Buttons with image map

                                             WWW www.wordcorr.org                                    Home


Wordcorr Home > Linguist > How to use
There are four main kinds of things you do in Wordcorr. Usually you jump back and forth among them; sometimes you do them in order. Click to read more about each one. The Wordcorr Help facility goes into much greater detail; this Web site just gives you an overview.

Enter and edit data. You can enter data manually. There is a convenient way to enter full IPA (International Phonetic Alphabet) notation. You enter multiple word lists for comparative analysis into a data collection.

You can also import data from another computer, via

  • Wordcorr export file
  • spreadsheet
  • WordSurv or PalmSurv.

Annotate the data to show your linguistic judgments for a particular view. You look at all the forms for one entry in your current view. If they all look as if they could have developed from a single original form even though they're not all identical (in other words, for now you consider them all as possible cognates), you give them all the same one-letter tag, like "a". If it looks like two earlier forms must be involved (i.e. two possible cognate sets), you tag some of them with an "a" and others with a different letter like "b" and so on. Use any letters you like; you'll get the same effect with "c" and "r". Different tags just mean different cognate sets.

Then you look at each tag group in that entry, all the a's together, then all the b's, or whatever you've used for tags. If the corresponding sounds don't match up, you stretch the shorter words by putting in a special symbol "/" called "Indel"; we'll explain it later. You can also make some symbol sequences act as if they were units; they're called "grapheme clusters" in Wordcorr.

Tabulate the annotated data to turn an annotated tag group into correspondence sets, one for each position in the group. When everything is lined up so it shows the regularities you've noticed, Wordcorr produces the correspondence sets that your annotations imply. You decide tentatively (Wordcorr won't be miffed if you change your mind on things as you go) what protosegment of the parent language each correspondence set might be evidence for. You also add a brief code to show where the set was found in relation to the sounds around it.

Refine the results. It's normal if many of your initial guesses are not quite on target, because it will be a while before you figure out what the target is. But as you tabulate more and more entries, you begin to see patterns emerge that point to regular sound changes. You move correspondence sets around to make the evidence more clear, and you revisit the judgments you made on your earlier annotations. Refining gives you a means of backing out of lines of thought that made sense when you started, but with more data you see they won't work.

For more remarks about how to use Wordcorr effectively, click here.

These are TASKS, not STEPS.

If you entered all your data as Step 1, then went back and annotated everything as Step 2, then tabulated it all as Step 3, then refined everything as Step 4, you'd bore yourself to death and make some confused choices because you wouldn't have the broad picture.

A better way to begin is to

  • enter the data for just one entry
  • annotate that entry.
  • when you have the group tags assigned and the alignment done, tabulate the groups that have enough varieties in them to meet the threshold for your view
  • click over to the Refine panel and immediately make any changes that suggest themselves.

Then do the same for the next entry.

At first it may seem silly to work through all four tasks for every entry. But as you go, you'll see howlinking everything together keeps you connected with all the data, and with all the analysis at the same time.

If you do it that way,

  • you won't be bored;
  • you'll be swimming in all the data;
  • you'll know where everything is;
  • you'll see patterns emerge;
  • you'll spot anomalies immediately.

Try it; you'll like it.

Up                               Site map                                              Home