Exporting data
After you’ve explored APLS, conducted a search, and found data you’re interested in, the next thing you’ll probably want to do is export that data out of the corpus for further analysis.
This page is meant to help you understand what you can export and how you can export it after you’ve conducted a search or have a search results .csv file. If you’re unfamiliar with searching APLS, you should start with the Searching the corpus documentation pages and return here after you’ve completed a search.
On this page
- APLS pages that can export data
- What you can do on the Search results page
- Viewing metadata on your search
- Browsing matches
- Viewing matches in-context in their transcript
- Exporting matches
- Exporting matches to
.csv - Exporting utterances containing matches
- Exporting utterance audio files containing matches
- Exporting a list of matching words or annotations
APLS pages that can export data
This documentation page focuses on the Search results page, but there are other pages in APLS that let you export data as well. These pages and the data that they export are:
- Participants page
- Lets you export participant attributes
- Transcripts page
- Lets you export transcript attributes, formatted transcripts, and audio files
- Individual Transcript pages
- Lets you export formatted transcripts and individual utterances as Praat TextGrids
- Process with Praat page
- Lets you extract acoustic data from transcripts using Praat
- Extract data based on search results page and Extract annotations based on time intervals pages
- Let you upload a search results
.csvfile to extract additional information about your search result matches
- Let you upload a search results
What you can do on the Search results page
The Search results page allows you to:
- View metadata on your search
- Browse matches
- View matches in-context in their transcript
- Export matches to
.csv - Export utterances containing matches
- Export utterance audio files containing matches
- Export a list of matching words or annotations
Page layout
Here’s what the Search results page looks like when you search the orthography layer for hobbies:

- The match list (highlighted in red above). This is where you can…
- The results header (highlighted in blue above). This is where you can…
- The export menu (highlighted in green above). This is where you can…
You’ll probably see a different number just after
threadIdin the URL than what is shown in the screenshot. You’ll probably also see a different number just beforemillisecondsin the header.
Accessing the Search results page
The main way to reach the Search results page is to create a search on the Search page. A new tab with the Search results page should pop up once the search completes (as long as there’s 1+ matches to display).
If a new tab doesn’t pop up, then either:
- The search is still in progress.
- The search stopped because it was cancelled or yielded an error.
- The search returned 0 matches.
- Your browser blocked the pop-up.
- The No matches, only a count of results display option checkbox is selected on the Search page.
You can also reach the Search results page by uploading a previously-downloaded results .csv file to the Extract data based on search results page.
The Search results page can’t be permalinked
Unlike all the other pages in APLS, you can’t permalink Search results pages. For example, the URL in the screenshot above, https://apls.pitt.edu/labbcat/matches?threadId=1351&wordsContext=1, won’t work if you try to load it in APLS. Instead, you’ll get the error message Invalid ID: 1351 and a blank page:

This is because APLS is not designed to hold onto a search task (information about search and its matches) forever. If APLS held onto every search task users ever completed, this would strain the server’s processing resources, storage space, or both! As a result, APLS releases search tasks when users are done with them.
As long as you have a browser tab open that has a search task (on the Search or Search results page) and you’re connected to the internet, APLS won’t release your search task. If you have multiple Search/Search results tabs open, APLS will hold onto all those search tasks. However, if you navigate away from the Search and Search results pages, APLS will release your search task(s). Likewise, if you lose internet access (including if your computer goes to sleep), APLS will release your search task(s).
- Log into APLS
- Load the Search page
- Create a search
- Put your computer to sleep
- Go for a walk
- Wake your computer up and click Display results at the bottom of the Search page
You should get the error message
Invalid IDand a blank page.
Normally, this isn’t a big deal—you’ll often create several searches to explore before settling on the one you want, and rarely will it be necessary to hang onto all those searches. However, if you’re worried about losing access to your matches, export them to a .csv file. You can always re-upload the file to the Extract data based on search results page to basically restart the Search results page where you left off. Similarly, if you want to share matches with a colleague, share the exported .csv file, not a Search results page URL.
Coming soon to APLS is a feature created by APLS admin Dan Villarreal: search history
. This adds a new History tab to the Search page, which keeps a record of previous searches in the same browser tab:
Search histories can be exported to JSON for reproducibility, debugging, and/or performance benchmarking; these JSON files can then be re-uploaded so users can repeat the search:
This feature was integrated into LaBB-CAT in October 2026 as part of LaBB-CAT version 20261012. It will be added to APLS in upcoming version 0.6.0.
Viewing metadata on your search
The results header contains metadata on your search. For example, here’s what the header looks like when you search the orthography layer for hobbies:

Let’s break down what we’re looking at:
orthography=hobbiesis the search name, a shorthand representation of your search pattern.- The search name in this example,
orthography=hobbies, is pretty straightforward because this was a pretty straightforward search pattern, but search names can get more interesting if your search pattern is more complicated.
- The search name in this example,
Found 8 results in 4485 milliseconds (Total utterance duration: 0:29.003)contains 3 pieces of metadata:Found <number of results> results in <how long it took search to run> milliseconds (Total utterance duration: <total utterance duration>).- If you run this search on your own, it probably won’t take exactly 4485 milliseconds—but it probably won’t be much quicker or slower, either!
- More complicated searches take longer to run. All searches take longer to run if there’s a lot of people using APLS all at once, or if APLS admins happen to be adding a feature or doing maintenance while you’re using APLS.
- The total utterance duration is what it sounds like: it’s calculated by taking the durations of all the utterances that contain all the matches, then summing them. (Each utterance is only counted once, even if it contains multiple matches.)
Browsing matches
You can browse matches in the match list, which has a few parts:

As you can see, the match list displays each match (here, each instance of the word hobbies) on its own line. Each line consists of:
- A match checkbox (highlighted in red).
- This allows you to select which matches you want to include in your exported
.csv, utterances, or utterance audio files. By default, all are selected when you load the Search results page.
- This allows you to select which matches you want to include in your exported
- 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.
In a little bit, we’ll also discuss:
- The context selector (highlighted in orange)
- The “more matches” buttons (highlighted in brown)
The match list is broken up by transcript and participant; here, you can see that there’s 1 match in CB01interview2.eaf from speaker Barbara Johnstone, 2 in CB10interview2.eaf from Barbara Johnstone, etc. The match list is sorted first by participant, then by transcript, in alphanumeric order. As you can see, this means that matches from Barbara Johnstone are displayed first, then Interviewer HD, and finally LV11. This also means that Barbara Johnstone’s match from LV11interview1.eaf is separate from LV11’s match in the same transcript.
Barbara Johnstone is the first participant alphabetically in APLS, so if your search matches any of her data, she will always appear at the top of the match list. Similarly, Interviewer HD will always appear above any interviewees from Lawrenceville (whose speaker codes always start with
LV). If you want to exclude interviewers’ matches from a search, use either the participant filter or the Only include matches from the main participant in a transcript match option on the Search page.
You might have noticed that the match links don’t all look the same for this search. Some are hobbies ? with a question mark, some are just hobbies. That’s because regardless of which layer(s) you include in your search matrix, the match list will display matches (and context) as they appear in the word layer. The word layer includes punctuation that is useful for when you’re reading the transcript, such as . for short pauses and ? for questions or question intonation. However, these bits of punctuation are not useful for searching, so the orthography layer strips them out. In short, the difference between hobbies ? and hobbies is just on the surface—all these words share in common that their orthography annotation is hobbies.
Matches larger than one word
If your search matrix is wider than one word, the matches will be wider than one word, as well. For example, here’s what it looks like when you search the orthography layer for not followed immediately by many:

As you can see, the match links are all 2 words wide.
This principle also applies to the followed up to 1/2 words later by options. Here’s what it looks like when you search the orthography layer for not followed up to 2 words later by many:

As you can see, the match links are 2 (not many), 3 (not too many, not very many), or 4 (not m~ very many) words wide. This reflects how wide the matches themselves are.
Matches smaller than one word
The Search results page displays match links as complete words, even if the matches themselves are smaller than one word. This applies to searching segments within a word or searching layers with both word scope and sub-interval alignment. For example, here’s what it looks like when you search the segment layer for f:

As we would expect, there is one line in the match list for each match, even if there are multiple matches in the same word. We can see this in matches #6 & 7 above; both are in the same instance of the word fifteenth, which contains 2 /f/s. Although these lines of the match list look identical, APLS is keeping track of the fact that #6 refers to the first /f/ and #7 to the second.
As mentioned above, the match list displays matches and context as they appear on the word layer. This is why match links are always complete words, even if the matches themselves aren’t.
Viewing more matches
When you first load the Search results page, the match list will show up to 20 matches. As you can see in the previous screenshot, there are 17777 f annotations on the segment layer, but only 20 are showing at first.
Clicking đź”˝ 20 More Matches will extend the match list:

You can load all remaining matches at once by clicking the ⏬ Remaining Matches button.
We don’t actually recommend clicking ⏬ Remaining Matches for this particular search, since it returns a very large number of matches. Not only does the page take a while to reload, it may actually be too big for your browser to display correctly.
Context selector
When you first load the Search results page, the before-match context and after-match context consist of up to 1 word apiece.
We say “up to” because APLS will only show context that is in the same turn as the match. For example, in the search for f segments, match #9 has no before-match context because for is the first word in its turn:

The context selector allows you to display more (or less) context: 0 words, up to 1/5/10/20 words, or the entire utterance. For example, here’s the same match list with up to 10 words selected:

Because a turn can contain multiple utterances, sometimes the larger up to N words options will actually display more words than the entire utterance. For a refresher on the turn and utterance layers, read this section of “Viewing a transcript”.
The context selector also affects .csv export. The Before Match and After Match columns contain whatever context is displayed on the Search results page when you click CSV Export (or would be displayed if all matches were visible).
Viewing matches in-context in their transcript
You can view a match in-context in its transcript (specifically, on the Transcript page) by clicking its match link. Here’s what it looks like when you search the segment layer for f, then click the 4th match link (informal):

