the user`s manual.
Contents
1. on the second 437 on the third and so on The last column provides the sum of activation that a unit has received over the course of reading For example the unit poor received 10 52 total activation over the course of reading of the entire text Each column represents a reading cycles and shows the activation values each text unit received In this example the first reading cycle had a total activation of 15 evenly distributed among three text units poor old and woman IV Connection In this output users will find a matrix table that includes the strength of the connections between all the text units Figure 5 These connections create a network between all the text units Connections between units vary in strength as indicated by the different connection values in the matrix In this example the text unit poor is connected with the other text units at various degrees first row For instance poor is more strongly connected to old 1 392 than to hill 201 The main diagonal is the connection strength value for each concept with itself For instance the auto connection strength for old is 1 941 for woman is 6 278 and so on These auto connection values are a function of the total activation these units received over the course of reading but also the degree to which they have become connected to other units Landscape Model v 1 0 12 Ez Matrix Table m x qp2. Area Once there was a poor old woman Together with her husband she lived in the forest One day she was walking in the hills She came upon the entrance to a tiger s cave She wanted one of the tiger s whiskers She wanted to make a medicine of the whisker for her husband She put a bowl of food in front of the entrance to the cave She softly sang a song which put her the tiger to sleep The old Woman quickly pulled out one of the whiskers Very quickly she ran down the hill Panting she came home She was very pleased Unit Area poor old woman live forest husband walk hill came upon entrance tiger cave want whisker make a medicine put bowl of food in front sang a song put sleep pulled out ran pant came home pleased Activation Type Area m m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3 e e r m m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 r e e 0 m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 c 0 0 0 6 e m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 c c m r m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 m 0 0 e e e e m m m 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 m 0 m c r c m m m 0 0 0 0 0 0 0 5 0 0 c 0 0 0 0 0 0 0 m 0 0 0 r e e e m m 0 0 0 0 0 0 m m 0 0 0 0 0 0 0 0 0 c m r 0 0 0 e e m 0 0 0 0 0 0 c 0 0 e e m 0 0 0 0 0 c 0 0 0 0 0 0 c m 0 0 0 0 0 r 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 r m m 0 0 0 c 0 0 0 0 0 0 0 0 0 0 c 0 0 0 0 0 0 c 0 r r m 3 Landscape Model v 1 0 17 ERROR MESSAGES Most of the error mess
3. THE COMPUTATIONAL IMPEMENTATION OF THE LANDSCAPE MODEL VERSION 1 0 USER S MANUAL Paul van den Broek Yuhtsuen Tzeng Panayiota Kendeou CONTACT INFORMATION Paul van den Broek University of Minnesota 211 Burton Hall 178 Pillsbury Drive Minneapolis MN 55455 E mail pvdbroek umn edu Yuhtsuen Tzeng National Chung Cheng University E mail ttcytt ccu edu tw Panayiota Kendeou University of Minnesota 75 East River Rd 314 Hlliott Hall Minneapolis MN 55455 E mail kend0040 umn edu Last Updated 17 December 2004 Landscape Model v 1 0 TABLE OF CONTENTS Page InttOgucttOn x22 en eb ae 3 System Requirements and Installation 4 Th Model DUC oorpore rt inve ta Dot ds 5 Ru n ing the M del nren nisi oben iias 7 The Model OUIDIL nuron a e A P EUR 10 Saving the Results ccccsvceyedavavatersdeades seaccesstvceaasesannacsesnne 13 Other Features of Version 1 0 sess 14 Example 1 4 o ute ee ho Behe S 15 Ex AMPS 2 uereg iinn ukt a T A leadeaeaaeenaa eas 16 Index of Error Messages eere ente enne nn 17 Landscape Model v 1 0 3 INTRODUCTION The complexity of text comprehension demands a computational approach to describe the cognitive processes involved The computational implementation of the Landscape Model of reading was designed to meet that need This model captures both on line comprehension processes during reading and th
4. ackage will install the program automatically and generate the Landscape folder in the Program Files folder with the executive file Landscape exe and several demonstration files Demol Im and Demo2 lIm For now the interface of the program is in English but it will take any languages as model inputs as long as those languages are supported by Microsoft Windows Landscape Model v 1 0 5 THE MODEL INPUT The input to the program consists of a single file that includes three components that should be labeled Text Area Unit Area and Activation Type Area including the square brackets see example on page 15 Demol Im These are the three components that the Landscape Model reads and simulates Figure 1 We describe each component in detail below Data CAWINDOWSXDesktopMigure 1 of SCiP paper Im Text Area Once there was a poor old woman Together with her husband she lived in the forest One day she was walking in the hills Unit Area poor old woman live forest husband walk hill Activation Type Area m m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e e rm m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 0 m m 0 0 0 0 0 0 0 0 9 0 0 0 0 0 0 0 0 Figure 1 The Model Input I Text Area In the Text Area component users can type any text without affecting the results of the simulation Users can type in memos comments or as we recommend the text they use for the simulation II Unit Area In
