Plaid IGT Guide

Interlinear glossed text is organized at three scopes, and the annotation fields and the Analyze grid are built around them:

A word is divided into one or more morphemes. A simple word is a single morpheme, while a word like English cats is two, cat and -s. Every field in a project belongs to exactly one scope, which decides which row of the grid it appears on.

The Documents tab lists a project’s texts with their word counts, when each was last changed by anyone, and when you last changed it yourself. Click a column heading to sort by it, and again to reverse it. Sorting by Your last edit is the quickest way back to what you were working on. A document you have never edited shows a dash. Once there are many, the list can be searched and paged. How a list is sorted is remembered for next time, and for the document list so is the page you were on, both per project. New document creates an empty document and opens it on its Baseline tab. A document is renamed, copied, and deleted from its own Details tab.

Each document is built up through a handful of activities, one per tab. The rest of this chapter walks through them in the order you would normally use them.

The Details tab holds the document’s name and its document-level fields, such as speaker, date, and notes. Each part has its own Save. Which fields appear is configured per project (see Annotation), so a project records exactly the information its corpus needs. A metadata field can be held to a tagset, for example so that a genre is always one of a fixed set (see Tagsets).

Copy makes a second document in the same project under a name you give it, holding the same text, annotations, vocabulary links, and recording. Comments are not copied.

Text direction says which way the document’s data is laid out. It is set to Automatic, which reads the baseline text: a document written in Arabic, Hebrew, Thaana, N’Ko or any other right-to-left script lays its words out right to left in the Analyze grid, with the row labels on the right. Set it by hand for a document the text cannot speak for, such as a right-to-left language written here in a Latin transliteration.

Direction affects the way words are laid out, not the values in them. Every field decides for itself, so a gloss written in English reads left to right in a column standing under an Arabic word.

The Baseline tab holds the raw text you are analyzing, in whatever orthography you work in. Paste or type it, one sentence per line, and save. On the first save each line becomes a sentence. You can move sentence breaks afterwards on the Tokenize tab.

You can edit the baseline later too. Each change is saved where you typed it. Existing sentences, words, and annotations are kept and adjusted to fit the new text. A change inside a word, or touching it with no space between, changes that word, and it keeps its analysis. A space typed inside a word leaves one word with a space in it. Split it on the Tokenize tab to make two words. On a project whose words are shared with an app that splits a word at a typed space, a space typed inside a word splits it there: the word and its analysis stay on the part holding most of its letters, and the rest is untokenized text. Deleting the space between two words keeps them two words. Merge them on the Tokenize tab to make one. New text with a space between it and every word is in no word. The Tokenize tab makes words of it. A word goes, along with its annotations, only when all its text is deleted. Text typed right before a sentence’s first character joins that sentence, unless it holds a line break: a line added in front of a sentence joins the sentence before, so splitting it off leaves the sentence its translation. A line typed in front of the first sentence becomes a sentence of its own. Spaces typed there alone join the first sentence. Text typed right after a sentence’s last character joins that one. Move a sentence break on the Tokenize tab.

When someone else saved the baseline after you pressed Edit text, Save changes puts your changes onto their text. When you both changed the same passage, the save is refused, Changed elsewhere in the same passage. shows under the box, and your text stays in the box. Cancel then shows the saved text, and you make the edit again. The save is also refused when the text has passages that read the same and your change could be in either of them.

A project may define additional orthographies, alternate representations of each word such as a phonetic transcription or a practical spelling. These appear as their own rows in the Analyze grid (see Text and vocab).

The Media tab attaches an audio or video recording to the document and holds its transcript as time-aligned segments. Each segment is a stretch of the recording with its text, which becomes part of the baseline, and an optional speaker label. Media is optional, and a document with none is simply a text. Common audio and video formats are accepted (MP3, WAV, M4A, MP4, WebM, and others), and an upload shows its progress.

