Coder Mode is useful for researchers who have defined a fully structured coding scheme that they wish to apply to all the utterances in a transcript. To begin Coder Mode, you need to shift out of Editor Mode. To verify your current mode, just double- click on a file. Near the bottom of the text window is a line like this:
CLAN [E] [chat] sample.cha 1
The [E] entry indicates that you are in editor mode and the [chat] entry indicates that you are in CHAT Mode. In order to begin coding, you first want to set your cursor on the first utterance you want to code. You can use an file to do this. If the file already has %spa lines coded, you will be adding additional codes. If none are present yet, Coder’s Editor will be adding new %spa line. Once you have placed the cursor anywhere on the first line you want to code, you are ready to leave CHAT Mode and start using Coder Mode. To go into Coder Mode, type Esc-e. You will be asked to load a codes file. Just navigate to your library directory and select one of the demo codes files beginning with the word “code.” We will use codes1.cut for our example.
5.1.1 Entering Codes
Now the coding tier that appears at the top line of the codes1.cut file is shown at the bottom of the screen. In this case it is %spa:. You can either double-click this symbol or just hit the carriage return and the editor will insert the appropriate coding tier header (e.g. %spa), a colon and a tab on the line following the main line. Next it will display the codes at the top level of your coding scheme. In this case, they are $POS and $NEG. You can select one of these codes by using either the cursor keys, the plus and minus keys or a mouse click. If a code is selected, it will be highlighted. You can enter it by hitting the carriage return or double-clicking it. Next, we see the second level of the coding scheme.
To get a quick overview of your coding choices, type Esc-s several times in succession and you will see the various levels of your coding hierarchy. Then return back to the top level to make your first selection. When you are ready to select a top-level code, double-click on it with your mouse. Once you have selected a code on the top level of the hierarchy, the coder moves down to the next level and you repeat the process until that complete code is constructed. To test this out, try to construct the code $POS:COM:VE.
The coding scheme entered in codes1.cut is hierarchical, and you are expected to go through all the decisions in the hierarchy. However, if you do not wish to code lower levels, type Esc-c to signal that you have completed the current code. You may then enter any subsequent codes for the current tier.
Once you have entered all the codes for a particular tier, type Esc-c to signal that you are finished coding the current tier. You may then either highlight a different coding tier relevant to the same main line, or move on to code another main line. To move on to another main line, you may use the arrow keys to move the cursor or you may
automatically proceed to next main speaker tier by typing Control-t. Typing Control-t will move the cursor to the next main line, insert the highlighted dependent coding tier, and position you to select a code from the list of codes given. If you want to move to yet another line, skipping over a line, type Control-t again. Try out these various commands to see how they work.
If you want to code data for only one speaker, you can restrict the way in which the
Control-t feature works by using Esc-t to reset the set-next-tier-name function. For
example, you confine the operation of the coder to only the *CHI lines, by typing Esc-t and then entering CHI. You can only do this when you are ready to move on to the next line.
If you receive the message “Finish coding current tier” in response to a command (as, for example, when trying to change to editor mode), use Esc-c to extricate yourself from the coding process. At that point, you can reissue your original command. Here is a sum- mary of the commands for controlling the coding window.
Function
finish coding current code finish coding current tier
finish coding current tier and go to the next restrict coding to a particular speaker go on to the next speaker
show subcodes under cursor
5.1.2 Setting Up Your Codes File
When you are ready to begin serious coding, you will want to create your own codes file to replace our sample. When editing your codes file, make sure that you are in Text Mode and not CHAT Mode. You select Text Mode from the menu by deselecting (unchecking) CHAT Mode in the Mode menu. To make sure you are in Text Mode, look for [E][TEXT] in the bottom line of the Editor window. If you decide to use another editor or if you do not use Text Mode in CLAN, you will probably have problems.
You will probably find it useful to use the samples in the /coder folder in the /lib folder included with the CLAN program distribution. In the next paragraphs, we will explain the construction of the codes-basic.cut file in that folder. The first line of your codes-basic.cut file is:
\ +b50 +d +l1 +s1
The options on the main line were described in the previous section on editor options. In this example, the +b option sets the checkpoint buffer (that is, the interval at which the program will automatically back up the work you have done so far in that session). If you find the interval is too long or too short, you can adjust it by changing the value of b. The +d option tells the editor to keep a “.bak” backup of your original CHAT file. To turn off the backup option, use –d. The +l option reorders the presentation of the codes based on
their frequency of occurrence. There are three values of the +l option: 0 leave codes without frequency ordering
1 move most frequent code to the top 2 move codes up one level by frequency
If you use the +s option, the program assumes that all of the codes at a particular level have the same codes symmetrically nested within them. For example, consider the codes- basic.cut file: \ +l1 +s1 +b50 %spa: " $MOT :POS :Que :Res :NEG " $CHI
The spaces in this file must be spaces and not tabs. However, there must be a tab following the colon on the %spa: tier. The above file is a shorthand for the following complete listing of code types:
$MOT:POS:Que $MOT:POS:Res $MOT:NEG:Que $MOT:NEG:Res $CHI:POS:Que $CHI:POS:Res $CHI:NEG:Que $CHI:NEG:Res
It is not necessary to explicitly type out each of the eight combinations of codes. With the +s1 switch turned on, each code at a particular level is copied across the branches so that all of the siblings on a given level have the same set of offspring. A more extensive example of a file that uses this type of inheritance is the system for error coding given in the codeserr.cut file in the /lib/coder folder distributed with CLAN.
If not all codes at a given level occur within each of the codes at the next highest level, each individual combination must be spelled out explicitly and the +s option should not be used. The second line in the file should declare the name for your dependent tier. It should end with a tab, so that the tab is inserted automatically in the line you are con- structing. A single codes.cut file can include coding systems for many different dependent tiers with each system in order in the file and beginning with an identifier such as $spa:.
Setting up the codes.cut file properly is the trickiest part of Coder Mode. Once properly specified, however, it rarely requires modification. If you have problems getting the editor to work, chances are the problem is with your codes.cut file.