>  !"#$%&'()*+,-./023456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[\]^_`abcdefghijklmnopqrstuvwxyz{|}~RdO)ĺM@Pictures ZPowerPoint Document(1SummaryInformation(LǮ.Cseq8ԭD lw/V1:ي3 yBK3bE$e’w?=*mӀAl[ ]MM(wdN*I3aA3O ]x <&F#Ӻ6|nrs®<_?sa . k֙Gv׉sv8w& [i9mYp ܛ:0 Xb*m&}nYkڮ)l‰&F{c::D PqkFh<,p:WȊUxU-WOX: s*~5 QAι I$,O"B|HUdKDՑWo`^( yH$G^$R=Xp|ZcpHӱ !/ Gޕ==e ϐj\Zqq ߳aJ)"~~+;vm+>ƲׁʬL]gf-w"0? 03Z2T;>I^mTT*T1P:Yg6Ɏ3{x>ߤqހڃW{ڜ󗀗ަw}n=𩛃f$?~gf~髫6 8&U}/ s3soXU**WBT ooo)MtE|Q(?R1q4O>'O>Q%'O>">'|D|]O>">௔7(_/«U#O>F'GG|!5|DN>">͕G>௔g |GOϐ}F(_T}Q(^$|D|ڷDҊG'H!|$|N>:Q|D|Ҋ(Ҋ'E-c讙">௔="|tU#O>z|Ҋ߁'轚,'|vi| > ҊG {ڷҊ'\$' w۸('B|yyCN/_/1|SE>*:3*:3*:3*>/qAAATTKMTKMTKMTKMTK.dlT+.^~ލH~8/N JRu<1krWM*rT6Z[ /R5AΞ,di{eʺucxiԝk'p|ܝNuZknʙZ[8^4[s _mT&kO8fWe4O|rWi“V~inC5i}]]ҡ;/'%V'guY k_ݼd-WSгý\MSZz>mIj);6qszH/2ΖəܙjSZȺ؏L}|魯d9MכmYw}n斩0vUt~þ/^Yu' M\'~Ϻ^|Պ T&P҆Y0W>hlQĨ\ܧ&x}y\ygq;A'q3I&CI&8qgf3OrsN&9@bG" ,XB+C٥^+-U{ΖMXʹV}Un-UqOQ??tuή[~LMQum㥐ҕ#sԓ*Tfә|5 ='җ}!_ڗOҗ;Ψt6<}<ߔmW~.ްU^jFK;ΫLyСl>+Be6e]RZKUQw„ ߠW rR-Ju>f.c@x/*GV6]T1 ٙj>Jʢ5yp` cImRF|8 nq "H;#"i"W& vO:Kܡ/ y>J1 n2UZ*eS@'yGh9h<$:_Af]d&vlƌA,d'I^e]Q +HPė/AZFb*k ӏ GUM@ZS^e' ZEm"^TrGMbMZAHCmy"ݱ-r"GK$7cqeF&¦UÕ Fє > G)'bx(L{]NmUݢ@Y4. hTW &;iTY '˧W{cl*3)1OeMb^5*: pAA_Ҍ s{ $f\ )dZPmIILOi$%'R}k #6@Uԓ>Jע۷x`ZE-hmH!8Xoɀoаǚ 4%?)cLB;)Eϫ ك)5&Y@P]螈'R\"ph1ʽ ~(,ۍ0C 1u71ao4-\)7uL>K?(?o xi4ㆤE׀-Iuiw(uG xC0^<]kRz"? FO`_^w+xpR2|QEhk^a=Aψ^~څSRqpbS&qἡǦJ]jcK[tGBou;?._W'ڨ詝n0Ν2DeZ:D\p]U FY`>LkY˜vL?w>##x7 :? 4t~-$𕱵=̳֦C i<kAu[AwGc.:@IМtn& /n xy|eMJATl"JQGҬx QQ/TZQ+Rkm1c1ua:y<Ը&l6&I;s՚>3Ժ|֢T9ϼ\'Γ.qXrՒDac}c׋q!Hm/*B*E5fgn[Jut}>ѣTg'I=):8$KCd}n B6o-+Hg_8ٛxpx8|.㿉C'.O )iJt Q*p5|4x&< <| xXqs'οyۡ  'O/7O>|LYĦ/n |!"K/_n_|j5kW|fJ-[o|n={-?~0x x-:m''߿g7? ~"x %W_o~&x-vN;.?w? Sg_ҏ{_{_ɚĚNn=Կ{]n0z{?}~/7p;8 ėE_` 7X+AVC>a0GA ?=s"Oف00t @(͖#mT;kZ2nMQFA;BoI2uՓ;IE ЩJS dnJU>f4~I׃cȥ6ߓP*[@YTCt!>:W(Qrz¾6R6_*#G'{ n3VQk8gYV;:[x%kiخ)izWsl\Z&Ͽ7hiEQWgo8;ZګTz,ۥ;%/ęրW7M&nuZﵛjA D ,{urAs٪%U=f'h@ث =T4U}ᗢWꎒ頻Ŵ%/)qt~`6+?1B/B:ΧBbe}D6)>?#Ҍ/\oQ@yoNGlƝyl4fӳ%k+C9[( M:| CIEb_p<Xe<'A/۬C4 AeXa~&v0G#H=qƣoiXcL=DH['G.ua-܇p X?n0JFoVM_er im02|^hqzQi * c%зD&'a}I}9 DAF4&}Yb*[|;!gpDM*w$'Dc&R)td`ޙ#8^6nUО#aHW(02Zpzէ!>ࠥXD UZAtmJFjFgϽZ;|RUOb(C8mm3tYKcVPIU0MUA/F;r346fv0ouK"w<ξa8sF3:u;)#- 's^R̪3a:ې%KB6ש }d~jû|T?V耇4[G PI;?fjtHƙ{@V$QU \mJI#hW|k)©" n(hmmVVjiQS)AQ"G n(K--nJTs ȶ$R,˒n剎1ʯ.J+'B9l=$OgѤ 3EE@:r^.D_];='"q^dea,Ռn*-,q T>r` IŢe5 nx xUMlTU{ ""* pjaNk`Z&4` )mCQib$1])F L# <]M4Z 7'dlp#D=YG2;0=YKk^ՓxrjQƝ AZMyE}ov-Fϯш- +/3 86" 9ܥ}Ы)Ovq9Ǵ ]i ΐqD(5"fس>㧋)*vk[`J [4f=b`z4Oo.f%<{c*cӴ<(+4SUs_]qEy\?X$ ~/eԻ)n*V {w)*ڡ 3f5iw>U7z"MW~VHخ#ez}CDU0Ҁ&-~$(O* WVQYS%zjǤp*}n ~!TWٺ>NyJ驎;7zq7o74ꐴbS=;UtGvp}ZWw?ӭ;VQ‘םNO^hþ^5z=թ^\(fB in.tzA93z!ˉ9ZfGx*7-^#r+ TZ:{80 ys@Gx] $UuY@GqC4f4QiW *Q4nA-&&1!FTTD\6cTM q"!8􋩙?Թ]3ëi};暏TUuR{ߨkb-<Ekur_i͢d;㓸Sr;ɑ;l[yq}iҚʹMqtH_r"=HV Q8%~U&~JYKiDI)w&~*r ]I?5w#ӧC'^J I8h/Dk+_%~=)<ćΧ?`9*8^y()#Z+HRA 맿?G Q?5A ?V[9܂8RO H8TDH)#?Ṟ=O#EI)$"8z)&wH)9ţ(m<.)Rc)O HQBR_D [%H))H 2mCV)$H5)QmD^KmĝcRuJ Hŝ3Z5N"/H=)I ;b?%M RVGA'EM™S L4fRB [I8)󻤊I ;HҗNR EtM sHj!>jţ#uAVS 8K⨖pQ}ڑR6V7ֳ1"6*A@vWdd9 /E D@`0 "O S}Vr! C"G[] H3$VWNRGhG0TPyĘIס!wG>sAƑNfdL UZy  :be&UgFVo P0N>SJ`;nϟ<0'\.{yp@(QHW#5\"FbE8>[@=br~{Gܩk @ ԡ9/%CTWFd́5 4f!~z3gku 8\KDɅE옆eC}'j9h"3L9j@+j`h2cclKa {6e\ZrO9R-QPC_nsj[f*Q Z+8bP&}@Hfn;~dh2 %Z1@X!834U_ Nx uVITf=i*3}NC_ QQV!h|P4:nT[¥Wc4`$H筈BxeUb|h Ygb]ӢtJHnnT.΁?"ڠe*0 'UF.Ag#TU}K8#=r1'A{L};f𦁀hn]א#-\ .f3gfzޢSHauT̓[Z)a2dQ%0yt3\FItfjfLt8ք49WAa􃏄>IikfPW6EdhʦnXq!fYbI35'6Lַ䒩,t[).wWpxg>Kʉr^-'3{@~|! VL?Q7h7b4ίͮbcn:" f"9+Y=*K5(3AU5mso؟' Yf*T3!**}vrM49N #:` ҩce0ھ4=jA:갽A̋8K#mLlMCjd2‹fXUh2)i khO[ulߋ39peR4gG7Kyb,wq+9_ XG[3 #`Wdc L+d3uM1.ӽ;|SYc-b"uK˾}ƄΓ}"k,#)6){ DOf^{Fbzh'Ĉ`ExQ0w@҃cFmbvS=WG4=r#?zpGL3t|їl)^82ghJpY a?Ω3 ߉zc>k4NjI-k} |AXsM]?1 tEoIhЮ#皵1-E6T"%qcVUJ`(#lAt2X,1[Ʃly5vM70<Ðϩn_/ؘh{@,1(ZʰEV򁨡h1ܻ&Pﵡz]yBe |Mr5xM2̀BSP=FS,(qx5el)=pOOh0i$J&XHxr&e"H6b4sR^%6YBɁ-9h[KA^"[C;vUM~|,sE޾υ.2kR`} „d_ha/Ӿ2 G dJs~{z'._/Xǝ͕6ACRIW eȢz4 1m]׿ ?hسb|v L6/1s վoƠw n幀gߦ?.;'LҼWr}}["R:!n|As&I051h_ˢc8Mi-J{j>/!_'O#\Th|(KWFwhh=` ϣZw׳bY7܏rgKY$wM"2)܏ ;|fnr:zf)n%%̏E>42%=|2E=#٥2.