5. ages occur in user s attempt to parse the model This is because of errors in the proposition or activation value areas Error Message Some activation value is not correct This indicates an error in the Activation Types Area It is possible that you have an extra space or you miss a space Error Message There are not enough numbers of activation values in some cycle This is the most common error message and it could indicate a number of things l 2 3 4 A comma was omitted A commas was inserted at the end of a cycle The number of activation values per cycle is not the same The number of cycles in the Activation Types Area does not match the number of cycles indicated in the Unit Area or vice versa Inconsistent space between commas and activation values e g c 0 m instead of c 0 m Error Message Some user defined Activation Values have not been set This error message will occur when you attempt to run the program and there are undefined activation variables You will be automatically prompt to define those variables
6. and retrieves the text unit with largest connection strength with the first retrieved unit This searching and retrieving process iterates until it exhausts all the text units or it reaches the retrieval threshold The output gives two columns The first column includes the text unit whereas the second column includes their connection value Landscape Model v 1 0 13 SAVING THE RESULTS We recommend saving the results in the Excel format and opening the files in Excel for further analysis To save the input file with extension lm such as the one in Figure 1 users need to select File at the menu bar and choose Save As to specify a file name Users can open this file for later editing To save or edit results of the LS model such as input matrix activation matrix the resulting connection matrix or the retrieval matrix users can use the Launch Worksheet Designer under the Display function of the Menu bar This will work like Excel spreadsheet Landscape Model v 1 0 14 OTHER FEATURES OF VERSION 1 0 This version of LS model is equipped with several additional features to give suggestions for additional features please contact the authors Reading Delay The model has delay capabilities that allow one to simulate the hypothetical situation that readers wait for a while before continuing reading the next cycle The program allows users to simulate this by attaching an integer enclosed by a pair of parenthesis at the end of a
7. del Architecture Delta rule Reduced Expectancy Working Memory Enable Wh CAP CAP Value 20 Retrieval Enable retrieval module Activation function Sigmoid v r Reading Cycle Initial Connection Matri C Epochs None C Load File C Current matrix C Cycles less than 1 Epoch Run X Cancel Figure 2 Running the Model I Sources of Activation The program contains several default types of activation referential causal and enabling connections the activation values associated with these default sources are based on prior research However users can modify the values for their own theoretical purposes as well as add other types of activation If there are any undefined activation variables and you attempt to run the program you will get an error message Some user defined Activation Values have not been set The program will pick out the undefined activation source variables and ask you to define their activation values one by one All you need to do is to type in the numerical value that represents the activation value for each variable name if no values are desired hit OK and the model will Landscape Model v 1 0 8 assume zero values To avoid this error you may hit the Create more Sources of activation button and the program will prompt you to define the sources of activation that are undefined II Free Parameters The program contains several defau