A large recording can be sent as audio instead of itself. Choosing one offers to convert it to MP3, naming the size that would produce. The conversion runs in the browser, mono at 16 kHz, about 14 MB an hour, and that is what is uploaded. The timing is unchanged, so segments and alignments are the same either way, but the picture is dropped and the sound is no longer good enough for phonetic measurement. A server refuses a file over its own size limit, and the offer says so before the upload starts rather than after it.

The Tokenize tab divides the baseline into sentences and words. Words are highlighted, and text that is not yet part of any word is plain. Nothing is annotated here: that happens on the Analyze tab.

You can tokenize by hand:

The two tabs hand over to each other: Analyze opens on the last word you clicked here, and this tab opens on the sentence you were in on Analyze, with the word flashed, whether you switch by Alt+click or by the tab itself.

Or run a tokenizer over the whole document with Tokenize, in the Tokens header. Words already there are kept, so text added after a first run can be tokenized by running it again. The built-in Rule-based punctuation splitter finds words and leaves sentence breaks as they are. A tokenization service can replace it with smarter segmentation of both sentences and words (see Services). A service run that stops partway says how many sentences and words it wrote. Run it again to finish. Clear tokens and Reset sentences, beside it, start over. Split at segments adds a sentence break wherever a segment from the Media tab starts, so a recording cut into utterances becomes one sentence per utterance. It only adds breaks, never removes one, and a segment that starts inside a word is left alone.

A project can mark certain tokens as ignored, usually punctuation. Ignored tokens carry no word-level annotation and host no morphemes, so they stay out of your way in the grid (see Annotation).

The same rule decides where the built-in splitter breaks a word: it breaks at punctuation, except the characters your project lists as behaving as letters. Many orthographies write a glottal stop or an ejective with an apostrophe or a similar mark, and unless it is listed, a word spelled with one is split in two. List it under Settings → Annotation and k’a stays one word, keeps that character in the lexicon, and can be annotated like any other.

The Analyze tab is where interlinearization happens. It shows each sentence with its words, the morphemes under each word, and a row for every annotation field. Long documents are shown twenty-five sentences at a time, with a pager above and below the grid.

What you do here:

Move around the grid with Enter or Tab (next cell, Shift for the previous one), ↑ and ↓ (rows), and ← and → from the ends of a value. Alt+click a word to open it on the Tokenize tab. The full set of gestures is collected under Keyboard Shortcuts.

Any sentence, word, morpheme, or annotation value can carry comments: a conversation in the margin of the text, for a question to the speaker, a doubt about a gloss, or a note to a classmate. In the Analyze tab, hover a sentence’s label, a word, or a morpheme and click its comment badge to read the thread or add to it. A badge with a number marks a thread that already has comments.

The Comments tab lists every thread in the document, newest activity first, each collapsed to its latest comment. Click a thread to open it, and click its heading to jump to its sentence. Search finds a word, an author, or a phrase from any comment, and the list can be sorted in text order instead. Comments take Markdown. You can edit your own, and a maintainer can remove any. Comments other people write appear while the tab is open.

A comment stays even when what it was about is edited away, for example a word that is merged or retyped. Such a comment sits behind the tab’s Outdated filter, headed by what it was about, until someone deletes it.

Comments are notes about the text rather than part of it. They are not versioned with the document (see History) and do not appear in exports, except in the native archive, which carries them along.

An entry in a vocabulary takes comments too. Open the entry to read or add to its thread, and the vocabulary’s own Comments tab lists every thread. Anyone who can edit the vocabulary’s entries can comment on them.

The Vocabularies area lists every vocabulary you can see. The list of vocabularies can be searched and sorted like any other. A vocabulary’s entries are shown on its Entries tab as a table you can search, and sort by clicking a column heading. The search box reads every column, or one field chosen beside it: pick Gloss and type go to see the entries glossed with it. With a field chosen, a count of the entries that have nothing in that field appears under the box, and clicking it shows just those, which is how a lexicon’s gaps are filled. A Uses column counts the words and morphemes linked to each entry, and a Concordance on any entry shows every place it is linked, with a link to each. Two entries may share a form, for homonyms. A number after the form tells them apart, in the order they were created: a 1, a 2. See Headwords and senses.