ϐrspXߤ6wŨ{,SS1=vt˨2}bV6b0v]~u,JpgV ET9t/B_qr5nSM뚛:~JQq{ͧJ>;r si0z4WƀGFe. FoɼkIYK wyԁ9À e@6ρyk~_:|;X/k/[cOW߲)Ce3ç_p 0Էko\PKCnCpTz%n.vN.19l @㲥kOx<iы>"k1InQwrf=3 d+3SRYXfV$޻#iH[zwzwC`meB3l\"=Wa:h]DŽzjɶLZVA+il>"w)F {: /~ٯL_ /u^c]HZ. :+['#*iuζroؚ;Fob$w1ud*5#؋Jn%;Fo0;gY,*betwY3{|^Vz?;)ۀSۖ[v6Jrݮ 7\2en̬5˟?ZcOmy#CO,ICWнIaode{E4õa_=}ѹg9Դ3_h+fWl^J_y:Jn"_Ч;)2~baʞlk%!C,1\_>7]MFDU ;ɹʸZ(Nb]l&VŭEH"rU%W!OP27I7hGfؼ~sYo-s**Ip]Z{ KJG{y=1Њ-7(EUU%c)|#~Ϛ2"N`N^{+ovNfafAj)N߇Y5Vn++I* v"jk ^Y43rV+-b~ TP&!3ar."lp"xW]hU>sgh&21(ߍ`@XZ%Od vTO'A >BA"+Q"CcDH!v:sLHMl}pa{Ә=ww=&z/Oheݛ #/C4o-ݡ8C"?A'p3<˭ ÝFgKYA#:ݼ%Rї檘'M@kh^[KdKаX+H^۝v9UQ"?cfs'@f?Tvӧiv ݀Vd+|_V#Ez=ʝm 04}*bYA= 1B|VGA3E+XOH 냂[35T*q叩+fkT(J 40:[_ ~NkBU'k y,$bJ"=_МMOA-3NKrj:ފeߌ9u3ea=um);m=Q/z3F.B"zױ%D!E&GaBKb>f Al&bO~f n"bwv{1sby89ɻS(^S#@cљSo*zćFF_Tn㩇y941T Z;,jRz5;Hl_h|ڥ ED<$pxL?8[k9_ }^ӋM/.i-KH|G%:M_t|5pRl 8jR؅=6{x".^?fFߋ->`jnuxCAtCֻCG,Blcxb{w>@!ND~)b9,Clc3<2v>ۤE#a %z8w7{yXyJz^F =;X*3VWCdjwˮkuLU]}r:~@N'uCO/N!]6N9زrc7F߲~SDqѥIx>ɶuN>?:+݊208<\D=!ob/ Caù =(33#!K  WI}b*b3:$D9.w :m(QH}6m۲3n6#0B Eֿ 0<r-({hrNR_g|+nWVΨ5 P=( B }XG#kPτ6}ʚ6_Rs*}XJ\5!|9&gcV-?HrbWz/,gM?LT]5i8IsH.ȶ?W:sU-,,'"~>&=]Z,`Y`~|~Mc?iMRg]E~K#4NAFl/ݬ^jgrqȂ~(q:{)|SUZ'~,h$u!v,Ѣ͞or&VyAMGo<VQnKtP(̒tCYm}!$=G *mdjM*)D hSM툞?*w84YxƉORP%ͦx+,2Z,y={+͑rg9ۤO.9ŕb"w* \wHSU=I1\#H98 - ́H-dٍ+ )sNu)ue^,<2U_VbX9`(GX"\x [BpUF){CxbigV`ƴEcü[QT.)+Lk"l!WCk^h.)^JUpnNZq8$Ze!ܼɵJbEPzWᮤTĐEG^W%|"[6'cp)EK7Od {uTմAO`';DNV\.4̗Ft*?b#*|A\e\\r<-R hMu!ZT: ?H.QUCcb^2W/p#QO\<.@s<_EeUZB[/@WRhTnkoy)CjϵL#t5Fa~}&}3?q8su n'SiU>Ov^#]%bFf@ooNCh{~萣]Ӈ4y^1 먕TyFayݹO7@g!@$]%,SaG|~b҇pyyԣ5$^BN3rȘ=(D^y'x#)0lǾ#5]yƊml!=ZjGhJmf}4Rr~oh6tF0#sr2q31b(m売hu,rce4>Oy9o:6:M<88zk_1*nR آ { ^N 4ͲbNKe<,뵪=CpbGfvUU"_b$XܺU9i:OaN_RbJqyu? +"m\O#7%n{#v;rrr9$su|t04پ39PRvlo"(x^p$'1DV"%sPR)C`jg>y$cRY2AO$d T +yqK,bxV6y hǀ:7k yeC\5Ѫ(D0zqǫ$ܦx>6৖3ȁZSS^;=7FI(: {:S)Uٞ{fi ffc鼽nʛQ>j/ٸRPeOJKqnY+Y9M{`=tI t#JN .xBC>g1a%+.6Ђjf((j(]ؙa|"Z)s?"1~LRje*<} G+SKFA9Nɓ>Tv%)>)UJq P+NAuMFIF??k.]3%5Eà#BDw)'y@D7ԍgĎ)vnm=E;Ml޶mpR3*1#J!;/jH9Q^3-Ef&ߴ|KUI趤5Ғ7Db6s[TAMA\^On{>5LM³nc$tK> =HJŤuc>)j54bOI\fߏ,4CPg*AQ;a.S%МKQNmRv *0Iλ.QP^xƗ<*j[eK}V**_שR++A}Uk xW-B.&#hZ1<0볌tcOsmc^9";&l%$Qvx0+E|ԣ^O;O"קmgGFm'1PtO`+Iȗ4TԮ{6n-yyLNGa(<\JqwOvߊUƖtQtf8J UN*p? &ԟ[nOYeOKJxbtwG>6wm봩-q>ʯ\ g%kO\_! ~n]My _T*gp}\m-^9Zaa[7?{ӶhcR ~Kl2'U3_*.3XV>.w-`8g]u'yKMɀVYޢogJzǴD<⊅jӗFx9ە&z2nE)6:mu " huta48ϮnB;ӍL"٭gz:fT^˕37 C<4*v˃&+0~@1f96WPT%rBǼITcy0xĿMYOMӶZ@0:=a(2~KA0Z{V!0cwz >>˨:?= 㦢 1@ գ՗2giaa;֨w36;f1COb~f2:7=y1cQgb+`(p |}˫ogq`0 #i͒iA[Yl2}[XU$ :lȔpx 1C3ݡ/cQbK쥸lnb hi/R7[|hs<_0SYj-AXR.iy&چS2+H}hJgZܧ5.#n2 ׆|P9D,5p.ʱHz})3:_Nm˻gBJ ;yX@騈CtHZ@~%=mP@A3WtF5GEGRu #z~T˺ SWf tz&GCԟzTZ^wsPs@$W]Z0gQՠ{T5G_DN/YDŽo>H͏ >_ߎoX5kӳF5"Eʾ<O̙Tpߣc@bhwv|NnjFGᣈp#o0{ەSJKLoznpr*=~fР:yhgQғmg̠6̔K.e$Q>x9#.=y˜AhWZv'*{Tv)+Џ|\SƢZK\Ze=̚i8vw4}ts)K됚FSƴ*jKI4:ΟՔ-tX%n%mD!/}Ϋƨ8E-Y}WZs` 6ߙ}*.^P[O(zk_% DVϪʥӎԮhH6UwiRĚ=Nu']& u1hL {XQ ϱu(i^To8U&_\׶X7>/tkMgD:'Dv u{ t=<)Z[4_w7{̩l-^zuhS ACcOCmmW!Nt@_3iRp}E6-\h>49v}Zb:DQC!u"D>!NNofIdkp4Mp[sf~1E`_ z㴗^vZb+΋cC\e˞IA[ݪȠz{8]9C( <eyd@gl[?vTIZ-ŭ@pIDk37 M3[8n{^|6>Yq`G]W 5(`X  X d=-Picture Word.Picture.80,Microsoft Word PicturedUPicture Word.Picture.80,Microsoft Word Picturep/0(  0;[0 0 000$([\{b00 000000000  0=] 0 0 0000 2 3 !A0C0E0G0I0c00000000000000000!%),.:;?]}acdeghijklmnopDTimesngs 2A9 i eq \hlp elpeSDZapf Dingbats i eq \hlp elpeS DSILDoulos IPA93 eq \hlp elpeS0DWingdings 3A93 eq \hlp elpeS@DWingdings3A93 eq \hlp elpeSPDWingdings 2A93 eq \hlp elpeS@  @@``   @n?" dd@  @@`` $<!  q    B  "#    B$N- `p>@IМt dh$dhB$r` IŢe5F dh$dhB$Z:{80 4dhB$P&!3ar )dh$dhB$"ggDͻOj.dhB$P҆Y0W>h&3dh `1?@xÝ@ > g4ddddpl4ippp0 <4BdBdgi e g4xdxdpl4ip p,? %( ,Is print-to-sound conversion based on rules?,  hMarielle Lange & Alain Content Universit Libre de Bruxelles http://homepages.ulb.ac.be/~mlange/escop99/&i=+<                 8Quasi-systematicity of the grapho-phonological relationsL          \P AI R P /p/ but P silence in COUP /ku/ AI /E/ but AI // in FAISAN /fza)/ AI /e/ in BAIE /be/ R /R/ but R silence in GARS /ga/f4 P+ 2& * P     &                                                       Questions   How are these quasi-systematic grapho-phonological regularities represented in the human reading system? How do they influence human language processing? 