8. e Cohort or Learning Rate parameters accordingly III Model Architecture The program contains several default values in this section as well In the box of Delta rule users have the choice of reduced versus normal expectancy Expectancy is a property of the delta learning rule reduced expectancy results in a more stable performance of the model Users also have choices as to whether to enable working memory capacity constraint and the retrieval component Once a user enables the working memory module a choice of capacity value can be made The Activation function also gives users a choice between a sigmoid or a linear function for the learning curve As noted above unless there are strong theoretical reasons to do otherwise we recommend using the default options for each of these parameters Another important feature in this section is the Reading Cycles options By selecting Reading Cycles the user will be prompted to enter the number of cycles that are read By changing this value readers can model reading of a subset of all possible cycles For example there are 12 reading cycles in the sample text If one enters 10 in this menu the model will read only the first 10 cycles If users intend to model rereading of the same text for several times they need to select the Epochs Users will be prompted to enter the number of epochs One epoch corresponds to rereading once two epochs to rereading twice and so on The final feature in this secti
9. e off line memory representation after reading is completed incorporating both memory based and coherence based mechanisms of comprehension Several studies comparing computational and behavioral data indicate that the implemented model is able to account for cycle by cycle comprehension processes and memory for a variety of text types and reading situations For a detailed description of the model s architecture basic settings and applications we recommend the paper below Please contact us if you need additional information van den Broek P Young M Tzeng Y amp Linderholm T 1999 The landscape model of reading In H van Oostendorp amp S R Goldman Eds The construction of mental representations during reading pp 71 98 Mahwah NJ Erlbaum Linderholm T Virtue S Tzeng Y amp van den Broek P 2004 Flunctuations in the availability of information during reading Capturing cognitive processes using the Landscape Model Discourse Processes 37 165 186 van den Broek P Rapp D N amp Kendeou P 2005 Integrating memory based and constructionist approaches in accounts of reading comprehension Discourse Processes 39 299 316 Landscape Model v 1 0 SYSTEM REQUIREMENTS AND INSTALLATION The program designed with C programming language could run on Microsoft Windows 95 98 ME XP and 2000 with 64MB minimum of RAM Users who have obtained the program can click on the Setup exe file and the p
10. f for referential and so on Units that are not activated in a given cycle are assigned a zero 0 As in the Unit Area section each type of activation unit must be separated by a comma and each reading cycle by the ENTER key Units at the end of each reading cycle do not need a comma The number of reading cycles in the Activation Type Area and in the Unit Area sections should be the same Landscape Model v 1 0 7 RUNNING THE MODEL Running the LS Program When users enter all the required information in the three components Text Area Unit Area and Activation Type Area they may proceed to run the program by selecting Model Parse on the menu bar If all the required information has been provided the program will start running This is indicated by a running man icon that appears on the menu Users can proceed to set the model s parameter values by clicking on the running man icon A menu will appear Figure 2 that includes four important components Sources of Activation Free Parameters Model Architecture and Run or Cancel We describe the function of each component below Model x mSources of Activation Free Parameters 4 Cohort Effect o 03 Text Input m fa Learning Rate jag Peferential Inferences T 4 Activation Threshold 0 0001 Causal Inferences E 4 Retrieval Threshold po Enabling conditions e IE Create more Sources of activation gt Mo
11. ing Conditions Label e Activation Value 3 5 Delta rule Reduced expectancy 6 Working Memory Capacity 30 7 Retrieval module take 8 Cohort effect 0 03 9 jLeaming rate 0 9 10 Activation threshold 0 0001 Figure 3 The Model Output I Parameter In this file users will find all the information regarding the settings of the simulation These include the text input the types of connections and their values the delta rule working memory capacity whether the retrieval mode was enabled or not cohort effect learning rate and the number of reading cycles This file is important as a reference for the user II Input In the input file users will find the input matrix The input matrix is a matrix that includes all the text units and their values as defined by the user The first column contains all the cue words in the order they were entered in the Unit Area The other columns correspond to each one of the reading cycles in the order of input For example in the first cycle the text units poor old and woman were mentioned Because m mentioning was given the value 5 in the model these values will appear for each of these three concepts Landscape Model v 1 0 11 III Activation In the activation output users will find a matrix table This table includes all the text units and the activation values they received in each reading cycle This matrix show
