A Short Ethno Tutorial


Searching for the shortest series of real actions that I could use as an example of how Ethno works, I settled on something historical. Julius Caesar described his contribution to the expansion of the Roman empire in three words--veni, vidi, vici--which is a series of three actions: I came, I saw, I conquered. (Suetonius, in Lives of the Caesars , reports that "veni, vidi, vici" was inscribed on Caesar's Pontic Triumph.) I will ruin Caesar's parsimony now by analyzing those three actions extensively in order to illustrate the basic features of program Ethno.

Preliminaries

Starting up Ethno is done simply by clicking on the line, "The ESA Program," at the Internet site for Event Structure Analysis (URL: www.indiana.edu/~socpsy/ESA/ ), and then clicking the "Run Ethno" button on the next screen that appears. The program downloads to your computer, and begins running within your browser, assuming that you have Java enabled within your browser.

The opening screen is described in detail in the help message on defining sequences . For present purposes, it is enough to note that there is a space for typing names of actions.

Entering Data

It is important to enter actions time-ordered from earliest to last. The Ethno linking routine depends on this.

Julius Caesar says the first thing he did was veni: "I came". I typed that action and pressed the Enter key. The name box cleared, ready for another entry, and the action name "I came" became the first entry in the box at left titled Sequence.

I entered "I saw." Then I entered "I conquered." Thereupon the actions, "I came", "I saw", "I conquered" were the first three lines in the Sequence box.

At this point we can do two different kinds of tasks. We can choose Link actions from Ethno's Operations menu, and work toward a chart of the actions' prerequisite structure. Or we can use the Actions menu to provide more information on each action and then determine how the actions associate people. I chose to do the linking first.

Linking

The screen that appears after selecting Link actions from Ethno's Operations menu has a number of components that are described in the help message on linking. Bypassing the details, we see that Ethno is asking a question about the first two actions.

Does "I saw" require "I came" or a similar action?

Relational Questions

Ethno has to know how actions are related in order to build a diagram, and it depends on some expert to tell it. Ethno doesn't invent or induce action structures, it elicits them. Therefore, Ethno asks how each action relates to previous actions. Ethno can ask the question in several different ways, but the default relational question is:

Does the current-action require a specific prior-action or a similar action?

In other words, Ethno asks which prior actions are prerequisites for the last action entered, and in this way it taps our knowledge about the event in order to define the logical structure.

Ethno conducts complete and very efficient elicitations, asking every required question and never a needless question even in very complex systems with scores of actions. The program is smart enough not to ask if the last action is a prerequisite for any earlier action (that's why it's important to have the actions time-ordered). Once Ethno gets some knowledge it is even smarter, and it will not query about relations it can figure out on its own.

The phrase "or a similar action" is in the relational question for a good reason. Sometimes an action has substitute prerequisites, any one of which will do. We should answer "yes" to the relational question if the prior action can act as a prerequisite even if another action could serve as an alternative.

At least two kinds of knowledge get used in answering Ethno's queries. Sometimes we report what we know to be true of physical reality. Other times we report what has to be true within a given social reality. Events are structured at both levels. In fact, social constraints often get institutionalized and made into physical constraints (having to enter a Christian church in order to take communion, for example).

But back to the question Ethno is asking:

Does "I saw" require "I came" or a similar action?

The answer is "yes" because Caesar has to be someplace before he can see that place - a physical constraint. I answered by clicking the Yes button, and then the Next question button.

Ethno is anxious to show a picture and drew a chart even though the chart doesn't amount to much. Just a line from one dot down to another dot. This is a condensed version of the full-fledged chart we would see if we chose Chart from the Operations menu. The complete chart, a line going from "I came" down to "I saw", shows visually that Caesar seeing a new territory implies that Caesar comes to the territory. Note that lines connecting two actions can be verbalized in either of two ways. We can say the higher action is a prerequisite for the lower or we can say the lower action implies the higher.

After answering the first relational question I got another question:

Does "I conquered" require "I saw" or a similar action?