8    Dual-route model of reading   PI. Multiple GP associations  R1Dual-Route Cascaded (DRC, Coltheart et al., 1993)$2  2 4Letter detection study  `Task Ziegler & Jacobs(1995), letter detection paradigm. To detect a letter in briefly presented and backward masked pseudowords. In this kind of task, a homophony disadvantage is generally observed: more erroneous detections of the letter J with homophone pseudowords like GEUDI than with their orthographic controls like BEUDI. Critical manipulation\F K a GLetter detection, Results"    1II. Rule strength   In a recent paper, Rastle and Coltheart (1999) state that  One refinement of dual-route modeling that goes beyond DRC [Dual-Route Cascaded model] in its current form is the idea that different GPC [Grapheme-Phoneme Association] rules might have different strengths, with the strength of the correspondence being a function of, for example, the proportion of words in which the correspondence occurs. Although simple to implement, we have not explored the notion of rule strength in the DRC model because we are not aware of any work which demonstrates that any kind of rule-strength variable has effects on naming latencies when other variables known to affect such latencies such as neighborhood size (e.g., Andrews, 1992) and string length (e.g., Weekes, 1997) are controlled. H(Y              $                                          *                                , Naming study   Previous evidence Association frequency effect (Rosson, 1985). But possible confounds with orthographic variables. Our naming study Do grapheme frequency and grapheme entropy (a measure of association predictability) affect human naming times?Pb Pp bp ;  Naming, methodology  Materials 128 French pseudowords constructed by combining phonotactically legal consonant and vocalic clusters (dilve, juffe, ricte, tirve). Grapheme frequency: 32 pairs of pseudowords (Low vs High). Grapheme entropy: 32 pairs of pseudowords (Low vs High). Low and high items pairwise matched on: number of letters and phonemes, lexical neighborhood size, number of body friends, positional and non positional bigram frequency, grapheme segmentation probability, grapheme complexity  2V P'NY                                                                  Naming, Tasks,    Immediate Naming Task  Naming, Results,    |On Difference Scores Grapheme frequency , Grapheme entropy X <*  Z         Summary  Letter detection Study Homophony disadvantage for homophone-by-rule strings on false alarms (non significant on homophone-by-association items) and homophony disadvantage for both homophone-by-rule and homophone-by-association on letter detection times (correct responses). Naming study Significant effect of grapheme frequency and grapheme entropy on naming times (when numerous orthographic variables are controlled for). (  (,,/ KD       4Implications for models of print-to-sound conversion $    In Dual-Route Models, the conversion system is a one-to-one grapheme-phoneme mapping system insensitive to grapheme frequency and grapheme-phoneme association degree of predictability. In response to Rastle & Coltheart s (1999) citation, our data are clearly an invitation to think about a gradual representation of muliple associations in dual-route models.gf.  8             /789:;<@ A B C EFHQSPisx,,  3 :g{,,(d'h ` 33Ý` ` ff3333f` 333MMM` f` f` 3>?" dd@(~? " ddd@%  " @ `"  n?" dd@   @@``PR    @ ` ` p>> (     `wawa1 ?  V Click to edit Master title style! ! B  Z$wawa1 ?  RClick to edit Master text styles Second level Third level Fourth level Fifth level!     S   Zwawa1?OX lPage *      Zѯwawa1?_fX {Escop 4th Sept 1999 ,    B  s *@އh ? a( untitled 1 \T@( <I B  Zwawa1 ?n .*B  RClick to edit Master text styles Second level Third level Fourth level Fifth level!     S p  01 ?D  B  s *reP ? a( ` 0(   B  s *reP ? a(i P$(     r  S 8  r  S     H  0@އh ? a( + PHp(    3 rTwawa1 ?     3 rwawa1 ?uP  XB  0)?PPjB  BD1?%dB @ <D1?dB @ <DjJ?   jB  BDjJ?  dB   <Do?t jB   BDo?h h dB   <D1?OdB   <DjJ?cdB   <Do?<  H1?y VB AI R   H  0@އh ? a( 1 H@(    3 rwawa1 ?   XB  0)?PPr  S TPUP  H  0@އh ? a(   E = p@a ( 7@a@ @XB @ 0)?PPr F@ S    ^ [@ 6A 1?Hh^ ^@ 6?  ^ _@ 6?  d `@ <1? d v" a@ NG*H'I^;1?BgZ p" b@ HHIx01?- d p" c@ HHpڝIC1? d  d d@ <1? dR ^ m@ 61?P d e@ <1?4 g 8 ' Y@j TB I@B c $D  kTB J@ c $D X tTB K@ c $D^0 L@ C L@']  W Exc. BAIE  f  TB M@B c $D%U r V@ C V@T  W NonW BAIR  f   W@ C dW@   W Reg. PAIR  f  ^ j@ 61?@  i@ H1?/  HBAIR  d ]@ <1?,W  o@ 01?O ^ B AI R"    p@ 0D1?L  b E RB2  H @ 0@އh ??`e@]@a@d@_@b@`@^@c@ a(m     E ( 7@`@ ^  6A 1?ul  C    XB  0)?PPd  <1?:]:d  <1?e d  <1? d  <1? c v"  NG*H8IԸ1?]6 p"  HH>Iٸ1? e d  <1? -r d  <1? ^j   Twawa1 ?  & Concordance (Peereman, 1991): On pseudowords naming, more /Z/ pronunciations on BONGOUR induced by a discordant word neighbor (BONJOUR) compared to controls (MONGOUR).N;)+  p"  HH_II1?>    0d1? m  R associations      H1?   BONJOUR /b j u R/B(2  jB @ BD1? p8 H  0@އh ??`    a(&   ( Y ^  6A 1?ichr  S D   XB  0)?PPd  <1?^  61?(%H d  <1?@ 6 p"  HH0I;ɛ1?xd  <1?  0 d  <1?( p"  HH>EٝI_y1?t 6 d  <1?@ * p"  HHI>ѝ1?B * t d  <1?   Hщ1? t  zBONJOUR /b j u R/4  jB @ BD1? 0 ^2  61? ^2  61? H H  0@އh ??`   a(  % 4,|(  | | C x$wawa1 ?    | S ~Dwawa 1 ?$  XB | 0)?~ | BA ?1?K G? H | 0@އh ? a(  XP ( 0 p x  c $   XB  0)?^  6A1?PxpB  HD1? pB  HD1?0XX` pB  HD1?S    01?^ Y WHomophone-by-rule    0$1?^   ^Homophone-by-association    01?^  WHomophone-by-rule    0D1?^ vI ^Homophone-by-association  H  0@އh ? a(  H@t(  tx t c $   XB t 0)?PP t # lwawa1 ?  H t 0@އh ? a(  TLl( 7@`@ l l C xdwawa1 ?   XB l 0)?PPx l c $$P  H l 0@އh ? a(   H@ 4(  N  4 4 3 rwawa1 ?   XB 4 0)?PPr 4 S   H 4 0@އh ? a(   WO0P(  Pr P S    r P S D@  XB P 0)?PP P Zwawa1 ?   ;Delayed Naming Task   P Zdwawa1 ?P@8  @Difference Scores Immediate - Delayed latencies processing timesJ1 N         p P HA1?P 0 L p   P#  "  P H1?p  R *"F  N p   P p   P Hc1?p0  ^300 msec   "  P <tc1?p p@  <   P H4d1?p   ^200 msec   " P <d1?`  Sdilve   P HTe1?`  _ 2000 msec   p P HA1? 0  L   P# r  0" P He1? R *"F  bN   P   P Htf1? ^300 msec   " P <f1?0p <   P Hg1?0 ^200 msec   " P <g1?``  Sdilve   P Hh1?`Q _ 1500 msec   " P <i1? 0 <   P Hi1?l  d1200-1500 msec   " P <4j1? P  ] ,   P Hj1? P  _ 2000 msec   f P 61? @ ^ P 6A1?K H P 0@އh ? a(   | T( 7@a@ Tr T S T   r T S  n  XB T 0)? T <1? o  0ەە p < .001 , ە p < .05 8       ~ T BA =?1?-  =H T 0@އh ? a(D  @X( p Xr X S Tk   r X S k  XB X 0)?PPH X 0@އh ? a(8  P`x(  `XB ` 0)?PPl ` C l   l ` C tl8  H ` 0@އh ? a(i pK( Y X  C D     C nn .*B   S+ESCOP conference. Gent, 1-4 September, 1999 , H  0reP ? a(i B(   R  3 D     C Dn .*B   PIn alphabetic systems like English or French there are reliable associations between letters and sounds. Letters or groups of letters that form graphemes are generally pronounced with the same sound. Take the French word PAIR. P is regularly pronounced /p/, AI regularly pronounced /E/, and R regularly pronounced /R/. The characteristic of alphabetic languages is that a large majority of strings can be pronounced by applying a set of such grapheme-to-phoneme rules. A cue that this knowledge is represented in some way in the human reading system is the fact that when confronted to a string whose pronunciation has never been learned, a reader can exploit his knowledge of the print-to-sound relations to construct its pronunciation (for example, with pseudowords such as BAIR). However, in most alphabetical systems, this relation between graphemes and phonemes is quasi-systematic rather than systematic; graphemes are regularly but not necessarily always pronounced with the same sound. For example, as most consonants in French, the graphemes P and R are sometimes silent, as P in COUP or R in GARS. And AI is unfrequently pronounced // as in FAISAN and /e/ as in BAIE. This systematicity in the pronunciation clearly varies from grapheme to grapheme. For instance, in English it is generally acknowledged that the pronunciation of consonants is far more predictable than the pronunciation of Vowels. 9 P   H  0reP ? a(i ( Tt R  3 D     C n .*B   %Two basic questions in the domain of visual word recognition are: How are these quasi-systematic grapho-phonological regularities represented in the human reading system How do they influence human reading performance?   