Bulk Add takes rows pasted from a spreadsheet or read from a file. It lets you map the columns onto the entry fields, then reviews each row against the entries that already exist before anything is created, so a homonym stays separate and an existing entry is never overwritten unless you say so. When the entries change between the review and Import so that the import would write something else, nothing is written and the review is shown again. Replace finds text in one field of every entry, the form included, and replaces it: as plain text, as a whole value, or as a regular expression. Plain text ignores case. The other two do not. A fourth kind, is empty, matches the entries that have no value in the field and sets one, which is how a status is given to a lexicon imported without any. It has no counterpart in Bulk Edit: an annotation field has no value on a word until something writes one, so filling every blank means visiting every word rather than matching what is there. Use Validation to find the words a field is missing on, or ask the assistant. Each entry it would change is listed with its old and new value, and only the ticked ones are written. A value that changed after the list was drawn is skipped, and the message says how many. A value the field’s tagset refuses is listed but never written. Export downloads the table as a tab-separated file that opens in any spreadsheet.

A vocabulary’s maintainers edit its entries and fields here. A project’s writers link words to its entries and add new ones from the link popover, but only a maintainer renames or deletes an entry. Each entry carries a comment thread (see Comments).

Every entry has a Form, its reference form or headword, which is required. Beyond the form, an entry carries a set of fields. A new vocabulary starts with a few built in:

You can add your own fields, reorder them, and remove any except Form, Morph Type, and Gloss. The New vocabulary screen lists these before the vocabulary exists, so one that needs no editorial status can drop it there. A field marked inline shows both as a column in the vocabulary table and on each entry’s row in the link popover. Other fields stay on the entry’s own edit form.

A field can be held to a tagset (see Tagsets), so a part of speech is always n and never sometimes noun. A vocabulary keeps its own tagsets, under its Settings, because it is shared across projects. Define one there, then pick it on the field. A closed tagset becomes a dropdown on the entry form and refuses a value outside the list in Bulk Add. It also marks the entries already outside it: the table counts them, and a click shows only those. A lexicon imported before the tagset existed can thus be brought into line one entry at a time, or by adding the value to the list.

An entry in a vocabulary linked to a Plaid UMR project also has a UMR band, below its fields. Roleset is the concept the entry stands for on a UMR graph, such as lunch-01. Left empty, the entry stands for its headword. Arguments describe the roleset’s roles, one row each: a name (ARG0, ARG1, and so on) and a description (the one eating). A name that is not an ARG, or that another row already has, is marked under its row, and the entry cannot be saved until the name is mended. Plaid UMR offers the roleset in its concept picker and the descriptions beside the roles (see the Vocabulary section of the UMR guide).

