1: First Run
2: Second Run
3: Third Run: Randomness
4: Fourth Run: Rules
5: Fifth Run: Regions
6: Sixth Run: Social factors
7: Seventh Run: Multiple Variant Displays
8: Eighth Run: Clusters
9: Ninth Run: Persistence
10: Final Run
1: First Run
Open the simulation, or use the “reset” button to return the simulation to its original state.
In the “variant” box at left, click on each of several of the variants. You will see on the map the locations where that variant was elicited (NB: only 1162 locations out of the 8610 on the map can appear, the original survey locations).
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
With one of the variants displayed on the map, click the “DE-4bin (current)” tab above the map, which will change the display to Density Estimation using 8 nearest neighbors. Here, the 1162 original survey locations are used to predict the status of all the other locations, in 4 color levels from white (unlikely) through gray up to black (most likely). Other DE tabs show a 5-color map or a many-color map.
Click the Standard tab to return to the original display of locations. Click one of the more common variants on the “Variant” list at left (one with at least 25 tokens), then click the “Keep” button below the list of variants. On the right hand side controls, find the “Draw Interval” box and in it type numeral “1”. Now, click the “Play” button and watch what happens. To pause the simulation, click the “Play” button again or push the Space Bar. Pause after 1000 generations.
With the simulation paused after 1000 generations, click the “Age” tab above the map. Now click the “Fade Newborn” tab: all of the locations that have been alive for just one generation are turned gray. Note the presence of some clusters. Now click the “At least 10%” tab: locations alive for 10% of the run (e.g. 100/1000) are black, others in grayscale by age. Note that there are fewer black clusters.
Click the “+1000” button to run the simulation for another 1000 generations: note bigger clusters with the “At least 10%” view.
2: Second Run
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
This time, click the “T50” button below the list of variants, which reduces the list of variants to those that occurred at least 50 times in the original data.
Enter “1” in the “Draw Interval” box, and click the “+100” button to start the simulation and run it for 100 generations.
The simulation runs more slowly because it is calculating many variants at once: click on different variants at left to see on the map what each one is doing.
With the simulation paused at 100 generations, click the “Age” tab and look at different views for the variants. “Fade Newborn” and “At least 10%” views show clusters (10/100 for black in the latter view).
The simulation shows relatively uncommon variants filling the space of the map: any variant with sufficiently dense original locations will fill the space, while oncers or twicers disappear in the first generation.
Click “+100” again to run the simulation some more. Click the “Basic” tab, then the “Standard” tab to restore the display that shows all locations in black/white. You can explore the available options on the tabs above the map.
3: Third Run: Randomness
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Choose a single variant from the variant list and click the “Keep” button to process one variant.
Find and click the “More controls” link underneath the “Automaton settings” label at right. Find the “Random birth/survival” setting and enter “1” in each of the two boxes (for 1% randomness for birth and for survival).
Click “+1000” to run the simulation for 1000 generations. The display will update only every 100 generations. Note that only small clusters appear in the “Fade Newborn” and “At least 10%” views under the “Age” tab.
Click the “reset” button, and select the same variant you did before and click the “Keep” button so that the run processes only one variant.
This time click the “RND.03” button in the Controls at right, which inserts “.03” into the birth/survival settings for randomness. Click “+1000” to run the simulation for 1000 generations. The display will update only every 100 generations. Note that larger, and more stable, clusters appear in the “Fade Newborn” and “At least 10%” views under the “Age” tab. Too much randomness throws the simulation into a “chaotic” rather than a “complex” state; large stable clusters only form in the “complex” state, with randomness set to less than .06%.
4: Fourth Run: Rules
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Look at the bottom of the right side for the rule settings. The default is the “Bailey set” (2,3,4/5,6,7,8) button. There are other buttons for preset rules (GoL=Game of Life 3/2,3, R buttons use proportions and not counts), or you can use the sliders to select your own set of rules. For this simulation, please avoid the proportional rules (those work with full sets of variants).
Select one variant from the list at left, and click the Keep button.