H  0reP ? a( i 7( Ll R  3 D     C dn .*B   EOne response to these questions is known as the dual-route model of reading. In this view, the graded nature of regularities is not represented in the human reading system. Knowledge of the print-to-sound relations is represented in a set of all-or-none grapheme-to-phoneme correspondence rules. In this rule-system, each grapheme occuring in the language is mapped onto one phoneme, a speech sound, in a one-to-one relation. Thus, for the grapheme AI, only one pronunciation, the regular /E/ pronunciation is represented in the system. The correct pronunciation of words that contain irregular or minor associations, such as BAIE, is obtained by a second system, an instance-based system that can address the pronunciation of all known instances. This view was challenged on several grounds in the mid eighties. At the center of many attacks was the notion of rule, that is, the idea that only the most frequent phoneme association is listed in the conversion system. Js H  H  0reP ? a( i | t  ( LlX R  3 D   z  C n .*B   One logical implication of the multiple association hypothesis defended in our first study is the idea that the conversion system contains information about the probability of occurrence of each association. This consistutes a further challenge for the DRC model. Hence, Rastle and Coltheart in a 1999 paper state that:  One refinement of dual-route modeling that goes beyond the Dual Route Cascaded Model in its current form is the idea that different Grapheme-Phoneme correspondence rules might have different strengths, with the strength of the correspondence being a function of, for example, the proportion of words in which the correspondence occurs. Although simple to implement, we have not explored the notion of rule strength in the DRC model because we are not aware of any work which demonstrates that any kind of rule-strength variable has effects on naming latencies when other variables known to affect such latencies such as neighborhood size and string length are controlled.   H  0reP ? a(i @( Ll R  3 D     C dn .*B   NAs a matter of fact, empirical evidence for an effect of grapheme or grapheme-phoneme association strength though rare is not completely absent. For instance, Rosson reported differences in naming times between words with highly frequent, midly frequent or unfrequent grapheme-phoneme associations. Though they did not bring the controls that are asked for. The aim of our second study was to bring further evidence for an association strength effect on naming latencies. Two variables were manipulated, grapheme frequency and grapheme entropy. Grapheme entropy is a graded measure of the predictibility of the grapheme pronunciation. It reflects both the number of alternative pronunciations and the respective probabilities of the alternative pronunciations.$  H  0reP ? a(( i x( Ll R  3 D     C p    Here is the methodology. Pseudowords were used because they maximally sollicit the conversion route. We had 128 monosyllabic French pseudowords in two lists. 32 pairs of pseudowords of low and high grapheme frequency; and 32 pairs of pseudowords of low and high grapheme entropy were independently selected. Each pseudowords was a combination of phonotactically legal consonant and vocalic clusters of French. Special attention was drawn on the control variables. To the classical ones like length in letters, length in phonemes, lexical neighborhood, number of body friends, positional and non-positional bigram frequency we added two new variables, grapheme complexity and grapheme segmentation probablity; variables never controlled in other studies despite their importance. Hence, grapheme complexity is important to unconfound with grapheme frequency since multi-letter graphemes are on average of lower frequency than single letter graphemes. Similarly grapheme segmentation probablity is important to unconfound with grapheme entropy, since we are specifically interested in the late conversion process. And obviously, grapheme frequency was controlled in the manipulation of entropy and vice versa. 