Every vocabulary is a dictionary. An entry is a headword, and it can have senses under it, each of which can have senses of its own. Numbers say what is what: a headword with senses is adidi 1, its senses adidi 1.1 and adidi 1.2, and a sense of the second adidi 1.2.1. Entries that share a form are told apart by that first number, so two entries spelled a are a 1 and a 2, and the senses of the second are a 2.1, a 2.2. One number is always a headword and two are always a sense. A headword that is alone and has no senses has no number at all. Whether a headword’s own gloss is a meaning that words link to, or the headword only stands over its senses, is up to you: a word with one meaning is simply an entry with a gloss, and a headword that leaves its gloss empty is one nothing needs to link to. The number is written as a subscript after the form, as FieldWorks writes a homograph number. That is the name an entry goes by everywhere: the list, the picker, the link popover in the interlinear view. It is written out in full, with the number after a space, wherever an entry is named in plain text rather than on screen: an export, a machine-readable file, a change in the history. Any entry can be made a sense of another: open it, choose Set parent, and find the entry it belongs under. Add headword in the same place raises a new entry over the open one, with the same form, and makes the open one its first sense: how a word with one meaning gets a second. What belongs to the entry rather than to the meaning goes up with it: its number among the entries spelled alike, and any headword-only field. The morph type and lexeme form are shared, so the new headword and the sense both keep them. Every item in a vocabulary is an entry: a headword, or a sense under another entry. A sense says where it sits on the line under its name: Sense 1.2 of its entry. Beside it, Senses unfolds the whole entry as a tree, numbered, each sense a link, and Add sense writes a new one directly under the open entry. A sense with no morph type of its own goes by its headword’s, on the interlinear line and in auto-linking alike. Senses are rearranged by dragging in that tree: drop one above or below another to reorder them, onto one to make it a sense of that sense, onto Make separate entry to make it an entry of its own, or onto Move to other entry and pick where it goes. A whole entry can be dragged the same way, senses and all. Set parent works for any entry, dragging or not. When several entries share a form, the number after the headword is a button: it opens the list of those entries, and dragging them into a new order gives them their numbers. The entry list opens in a By entry view, beside Flat, that draws senses indented under their entry. A search shows every matching entry in either view. In the By entry view a sense that matches is shown under its entry, which is dimmed when it did not match itself. The open entry has three tabs: Entry for its fields, senses, examples and references, Concordance for every place it is linked, and Comments. In the interlinear view, the link popover lists a dictionary the same way: each headword once, its senses numbered under it.

A field can hold a reference to another entry. On the Settings tab, set the field’s Type to Entry for one reference or Entries for a list, and name it for what it means: Variant of, See also, Compound of. Widening a field from Entry to Entries keeps every reference it holds, and narrowing it back keeps the first of each. A value the new type cannot hold at all, such as text under Entry or a reference under Text, is cleared as part of the change, which says how many and asks first. A sense is an entry too, so a reference can point at one. On the entry form such a field is a picker: type part of a form or gloss and choose. Each chosen entry is shown as a link to it, and the entry it points at lists it under Referenced by. References stay within one vocabulary. Whether a variant is its own entry pointing at the main one, or simply another entry, is up to you.

A field can also be set to Headword only, so it is offered on a headword and not on its senses: an etymology or a pronunciation belongs to the entry, a gloss to each sense. A sense that holds a value in such a field shows it all the same.

Examples are chosen from the text, never typed in. Beside each row of an entry’s Concordance is Use as example. The chosen sentences are listed in the entry’s Examples panel, in context, with their translation, each a link into the document. An example is a reference to that word in that document. When the word is deleted, or linked to a different entry, the example says so and can be removed. A sentence worth citing that is not in any text belongs in a document, where it is glossed like any other, and is chosen from there. Examples brought in by a FLEx import are shown in the same panel as text.

Deleting an entry frees its senses, which become entries of their own, and clears the fields that named it, and the confirmation says how many entries that touches. The confirmation counts the entry’s links again when you press Delete entry. When a word was linked to it meanwhile, the entry is not deleted, and the confirmation stays open on the new count with Its links changed while this was open. Links in projects you cannot open are not in the count you see first. When there are any, the confirmation opens again with the full count and says how many of them are in projects you cannot open. Merging entries with Bulk Edit points those references at the surviving entry, hands the merged entries' senses to it in their own order, and, when the survivor was a sense under one of them, makes it an entry in that one’s place.

A FLEx import keeps the sense structure of the lexicon: an entry with several senses becomes an entry with them under it, numbered as FLEx numbered them, and an entry with one sense is that sense. A CLDF import does the same with its entry and sense tables, and recreates each vocabulary the dataset names. FLEx’s variants and complex forms come in when you tick Variants and complex forms on the review screen: a variant gets a variantOf field naming what it varies, a complex form a components field naming what it is built from, and each keeps FLEx’s name for the relation (Spelling Variant, Compound) beside it.

