242App<strong>en</strong>dice INotwithstanding, a nontrivial question regarding falsifiability can be answered here : Is the MROM-Pfalsifiable, at all? Giv<strong>en</strong> the repeatedly appearing critique of connectionist models as being too powerful, andtherefore not falsifiable in any easy way, this is not an idle question. Following other theoreticians, wetherefore have proposed that any A-type model should come with a clear answer to the following question :What cannot be or happ<strong>en</strong>, if the model is correct? In other words, which effects or ph<strong>en</strong>om<strong>en</strong>a does themodel exclude? An example is giv<strong>en</strong> in Grainger and Jacobs (1996). They show that if the MROM is correct,a facilitatory effect of orthographic neighborhood d<strong>en</strong>sity (as measured by Coltheart's N) is possible inboth the yes/no and go/no-go variants of the LDT, but not in the perceptual id<strong>en</strong>tification task. Thus, replicableexperim<strong>en</strong>tal demonstrations of a facilitatory N effect in the perceptual id<strong>en</strong>tification tasks would falsifythe MROM.Another example is giv<strong>en</strong> above (see Figure 8). According to the MROM, as included in the pres<strong>en</strong>tMROM-P, a phonological pseudohomophone effect in the LDT is not possible, because it possesses nophonological processing units whatsoever. Thus, although the MROM was explicitly <strong>des</strong>igned to deal withorthographic processing in the LDT and other reading tasks that do not include pseudohomophonic stimuli,the simulation data in Figure 8, for example, pres<strong>en</strong>t a falsification of the (non-phonological) MROM. Evid<strong>en</strong>tly,simply falsifying a model by using conditions that are outside of its explicitly stated validity space(domain of application) is not necessarily useful. As we have demonstrated in this chapter, using theMROM as a null-model against which to test models of phonological coding is a more useful variant of"falsification studies".At any rate, in a field that lacks universal laws, we cannot expect models to have universal validity (cf.Newell, 1990). On the other hand, we can hardly want to continue with models that can accurately accountonly for a single effect, as measured by a single variable in a single task, but whose validity stops there (cf.Jacobs & Grainger, 1994 ; Newell, 1990 ; Roberts & Sternberg, 1993).The MROM-P is also falsifiable in several nontrivial respects. Like the MROM, it allows to makequalitative predictions that can be tested in a straightforward way. An example is discussed in Ferrand andGrainger (1996). They used a pre-quantitative version of MROM-P -they called it a bimodal ext<strong>en</strong>sion of theMROM- to make qualitative predictions concerning the exist<strong>en</strong>ce and direction of priming effects in amasked priming LDT manipulating prime type (homophones, pseudohomophones, or unrelated controls)and list composition (pseudohomophones, legal pseudowords, or illegal nonwords). The strongest qualitativeprediction of MROM-P, i.e., the one most easily falsifiable, was that it predicts a null effect with homophoneprimes in the pres<strong>en</strong>ce of illegal nonwords 9 . The rationale for this is that i) the pres<strong>en</strong>ce of illegalnonwords <strong>en</strong>courages participants to use the ∑ criterion, since such nonwords can easily be discriminatedfrom words on the basis of summed lexical-orthographic activity. ii) homophone primes g<strong>en</strong>erate high levelsof orthographic inhibition wh<strong>en</strong> read-out is from the orthographic M criterion. The facilitatory effects<strong>du</strong>e to increased use of the ∑ criterion (i.e., the fast-guess mechanism pro<strong>du</strong>cing decreases in RT) will becanceled by the inhibitory effects <strong>du</strong>e to homophone primes. A null-effect is the predicted result. In Ferrandand Grainger's (1996) experim<strong>en</strong>t, this was the case.DISCUSSION AND OUTLOOKTo summarize : While a complete, criteria-ori<strong>en</strong>ted evaluation of the MROM-P is not possible at pres<strong>en</strong>t,the results of our partial evaluation stand the g<strong>en</strong>eral test criterion that we had fixed as our objective,that is, whether the pres<strong>en</strong>t MROM-P is an appropriate "prototype" for developing a g<strong>en</strong>eral model ofphonological coding in visual word recognition. The model pres<strong>en</strong>ted here is definitely a prototype, not inthe s<strong>en</strong>se of repres<strong>en</strong>ting an ideal, but in the s<strong>en</strong>se of being a "working model". If one accepts the principlesof model developm<strong>en</strong>t we adhere to, it has some virtues. Within the constraint of nested modeling, it repres<strong>en</strong>tswhat we think