I might have answered yes by attending to the "similar action" phrase and thinking that seeing or getting detailed intelligence reports is essential to conquering. I also might have answered yes because the social reality of war in ancient times perhaps required Caesar to see a territory before he could conquer it.

However, I answered No to this question because that answer makes this first experience with Ethno more interesting. The "no" answer isn't totally eccentric: after all, contemporary generals usually don't see the territories they conquer.

After I clicked the Next question button, Ethno displayed another relational question:

Does "I conquered" require "I came" or a similar action?

Even though I told Ethno that seeing is not required for conquering, the prerequisites of seeing might be prerequisites of conquering, so Ethno asked the question. (Note: if I had said that seeing is a prerequisite for conquering, then Ethno would NOT have asked this question: the program understands that prerequisites of prerequisites are prerequisites once removed.)

Well, we're into the murkiness of social realities. I assumed a Roman commander does have to come to a territory in order to lead the battle and conquer, thus the answer was yes.

On getting this answer Ethno redrew the preview diagram as a dot ("I came") with lines down to two other dots ("I saw" and "I conquered"). This diagram shows that Caesar's expansion of the empire involves him in sightseeing and in conquering, either of which implies he visits the territory.

That's an Ethno model describing expansion of the Roman Empire! Don't laugh! That diagram is in the best tradition of sticking close to the data. It's based on Caesar's own report, and it is simple because Caesar said it was simple!

Saving Information

That completed the questioning. Ethno popped a little window on the screen saying I'd finished specifying links among actions, and the program would save the record of questions and answers, which had been written in a scrollbox at the bottom of the screen. This record night help me recall my reasoning as I linked actions.

It is a good idea to save the data at this point. I did so, allowing me to restore the model anytime.

Java programs running on the Internet cannot read or write files on your computer - a security measure to protect your computer from viruses. Ethno provides a substitute method of saving data from analyses. Select Import-Export from Ethno's Operations menu, and you will be taken to another Ethno screen. When you arrive, a box is filled with a record of all the information you have provided. Copy this text to a text processing file as described in the import-export help message.

Testing a Model

So far I had entered two different kinds of data. First, I entered an action series - a correct ordering of actions, though not necessarily the only permissible ordering. Second, I entered information about prerequisites - the information which Ethno processed into its diagram of action structure. Ethno implicitly adds some metatheoretical assumptions to turn the logical structure into a grammar for generating an action series.

Now we can run a test to see if the two kinds of data concerning the same actions are consistent with each other. That is, the questions now are: Does the logical structure (with additional assumptions) explain the sequence of actions? Do we now have a grammar of action that accounts for action order?

The test is conducted by selecting Chart from the Operations menu, which brings the Ethno diagram to the screen. We also get a new Fumctions menu which allows selection of Testing.

After I selected Testing mode, the diagram was almost but not quite the same as it first appeared in Inspecting mode. The name "I came" was printed on a red field. The red indicated that this action - the first one in the series - was "occurring".

Ethno had begun to work through the action series to see if the action structure used as a grammar could account for each action in sequence. There was no problem with the first action - there never is.

 Clicking anywhere on the chart causes Ethno to process the remaining actions, determining their various consequences for other actions in the system. I clicked, and no special messages appeared for the second action because there was no problem with the second action either. However, the diagram changed in appearance.

The action "I came" was colored aqua, meaning it happened and nothing yet had occurred to undo its consequences. The action "I Saw" was red indicating that this was the next action in the series. The action "I conquered" was yellow, indicating that it could be happening, now that Caesar had come.

Events occur automatically, so after "I saw" Ethno moves rapidly.

Interruption

Suddenly Ethno threw a dialog window up on the screen, telling me I had a problem and suggesting ways I might deal with it.

Let's hold off considering the logic of the problem and of the solutions for a minute and see what Ethno's doing first.

Ethno reports the kind of problem, in this case that we have an unfulfilled prerequisite. The other kind of problem that can arise is repeating an action without making any use of its consequences between times.

