Hebrew unvoweled con. lutely necessary for lexical access Apparently this result sonant strings. suggests that when the element of ambiguity is eliminated Previous studies in Hebrew employed only unambigu. even by adding unfamiliar vowels lexical decisions are ous words that is consonant strings that represented only. based on a full analysis of the graphemic and the pho one lexical item and could be pronounced correctly in only. nemic codes However the enhanced effect of word fre one manner It has been reported that naming voweled. quency in the obligatory condition suggests that such a words was delayed when the vowel marks were incom. simple conclusion might be premature Recall that in the patible with the correct sound of the word even if the. optional condition discrimination between words and non subjects were instructed to ignore the vowels but it was. words could have been accomplished even if the vowels equally fast ifunvoweled words were compared with cor. were ignored Therefore a relatively reduced effect of rectly voweled words Navon Shimron 1981 In a. frequency may have resulted if some of the decisions were more recent study however vowels were found to speed. based only on the consonant strings which were identi up naming although they had no effect on lexical deci. cal in the high and low frequency groups whereas other sions Koriat 1984. decisions involved processing of the full phonological Recent data from our laboratory revealed that in con. code However an alternative explanation is possible The trast to more shallow orthographies such as Serbo. overall slower RTs in the obligatory than in the optional Croatian and English in Hebrew making lexical deci. condition suggest that the word nonword discrimination sions for unvoweled unambiguous consonant strings was. was more difficult in the former than in the latter condi faster than reading the same words aloud Frost et al. tion Obviously the words and the nonwords in the in press Furthermore significant semantic priming was. obligatory condition were more alike reducing the cer found for naming in Hebrew but not in Serbo Croatian. tainty level of the subjects The uncertainty could have These results suggest that although naming voweled. been greater for the low than for the high frequency Hebrew words can in principle be based on phonetic cues. words since the former were subjectively more similar generated via a process of grapheme to phoneme trans. to nonwords This interpretation is supported by previ formation naming unvoweled words is always mediated. ous studies that revealed that low frequency word deci by the lexicon Therefore lexical ambiguity would in. sions are facilitated more than high frequency word de fluence naming because it probably requires a choice. cisions by the essence of unpronounceable nonwords among several lexical representations In the present ex. Duchek Neely 1984 cited by Balota Chumbley periment we compared the naming of ambiguous and un. 1984 James 1975 Thus the interaction between the ef ambiguous voweled and unvoweled stimuli This com. fect of frequency and the nature of the nonwords may be parison we hoped would shed additional light on naming. explained as a postaccess decision factor rather than a words in a deep orthography in general and on the rules. different manner of lexical access of disambiguation of Hebrew homographs in particular. An insight into the origin of the interaction between the An additional aspect of word processing that might be. nonword type and the word frequency effect can be influenced by lexical ambiguity is the word frequency ef. achieved by comparing the condition effect on the am fect on naming Several studies suggested that frequency. biguous words with the effect on the unambiguous words effects are smaller in naming than in lexical decision tasks. Recall that for unambiguous words the word frequency Andrews 1982 Balota Chumbley 1984 Frederik. effect did not interact with the effect of the nonword type sen Kroll 1976 The same relationship was recently. If the difference in the magnitude of the frequency effect found in Hebrew Frost et al in press In that study. in the obligatory and optional conditions was not related however only unambiguous words were employed Nam. to word ambiguity it should have emerged with unam ing ambiguous words on the other hand might be af. biguous words as well Therefore we conclude that the fected by frequency both during lexical access