22 first year students took part in this experiment for course credits. RG   H  0reP ? a( i  B( Ll R  3 D     C ѭn .*B   PSubjects participated in two successive naming tasks, controlled by a computer. In the immediate naming task, participants are instructed to pronounce the pseudoword as rapidly and as accurately as possible. In the delayed naming task participants are instructed not to pronounce the item immediately, but rather to fully prepare the pronunciation of the string and to wait for a go-signal to utter the response. This delayed naming task is used as a control experiment, to reveal differences in terms of articulatory characteristics between the experimental lists. And following a simple additive logic, difference scores are computed by substracting the delayed naming latencies to the immediate naming latencies to have a more direct indication of processing times.   H  0reP ? a( i R J 0 ( Ll꣏ R  3 D   P   C n .*B   vHere are the results: Grapheme frequency For the grapheme frequency manipulation, a significant difference between high and low grapheme frequency pseudowords was found on immediate naming latencies. No difference was found on delayed naming latencies. Grapheme entropy For grapheme entropy, a highly significant difference between high and low grapheme entropy pseudowords was found on immediate naming latencies. Note that since entropy is a measure of uncertainty, the Entropy scale goes in the opposite direction than frequency. Thus, the lowest the value, the highest the predictability of the pronunciation. Here, a significant difference was also found on delayed naming latencies. This indicates that part of the effect on the immediate naming latencies is due to differences in terms of articulatory factors. However, despite these differences in articulatory factors, a significant effect was observed on Difference scores. In sum, [in response to Rastle & Coltheart s citation], both grapheme frequency and grapheme entropy had a significant effect on difference scores J<- < H  0reP ? a( i f( Llp R  3 @     C n .*B   txThe aim of our first study was to disentangle the multiple associations and multiple phonemic codes activation hypotheses. A letter detection task was used. In this task, participants are asked to detect a prespecified target letter in a briefly presented and backward masked pseudoword. Ziegler and colleagues found that when the letter is absent from the pseudoword participants produce more false detections when the pseudoword is homophone with a word that contains the target letter. For instance, if they have to detect J in GEUDI, as compared to another baseword neighbor, BEUDI. The critical difference we introduce in this study is to use both homophone-by-rule and homophone-by-association pseudowords. Homophone-by-rule pseudowords are homophone of the base word when relying on the regular and dominant grapheme-phoneme associations. Homophone-by-association pseudowords are homophone of the base word when relying on any legal phoneme association of the grapheme. Following the rule hypothesis, only GEUDI is susceptible to produce a homophony disadvantage, that is longer detection times for homophone strings as compared to orthographic controls. Following the multiple association hypothesis, both GEUDI and BONGOUR are susceptible to produce this effect, because for both pseudowords, the /j/ phoneme is partially activated, as a dominant association in GEUDI and as a minor association in BONGOUR. The experimental stimuli used in this study were derived from words containing the letter J or S in which this letter was replaced by a G or C (GEUDI, BONGOUR, PENCER, PINCON). Pseudowords fillers were added to mask the manipulation. y. N       u   X   y H  0reP ? a(ji *"0( Ll R  3 D   (  C n .