As you can see, the word informal is highlighted in yellow. Highlighted in green are other words that contain matches for the same search are highlighted in green: effort, for, fifteenth. While fifteenth appeared twice in the match list because it has two f segments, there’s only one word to highlight on the Transcript page.
In addition, since this search involved the segment layer, the Transcript page loaded with segment pre-selected for display.
Exporting matches
As mentioned above, the Search results page gives you several ways to export matches. Each of these will be discussed below, but they have a few things in common:
- You can click the CSV/Utterance/Audio/Dictionary Export button to download matches using APLS’s default options.
- You can click the small ▼ button next to any of these buttons to open that export’s options menu.
- For all but Dictionary Export, you can select which matches to download (or not) by selecting (or deselecting) match checkboxes in the match list.
- When you load the Search results page, it defaults to having all matches selected for export (even those that aren’t showing). To select individual results, deselect the Select all results checkbox—but note that in this case, only matches that are selected and showing will be included in the export.
- CSV Export and Dictionary Export will always download a single file.
- Utterance Export and Audio Export will usually download a
.zipfolder containing multiple files.- If there’s only one file to export, Utterance Export and Audio Export will download that file on its own (not in a
.zipfolder).
- If there’s only one file to export, Utterance Export and Audio Export will download that file on its own (not in a
Exporting matches to .csv
Clicking the CSV Export button will download a .csv file with one row per match. This file will include all the matches whose match checkboxes you’ve selected.
.csvstands for “comma-separated values”. It’s a standard file format for data that’s organized into rows and columns. You can use any standard spreadsheet program (e.g., Microsoft Excel, Google Sheets, macOS Numbers) to open, edit, and save.csvfiles. It’s very likely that your operating system already has a default program for opening.csvfiles.
Certain information about the matches will be included in the .csv by default. In the options menu (that is, by clicking the â–Ľ button next to CSV Export), you can select what data you would like included in the exported .csv file. You can add metadata, participant and transcript attributes, and layer annotations by clicking the checkbox for the data you would like to include.
The Fields column is unique to the CSV Export layer picker, and it allows you to include information about your search and the relative location for where your search matches were found. Database title, APLS version, Search name, Match ID, and URL are always included in the exported .csv file and cannot be toggled off.
The participant attributes, transcript attributes, and layers are described on their respective field guide pages. As with layer pickers on other pages, you can also hover over the names of each checkbox to make a brief tooltip description appear.
Multiple peers
Layers that allow horizontal peers or vertical peers will also display numeric boxes when you select them.
This box allows you to choose the number of peers to export, and how many columns they should be in. Again, horizontal peers divide up a timespan (such as multiple syllables in a word), while vertical peers occupy an identical timespan (such as multiple possible lemmas for a word).
This layer picker hides layer icons by default to save space, but you can click Show layer icons to toggle them on. These icons have the same meaning as other APLS pages. Layers that allow horizontal peers have the
icon; layers that allow vertical peers have the
icon.
By default, this box displays all. This means that all annotations will be exported in a single column for that layer, with peers separated by a space.
- Go to the Search page.
- Enter
readinto the orthography input field.- Click the Search button and wait for the Search results page to appear.
- In the CSV Export options menu, select morphemes.
- Click the CSV Export button.
- Open the
.csvfile when it finishes downloading.The
morphemescolumn of the.csvwill containreadIRR readin each row. This indicates that each match has two vertical-peer morphemes annotations:readIRRandread
If you enter an actual number into the box, APLS will export that many peers, each in their own column. For example, if you enter 3 for morphemes, the exported .csv will have a morphemes 1 column with each match’s 1st morphemes vertical peer, a morphemes 2 column with each match’s 2nd, and morphemes 3 with the 3rd. If there are fewer peers in the layer for the match than the number in the numeric box, then the .csv will still contain the number of columns you specified but the columns will be empty.
- Return to the Search results page from the previous try-it.
- If you haven’t already, select morphemes in the CSV Export options menu.
- Enter
3into the numeric field for morphemes.- Click the CSV Export button.
- Open the
.csvfile when it finishes downloading.There are entries in the
morphemes 1andmorphemes 2columns of the.csv, butmorphemes 3is empty. This is because the morphemes layer only has two vertical peers for each instance of the word read.
For layers that allow horizontal peers, if you’ve entered a number into the box, APLS will export 3 columns per peer: the annotation’s label, its start time, and its end time. These columns are named (for example) part_of_speech 1, part_of_speech 1 start, and part_of_speech 1 end.
- Go to the Search page.
- Enter
wouldn'tinto the orthography input field.- Click the Search button and wait for the Search results page to appear.
- In the CSV Export options menu, select part_of_speech.
- Enter
2into the numeric field for part_of_speech.- Click CSV Export.
- Open the
.csvfile when it finishes downloading.You should see 6 columns corresponding to the part_of_speech layer:
part_of_speech 1,part_of_speech 1 start,part_of_speech 1 end,part_of_speech 2,part_of_speech 2 start, andpart_of_speech 2 end.
Layers that allow horizontal peers will also display unlocked lock icons () on the sides of the annotation count selector. If you click the lock icon (which changes it to
), APLS will export the annotation(s) for that layer to the start or end of the targeted annotation. You can use this to, for example, only export the last syllables annotation in each of the words matching your search pattern.
The lock icons will restrict the exported data to only the peer at the start or end of the targeted annotation.
This means that regardless of the number of annotations specified in the numeric box, using either of the lock icons will cause that layer to only export data for one peer. Selecting both lock icons will cause that layer to export data only for annotations that do not have any horizontal peers.
Exporting utterances containing matches
Clicking the Utterance Export button will download all the utterances that contain the matches whose match checkboxes you’ve selected.
You can export utterances in a variety of file formats, which all differ slightly in terms of what information they do or don’t represent. The default file format is Praat TextGrids (denoted by the icon
). TextGrids can represent most of the information that you’ll likely want to know, and they are widely used by linguists. You can choose a different file format in the options menu (that is, by clicking the ▼ button next to Utterance Export).
The way that APLS formats exported transcripts and utterances is the same regardless of whether you’re exporting them from the Transcript, Transcripts, or Search results pages. For more information about what exported transcripts look like and how layers are represented in different file formats, see the “Viewing a transcript” documentation page.
Here’s what it looks like when you search the segment layer for f, select match checkbox #1 & #2, click Utterance Export, and open the downloaded TextGrid in Praat:

As you can see, there are 4 tiers: Barbara Johnstone (which corresponds to the utterance layer), word - Barbara Johnstone (corresponds to word), segment - Barbara Johnstone (segment), and target - Barbara Johnstone (discussed below). By default, APLS will export the utterance and word layers, plus whichever layer(s) were included in your search pattern (in this case, segment). In the options menu, you can select which layer(s) you would like to include in the exported utterances.
You can select layers with phrase, word, or segment scope, but not span layers. This is because span layers’ timespans can extend beyond a single utterance. If a span layer’s annotation was included in the exported file, this might give the mistaken impression that it began or ended at the utterance’s start or end timepoint.
By default, APLS exports one file per utterance. If multiple matches come from the same utterance, you’ll have fewer files than matches. We saw this in the previous screenshot, which contained two f segments (in information and informal), both of which were in the same file. This is not the case if you select Add prefixes to file names.
Add prefixes to file names
Let’s say you want to run some automated process that involves comparing data in an exported .csv file to exported utterances. In that case, it might be convenient to have an easy way to match rows in the .csv to transcript-fragment files. This is the purpose of the Add prefixes to file names checkbox, which is deselected by default).
We saw in the previous section what happens when you search the segment layer for f, select match checkbox #1 & #2, and click Utterance Export (without clicking Add prefixes to file names). APLS exported 1 utterance: CB01interview1__4.359-12.529.TextGrid. However, if we click Add prefixes to file names then Utterance Export, we get a .zip folder containing 2 files: 00001-CB01interview1__4.359-12.529.TextGrid and 00002-CB01interview1__4.359-12.529.TextGrid. Here’s what the first one looks like when you open it in Praat:

This TextGrid is almost identical to the previous screenshot, except for the target - Barbara Johnstone tier. The previous TextGrid had one small target - Barbara Johnstone interval in information and another in informal, corresponding to matches #1 & #2. However, this TextGrid (with the 00001- prefix) is only for match #1, so it doesn’t have the small target - Barbara Johnstone interval in informal.
APLS will “zero-pad” the match’s result number based on the number of matches returned by that search. For example, the prefixes in this particular example have 4 leading
0s because the search turned up a 5-digit number of matches ().
Finally, the Audio Export option menu has an Add prefixes to file names checkbox, too. These checkboxes are linked, since if you want prefixes for one type of export, you probably want them for the other one, too. For example, if you select the Add prefixes to file names checkbox in the Utterance Export option menu, APLS will automatically select that checkbox for Audio Export (even if you don’t have that option menu open).
Include “target” tier
If the annotations targeted by your search pattern are short (like individual segments), finding them visually in an exported utterance can be like reading a Where’s Waldo? book. This is why the Include “target” tier checkbox exists. When this checkbox is selected, exported TextGrids include an additional target tier, in which targeted annotations are marked with an interval that is labeled with the match’s result number. (Like file-name prefixes, result numbers are zero-padded so they’re all the same number of digits.)
We’ve already seen these intervals in the previous two screenshots. Here’s what you’ll see if you zoom into the interval marked information on the word - Barbara Johnstone tier:

As you can see, the f in information is marked 00001, meaning it’s the targeted annotation within match #1.
Include “target” tier is selected by default when you load the Search results page. However, this option is currently only available for exporting utterances to the default Praat TextGrid format.
Exporting utterance audio files containing matches
Clicking the Audio Export button will download .wav audio files for the utterance containing the matches you’ve selected. If there are multiple .wav files to export, APLS will package them in a .zip folder.
There is only one item in the Audio Export option menu: an Add prefixes to file names checkbox. This works identically to the same checkbox in the Utterance Export option menu. Instead of exporting file names like CB01interview1__4.359-12.529.wav, with one file per utterance, selecting this checkbox exports file names like 00001-CB01interview1__4.359-12.529.wav, with one file per (selected) match. As mentioned above, these checkboxes are linked; if you (de)select one, APLS will automatically (de)select the other.
Exporting a list of matching words or annotations
Clicking the Dictionary Export button will download a .txt (plain text) file containing a word list that is based on the search results, with the option of adding annotations from another layer. (You can open .txt files in any standard text editor, like Windows Notepad or macOS TextEdit.) The word list consists of every unique orthography annotation included in every match. For example, let’s revisit our earlier search for hobbies on the orthography layer. Here’s what happens when you click Dictionary Export, and open the downloaded file in Notepad:

As you can see, the file consists of a single line, hobbies, since this is the only unique orthography annotation returned by our search.
To find all unique words that follow the word not in APLS:
- Go to the Search page.
- Enter
notinto the orthography input box.- Click the large
button on the right side of the pattern input to make the search one word wider.
- Enter
.*into the second orthography input box.- Click the Search button and wait for the Search results page to appear.
- Click the Dictionary Export button and download the
.txtto view the unique matches for the search.
The default for Dictionary Export is the word list by itself. However, in the Dictionary Export options menu, you can choose to export a list of all the unique combinations of annotations (in both orthography and another layer) that exist within your search matches. Here’s what it looks like when you search orthography for a|the, click the ▼ button next to Dictionary Export, select phonemes from the drop-down, click Dictionary Export, and open the downloaded file in Notepad:

As you can see, there are 5 lines, each consisting of an orthography annotation and a phonemes annotation (in the DISC phonemic alphabet). From this file, we can conclude that there are 3 different phonemes annotations for the word a in APLS and 2 for the. These are the only combinations of orthography and phonemes matched by this search; for example, there aren’t any instances of the that have 1 (IPA /eɪ/) for their phonemes annotations.
The layers that are available for Dictionary Export are those that have word scope and complete interval alignment.
This example also illustrates one way a word list might be useful: for checking whether you need to refine your search pattern. The reason a has Q (IPA /ɑ/) as a possible pronunciation in APLS is phrases like a capella or a la mode. This suggests that if you’re only interested in a as an indefinite article (as in the phrase a boy), you might want to add a part_of_speech condition to your search pattern.
These layer options will include the information that is present for that layer. This means that if there is not any information in the selected layer for your search matches, then the Dictionary Export .txt will not include that match in the file.
To find words that start with cranberry and have a custom DISC pronounciation on the pronounce layer:
- Go to the Search page.
- Enter
cranberry.*into the orthography input field.- Click the Search button and wait for the Search results page to appear.
- Click the â–Ľ button next to Dictionary Export and select pronounce.
- Click the Dictionary Export button and download the
.txtto view the unique matches for the search that have information on the pronounce layer.
- The
.txtfile should not have entries for words like “cranberry” or “cranberry’s” because those words don’t have a custom DISC pronounciation in APLS.
Finally, unlike the other export options on the Search results page, Dictionary Export is not affected by which match checkboxes are selected or deselected.

