Adding more fields
The program in on the previous page only had one field, text. It's not very useful with just one field!
But it's still someone useful! If you had a lot more forms, you could make a spellchecker from it. The client program could go through every word in a document, ask "Is this word in my database?", and if yes, it leaves it alone, and if no it underlines it with a red squiggle.
In "finite-state automata" terms we'd call a program like this an "acceptor". It doesn't turn something into something else (that would be a ``transducer''), it can only say "yes that's one of the words I cover" or "nope not one of mine".
(Side note: You'll probably often hear us call fields 'tapes'. This is using the metaphors of a Turing Machine, picturing computer programs as reading and writing to long strips of tape. In the finite-state machine world, we often use this metaphor, speaking of our programs as "reading a symbol from the input tape" or "writing a symbol to the output tape". An acceptor is a "one-tape automaton", it can only read from one tape; a transducer is a "two-tape automaton" reading from one and writing to another.)
So let's add another field and turn this into a transducer.
From acceptor to transducer
Let's take the previous program and add another field, gloss. (By the way, if you don't know this word, "gloss" is what linguists call a labeled breakdown of the parts of a word, like 1-past-pend.)
We have to decide what our gloss is going to look like, so let's say jumps comes out as jump-3SG.PRES, jumped comes out as jump-PAST, etc. (But there's nothing special about these labels. I chose linguist-y labels, but you could use any labels you want. Gramble doesn't understand what what text or gloss or jump or 3SG.PRES mean.)
We add the gloss field to each table, associating each piece of text with the appropriate piece of gloss. (If you're following along in the interface, this is Example 2 in the Gramble menu.)
| Root = | text | gloss |
|---|---|---|
| call | call | |
| jump | jump | |
| Suffix = | text | gloss |
| s | -3SG.PRES | |
| ed | -PAST | |
| ing | -PRES.PROG | |
| Verb = | embed | embed |
| Root | Suffix |
(There's nothing special about that hyphen either, by the way. It's a character like any other, and I'm just putting it there to split up the gloss because we decided that was how glosses are going to be split up.)
Also, there's a shorthand we can use when a table like Root has two fields that are exactly the same, we can join them with a forward slash like so:
| Root = | text/gloss |
|---|---|
| call | |
| jump |
That means we don't have to write the same thing twice. (Writing things twice can be bad because the two version sometimes go out-of-sync with each other. This makes sure they're always the same.) After all, not-writing-things-a-bunch-of-times is the whole reason we're writing a program!
Trying out a transduction
Let's put that into your sheet. Then hit "Sync & Validate" in the sidebar, which makes the sidebar interface update to use your latest changes.
<YourSheetName>.Verb should still be selected in the dropdown menu, but if not, select it again. Now if you sample you'll see there are two fields in the results. Make sure you're generating all six forms, and that the correct gloss is matching up with the correct text!
Now let's look at another part of that interface that we haven't used yet. Below <YourSheetName>.Verb but before the buttons, you can see text input areas labeled with Verb's possible field names. Anything you type into these restricts the generation/sampling -- or put another way, it lets you make a database query.
So type called into the text field. Now it you sample or generate, the only possible result is {text:called, gloss:call-PAST}. We've done a transduction!