Your first search - orthography

If you are a newcomer to APLS, this page will help you get acquainted with the Search page by walking you through a simple corpus search. After following along with the example search on this page, you can visit the other pages in this section to learn about the more advanced Search page features.

On this page
  1. Your first search
    1. Allow pop-ups
    2. Search results
  2. Search orthography, display word
  3. One word, the whole word, and nothing but the word

When you first load the Search page, there is a pattern input for the orthography layer. This is highlighted in red below. (We’ll break down the full layout of the Search page on the next page in this section.)

Let’s search for all instances of the word test in APLS.

  1. Go to the Search page.
  2. Enter test into the orthography input box.
  3. Press the Enter/Return key, or click the Search button.

Here’s what that should look like:

As you can see, a progress bar and Querying... have popped up below the pattern input.

Some searches will return results quickly, but more complex searches may take a longer time to complete. As long as the search progress bar is being displayed, it means that APLS is working on doing your search!

In a few seconds, the search should complete; the progress bar will fill up, with Found 13 results and a Display results link below the bar. Here’s what that should look like:

Allow pop-ups

At this point, you’ll either see a new browser tab appear, or you’ll get a message from your browser that a pop-up was blocked. If so, select the option to Always allow pop-ups and redirects for apls.pitt.edu, or something similar (the wording will be slightly different depending on your browser). Here’s what that looks like on Google Chrome:

On some browsers, choosing this option will automatically load the new tab. On others, you will have to manually click Display results below the progress bar.

Search results

The new browser tab that appears is the Search results page. We won’t get into all its functionalities here—that’s covered on the Exporting data documentation page—but it’s useful to go over the basics so you can understand the effect of different search patterns. Here’s what that should look like:

Hide outlines

You’ll probably see a different number just after threadId in the URL than what is shown in the screenshot. You’ll probably also see a different number just before milliseconds.

Let’s break down the things we need to know about:

  • The match list (highlighted in red) displays the search results.
    • As you can see, the match list includes 13 matches. This is consistent with the Found 13 results message that appeared below the progress bar on the Search page.

Each line in the match list consists of:

  • The before-match context (the first few highlighted in blue).
    • This displays the word immediately before the word(s) that match the search pattern.
  • The match link (the first few highlighted in green).
    • This displays the word(s) that match the search pattern.
  • The after-match context (the first few highlighted in purple).
    • This displays the word immediately after the word(s) that match the search pattern.

For this simple search, all of the match links display test. This makes sense—our search pattern asked APLS to find all words whose annotation on the orthography layer is test. This isn’t the case for all searches, however, as we’ll explore in later pages in this section!

In general, anytime you search APLS, you should check the match list. This can help you understand whether the phenomenon you wanted to search for is the same as the search you actually created.

To see how many times Steelers appears in APLS:

  1. Go to the Search page.
  2. Enter Steelers into the orthography input box.
  3. Click the Search button.

Search orthography, display word

When you first load up a transcript on the Transcript page, the main layer that you see is word. So why does the Search page select orthography by default, not word?

You might remember that when you’re reading the transcript, there are some bits of punctuation in participants’ turns: . for short pauses, - for long pauses, ? for questions or question intonation, and ___ for redactions. These punctuation marks belong to annotations on the word layer. For example, the last word annotation in this line (from transcript CB01interview3.eaf) is whatever ?, including the space and question mark.

While these punctuation marks are useful for understanding what’s going on in the transcript at a glance, they’re not very useful for searching the corpus. For example, if you want to find all instances of the word yinzer, you probably don’t want to restrict your search to instances where there is (or isn’t) a pause after the word or question intonation. (The second-person pronoun yinz is popularly imagined to be unique to Pittsburgh, so yinzer is an informal way to refer to Pittsburghers.) Another way to think of it is that yinzer is the same word regardless of whether or not it’s followed by a pause, so any search for yinzer should ignore this irrelevant fact.

You might’ve noticed that there was one punctuation mark we left out above: ~ for an incomplete or hesitated word. This is because ~ characters are in both word and orthography annotations. For the purpose of processing speech data, it’s often convenient to treat an unfinished word as distinct from its finished counterpart (e.g., bec~ is not the same word as because). As a result, APLS preserves ~ when it creates orthography annotations from word annotations.

So the orthography layer is the default for search because it:

  • strips out the punctuation marks ., -, and ~ from the word layer,
  • is always lowercase, and
  • has no annotation for redacted words (those whose annotation is ___).

To demonstrate what we mean, when you enter yinzer into the orthography input box on the Search page, here’s the Search results page that pops up:

As you can see, the match links in the match list don’t all look exactly the same: yinzer ?, yinzer, yinzer ., etc. That’s because regardless of which layer(s) you include in your search matrix, the match list will display matches (including before/after-match context) as they appear in the word layer.

Similarly, when you search for market on the orthography layer, the matches include both lowercase market and uppercase Market. (The uppercase Markets refer to the now-closed Market House in Cranberry Township, Downtown Pittsburgh’s Market Square, and other proper nouns.) Here’s what that looks like:

To confirm that the orthography layer is case-insensitive:

  1. Go to the Search page.
  2. Enter Steelers into the orthography input box.
  3. Click the Search button.
  4. Now do the same for sTeElErS. Do you get the same matches?

One word, the whole word, and nothing but the word

In the yinzer example above, you might have noticed that there were no matches for yinz or yinzers. This is because APLS matches the entire annotation—not more, not less—and orthography annotations are one word. To find yinz, enter yinz into the orthography input box; to find yinzers, enter yinzers. If you want to find yinz and yinzer and yinzers all in the same search, you need to use regular expressions.

Similarly, a common mistake that new APLS users make is to enter multiple words in a single input box. For example, you might think that you can enter a few in the orthography input box. If so, you will probably be surprised when this search returns 0 results (especially because a few is in the first line of CB01interview3.eaf)! However, there is no single word in any transcript whose orthography annotation is a few.

If you want to find multi-word phrases, you need to widen the search matrix, as explained in the Searching multiple words section of the Searching complex patterns documentation page.

Next page: Search page layout