and at post. interaction between the nonword type and the frequency lexical processing stages Balota Chumbley 1985. effect was related to the ambiguity factor Although other Forster Bednall 1976 Simpson 1981 Therefore it. explanations are possible we propose to consider this in is possible that naming voweled ambiguous words might. teraction as corroborative evidence that lexical decisions be affected more strongly by the relative frequency of their. for ambiguous Hebrew words do not require phonologi alternative phonological representations than naming. cal disambiguation voweled high and low frequency unambiguous words. LEXICAL AMBIGUITY AND WORD RECOGNITION IN HEBREW 19. A secondary purpose of Experiment 2 was to test this All RTs were normalized by excluding responses that. hypothesis were above or below two standard deviations from the. subject sor the word s mean Lessthan 1 5 of the RTs. Method were excluded, Subjects The subjects were 64 undergraduates who were paid The time to initiation of naming ambiguous consonant. for their participation They were all native speakers of Hebrew strings in the high andlow frequency presentations Lists. with normalor corrected to normal vision None of them had par. ticipated in Experiment I A and B respectively and in the unvoweled condition. Stimuli and Design The stimuli were identicalto those used in List E 5 was analyzed by one way ANOVAs across. Experiment 1 Only List A voweled high frequency ambiguous stimuli and across subjects see Note 3 In contrast to. words low frequency unambiguous words and regularnonwords lexical decision performance the unvoweled consonant. List B voweledlow frequency ambiguouswords high frequency strings werenamed as fastas the high frequency voweled. unambiguous words and regularnonwords and ListsE and F un alternatives but both groups were namedfaster than the. voweled replicas of Lists A and B respectively were used each. presented to a different group of 16 subjects The assignmentof. low frequency alternatives Figure 4A This pattern of. subjects to lists was random performance wassupported by the ANOVAs followed by. Procedure The conditions of Experiment I were repeated in this Tukey A post hoccomparisons forthe stimulus analysis. experiment The subjects were instructed to read aloud as fast as F 2 30 25 79 MSe 2 268 p 0001 for the sub. possible words and nonwords that were presented to them on the ject analysis F 2 45 5 40 MSe 12 679 p 008. CRT screen In thevoweled condition Lists A and B subjects were minF 2 65 4 46 p 025. toldto read thestimuli as voweled In theunvoweled condition Lists The comparison between voweled and unvoweled un. E and F subjects were told to read aloud as fast as possible the. words and the nonwordsthat were presentedon the screen Since ambiguous words is presented in Figure 4B Statistical. in Hebrew reading withoutvowels is the rule rather than the ex significance of the differences was assessed by vowels. ception no additional instructions weregiven or solicited for read voweled unvoweled x frequency high low ANOVAs. ing the words Subjectswere told however that there was no cor across stimuli andacross subjects Unvoweled words were. rect or incorrect way to read the nonwords and that they could named 30 msec faster than voweled words This differ. pronounce them by arbitrary assignment of vowels to theconsonants ence was significant for the stimulus analysis F 1 30. Subjects responses were recorded by MuraDX tt8 microphone. which was connected to a Colbourne Instruments voice key RTs. 16 47 MSe 13 718 P 001 but not for the sub, were measured in milliseconds from stimulus onset by the com ject analysis F 1 30 1 Consequently the minF was.
puter responses were recorded on a magnetic tape for off line not significant The high frequency words were named. analysis 26 msec faster than the low frequency words Thisdiffer. ence was not significant either for the subjector for the. Results stimulus analysis The interaction between frequency and. As in Experiment 1 the RTswereaveraged across sub vowel conditions was not significant. jects for each word and across words for each subject Noneof the unvoweled wordswaserroneously read as. Ambiguous words Unambiguous words,High Low High low High low. Fr eq Fr ec Freq fOreq Freq Freq,Unvoweled Voweled Unvoweled Voweled. Figure 4 Mean reaction times and standard errors to voweled and unvoweled words in the. naming task A ambiguous words B unambiguous words,20 BENTIN