Exports follow the same structure. A LIFT file has one entry per headword, its senses as senses and their senses as subsenses. A CLDF dataset puts headwords in its entry table and every sense under them in its sense table, each sense with its own part of speech and fields. The vocabulary’s TSV carries each entry’s number and names the entries a reference field points at. A reference field is a relation in LIFT, pointing at the entry or sense it names, and the relation’s name is declared in the ranges file beside it. Promoted examples go out with the lexicon: LIFT carries each one’s sentence and translation, and a CLDF sense names the example rows its sentences became. An example whose document is outside a LIFT export is fetched for its sentence, and one whose word is gone is left out and counted in the export’s warnings. The native archive keeps everything, senses, references and examples included, and brings it back on import. Bulk Add and Replace leave Entry fields alone, since those hold references and not text.

Some lexical units are longer than a word: an idiom, a phrasal verb, a compound name. You can link such a multi-word expression to a single entry by gathering its words:

A linked multi-word expression draws as a line under its words, inside the gray word band, labelled with the entry’s form. The line runs on across a line break, and a skipped word (or punctuation in between) shows as a dotted stretch. A word can belong to several multi-word expressions at once. Overlapping ones stack, and the band grows a line for each. Each word keeps its own entry link and its own morphemes regardless.

Click a line’s label to change the entry, unlink it, or accept one that Auto-analyze made. (Auto-analyze links a run of words when their joined form matches a multi-word expression entry, or one this document already has.) While that popover is open, Shift+click a word to add it to the expression or take it out. The popover for a single word lists the multi-word expressions it is part of, and offers Gather into a multi-word expression to start gathering from there.

Each step remembers whether you left it on, and a service step is unavailable when no service for it is online. A service’s options that name one of your fields, such as the field its glosses go into or the translation it reads, are chosen from the project’s own fields, so you pick Gloss (en) from a list rather than typing it.

A run over a long document takes minutes. While it goes, the dialog lists the steps it will take and ticks them off, and shows the step it is on, how far that step has got, and how long the run has been going. Steps that report no percentage of their own, such as a model pass, still show their name and the clock. You can close the dialog and watch the document instead: the run carries on, and the Auto-analyze button on the toolbar shows the elapsed time until it finishes. Editing is paused for as long as the run writes, and the document says so at the top. It comes back when the run ends. Reloading the page does not stop it, but it does end the run early: the step that was in flight is picked up and finishes, and the steps after it do not run. Stop ends the run at whichever step it is on, whether that step is a service or one of the built-in ones, and the steps after it do not run. What the stopped step wrote before the stop stays, and the document shows it. When the run finishes, a message says what was done: how many translations were proposed, how many words received a copied analysis, and so on. It also says when a service went without a field it reads for context, such as the translation an analysis service is shown. Then review the result with the gestures under Analyze: Ctrl/Cmd+Enter accepts a whole word, Ctrl/Cmd+Backspace discards its proposal, and Ctrl/Cmd+Shift+↑/↓ jumps between the words that still have unconfirmed material.

Step 2 reuses your own work, in the manner of FieldWorks' guesses from previous analyses.

A word receives a copy only when it is unanalyzed: it is still a single morpheme identical to the word itself, it carries no annotation values, and neither it nor its morpheme is linked to the lexicon. A word you have segmented, glossed, or linked, even partly, is never touched. Ignored tokens such as punctuation are skipped.

For each such word, Plaid IGT gathers every other occurrence of the same word form across the project, in this document and in the others, and looks at how each was analyzed. An analysis here means the whole package: the morpheme segmentation with each morpheme’s form and type, the lexicon links, and the annotation values on the word and its morphemes. The analysis given most often is copied. If two different analyses are tied for most frequent, the form is contested and the word is left alone for you to decide. A form is first matched exactly and then ignoring case, so a sentence-initial capital still finds its precedent.

Only analyses a person made or confirmed count. A machine analysis that is still violet does not, so one machine pass can never feed the next.

What a copy carries is set per project under Settings → Services, in the Auto-link section: the morpheme segmentation (forms and types), the lexicon links, and the annotation values, in any combination. The same setting says whether this step is ticked when the Auto-analyze dialog opens.