Ethno's solutions were listed in a scrollbox. Looking through them, I found that the first solution suggested changing the logical structure. Ethno's second solution was something about the way actions deplete one another. The third solution suggested that things didn't really happen the way I said they did, that the action series should be changed.

It's a good idea to examine all of Ethno's suggestions before settling on one. Ethno usually figures out several ways of solving a problem, and Ethno won't suggest a solution unless it really would work, so it's sensible to see what the possibilities are.

The dialog window can be dragged to view the chart and see the effects of the last action, which helps in understanding how the problem arose. Let's consider that now.

The action "I conquered" was highlighted with a black field indicating that it was the problematic action. The prior action, "I saw", was aqua indicating that it happened and its consequences still were unused. But "I came" no longer was aqua, just yellow indicating it could happen again! The problem was that Ethno supposed "I saw" used up "I came", and therefore Ethno thought "I came" had to happen again in order to prime its other consequence, "I conquered".

The Extra Assumptions

It's time to mention the extra assumptions which Ethno uses in order to turn a logical structure into an action grammar.

First, Ethno assumes that an action cannot occur until all of its prerequisites have occurred.

Second, Ethno assumes that an action depletes its prerequisites, uses up the conditions which the prerequisites created.

Third, Ethno assumes an action isn't repeated until the conditions it created are used up by some other action.

These assumptions along with the logical structuring of actions constrain the action sequences which can occur. For example, an action can't occur before its prerequisites occur: if that happened then Ethno would see it as a problem. Also, an action isn't supposed to repeat unless some consequences occur between times. So if an action repeats without intervening consequences, then Ethno considers that a problem.

The way we fix a problem is by changing the logical structure or by changing the extra assumptions or by changing the record of actions.

Structural Changes

Now let's go back to Ethno's suggestions and consider them more carefully.

The logic of Ethno's first suggestion - to change the logical structure - goes like this. If "I came" isn't a prerequisite for "I conquered" then it wouldn't make any difference that "I saw" depleted "I came" because Caesar's conquering would not require "I came". On the other hand, if "I came" isn't a prerequisite for "I saw", then Caesar's coming wouldn't be used up by "I saw", and "I came" still would be available to prime Caesar's conquering.

These changes in the structure were not attractive, but sometimes analysis of the series does convince us that we got the structure wrong and it should be changed.

Observational Errors

Ethno's second suggestion was the one I used so let's skip it for now and go to the third suggestion next. In the third suggestion Ethno reasoned that since "seeing" depleted Caesar's "coming", he'd have to come again in order to conquer. If we asked Ethno to implement this solution, then Ethno would insert a second occurrence of "I came" just before "I conquered" and then there would be no depletion problem.

Not a plausible solution in this case, but the approach does make sense sometimes. For example, William Corsaro and I have worked on models of peer culture using data reported in Bill's book, Friendship and Peer Culture in the Early Years (Norwood, NJ: Ablex Publishing Corp., 1985). At one point we ran into a problem and Ethno's solutions simply weren't acceptable - at least none were other than sticking in an extra action. Bill and I didn't feel comfortable just telling Ethno go ahead, stick an action in. But Bill had the original video tapes, and we checked them. And there in the right place was a child doing the expected action! The act had been passed over in transcription. Thus changes sometimes can make data better rather than worse.

But let's put it in the worst light. If we accept this kind of suggestion from Ethno then we actually would be fudging the data in order to make the data fit the model! This sounds horrible, totally unscientific. Actually, though, we are simply adding to qualitative analysis something it never had before, a means of dealing with observational error.

The general principle is this: if a model corresponds to general patterns in data but not to a specific datum, then the datum might be viewed as an error and changed to what is required in order to continue the analysis. Quantitative researchers have been using this idea all along - assuming that every measurement is fallible and that true values are defined by overall patterns in data as fitted to a specific model - so the idea is not new except in its application to qualitative data.

But you say, Wonder if someone else develops a superior model that doesn't require treating the action series as erroneous? That's a familiar risk: it is like a quantitative model being superceded by another one that explains more variance.