AND FROST. a nonword and except four occasional errors all voweled to that for naming unambiguous words originates from. words were read correctly All four errors were on low postaccess processing In agreement with Forster 1981. frequency alternatives For technical reasons errors made see also Kinoshita 1985 we suggest that the delay in. while reading voweled nonwords could not be recovered naming voweled low frequency ambiguous words reflects. from the raw data the time spent in evaluating the initially generated high. The analysis of the words that were actually pronounced frequency phonology vis a vis the presented vowels and. by each subject when unvoweled ambiguous words were rejecting this alternative in favor of the low frequency. presented revealed that for 4 of 16 consonant strings the phonology. high frequency alternative was unanimously chosen, by 32 subjects and for 8 strings the high frequency al GENERAL DISCUSSION. ternative was pronounced by more than 75 of the sub. jects and in no case was the high frequency alternative In this study we investigated the process of disambig. chosen by less than 45 of the subjects In contrast 7 uation of phonemically and semantically ambiguous. low frequency alternatives were not chosen by any sub Hebrew printed words and the effects of this deep orthog. ject and 7 were pronounced by less than 10 of the sub raphy on lexical decision and naming performance The. jects whereas none of the low frequency alternatives was results suggest that when unvoweled consonant strings are. chosen by more than 40 of the subjects presented lexical decisions are based primarily on the am. The nonwords in the voweled condition List A and biguous grapheme lexical disambiguation is achieved in. List B and in the unvoweled condition List E and List F parallel but has little influence on the decision processes. were compared by a t test In contrast to lexical decisions per se In contrast phonemic and therefore semantic dis. naming of unvoweled nonwords was significantly faster ambiguation must precede naming of unvoweled con. than naming of voweled nonwords 757 msec vs sonant strings and we suggest that the process of disam. 843 msec t 62 2 24 P 03 biguation is based on a postaccess race of the different. phonemic semantic lexical representations to which the. Discussion specific consonant string is related The result of this race. The naming time data revealed that the only significant i e the word that is pronounced is determined to a great. difference between naming voweled and unvoweled words extent by the relative frequency of the alternative lexical. was the slower naming of the voweled low frequency representations These conclusions are based on the fol. alternatives of ambiguous words relative to naming of lowing pattern of observations. the unvoweled alternatives In addition the frequency. effect for naming ambiguous voweled words 99 msec Stimulus Ambiguity and Lexical Decision. was as large as the frequency effect in the lexical Presentation of the vowel marks in conjunction with the. decision task and considerably larger than that for consonant letters disambiguates the printed Hebrew word. naming unambiguous voweled words 16 msec Note that Even though word perception should have been facilitated. for voweled unambiguous words the relationship between by exclusion of the ambiguity factor our data revealed. frequency effects in naming and lexical decision tasks that addition of vowel marks significantly delayed lexi. replicates our findings with Hebrew unvoweled unambig cal decisions for ambiguous words The direct implica. uous words as well as findings from previous studies in tion of this result is that in Hebrew the information. English provided by the vowels is not absolutely necessary for. These data suggest that vowel marks did not facilitate lexical decisions A possible explanation is that the lexi. the naming of printed words Moreover vowel marks in cal decision for unvoweled homographs is normally based. terfered with pronunciation when they imposed an unex on the ambiguous string of letters that may be recognized. pected interpretation of the grapheme Therefore we pro as a word or rejected without access to any of the alter. pose that the subjects might have initially generated a native meanings phonological disambiguation is accord. phonological code on the basis of the consonantal infor ing to this hypothesis a postaccess process This hypothe. mation In the