to be the simplest possible localist connectionist network that allows an account oftwo?? critical empirical effects indicating the influ<strong>en</strong>ce of phonological processes in what is still the mostwidely used reading task in experim<strong>en</strong>tal psychology and psycholinguistics, i.e. the LDT.Moreover, the MROM-P, as our other work involving A-type modeling, is ess<strong>en</strong>tially a heuristic devicein the s<strong>en</strong>se discussed in Grainger and Jacobs (pres<strong>en</strong>t volume) : It provi<strong>des</strong> a heuristic, algorithmic <strong>des</strong>criptionof phonological coding, but -needless to say- it falls short of pres<strong>en</strong>ting a computational theory in thes<strong>en</strong>se of Marr (1982). It is not difficult to admit this : Few theoreticians in the field of cognitive sci<strong>en</strong>cehave achieved (or come close to) a computational theory (Marr, 1982 ; see also Jacobs, 1994 ; Pylyshyn,9 A word on null-effects and their significance for theory building is in order, because many psychologists arefirm believers in the virtues of null-hypothesis testing (but see Giger<strong>en</strong>zer & Murray, 1987 ; Rouanet, 1996 ;Van Ord<strong>en</strong>, Aitchison, & Podgornik, 1996). Wh<strong>en</strong>ever one possesses models permitting quantitative predictionswith reasonable precision, the prediction of a null-effect is actually a strong prediction to make. Perhapsthe most famous example is the prediction of the null-effect concerning the speed of light in Michelsonand Morley''s experim<strong>en</strong>ts by Einstein's special theory of relativity (Spielberg & Anderson, 1985). This isnot to say that any known psychological A-type model can be compared with Einstein's theory. We simplywant to make clear that the exist<strong>en</strong>ce of theoretical tools allowing quantitative predictions concerning empiricaleffects frees us from the use of null-hypothesis testing as exclusive infer<strong>en</strong>tial method. Thus, contraryto standard practice, accepting the null-hypothesis can become a valid infer<strong>en</strong>ce wh<strong>en</strong>ever one has suffici<strong>en</strong>tfaith in the validity and precision of a formal model or theory.
App<strong>en</strong>dice I 2431989), and it can be argued that the fields of memory or reading research are still not fully prepared for suchan <strong>en</strong>terprise (Humphreys, Wiles, & D<strong>en</strong>nis, 1994 ; Jacobs, 1994). This is not to say, however, that theyare not set up for some fair amount of theoretical unification, as argued below.Neither the rec<strong>en</strong>t special section of the JEP :HPP (1994) on modeling visual word recognition, nor anyother literature we have come to know since provi<strong>des</strong> a computational model that can formally account forthe pseudohomophone and bi-directional consist<strong>en</strong>cy effects in the LDT we have simulated here. However,there is, at least, one other model that has the pot<strong>en</strong>tial to provide such an account, the <strong>du</strong>al route cascadedor DRC model (Coltheart & Rastle, 1994). Giv<strong>en</strong> the fact that both the pres<strong>en</strong>t model and the DRC belongto the family of localist connectionist models (and therefore are more easily comparable), but differ with respectto one crucial structural feature -the pres<strong>en</strong>ce of a GPC rule mechanism in the DRC- we may have thepromise of some exciting strong infer<strong>en</strong>ce studies ahead. However, for such studies to be effici<strong>en</strong>t, certainmethodological issues have to be solved, as argued below.Future work concerning the MROM-P will involve adding phonological repres<strong>en</strong>tations to multilingualmodels such as the bilingual interactive activation model pres<strong>en</strong>ted by Dijkstra and Van Heuv<strong>en</strong> in the pres<strong>en</strong>tvolume. This raises the interesting question of how multilingual speaker-readers deal with the differ<strong>en</strong>tsets of spelling-to-sound correspond<strong>en</strong>ces in each language. If one postulates, as in the pres<strong>en</strong>t work, thatthe phonological coding compon<strong>en</strong>t of the MROM-P is automatic and strategically non-modifiable, how dobilingual readers deal with the pot<strong>en</strong>tial interfer<strong>en</strong>ce caused by automatically g<strong>en</strong>erating all correspond<strong>en</strong>cesin both languages? The notion of language node intro<strong>du</strong>ced by Grainger and Dijkstra (1992, see also Dijkstra& Van Heuv<strong>en</strong>, pres<strong>en</strong>t volume) provi<strong>des</strong> one solution to this problem. Top-down inhibition from the unatt<strong>en</strong>dedlanguage node to the corresponding word units (i.e., all words in the