Qualitative research has a tradition of taking its empirical data seriously, even a bit too seriously since you can't develop a theory of error unless you allow that observations are fallible. Ethno allows questioning the transcription of actions but still keeps faith with the scrupulous qualitative tradition by requiring individual consideration of every supposed error. Obviously, the record of actions is changed only if the change is plausible.

Non-depleting Relations

Now let's return to Ethno's second suggestion of changing the assumption about actions always depleting their prerequisites. (The dialog comes up with the structural-change solution ready to implement. We get the second solution in implementable form by clicking a button labeled Go to next solution .)

The solution said, "The action 'I came' is not used up the action shown here (I saw)."

Yes, this was the answer, because Caesar's act of seeing did not magically transport him home. I clicked the button labeled Adopt this solution . (The action that does deplete "I came" would be "I left". I could have added that action to the dataset, but I decided to stick with what Caesar said.)

Ethno reset the analysis at the beginning and rapidly replayed each action to make sure the solution didn't mess something else up. It didn't. Ethno breezed through without interruption to the end of the action series.

Incidentally, if I had changed the model and included "I left", then I would have introduced a loop. Coming is required for leaving, and leaving is required for coming again. Loops are discussed as "commutative relations" in the help message about the chart .

Editing Diagrams

Let us try one other option in the Functions menu - Editing . When I chose this option, the diagram returned to its usual appearance without coloring.

I decided to change the structure. Perhaps Caesar meant "I saw" as a prerequisite for "I conquered". Then "I conquered" should not have a line up to "I came", instead it should be below "I saw" on the diagram.

I clicked on "I conquered", then on "I came". That caused the appearance of a dialog window in which I could click a button to delete the link between these two actions. The diagram was redrawn with "I conquered" an unconnected action.

Next I clicked on "I saw", then on "I conquered". Again the dialog appeared, this time allowing me to add a link between the two actions. Ethno redrew the diagram the way I wanted: "I came" at the top with a line down to "I saw", which in turn had a line down to "I conquered".

Event Composition

So far we've looked at how Ethno can help in understanding relations between actions. Ethno also can be used to examine the internal structure of actions. Here, too, the basic principle is that the analyst has the knowledge, and Ethno simply helps in seeing implications of that knowledge.

Every action has to be coded in a composition analysis. I used Ethno's Event menu to go to the first action, listed alphabetically - I came .

The screen for an action has multiple features that are described in detail in the help message on actions . Of immediate interest is an array of buttons and boxes at the right of the screen, presenting the eight categories of an Action Frame. Coding an action's composition involves specifying these elements, at least the ones that seem significant.

The Action Frame
Element Definition
Agent The instigator of a happening.
Act The verb that fuses action-frame elements into a happening.
Object The entity that is moved or changed, such that a repetition of the happening requires replacement. People can be objects.
Instrument An entity used by the agent to causally advance the happening while not being significantly changed by the happening. People, social organizations, and verbalizations can be instruments.
Alignment The specific place or time at which an instrument is applied to an object or in a setting.
Setting A convergence of relatable agents, objects, and instruments within a space-time boundary.
Product An entity that comes into existence as a result of a happening and that enables or disables subsequent actions.
Beneficiary The agent of an action that intentionally is enabled or disabled by the agent in the focal action.

The agent of "I came" is Caesar. To get Caesar listed as the agent, I clicked the radio button labeled "Agents", then went to the people menu to find Caesar. Of course, Caesar was not in the menu, but the first line of the menu allowed defining a New person . I selected that option, and Ethno opened up another kind of screen - an action-element screen, described in detail in the help message on associations. All that need be said about this screen right now is that it has a box for entering the name of something, plus a button labeled Return to action . I typed Caesar in the box, and clicked the button.

On returning to the action screen for "I came", the people menu did contain Caesar . I selected it, and Caesar appeared in the Agents box.