case of ambiguous words the most fre sis implies that lexical decisions for unvoweled ambiguous. quent alternative was probably activated A subsequent Hebrew words are made without reference to their mean. consideration of the vowel marks had no significant ef ing and therefore apparently contradicts the well. fect on the processing time if they were congruent with established effects of word meaning on lexical decisions. the subject s initial response tendency as was the case e g the semantic priming effect We can however ac. with the high frequency alternatives or with the unam count for this apparent contradiction by assuming along. biguous words but vowel marks required a time with Chumbley and Balota 1984 that the effect of word. consuming revision of the output pattern if they were in meaning in lexical decision is attributable to a decision. congruent with the initial response Supporting this stage following lexical access We will elaborate our view. hypothesis all subjects chose the high frequency alter within the framework of a slightly modified version of. native in most trials while naming unvoweled consonant the two stage model of lexical decision performance pro. strings posed by Balota and Chumbley 1984 Briefly this model. This interpretation assumes that the enhancement of the assumes that letter strings words and nonwords differ. frequency effect for naming ambiguous words relative on a familiarity meaningfulness FM dimension The. LEXICAL AMBIGUITY AND WORD RECOGNITION IN HEBREW 21. value of a particular letter string on the FM dimension voweledhigh and low frequency alternatives of the am. is determined by its orthographic and phonological biguousconsonantstrings were compared It is possible. similarity to real words Strings with very high or very that the addition of vowels reduced the orthographic. low FM values are classifiedas words and as nonwords familiarity of the stimuli and at the same time guided. respectively during a first stage of the decisionprocess the retrieval of the meaning Therefore the computation. If the computed FM value is not extreme a secondstage of FM valueswas basedon both orthographic and phono. in which a more detailed analysis of the stimulus is ac logic semantic analysis Alternatively the FM valuewas. complished determines the decision Obviously the dis determined by the orthographic analysis and the mean. tributionof the FM valuesfor words and nonwords over ing wasdetermined during the second stage analysis The. laps The amount of overlap is related to the second possibility implies that lexical decisions for. discriminability between words and nonwords in the par voweled words require the complete two stage analysis. ticular stimulus list in many more cases than required for unvoweledwords. The relative contributions of the phonological semantic This hypothesis might also explain the observed differ. meaningfulness and orthographic familiarity to the com ence between voweled and unvoweled words Note. putation of the FM value was not specified by the two however that even if this explanationis true it does not. stage model In agreement with Balota and Chumbley invalidate our claim that for unvoweled words and par. 1984 we suggestthat the analyses of the orthographic ticularly for homographs the lexical decision is based. familiarity and of the meaningfulness of the stimulus over mostly on the first stage analysis whichdoes not include. lap in time to a great extent However the determination phonological disambiguation. of an FM value need not wait until both analysesare ex The only difference between the optional and the. hausted whenever enough information accumulates obligatory conditions was that the orthographic similar. regardless of its source orthographic or phono ity between words and nonwords was significantly larger. logic semantic a value is set The relative contribution in the obligatory condition If indeed the FM value is. of each type of analysisdependson the familiarity of the determined primarily by orthographic familiarity this. orthographic cluster and on the availability of its mean manipulation should have shifted the nonword distribu. ing The orthographic familiarity of homographs is en tion along the FM value to the right increasingits over. hanced because the cluster of consonants is encountered lap with the distribution of words Following the logic. in several semantic contexts on the other hand their underlying Balota and Chumbley