*B   So, to summarize, In the letter detection study we found an homophony disadvantage for homophone-by-rule strings on false alarms (non significant on homophone-by-association items) and an homophony disadvantage for both homophone-by-rule and homophone-by-association strings on letter detection times (correct responses). In the naming study, using two measures of grapheme and association predictability, grapheme frequency and grapheme entropy, it was found that both variables had an effect on naming latencies. zE,/8$ 0  H  0reP ? a(i @V( Ll݆ R  3 D     C Dn .*B   dWhat are the implications of these results for models or print-to-sound conversion? Our data constitute a clear challenge to the description of the mechanisms proposed by the DRC model; namely the fact that conversion operates using all-or-none grapheme-to-phoneme rules without any sensitivity to grapheme frequency and grapheme-phoneme association degree of predictability in the conversion system. In response to Rastle & Coltheart s (1999) citation, the results from our two studies should [strongly] incite psycholinguists to explore the notion of muliple associations and unit or rule strength in the dual-route model of reading. Thank you!   H  0reP ? a(%i u(  X  C D     S n .*B   wGHere are the results: We observed more erroneous detections of the letter J in both homophone-by-rule and homophone-by-association items as compared to their orthographic controls. Although there was no interaction, the homophony effect was only significant for homophone-by-rule strings in contrast analyses. For what concerns latencies of correct no responses, we found longer detection times for both homophone-by-rule and homophone-by-association strings. The important point is that subject find it more difficult to decide that J is absent in BONGOUR than in BONDOUR. This result is at odd with the rule hypothesis. In this view G in BONGOUR should never activate its minor /j/ phoneme association. Thus, because by design experimental and control stimuli were neighbors of the base words and equivalent in terms of orthographic similarity, BONGOUR should in no way be different from BONDOUR in the dual-route theory. In contrast, it strongly supports the multiple association hypothesis that supposes that the minor association is partially activated during the conversion process. F 2 Uu}=1# H H  0reP ? a(i T(   R  3 D     C n .*B   bAn alternative hypothesis of multiple graded associations in the conversion system was introduced. It was supported by demonstration of the influence of rime inconsistency and grapheme discordance on naming performance. For example, Peereman found that people tend to make more errors on the pronunciation of the grapheme G in pseudowords when a close word neighbor induces a discordant pronunciation. There are more soft /j/ pronunciations in BONGOUR neighbor of BONJOUR, than in MONGOUR; and more hard /g/ pronunciations in GIRNIR neighbor of GARNIR, than in GIRLER. Peereman interpreted his results as an indication of the multiple activation of the G phonemic associations in the conversion system combined with the influence of lexical neighbours on the selection of the G pronunciation.   H  0reP ? a(Oi (  X  C D     S $n .*B   SInterestingly, a new and challenging version of the dual-route model, called the dual-route cascaded model was recently proposed by Coltheart and colleagues. This model steps back to the hypothesis that the conversion system contains one-to-one grapheme-phoneme correspondences. Only dominant associations are stored and no indication of strength is provided. It accounts for the consistency and concordance effects by the activation of orthographically similar words in the lexical pathway leading to the activation of multiple phonemic codes in an output phonological buffer. For example, BONGOUR activates its word neighbour in the lexical route which partially activates the soft /j/ sound in the phonemic buffer, G is converted into hard /g/ sound in the rule system and partially activates the hard /g/ phoneme in the phonological bufffer. TS T H  0reP ? a(Rfx\ p]u>GԫgY 6X,e=rL'0XldmRu30iHSH4$IC CgPH -mZ1_svtު,Ӌg={vϹwָ葛5B2Lխ(Gdd2>|e0`9&,k=`[[sg+lhk>skވg+W_ϑ0~ndS5QF!