Run the simulation for several hundred generations to see what happens.
5: Fifth Run: Regions
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation). Do not select a single variant; let the simulation run with all the variants.
Click the R6040 button in the rule settings. Run the simulation for 10 generations.
Click the Clusters tab, then the Regions button, to see the colored regions. To find out which variant belongs to each color, click the Context tab at right, and move the cursor over the different colors to update the context list.
If you want to try other rules that create regions, use the “reset” button to return the simulation to its original state. And click one of the other R rule buttons (eg 90/10, 80/20). Regions with fewer black dots, which indicate locations with multiple variants alive, are better.
6: Sixth Run: Social Factors
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Choose one variant from the list and click “Keep”.
Click on “more controls” under the label Automaton Settings at right.
Choose “KNN Score Rounded” from the Neighborhood Rules dropdown menu.
Click on one (or more, but best to start with one) of the social boxes like “Age” or “Educ” Fill in the ± value if you want (like “10” for Age), and change the weight to .66.
Run the simulation for 500 or 1000 generations (click OK when the similarity box comes up).
7: Seventh Run: Multiple Variants Display
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Click the T50 button under the variant list, which will clear all variants under 50 tokens. Run the simulation for at least 250 generations.
Click the Overview tab at right to see the frequencies of the variants on a chart. While in the Overview display, click on the one row of green and red numbers to bring up the table for all variants, and then scroll to the right to see how the numbers go up (green) or go down (red) . Observe that, once a variant has hit the 50% plateau, it no longer continues to go up all the time.
Use the “reset” button to return the simulation to its original state, and again click the T50 button under the variant list. This time, click the RND.03 button in the Controls section. Run the simulation for at least 500 generations.
Observe that, this time, many more variants survive and start to climb in frequency.
8: Eighth Run: Clusters
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Click on one variant, and click “Keep”. Run the simulation for at least 500 generations.
Click on the “Measurement” tab at the top, and then click on the “standard” button in the SAC section to calculate Moran’s I (a measure of spatial autocorrelation). You can also calculate it with the Fade Newborn and Age buttons.
Try the buttons in the “Clusters by Connectivity” section to count the number of clusters in the different views. Try counting clusters by Kernel with the button at top right.
9: Ninth Run: Persistence
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Select one of the variants and click the “Keep” button. Enter “10” in the “Draw Interval” box (at right under Play), and run the simulation for 1000 generations.
As the simulation runs, watch the continual movement of the matrix as locations constantly change from black to white and back again. You will see that small clusters often form and then disappear again.
As the simulation continues to run, watch for the emergence of clusters that remain stable and become larger over time.
When the simulation has reached 1000 generations, look at the patterns with the three views: the Standard button under the Basic tab, the Fade Newborn button under the Age tab, and the At Least 10% button under the Age tab. Can you see the clusters that are apparent under the Age views in the Basic|Standard view? Consider that you are actually seeing three different perspectives on speech: continual daily interaction (Standard), temporary clusters (Fade Newborn), and persistence that we interpret as regional variation (At Least 10%).
Click on the Overview tab at right. Scroll down and click on the Seed Statistics label. Scroll down the information there until you come to “Age profile - Individual ages”, then mouse over the bars for the different ages to see the number of cells with that age–as you move to the right, you can’t see bars anymore but you will still get figures for how many cells are that age. Age 1 is for cells alive for one generation only, Age 2 is for cells alive for two generations, and so forth. Looking at these numbers tells you something important about speech and time, about daily flux and persistence.
10: Final Run
Open the simulation, or use the “reset” button to return the simulation to its original state.
If you want to change the data set, use the drop-down menu above the map labeled “Question” to choose a different data set (g=grammar, l=lexicon, p=pronunciation).
Try out anything you like. Some trials would take longer than you want to stay: it will take over an hour to run an active memory simulation with only a few variants. To make things go faster, you can always choose a single variant and click the “Keep” button. Running the simulation with a large number of variants always takes a long time, but that is the only way to test memory and social parameters.