s 1984 model this. meaning is ambiguous and therefore not immediately manipulation shouldhave forced the subjectsto increase. available Consequently we suggest that the FM value the uppercriterion abovewhich a word is accepted without. for homographs is based mostly on the orthographic further analysis Sincemost low frequency words are lo. rather than on the semantic phonological familiarity cated below the original criterion raising this criterion. Therefore whenever the lexical decision is based on the should have affected the high more than the low. first stage computation of the FM value it is done be frequency words and therefore the net effect should. fore the phonology and meaning of the homograph are have been an attenuation rather than an amplification of. disambiguated Furthermore we suggest thatthis strategy the frequency effect This trend was indeedobserved with. is employed for most unvoweled words since even for unambiguous words the frequency effects in the obliga. single meaning unvoweled words the phonology is not tory condition 42 msec were smaller than those in the. immediately available in print Empiricalsupportfor this optional condition 67 msec see Table 2 However the. last hypothesis was provided in a previous study Bentin same manipulation with ambiguous words yieldedoppo. et al 1984 6 site results the frequency effect in the obligatorycondi. The interferenceeffect of vowelmarks on lexicaldeci tion 201 rnsec was twice as large as that in the optional. sion suggests that whenvowels are present subjects can condition 102 rnsec Table 2 One possible explanation. not ignore them This hypothesis is strongly supported of this resultis thatvowel markshavea largereffectwhen. by the significant frequency effect observed when the they indicate a low frequency rather than a high frequency. Summary of Reaction Times in Milliseconds in tbe Lexical Decision and Naming Experiments. Ambiguous Words Unambiguous Words,Voweled Voweled Unvoweled Nonwords.
Hi Freq 1 0 Freq Unvoweled Hi Freq 1 0 Freq HI Freq 1 0 Freq Voweled Unvoweled. Lexical Decision Optional Condition,670 772 616 635 702 610 678 764 739. Lexical Decision Obligatory Condition,775 976 748 790 1 055. 669 768 653 674 690 634 671 843 757,22 BENTIN AND FROST. alternative of an ambiguous consonant string Thus a new The main difference between lexical decision and nam. aspect of the disambiguation process is disclosed while ing unvoweled strings is that naming cannot be performed. processing a letter string subjects might automatically unless the stimulus is phonologically disambiguated Be. generate possible lexical representations of the grapheme cause the print does not provide enough phonemic cues. When the letter string is voweled generation of phonol naming requires postaccess processes of disambiguation. ogy might be initially based only on consonants indepen Therefore in contrast to other languages in Hebrew. dently of the specific vowels employed because the sub whenever lexical decision requires second stage processes. jects have little experience with reading vowels and as for example when vowel marks are presented. because the consonants are visually more salient At some semantic ambiguity should affect both tasks in a similar. stage however the top down generated lexical candidate way This assumption is supported by the remarkable. which provides unequivocal meaning and phonology is similarity of the frequency effect and on absolute naming. confronted with the bottom up analysis of the vowels We time and lexical decision time in the optional condition. suggest that at this stage vowels that indicate a high On the other hand when lexical decisions can be based. frequency lexical alternative have a different effect than on an orthographically generated FM value naming is. those that indicate a low frequency word We assume that relatively delayed. top down generation of meanings is influenced by fre In conclusion we suggest that the results of this study. quency High frequency words are more readily available revealed that phonological disambiguation of Hebrew 00. and therefore are generated first in a sequential or voweled words does not occur prelexically Furthermore. cascade type process Therefore the lexical candidate that at least for Hebrew consonant strings it appears that lex. is first confronted with the bottom up vowel information ical decisions are based on the ambiguous orthographic. IS a high frequency alternative