12. lt values for Cohort Effect Learning Rate Activation and Retrieval Threshold that are based on prior research The Cohort Effect parameter determines the cohorts that are formed among text units larger cohort parameter values produce bigger activation cohort effects During the process of building cohorts in readers working memory text unit interconnections are built through a delta learning rule This learning mechanism follows an asymptotic curve as reading elements co occur more than once in working memory The amount of learning for each co occurrence among reading elements is determined by the Learning Rate parameter larger learning rate parameter values mimic faster learning rate Activation Threshold is a very low activation value that is taken by the program to decide whether this amount of activation means no activation at all and therefore the program will end a loop It has no particular theoretical interest and is chosen arbitrarily Retrieval Threshold sets the minimum connection value between text units that can be retrieved during the process of retrieval It determines the number of text units that will be retrieved and therefore mimics readers criteria of retrieving All free parameter values can be modified according to the user s theoretical purposes but we recommend using the default ones unless there is a strong theoretical reason to do otherwise For example users may model slow or fast readers by changing the value for th
13. mx SS dem dE ETE dE TERES TESTE EDS TEE TECE Oe Ea poor old womarlive forest husbarwalk hill came tentranitiger cave want whiskemake a 1 392 1 792 0 67 0 67 0 587 0 174 0 201 0 125 0 145 0 102 0 09 0 124 0 203 0 09 eg P 1 941 2 26 0 685 0 685 0 694 0 187 0 207 0 163 0 192 0 142 0 133 0 559 0 742 0 547 4 2 347 6 278 1 537 1 537 1 887 1 843 2 077 1 884 2 237 1 737 1 458 2 137 3 105 1 867 5 0 65 1494 1 311 1 054 0 671 0 686 0 146 0 172 0 1 0 095 0 126 0 188 1 6 615 0 638 1467 1 311 1 275 102 0 663 0679 0 146 0 172 0 1 0 095 0 126 0 168 j 0 635 1 707 1 i 1 776 0 384 0 556 0 457 0 493 0 468 046 0 71 1 018 0 179 d i k i 0 503 0 197 0 163 0 196 0 145 0 135 0 763 0 544 0 048 0 048 0 048 20 0 372 8 9 10 1 12 13 i M 126 0 585 25 16 v 18 19 20 ZR 037 22 12 0 647 23 107 0 094 24 r 0 096 25 m 0 086 0 096 26 pleas 049 0 048 0 6 0 051 0 051 Q i 059 0 ma 289 1238 40 98 9 431 9 94 11 1313 15 187 N Parameter A Input A Activation A Connection A Retrieval 7 4 Figure 5 The Connection Matrix V Retrieval In this file the program suggests a retrieval order of the text units based on a two stage retrieval mechanism The program retrieves the first text unit based on the sum of their total activation values across all reading cycles Starting from this retrieved text unit the program searches for
14. ol Im Text Area Once there was a poor old woman Together with her husband she lived in the forest One day she was walking in the hills She came upon the entrance to a tiger s cave She wanted one of the tiger s whiskers She wanted to make a medicine of the whisker for her husband She put a bowl of food in front of the entrance to the cave She softly sang a song which put her the tiger to sleep The old Woman quickly pulled out one of the whiskers Very quickly she ran down the hill Panting she came home She was very pleased Unit Area poor old woman live forest husband walk hill came upon entrance tiger cave want whisker make a medicine put bowl of food in front sang a song put sleep pulled out ran pant came home pleased Activation Type Area m m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e e r m m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 r e e 0 m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 c 0 0 0 6 e m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 c c m r m m 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 m 0 0 e e e e m m m 0 0 0 0 0 0 0 0 0 0 0 0 7 0 0 0 0 0 0 m 0 m c r c m m m 0 0 0 0 0 0 0 0 0 c 0 0 0 0 0 0 0 m 0 0 0 r e e e m m 0 0 0 0 0 0 m m 0 0 0 0 0 0 0 0 0 c m r 0 0 0 e e m 0 0 0 0 0 0 c 0 0 e e m 0 0 0 0 0 c 0 0 0 0 0 0 c m 0 0 0 0 0 r 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 r m m 0 0 0 c 0 0 0 0 0 0 0 0 0 0 c 0 0 0 0 0 0 c 0 r r m Landscape Model v 1 0 16 EXAMPLE 2 Demo2 Im Text