The built-in linking rule of step 4 works word by word and morpheme by morpheme. For a form that has been linked before anywhere in the project, it follows precedent: the entry that form has most often been linked to, preferring what tokens of the same kind did, so a word follows words and a morpheme follows morphemes. For a form never linked before, it links to a lexicon entry with the same form, if there is one. A whole word is never linked to an affix or clitic entry. As with copying, an exact match is tried before a case-insensitive one.

Only unlinked words and morphemes, and links that are still machine-made and unverified, are touched. A link a person made or confirmed is never changed. Runs of words whose joined form matches a multi-word expression entry, or a multi-word expression already linked in this document, are linked as one (see Multi-word expressions).

The project’s name, its languages, and the option to delete the project.

Two languages are recorded: the object language, which the baseline text is in, and the meta language, which glosses and translations are written in. Each has a name, an ISO 639-3 code, a Glottocode, and a writing system tag, the label FLEx and ELAN files carry on their text (oni, pmy, en), and the object language can carry coordinates. An import from FLEx, ELAN, or CLDF fills these in from the file when the project has none yet. The language identity travels with the project into exports such as CLDF, where a dataset without a Glottocode cannot be linked to any other. A code that looks wrong is pointed out but still saved, since a project may document something the catalogues have no entry for.

Three things live here:

The annotation fields at each scope, sentence, word, and morpheme. You can add, rename, reorder, and remove them, and removing a field first tells you how many values it holds. The count is checked again when you press Delete. If it changed, the dialog shows the new count and asks again. Each field records the language its values are in, as a writing system tag such as pmy or nl: a FLEx, ELAN, or CLDF import fills it in from the file, a field named with a tag in parentheses (Gloss (nl)) is given that tag when the project is next opened, and FLEx and ELAN exports label each field with it. A field with no language goes out under the project’s meta language. Alongside the fields sits the ignored tokens rule, usually punctuation, so those tokens carry no word-level annotation and host no morphemes. Its exceptions are the characters that behave as letters: one character each, and each one joins the word around it instead of splitting it, stays on the word’s form in the lexicon, and can be annotated. An orthography that writes a glottal stop or an ejective with an apostrophe needs that character listed here.

This section also holds the project’s tagsets (see Tagsets) and the document metadata fields shown on every document’s Details tab.

The project’s members and their roles, invitation links, and API access. See Sharing and Permissions.

Review work, a checkbox per member, marks everything that person enters as contributed, in amber, until a verifier confirms it (see Provenance). It is independent of the role: tick it for a community member or a student whatever access they have, and leave it off for a trusted collaborator. The mark is shared with every Plaid app on the project.

Which service, or built-in method, each task uses by default: tokenization, transcription, speech detection, translation, analysis, and auto-linking (see Services). A default can carry default options too, and people can still switch per use. This is also where you say what Auto-analyze’s copy step copies from a word’s previous analyses: its segmentation, its links, its field values, or any combination (see Copying previous analyses).

The Assistant button in the header opens the same assistant in a panel beside whatever you are looking at. It is there wherever there is a project for it to be about, and on the general screens, where it asks which project to use. The screens for making a project and for importing one have neither, so it is not offered there. A handle on the right edge of the window, halfway down, opens it too: it widens under the pointer to show the assistant’s mark, and it is the same gesture as the history rail on the left edge of a document. The panel stays where it is as you move around, so a question can be asked without leaving what you are reading and an answer can be read while you get on with something else. The page narrows to make room for it. The panel can be dragged wider or narrower and hidden again, and its width and whether it is open are remembered. It starts closed until you first open it. A narrow window has no room for it, and neither the button, the handle nor the panel is offered there.

One conversation runs per project. The panel picks up the last one you had in the project you are in and keeps it as you move from document to document. Each question records where it was asked from, so a question about "this sentence" still means the right document when the conversation is read again later.

The panel’s history lists the conversations you have had in the project, with All projects to widen the list past it. Opening one from there keeps you on the screen you are on. ⤢, beside it, moves the open conversation to the Assistant tab and closes the panel.