If the vowels indicate a information without reference to meaning or phonologi. different word the subject must reject his or her first cal structure We propose that multiple meanings facili. hypothesis and generate or at least consider another one tate lexical decisions by increasing orthographical. This hypothesis which is not basically different from the familiarity and that the decisions are therefore based on. postaccess inhibition model suggested for naming by For this factor alone These processes are best explained in. ster 1981 is also supported by the naming performance the context of a multistage model of visual word recogni. in Experiment 2 tion such as the two stage model proposed by Balota and. Chumbley 1984 with only slight modifications and ad. Stimulus Ambiguity and Naming ditions, In the naming task we were able to know which alter It is difficult to comment on the generality of these. native was chosen when an unvoweled ambiguous con hypotheses for languages other than Hebrew Recall. sonant string was presented The data revealed that the however that Jastrzembski 1981 reported that among. high frequency alternatives were indeed chosen by the words with an equal number of derivations and an equal. great majority of subjects We have no immediate expla number of meanings those whose meanings tend to be. nation for those few cases in which low frequency alter associated with only one derivation were responded to. natives were selected but we tend to believe that these faster Furthermore Chumbley and Balota 1984 re. selections were caused by coincidental circumstances vealed that lexical decision RTs and RTs in semantic tasks. such as unusual individual preferences or phonetic prim are closely related independent of other factors If in. ing by the previous random stimulus At any rate we con deed RTs in semantic association tasks and clustering of. sider the analysis of the overt responses as supporting the meanings around only one etymological derivation are. word frequency guided order of meaning generation for measures of meaning availability these results suggest. ambiguous letter strings that our findings in the deep Hebrew orthography are a. The effect of vowel marks on naming was very differ rather extreme example of processing ambiguity in printed. ent from their effect on lexical decision Unvoweled am words. biguous words were named significantly faster than the. low frequency alternatives but were named only 16 msec. REFERENCES, faster than were the voweled high frequency alternatives.
Unambiguous words were named 30 msec faster if they ANDREWS S 1982 Phonological recording Is the regularity effect. were presented without the vowel marks than if they were consistent Memory Cognition 10 565 575. voweled This difference was small relative to the inter BALOTA D A CHUMBLEY J I 1984 Are lexical decisions a good. measureof lexicalaccess The role of word frequency in the neglected. subject variability and therefore was not significant decision stage Journal ofExperimental Psychology Human Percep. However the direction of the difference conflicts with the tion Performance 10 340 357. results of Koriat 1984 One difference between the two BALOTA D A CHUMBLEY J I 1985 The locusof word frequency. studies is that Koriat employed only unambiguous words effects in the pronunciation task Lexical access and or production. Journal of Memory Language 24 89 106, We do not have a simple model to explain how this differ. BENTlN S BARGA N KATZ L 1984 Orthographic and pho, ence might have affected naming performance but it seems nemic coding for lexical access Evidence from Hebrew Journal of. to us that the reason for the discrepant results should be Experimental Psychology Learning Memory Cognition 10. related to this difference between the stimulus lists 353 368. LEXICAL AMBIGUITY AND WORD RECOGNITION IN HEBREW 23. BENTIN S KATZ L 1984 Semantic awareness in a nonlexical graphic entries in the intemallexicon Effects of systematicity and. task Bulletin oftlu Psychonomic Society 22 134 139 relative frequencies of meanings Journal of VerbalLearning II Ver. CARPENTER P A DANEMAN M 1981 Lexical retrieval and er balBehavior 10 57 62. ror recovery in reading A model based on eye fixation Journal of SIMPSON G B 1981 Meaning dominance and semantic context in. Verbal Learning II Verbal Behavior 20 137 161 the processing of lexical ambiguity Journal of Verbal Learning II. CHUMBLEY J 1 BAWTA D A 1984 A word s