unatt<strong>en</strong>ded language) wouldblock resonance betwe<strong>en</strong> irrelevant phoneme units and word units in the unatt<strong>en</strong>ded language. Clearly muchexciting theoretical and empirical work is yet to be done in the multilingual domain.GOING BEYOND MROM-P : A GENERAL CHALLENGE FOR COGNITIVE MODELERS IN"WORD-NERD" WORLD"Sci<strong>en</strong>ce, ev<strong>en</strong> more profoundly than politics, is the art of the possible. It does only what can be don<strong>en</strong>ext" (Newell, 1990, p. 26).The 1994 special section of the JEP :HPP <strong>en</strong>titled "Modeling visual word recognition" gives an impressionof the empirical and theoretical richness of this classical field of experim<strong>en</strong>tal, cognitive psychology.The question we want to raise here is whether it is about time for some serious effort of theoretical unification?Who thinks that this is an idle question is invited to take a look at Table 1 in our editorial of that specialsection (Jacobs & Grainger, 1994), which gives a selective overview of 15 models of visual word recognitionstarting with Morton's (1969) logog<strong>en</strong> model. Originally, this piece of taxonomic work startedwith the ambition to give a synopsis -as complete as possible- of models of word recognition to be used forthe tasks of theoretical unification and developm<strong>en</strong>t of standards for model comparison and evaluation. Thatwas clearly too ambitious! Ev<strong>en</strong> the published version is far too complex to have a fair chance of being usedin the way we wanted it to (previous versions of the table, including more than 40 differ<strong>en</strong>t models of thepast 30 years were worse!).Notwithstanding, we continue to think that the answer to the above question is a clear "yes". Perhapspressure for theoretical unification in the word recognition literature is not as high as in the g<strong>en</strong>eral field ofcognitive sci<strong>en</strong>ce. As Newell (1990, p. 25) states : "In my view, it is time to get going on pro<strong>du</strong>cing unifiedtheories of cognition-before the data base doubles again and the number of visible clashes (i.e., betwe<strong>en</strong>theory and data) increases by the square or cube". On the other hand, if Newell (1990) can convincingly pres<strong>en</strong>tfour harbingers of unified theories of cognition, why should we not be able to agree on a limited numberof harbinger models that have the pot<strong>en</strong>tial to become unified models of word recognition? After all, weare only dealing with a small part of the cake of cognition (although perhaps with one of the most complexparts). Clearly, we t<strong>en</strong>d to think that models of the steadily growing IA family (see pres<strong>en</strong>t volume) areamong the harbinger candidates, and since we have giv<strong>en</strong> our argum<strong>en</strong>ts for this in Jacobs and Grainger(1994), we will not reiterate them here (see Grainger a Jacobs, pres<strong>en</strong>t volume). Suffice it to say that thepres<strong>en</strong>t results showing the g<strong>en</strong>eralizability of the original IAM to conditions involving phonological processingare clearly in favor of our view.However, to facilitate theoretical integration in the "word-nerd" field, the following chall<strong>en</strong>ge has to bemet : "Agree on a minimal set of standards for model comparison and evaluation". This chall<strong>en</strong>ge has twofacets. First, it involves agreem<strong>en</strong>t on a minimal set of standard effects and tasks, any model of word recognitionthat competes for "harbingership" should be able to predict in a way that can <strong>en</strong>ter into strong infer<strong>en</strong>cecompetition. In Jacobs and Grainger (1994 ; Table 1) we made a minimalistic proposal of four sucheffects. Whether they were the right ones, or whether they have to be augm<strong>en</strong>ted by other effects is a questionthat can only be solved by ongoing published sci<strong>en</strong>tific debate.The second, and more problematic facet of this chall<strong>en</strong>ge is to agree on a minimal set of criteria formodel evaluation, and on a standard way of applying them. Among other things, this implies the trickyproblem that complex A-type models have to be made comparable, at least on a number of critical featuresor dim<strong>en</strong>sions (e.g., the "curr<strong>en</strong>cy" problem discussed by Massaro & Friedman, 1990). This is a nontrivialissue (cf. Estes, 1975), as model builders who have problems of keeping differ<strong>en</strong>t variants of their ownmodel comparable over the years of developm<strong>en</strong>t will admit (for some examples, see Jacobs & Grainger,
- Page 1:
UNIVERSITÉ DE PROVENCE, AIX-MARSEI
- Page 4 and 5:
Ce travail clôt un cycle de format
- Page 6 and 7:
CHAPITRE 7 : LE FUM . . . . . . . .