15. on is the Initial Connection Matrix option This option allows the simulation of rereading or of reading of multiple texts Through this option the simulation will take a connection matrix as an input and build a new memory representation using the pre existing connection matrix as a starting point As a result the program allows users to open an old connection matrix and take it as an input When users select Current matrix they Landscape Model v 1 0 can simulate rereading of the same text When users select Load File they can specify as an input any other matrix The only limitation here is that the initial matrix must contain all concepts from both matrices even those set to zero IV Run or Cancel After setting the values in the Model menu the user hits Run to run the program The program will take only a few seconds to display the results If one wishes to cancel simply hit Cancel Landscape Model v 1 0 10 THE MODEL OUTPUT Looking at the Results The program gives five output files in an Excel Workbook format These files are labeled Parameter Input Activation Connection and Retrieval Figure 3 Landscape v1 0 Matrix Table i OF x File Edit Model Display Window Help 8 x DARSA TBZ A i T RT ee aul Text Input Label m Activation Value 5 2 Referential Inferences Lable r Activation Value 4 3 Causal Inferences Label c Activation Value 4 4 Enabl
16. particular reading cycle without a comma between the last proposition event and the parenthesis under the Activation Type Area section The integer will qualitatively mimic the waiting time interval and this integer will make the program insert a reading cycle with zero activation see Example 2 file Demo2 Im on page 16 Alternatively users can simply insert several rows of zero activation values at the proper places in the lm file and run the program This might have an impact on the results of the final mental representation of the text If a delay is inserted between text statements it simulates reflection on the last information in the last input cycle if is inserted at the end of the text it simulates the gradual decay of the final vector after completing reading of the story Rereading Additionally this version of the LS program has rereading or multiple text capabilities This means that the program will take a connection matrix as an input and build a new memory representation on top of the pre existing connection matrix As a result the program allows users to open an old connection matrix and take it as an input and reread Conversely the LS program allows users to save a connection matrix easily for later rereading or reading of a new text The only limitation is that the initial matrix must include all concepts that will appear in the current matrix even if they have to be set to zero Landscape Model v 1 0 15 EXAMPLE 1 Dem
17. s comprehension as it unfolds online Figure 4 and differs from the Input matrix that simply depicts types of activation 7 8 12 Sum of Cycle 0 437 0 284 0 231 0 151 0 145 0 167 0 265 10 52 0 437 0 284 0 231 0 151 0 145 4 036 0 277 14 35 4 4 3 644 2 737 3 284 4 036 4 45 44 3 0 398 0 243 0 208 0 143 0 162 0 273 10 19 3 0 398 0 243 0 208 0 143 0 162 0 273 10 19 0 322 0 234 0 093 0 18 0 186 0 186 0 315 13 55 0 359 0 145 0 15 0 159 0 296 12 46 0 359 0 146 0 15 0 159 0 338 14 23 3 644 0 395 0 452 0 287 0 347 13 45 3 644 3 422 0 517 0 363 0 391 16 95 4 555 0 358 4 105 0 321 0 364 13 25 3 644 3 422 0 488 0 326 0 349 11 71 4 555 2 737 0 606 3 228 0 416 16 8 4 555 2 737 0 606 4 036 4 24 09 2 737 3 284 3 229 0 353 14 78 3 422 2 463 0 087 0 051 6 151 3 422 2 463 0 087 0 051 6 151 3 422 2 463 0 087 0 051 6 151 D 4 105 2 421 0 123 6 949 2 421 0 123 6 949 4 036 4 11 89 0 321 8 729 make a mt 22 pulled out 23 2 3 3 4 5 5 5 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 NO OCOOOOOOOOOOcOGOdOoOoOotu un cmoooooooocooooco OCooooooooooo eojojojoiojiojiojojlimpoo 0 0 0 N 8 8 to to 3 28 D gt KX Parameter A input A Activation A Connection A Retrieval 7T Figure 4 The Activation Matrix Each row shows the history of activation for each text unit across the reading cycles In this example the text unit poor received 5 on the first cycle 3
18. the Unit Area component users provide the cue words for the individual parsed text units from the text with each cue word within a single input cycle separated by a comma Note that users can decide the unit of analysis in each cycle according to the theory under consideration e g propositions events idea units sentences and so on In the example discussed in this manual we use major propositions as text units but users can define text units according to their theoretical purposes e g concepts main ideas sentences and so on Reading cycles should be separated by hitting the ENTER key The final cue word in a reading cycle does not need a comma This format must be followed strictly Each row of this section is taken as a reading cycle and each cue word within the row is taken as a proposition within a cycle For a Landscape Model v 1 0 text with n cycles readers should type in n rows in this area with each row consisting of its constituent cue words The Landscape program will use this input to decide the number of reading cycles for the current text and the number of propositions within each cycle In addition these cue words will be used as output labels III Activation Type Area In the section Activation Type Area provide the types of activation each unit receives These types of activation are labeled using a single alphabetic symbol or a string of alphabetic symbols e g c or cau for causal r or re
Download Pdf Manuals
Related Search