A question asked on a document is about that document: name no document and that is the one it reads, though it can still look at the rest of the project when the question calls for it, to compare or to count. Ask, beside Copy on a sentence, puts that sentence into the question, and you can take it out again before sending.

Typing @ in the message box names something without leaving the message: the sentences of the document you have open, the entries of a vocabulary you are on, and the documents of the project. Sentences are matched on what they say, so @ followed by a word finds the sentence that contains it. Arrow keys move through the list, Enter takes the one that is highlighted, and Escape closes the list and leaves what you typed. Two entries spelled alike are told apart by their number, which is what @ writes for you. On a vocabulary’s Entries screen the vocabulary is what a question is about, and Ask, beside the tabs, puts the open entry into it. A vocabulary that more than one project links has no one project to file a conversation under, so nothing there is taken as the subject and the project is yours to choose.

On a screen with no project of its own, the panel keeps the conversation it has. Before you have opened a project at all it asks which one to use, because the assistant reads one project at a time.

A plan approved in the panel re-reads what is on screen, so the grid or the entry list shows the result without losing your place.

The panel’s header shows how much of the model’s limit the last turn used, with the counts behind it on hover. A conversation fills up as it grows, because every turn sends the whole thread again. Past 85 percent the panel says so, and a new conversation starts with room again. Where the model’s limit is not known the count is shown without a percentage, unless the assistant was started with its limit stated.

The assistant is a service that your server’s operator runs and picks the model for (plaid-igt-agent, installed from a checkout of the source tree). When none is connected there is nothing to open, so the tab, the button, the handle and Ask are not shown. Conversations already saved are unaffected and open from their own links.

Start a project from existing data on the New project screen. Five formats are supported:

A FLEx category is one thing with a name in each of the project’s analysis languages, v in English and kt.kerja in Indonesian, so one of those names is imported and not both. Where a file carries more than one, the review screen of either FLEx import lists them under Part-of-speech language, and the words, morphemes and entries take the name in the one selected. The rest are listed as not imported.

The review screen of an ELAN import has three parts, in the order you work. Files lists each .eaf with the document it becomes and its recording under it. Each document takes the name its file gives it, even when two files give the same one. When a file names a recording, the recording’s file name is kept in a Media file metadata field on the document. Clear Keep each recording’s file name in a Media file field to leave it out. Tiers is where each tier is given its role: sentences, words, morphemes, morpheme types, time alignment, an orthography, or a field. Preview draws a sentence of the first document as the interlinear lines the import would write, and redraws it as you change a role, so a gloss tier mapped onto words or a morpheme tier left off shows before anything is written. The arrows step through its sentences. Anything that needs a decision appears in an amber box, and the line beside Import counts what the run would add.

An ELAN import also runs into a project you already have: New document → Import from ELAN, on the project’s Documents tab. Each tier is written into a field the project already has, chosen from a list, and a tier can be given a field of its own instead, which its row then marks as a new field. A tier with no annotations is left out. Recordings are chosen in the same step as the .eaf files, and everything chosen is listed together with what will happen to it: which .eaf each recording belongs to, and which of them matches none. A recording is matched to the .eaf that names it, and a document whose recording you did not choose is imported without it. One too large for the server says so and says what converting would produce, and it can be converted to MP3 in the browser before the run (see Media), so the import does not finish without it. A file imported into this project before says so on its row, with a link to its document, and the screen asks once for all of them: keep that document, replace it (which deletes it and everything added to it since), or add a copy beside it, named with a number. A recording chosen beside a kept file is added to its document when that document has no recording yet. The rest of the project is untouched, and nothing is locked while the import runs.