- Page 8 and 9:
8IntroductionPour cela, notre domai
- Page 10 and 11:
10Introduction• au niveau lexical
- Page 12 and 13:
12Introduction• sa forme visuelle
- Page 14 and 15:
14IntroductionAprès avoir posé le
- Page 16 and 17:
16Méthodologiespulations sur les i
- Page 18 and 19:
18Méthodologies2.1. Protocoles exp
- Page 20 and 21:
20Méthodologiessi le stimulus se t
- Page 22 and 23:
22MéthodologiesCertaines études t
- Page 24 and 25:
24Méthodologiestes, on obtient des
- Page 26 and 27:
26Méthodologies1996 ; Peter & Turv
- Page 28 and 29:
28Méthodologiesles performances da
- Page 30 and 31:
30Méthodologies6 %8%10%15%30%50%80
- Page 32 and 33:
32MéthodologiesMatériel expérime
- Page 34 and 35:
34Méthodologiesentraîne le masqua
- Page 36 and 37:
36MéthodologiesLe même résultat
- Page 38 and 39:
38Méthodologies120100Situation Sta
- Page 40 and 41:
Chapitre 3Orthographe et phonologie
- Page 42 and 43:
42Orthographe et Phonologie3.1. Var
- Page 44 and 45:
44Orthographe et PhonologieLa Figur
- Page 46 and 47:
46Orthographe et PhonologieJacobs,
- Page 48 and 49:
48Orthographe et Phonologiedans la
- Page 50 and 51:
50Orthographe et PhonologieDans l
- Page 52 and 53:
52Orthographe et Phonologieteurs du
- Page 54 and 55:
54Orthographe et PhonologieGoldstei
- Page 56 and 57:
56Orthographe et Phonologietion est
- Page 58 and 59:
58Orthographe et Phonologierand, 19
- Page 60 and 61:
60Orthographe et Phonologieplus ad
- Page 62 and 63:
62Orthographe et Phonologie3.2.3.1.
- Page 64 and 65:
64Orthographe et PhonologiePlus ré
- Page 66 and 67:
66Orthographe et PhonologieUne autr
- Page 68 and 69:
68Orthographe et Phonologiedeux var
- Page 70 and 71:
Chapitre 4Modèles de la perception
- Page 72 and 73:
72Modèles de la perception visuell
- Page 74 and 75:
74Modèles de la perception visuell
- Page 76 and 77:
76Modèles de la perception visuell
- Page 78 and 79:
78Modèles de la perception visuell
- Page 80 and 81:
80Modèles de la perception visuell
- Page 82 and 83:
82Modèles de la perception visuell
- Page 84 and 85:
84Modèles de la perception visuell
- Page 86 and 87:
86Modèles de la perception visuell
- Page 88 and 89:
88Modèles de la perception visuell
- Page 90 and 91:
90Modèles de la perception visuell
- Page 92 and 93:
92Modèles de la perception visuell
- Page 94 and 95:
94Modèles de la perception visuell
- Page 96 and 97:
96Modèles de la perception visuell
- Page 98 and 99:
98MROM-pspécifier leur lien avec l
- Page 100 and 101:
100MROM-pphonèmes reliés par un r
- Page 102 and 103:
102MROM-pLorsque le modèle génèr
- Page 104 and 105:
104MROM-pque ce système artificiel
- Page 106 and 107:
106Unités de la lecturelinguistiqu
- Page 108 and 109:
108Unités de la lecture22606TR (ms
- Page 110 and 111:
110Unités de la lecturemes. Aussi
- Page 112 and 113:
112Unités de la lecturephonologiqu
- Page 114 and 115:
114Unités de la lectureelle-même
- Page 116 and 117:
116Unités de la lecture6.3. Expér
- Page 118 and 119:
118Unités de la lectureRead est qu
- Page 120 and 121:
120Unités de la lectureces modèle
- Page 122 and 123:
122Unités de la lecturechapitre su
- Page 124 and 125:
124FUMmultiples existant au sein de
- Page 126 and 127:
126FUMpar Berndt, Lynne D'Autrechy
- Page 128 and 129:
128FUMcessus de compétition et du
- Page 130 and 131:
130FUMgène et suit les principes c
- Page 132 and 133:
132FUMPseudohomophonesContrôles Or
- Page 134 and 135:
134FUM61023TR (ms) Seidenberg et al
- Page 136 and 137:
136FUMportementaux et les résultat