meaning affects Verbal Behavior 20 120 136. the decision in lexical decision Memory II Cognition 12 590 606 SIMPSON G B 1984 Lexicalambiguity and its role in modelsof word. FORSTER K I 1981 Priming and the effects of sentence and lexical recognition Psychological Bulletin 96 316 340. contexts on naming time Evidence for autonomous lexical process. ing QuarterlyJournal of Experimental Psychology 33A ui5 496 NOTES. FORSTER K I BEDNALL E S 1976 Terminating and exhaustive. search in lexical access Memory II Cognition 4 53 61 1 If the vowelmarksare presentedin conjunction withthe consonants. FREDERIKSEN I R KROLL I F 1976 Spelling and sound Ap usually placedbelowthe letters they disambiguate the phonology and. proaches to the internal lexicon Journal of Experimental Psychol in most cases unequivocally determineone word Thus in a fullyvowel. ogy Human Perception II Performance 2 361 379 izedsystem the reader mayusethe phonemic cues providedby the vowel. FROST R KATZ L BENTIN S in press Strategies for visual word marks to aid lexical access. recognition and 0I1h0graphical depIh A multilingual CODJIlIrison Jour 2 Note that the ambiguousconsonantalstrings in Lists E and F were. nal of Experimental Psychowgy Human Perception II Performance identical Therefore there were twice as many subjects who responded. JAMES C T 1975 The role of semantic information in lexical deci to unvoweled ambiguous words than to each of the voweled alterna. sions JournalofExperimental Psychology Human Perception II Per tives The RTs to unvoweledconsonantalstrings in Lists E andF were. formance I 130 136 not significantly different 616 msec and 621 msec respectively There. JASTRZEMBSKI J E 1981 Multiple meanings numberof relatedmean fore in order to obtain an equal number of observations in each cell. ings frequency of occurrence and the lexicon Cognitive Psychol we used only the RTs of the subjects presented with List E. ogy 13 278 305 3 Whenever appropriate both stimulus and subject analyses were. KINOSHITA S 1985 Sentencecontext effects on lexicallyambiguous performed and minF was calculated The stimulus analysis used a. words Evidence for postaccess inhibition process Memory II Cog within stimulus design and the subjectanalysisused a between subjects. nition 13 579 595 design, KORlAT A 1984 Reading withoutvowels Lexicalaccess in Hebrew 4 Initialanalysis revealedthat indeed the RTs to nonwordsin List A. In H Bouma D G Bouwhuis Eds Attentionand performance were similar to those in List B the RTs to nonwords in List C were. X Control of language processes Hillsdale NJ Erlbaum similar to those in List D and the RTs in List E were similar to those. KROLL I F ScHWEICKERT J M 1978 November Syntacticdis in List F. ambiguation ofhomographs Paper presented at the Nineteenth An 5 The RTs to ambiguous words in List E and List F were not sig. nual Meeting of the Psychonomic Society San Antonio Texas nificantly different 653 msec and 665 msec respectively For tech. MORTON I 1970 A functional model for memory In D A Nor nical reasons the RTs to unvoweled strings were collapsed across the. man Ed Modelsofhuman memory New York Academic Press differentphonological realizations that were indeedproduced However. MORTON J 1979 Word recognition InI Morton J C Marshall since the great majority of responses were high frequency words we. Eds Psycholinguistics 2 Structure and processes Cambridge MA assume that the average results are not biased significantly. MIT Press 6 Note however that according to our model stimulus analysis is. NAVON D SHIMRON J 1981 Does word naming involve not terIninated by the lexical decision As we suggested elsewhere Bentin. grapheme to phoneme translation Evidence from Hebrew Journal Katz 1984 words are exhaustivelyanalyzedto the deepest lexi. of Verbal Learning II Verbal Behavior 20 97 109 cal level provided that the task does not interfere with this analysis. NAVON D SHIMRON J 1984 Reading Hebrew How necessary Thus even thoughthe lexicaldecision was made the phonologicaldis. is graphemic representation of vowels In L Henderson Ed Or ambiguationcontinuesuntil at least one and possibly all meaningsare. thographies and reading Hillsdale NJ Erlbaum accessed. RUBENSTEIN H GARFIELD L MILUKAN J A 1970 Homo, graphic entries in the internal lexicon Journal of Verbal Learning. II Verbal Behavior 9 487 494 Manuscript received December 27 1985. RUBENSTEIN H LEWIS S S RUBENSTEIN M A 1971 Homo revision accepted for publication July 21 1986.