I created four kinds of non-human entities and one action as I continued coding "I came". Having populated the entities, people, and actions menus, it took only a minute to fill in the codings for the action by clicking radio buttons and making selections from the menus.

Agents : Caesar; Acts : transport; Objects : Caesar, army; Instruments : chariots; Alignments : dirt roads; Settings ; Europe; Products : Caesar, army; Beneficiaries : Caesar.

Composition analysis with the Action Frame typically requires specifying more detail than is given in the narrative. In this case, I set the act to "transport" to reflect my reading of what Caesar meant and in the process brought in an army and a geographic locale that Caesar didn't mention. (I brought in chariots and dirt roads, too, but those are glosses over my ignorance about how a Roman army moved.)

Most of my codings are straightforward, given definitions of the Action Frame categories. Maybe a bit mysterious is the assignment of Caesar and army to both the Object and Product categories. Caesar and his army are objects in that they are what gets transported. And a relocated Caesar and a relocated army are the products of the action. Caesar and army as products have extra properties - their new locations - over Caesar and army as objects. (Notes about such matters can be typed in the Comments box, and the comments are saved with the data.)

I coded "I saw" as follows:

Agents : Caesar, lieutenant; Acts : reconnoiter; Objects : defensive structures; Instruments : ?; Alignments : ?; Settings ; enemy territory; Products : Caesar; Beneficiaries : Caesar.

Lacking any idea of how intelligence was gathered in Caesar's time, I left instrument and alignment blank. The Product this time is an informed Caesar.

I coded "I conquered" as follows:

Agents : Caesar, lieutenant; Acts : subjugate; Objects : foreign leader; Instruments : Roman army; Alignments : enemy army; Settings ; enemy territory; Products : foreign leader; Beneficiaries : Roman leader.

The Product is a foreign leader converted from a principal into a patron.

Associations

Association analyses can be conducted after Action Frame codings are done for all actions. The results tell how a particular person or entity was associated with other people or entities through the actions described in the narrative.

I decided to check the results for Caesar. I selected Caesar from the people menu, which caused the screen for Caesar to appear.

I could have restricted tallies to actions in which Caesar appeared only in specified capacities - e.g., as the Agent, or as the Beneficiary. Clicking the button labeled Constrain associations would raise a dialog for making such specifications. However, I decided to forgo refinements in looking at this tiny dataset. By not using the dialog, I implicitly specified that I wanted to consider every action in which Caesar appeared, in any capacity.

I clicked the button labeled Compute associations , and got the following results.

Acts Entities People
1 reconnoiter
1 subjugate
1 transport
1 Europe
2 Roman army
1 chariots
1 defensive structures
1 dirt roads
1 enemy army
2 enemy territory
3 Caesar
1 Roman leader
1 foreign leader
2 lieutenant

Consider just the People column. All three actions related Caesar to himself in various capacities. One action connected Caesar to political leaders, both Roman and foreign. Two-thirds of the actions in this narrative linked Caesar to his military staff.

Caesar's veni-vidi-vici made his work seem simple to the point of purity, but I hear the clink of metal and the groans of men in battle in those words. I see the humiliation of defeated leaders and the supercilious satisfaction of leaders back in Rome. I see Caesar spending most of his time with his lieutenants and soldiers traveling hostile backroads of Europe. The association results from Ethno quantitatively materialize some of these nuances of my reading of Caesar's words.

Learning Ethno

Ethno is a lively program with striking visual displays, and you can appreciate the program best by seeing it in operation.

Practice is the best way to learn how to conduct Event Structure Analyses. Frankly, you have to commit a fair amount of time in order to learn all about Ethno. That's because qualitative analyses are time consuming, even with computer assistance. Rewards are forthcoming, though. You'll be prompted to think about questions that never occurred to you, and you'll gain new insights. You'll obtain models that allow you to explore incidents that may not have happened yet. And Ethno's diagrams and tables allow you to present your ideas to others in an objective manner.

 


URL: www.indiana.edu/~socpsy/ESA/Tutorial.html
1997, 2001, 2007 David Heise