- Page 138 and 139:
138FUMà une entité extérieure au
- Page 140 and 141:
Chapitre 8Des prédictionsau niveau
- Page 142 and 143:
142Des prédictions au niveau des m
- Page 144 and 145:
144Des prédictions au niveau des m
- Page 146 and 147:
146Des prédictions au niveau des m
- Page 148 and 149:
148Des prédictions au niveau des m
- Page 150 and 151:
150Des prédictions au niveau des m
- Page 152 and 153:
152Des prédictions au niveau des m
- Page 154 and 155:
154Des prédictions au niveau des m
- Page 156 and 157:
156Des prédictions au niveau des m
- Page 158 and 159:
158Des prédictions au niveau des m
- Page 160 and 161:
160Des prédictions au niveau des m
- Page 162 and 163:
162Des prédictions au niveau des m
- Page 164 and 165:
164Des prédictions au niveau des m
- Page 166 and 167:
166Les mots polysyllabiquesmots mon
- Page 168 and 169:
168Les mots polysyllabiquesTableau
- Page 170 and 171:
170Les mots polysyllabiques9.2. Exp
- Page 172 and 173:
172Les mots polysyllabiques19001890
- Page 174 and 175:
174Les mots polysyllabiquesnexe XI
- Page 176 and 177:
176Les mots polysyllabiques9.4. Dis
- Page 178 and 179:
178ConclusionConclusion« La grande
- Page 180 and 181:
180Conclusionplutôt un système o
- Page 182 and 183:
182Conclusiontester les prédiction
- Page 184 and 185:
184BibliographieAderman, D., & Smit
- Page 186 and 187:
186BibliographieBrysbaert, M., Vitu
- Page 188 and 189:
188BibliographieFerrand, L., Segui,
- Page 190 and 191:
190BibliographieGrainger, J., & Jac
- Page 192 and 193: 192BibliographieKay, J., & Bishop,
- Page 194 and 195: 194BibliographieMewhort, D. J. K.,
- Page 196 and 197: 196BibliographiePerea, M., & Pollat
- Page 198 and 199: 198BibliographieSeidenberg, M. S.,
- Page 200 and 201: 200BibliographieTreiman, R., & Zuko
- Page 202 and 203: 202AnnexesAnnexes
- Page 204 and 205: 204AnnexesAnnexe II : Temps de rép
- Page 206 and 207: 206AnnexesAnnexe IV : Temps de rép
- Page 208 and 209: 208Annexesd t 88 3 fixed fI%tdd d 3
- Page 210 and 211: 210Annexesoo u 3395 57 too tuoo $ 5
- Page 212 and 213: 212AnnexesAnnexe VI : Liste des 120
- Page 214 and 215: 214AnnexesAnnexe VII : Détail des
- Page 216 and 217: 216AnnexesSujets Stim. Erreur Sujet
- Page 218 and 219: 218AnnexesSujet Moyenne % Err ES Su
- Page 220 and 221: 220AnnexesAnnexe X : Stimuli employ
- Page 222 and 223: 222AnnexesAnnexe XI : Stimuli emplo
- Page 224 and 225: 224Appendice IMROM-P : An interacti
- Page 226 and 227: 226Appendice Ilanguages (French, Ge
- Page 228 and 229: 228Appendice Isummed frequency of f
- Page 230 and 231: 230Appendice IClearly, the ability
- Page 232 and 233: 232Appendice Ito an orthographic le
- Page 234 and 235: 234Appendice IFigure 6 gives an ill
- Page 236 and 237: 236Appendice Iestimator set study,
- Page 238 and 239: 238Appendice ISTEP 3. CRITERION SET
- Page 240 and 241: 240Appendice Iteractive processes o
- Page 244 and 245: 244Appendice I1994). Our stratagem
- Page 246 and 247: 246Appendice ILass, U. (1995). Einf
- Page 248 and 249: 248Appendice IAPPENDIXA1. Cleaning
- Page 250 and 251: 250Appendice IIA phoneme effect in
- Page 252 and 253: 252Appendice IIAs shown in Table 1,
- Page 254 and 255: 254Appendice IImay depend on their
- Page 256 and 257: Appendice III
- Page 258 and 259: 258Appendice IIIHowever, an unsolve
- Page 260 and 261: 260Appendice IIIgrapheme compared t
- Page 262 and 263: 262Appendice IIIRead, J. D. (1983).
- Page 264: 264Appendice IIIAppendix AMean resp