Related Books

CONVEYOR BELTS SAFETY PROCEUDRE - Tata Power

CONVEYOR BELTS SAFETY PROCEUDRE Tata Power

Conveyor Belt Safety Procedure TPSMS/GSP/CONV/002 REV 01 Date of Issue: 30-06-2016 6.1.3 Start/Stop of belt 1. If a conveyor belt needs stopping, it should never be stopped in load condition, except emergency. 2. If the high speed coupling in the drive is a fluid coupling, the number of start/stop of the conveyor belt should not be more than

SUPREME COURT OF THE STATE OF NEW YORK APPELLATE DIVISION ...

SUPREME COURT OF THE STATE OF NEW YORK APPELLATE DIVISION

N.Y.3d 902 (2015). The New York County Supreme Court refused to issue the order to show cause in Tommy II, in part, because it believed itself bound by Lavery, stating 2 CPLR Article 70 governs the procedure applicable to all writs of habeas corpus.

THE BRIDGE - tooleybuc-c.schools.nsw.gov.au

THE BRIDGE tooleybuc c schools nsw gov au

Lucy F Saoirse 4 Alister F Summer 4 Riley H F Emma E 5 Jakson F Dallas 5 Sienna 1 Tara 5 Felicity 1 Amy 5 Sara 6 Toni Mia 6 Up Coming Events Term 1 2015 Week 2 Timetable Week 2 Wed 29 May Breakfast Club Thu 30 Apr ATHLETICS CARNIVAL Week 3 Timetable Week 1 Accountancy Week Tue 5 May Parent Meeting 10:00 am L Bibby Wed 6 May Breakfast Club Thu 7 May Tooleybuc Cross Country Fri 8 May Primary ...

CHAPTER II CYBER CRIME AND ITS CLASSIFICATION

CHAPTER II CYBER CRIME AND ITS CLASSIFICATION

CYBER CRIME AND ITS CLASSIFICATION 1. Introduction Cyber crime is not an old sort of crime to the world. It is defined as any criminal activity which takes place on or over the medium of computers or internet or other technology recognised by the Information Technology Act. Cyber crime is the most prevalent crime playing a devastating role in Modern India. Not only the criminals are causing ...

Cyber Forensic Tools: A Review - IJETT

Cyber Forensic Tools A Review IJETT

Cyber Forensic Investigation includes the Capture & Analysis of digital data either to prove or disprove whether the internet related theft has committed or not. Earlier Computer are used only began to look at computers as sources of for storing large volumes of data & perform many operations on it ,but now a days it has expanded & occupied prior role in Crime Investigation. In order A

Edition 2.0 2006-12 INTERNATIONAL ... - webstore.iec.ch

Edition 2 0 2006 12 INTERNATIONAL webstore iec ch

IEC 61025 Edition 2.0 2006-12 INTERNATIONAL STANDARD NORME INTERNATIONALE Fault tree analysis (FTA) Analyse par arbre de panne (AAP) INTERNATIONAL ELECTROTECHNICAL COMMISSION COMMISSION ELECTROTECHNIQUE INTERNATIONALE XA ICS 03.120.01; 03.120.99 PRICE CODE CODE PRIX ISBN 2-8318-8918-9 This is a preview - click here to buy the full publication

ENVIRONMENTAL SCIENCE - Amazon S3

ENVIRONMENTAL SCIENCE Amazon S3

A Global Concern. ENVIRONMENTAL SCIENCE. A Global Concern. 14e. About the Cover. Coral reefs are among the most magnificent biological communities on Earth. They rival tropical rainforests in beauty and diversity. About one . quarter of all marine species spend some or all of their life cycle in the shelter of coral reefs, and in some areas, reefs provide three-fourths of the protein in human ...

DRAFT Red Hill Elk Management Plan - Boulder County, Colorado

DRAFT Red Hill Elk Management Plan Boulder County Colorado

The Red Hill elk is a sub-herd of the St. Vrain elk herd residing in northern Boulder and southern Larimer counties. This Red Hill herd was studied by Colorado Parks and Wildlife (CPW) and BCPOS from 1997-2005, around the time of purchase and opening of Heil Valley Ranch open space. The study helped identify important habitats in both summer and winter range, movement corridors, and the timing ...

AWS Billing and Cost Management - User Guide

AWS Billing and Cost Management User Guide

AWS Billing and Cost Management User Guide Related services Analyzing Costs with Cost Explorer The AWS Billing and Cost Management console includes the no-cost Cost Explorer (p. 37) tool for viewing your AWS cost data as a graph. With Cost Explorer, you can ?lter graphs by values such

IBM Tivoli Monitoring now available on Amazon EC2

IBM Tivoli Monitoring now available on Amazon EC2

IBM Tivoli Monitoring now available on Amazon EC2 Welcome You have joined early We will start at 11:00 am US Eastern Time We will record this event for replay A PDF of the presentation will be sent after the call Thank you for joining Audio conference: North America dial: 1-866-457-2759. International dial: 1-719-387-5502. Passcode 2522296972