Either way, the import asks what each tier becomes:

Tier names of the form Translation-gls-nl, which is how a corpus prepared for FieldWorks names them, are read: gls is the kind of annotation and nl the language, so such a tier is offered the matching field without your having to say so. A tier that needs a new field names it the way a FLEx import would, Translation (nl), or plain Translation when every such tier is in one language. A file made by ELAN’s Import FLEx names its tiers by level and item instead (A_phrase-gls-en, A_word-pos-en, A_morph-type), and those are read too: the fields are named as a FLEx import names them (Translation, Gloss, POS), a word’s text in another writing system becomes an orthography, and the morph type tier gives each morpheme its type. Each sentence’s text is written from its words, as FLEx writes a phrase, so a word the phrase line leaves out, such as [0] for a zero argument, keeps its place and its annotations, and the text shows the words' own spelling. The segment number tier is left out. So are the times of such a file when it names no recording and every time falls on a whole second, since that import makes them up.

An import that stops before it finishes, whether you cancel it, it fails, or the tab is closed, leaves a project that is only part filled. Opening such a project takes you back to the import instead: choose the same file or files and it carries on, keeping the documents and entries already there and finishing the ones it never reached. A document is recognised by the file’s own identifier for it, not by its name, so two texts called the same thing are kept apart, and a document you made by hand is never taken for one of the import’s. A FLEx import writes into the lexicon it started with, and finishes placing every sense an earlier run left flat. Importing the same backup afresh into a lexicon you have since arranged places only the entries it adds. If what arrived is enough, Use the project as it is on that screen puts the project back in your hands.

Exports are driven by presets. A preset fixes a format and which orthographies, fields, and options it includes, so the same export can be run again later, or by someone else, without choosing everything afresh. Maintainers create presets on the project’s Export tab. Anyone can run one from there or from a document’s Export tab, choosing a scope: this document, selected documents, or the whole project.

Six formats are supported:

A FLEx preset also carries the writing-system tags FieldWorks will see: one for the baseline text, one for each alternate orthography, and one for glosses and translations. Use the codes your FieldWorks project already uses, or FLEx will offer to add a new writing system when you import.

Document metadata goes out with the texts. A field FieldWorks keeps in a place of its own lands there: Title, Abbreviation, Source, and Description, which FieldWorks shows as the text’s comment. A field named for a writing system, such as Title (en), goes out under that one. Everything else, the researchers and participants among them, joins one comment naming each field it came from, since a .flextext has nowhere else to keep it.

Each field goes out under its own language where that is known. A field imported from FLEx knows it, whatever it is called: an import records the writing system each field’s values are in, so a project glossed in three languages exports all three correctly without anything being set here. Failing that, a field whose name ends in a tag, such as Gloss (nl), goes out under that tag, and a field with neither goes out under the tag for glosses and translations. The box beside each field overrides all of it.

A vocabulary has a history of its own. Open it with History beside the vocabulary’s name, on any of its tabs. It lists every change to the vocabulary and to its entries: an entry added, edited, moved or deleted, a field or tagset changed, a maintainer added. Links are not in it: linking a word is annotation, and shows in the document’s history. Its entries are marked by kind as a document’s are, and a Replace is marked Bulk edit.

Selecting a history entry shows the vocabulary’s entries as they were right after that change, read-only on the Entries tab, entries deleted since included. History on an open entry lists only the changes to that entry, and keeps it open, so each state you select shows that entry as it was then. Opening another entry lists that entry’s changes instead, and History beside the vocabulary’s name lists every change again.

To put one entry back as it was, open it in the past state and click Restore. A dialog lists what changes: the form and each field set back, or, for an entry deleted since, the whole entry. The entry keeps its comments. Its links in documents do not come back with it. Each document’s own restore, to a moment when a link was there, brings that link back once the entry exists again. The message that confirms the restore offers Undo, which sets the entry back again, or, for an entry the restore brought back, deletes it. Restoring an entry is for the vocabulary’s maintainers.

A morpheme with a grammatical job and no sound is written with the empty set sign, ∅. It has a code of its own (\00) and a shortcut in the Analyze grid: Alt+0 in a morpheme form types it. Write it as ∅ rather than a capital O with a stroke or a zero, so it exports as a zero morph. The Validation tab lists any morpheme that looks like a zero written another way.

The arrow button above a sentence merges it with the previous one.