|OXF}%5oPy { ΅P{}RjzMBmANɸ@Q2-"~l%>ͼ>:M-\piXLc(w,:E5QQ#wi[deY-dHMdDF.l7vY<,3|&v~>zr j1͟QkHf^"^< zQ$+2٢WtkoL~>q4 ?L{ c4`cJROOepș`aY5X&ɀBa kM&|4=X+FYvQ 48DZ܌?#3(=׈ec6NFn,;#e[tN8*Sx0-c8~/oޡa68/_Jt|'+)LNߔ_|=69㚩;skBFdyUx.;^$Ș9OVO]µK6{`wlL,6u{Y6U푃x"d/OvwogSx\T6(7lbu:/Eȏgsn_ʹ&:xZ܃ !Y9lӀ8NYB jP&ڹrWiz%"6X<9% !YA9lDJJ2,+dp )e8X%B\5NI$(plH~.ls (47}5mjx ?xs,(xFivяN17N5O96Cީ<͙U|z4>cy@Yۨ;5icoV^rBVTTXv߉E[a*dHfS@Z:́KٺJU-g\]вͨM[iXHglj:|sV֖~*+9T>m)GZl+9+d 3T?Ϣ,-g"+0R43޷eUf?^z[bx9. /=؇?}ucdIc aȴr`U2ށzV-:uuVNw SR9ߍcX{ )ղ1H=~Ǐ) 5O֧!<4ny-XbQѝ~ErUI>?:p87l{o2_d2wm^aK!|ӵbᜣ,DC~얃NO6|=H?حK5}VZ古o,>'=?9W&Lr;NDz8XOi{71w{˘`pSgҖxdcgc.ӑc|C#5qsmw,,mHi{W9t`s{;NE44Ұ#!3B4<78xH18Aǡ.D+ɕ\5⁒]xW0<` bɔ3>ł$KQ`pVwbŪ{lu?]?͑̿~ >1iO/jUpxo͍y#O@~E a;:tH\_vUU\h$={LZUJ>$B3 ,.^¾5.x݅<1bw^7b\OJb N(1K j4w>lV_F*KwPØY}dOJ/ŋp\@]pvs8to'9waҷL0ΙN.יYW㱫-¶%+tH x6HqdL%7P2-SHwʑK".zNVgf\,5vhag><Lˇ&Λڪp <MӟnDJ>:zWR1vj4\=M[77|lhaO.̧>gĝ}8{%媁'<z&R䇒6RJ>^}f}|\/+N5DN7,8| - <_pk{ژYʙ;!bm5 ؆\"$Kr'ncpoV>|hEa|ulIA^P /?Q?MӜ︻s{ҒY=֚K]pp)_5-}1e̷曀-[Y[oƓ;H)}Ӟ>DͲI}O~,l'J7C37y܆{~cP:WwFb%%>A]N ⑖:׍>_Tۛe?Y"ݳΜ'Jz> w!ΕaV|||?›Wp_>::H͸R>|f?BqRp&[_&}'Sۍ;؈ČߥQi[vtmZ%y _EL<ԤN?n;_+}8>"zij_ 眐aʆ6\7: ?* ZMeak+_U"yUU@ʒd/?>>>>@aC Xx L]ž|ޫ{ <(?R|‹)- fsؓQmN6l.3$&:`k3T4nY].]ZèqLӑ9w6=ԗ;~}z3С_.xFA0lz!'246M6Ǧg1LHcF zST!K ?,9s!W9_x-quNK".CG r>ʋ3zݰX ؐC^:?O':}'&"dp4A l i>@iS?Ҕi9Iނ*NJ9:'?A>bٸDh즽qqq8~jAh$^>p$%вoh𲲑X*BfDցp%D|bCr^V p'<8i24\1aB"ŖB#e065nɕ($8'͂jjݯY`4i>+iuFISj%jprfP^׊6C\az DWBL(7qFk:{ч!C}m~yUy%*j3fB>ĸF zhܤpDf9R o-yyhv" D 699P쟳]:yLC?UW׿]<±܀1ϳ#p#/1ADN;ϲ^vb^{7ܲID5Sp|(HAǻ<9垤2Oi1?x@;Z$4;အC!CfCmiD1WǞ5χֱMC>s=OOԦ켷sh$A\=FFfY݉8H)htߜ.zE1x:s IR]Ь6 KlwP3LKJMD=7*[FL f!7p32րiЂk(OaJ VKًIQr!FDK>T>7)n&TP(B z5n3#oTau͸1VL1l2_3gpC sTL +.eHH$Oڮ@d '/7&I&:mnD\5d4 { yqu\ Y%fP:2S[AI;$e4bZ'&.Zpwl%Gs ˋf;R:Zdh [/EcoicZNJC^ӿg/Rb~\9:yGOwk>r7O1&"]^0<6qMᵱ `@9Sh;' ­Hmc.rqw>lgЫN"͗8V|d_y3.G43\Ppݴ{ k?C;ZAޑ>vNlZVh/s"PSy?-iG=U|,[v *1lsh-{'`2G7yotwggb՗F7zVYf^@37Nk|`駺Oq5?@'l11[ˑab"|%NOIN(N$N2VI٬ +Ӯ_H]ǿ^!^ҘcDZ*ǁYB}]{Ph+Tt^ y3mD׺ק+5OU:cȹb/_Ss7-1*~zX5)gF51lLXwHVOq );Iكt@ OϷt3\/:.ݎS} nhD(O3iiehTGHdG'Ktʮ^ Gj9ӈ^nh5D}\`͌ ^3Tn ܮSiS'-WULSʽ: clAlv<{bl<esbl<esb1{Tlq5UtE#7t񉯭h-Dӱ^^HW_j!EzC;29r*T=$'-Ta4R f 0[|m+v+#ude9WPa~ȁH z)_PS".HFlz{x\ w&`QFkdZف/(kB\L({Jቌt"Gd>$s0 ^8jH__Ѫ,wkqm׹u=}=wUP[[z;cJ{zw wF!KSh!_O+XsssE\FW"" "E\X֯5=r. 5g,uN'cGi +j۰1V*l|21IEZjXnGtV_OG '_. /Tcؐ*5Mjli\?+{1a6[e{o&~w4?1p<97Ohv*17fcyM5浻XKAzZmR,f>܆H-KX Tsrx>s4!Dڦ#>WK\'2:0-<yp:P# [Ey=%sKyw'r zVaX/jj9>Xh%B+sX>'|[8 Sņ/P~KB֘+D^~+@S}D?\ )}Cw>%;~?E<C]O G@ȑ>| ]k Oy["ΛCW֤=ݜH0gF?7U7uvX0r间R%_JS65lV2&}9uXUW1VbN䕼V_%cͤb,+k1xpppp %*r0>6A 05B D K|N40 , |1z4o7`KNH@@ڻE@hs$P@!\Qf]UUP ژOh+'0(HT`      'zLength effect on pseudoword reading: Letters or graphemes? Marielle Lange & Alain Content Universit Libre de Bruxellest engPSYCHO.EXPERIMENTALEdowSYCSYC-SYS:Microsoft Office:Microsoft PowerPoint 4:or Lange Mariellef297Microsoft PowerPoint 4.0oso@wo@ѥ-@YKu@ 䵗 B GPICT @@} }ĥ}HH`xff33Ýff33ff33ffffffffffff33ff33333333ff333333ff33ÝÝÝÝffÝ33ÝÝff33ff33ffffffffffff33ff33333333ff333333ff33ff33Ýff33ff33ffffffffffff33ff33333333ff333333ff33ffffffffffff33ffffÝffffffffff33ffffffffffffff33ffffffffffffffffffffffff33ffffff33ff33ff33ff33ffff3333ff33ffffffffffff33ff33333333ff33333333Ý333333ff33333333333333ff33333333ff33ff33ff33ffff33ff3333ff3333333333333333ff333333333333333333ff333333ff33Ýff33ff33ffffffffffff33ff33333333ff333333ff33wwUUDD""wwUUDD""wwUUDD""wwwwwwUUUUUUDDDDDD""""""}}u | x g   "$! !)FFA )E D= >7VVVVVVDocumentSummaryInformation8Current User. չ.+,Dչ.+,$   Y' A4 Paperio7zb TimesZapf DingbatsSILDoulos IPA93 Wingdings 3 Wingdings Wingdings 2 untitled 1Microsoft Word Picture-Is print-to-sound conversion based on rules?9Quasi-systematicity of the grapho-phonological relations QuestionsDual-route model of readingI. Multiple GP associations2Dual-Route Cascaded (DRC, Coltheart et al., 1993)Letter detection studyLetter detection, ResultsII. Rule strength Naming studyNaming, methodologyNaming, TasksNaming, ResultsSummary5Implications for models of print-to-sound conversion  Fonts UsedDesign TemplateEmbedded OLE Servers Slide Titles 6> _PID_GUID'AN{8CFE5A00-4B2C-11D3-A848-000502D61724}&_Lange Marielle