General Info
This web application enables you to browse several lexicons, or word databases, of connectives,
also called discourse markers.
The app can be used with recent versions of Chrome, Firefox, Edge, Safari, and Opera.
Connectives
Connectives are words or short utterances which set two phrases into a relation, e.g. "because", "while", "even though".
They belong to various word classes, but are usually either conjunctions (coordinating or subordinating),
adverbs/adverbials, or prepositions.
Lexicons
This webapp was created at Universität Potsdam to integrate their own DiMLex lexicon for German
connectors and similar lexicons for other languages created at other universities. Each lexicon has a few
hundred entries; each entry contains a connective and associated linguistic information, such as syntactic
categories (word classes), discourse relations (semantic senses), synonyms and/or examples of their usage,
among others.
How To Use
Lexicon Selection
The Lexicon Selection section of the sidebar shows all available lexicons. Select one or more lexicons
to browse their entries. Click about for more info about a lexicon.
Search Options
You can filter the displayed lexicon entries according to several criteria:
-
Filter: Enter a word in the text field labelled Filter to only display lexicon
entries which contain this word. In the dropdown menu next to the text field, you can select where this
word may occur in an entry:
- Word (default): Only display results whose connective contains the filter word.
- Word + Synonyms: Only display results whose connective or synonym list contains the
filter word.
- Word + Examples: Only display results whose connective or example list contains the
filter word.
- All: Display all results which contain the filter word in their connective, synonym
list, or example list.
- Syntactic Category: Select one or more word classes to filter by. Hover the mouse over
an abbreviation to see which word class it stands for. If none of the checkboxes are selected, the behaviour
is the same as if all of them were selected, i.e. connectors of all word classes are displayed in the
result list.
- Discourse Relation (PDTB3): Select one or more semantic senses to filter by. As with the
syntactical category, the behaviour if no senses are selected is the same as if all senses were selected.
The set of discourse relations we use is the Penn Discourse Treebank 3 (PDTB3) tagset:
A discourse-annotated corpus of conjoined VPs.
Bonnie Webber, Rashmi Prasad, Alan Lee, and Aravind Joshi.
Proceedings of the 10th Linguistic Annotation Workshop, ACL, Berlin, Germany. 2016.
The Result List
The result list contains all those elements from all selected lexicons which match the filter settings. The interface is
pretty self-explanatory, but there are some not-so-obvious features which are explained here.
Word & variants: A connective can have multiple spellings. Some connectives consist
of disconnected parts. These can be specifically searched for by entering ... into the
Filter field. Words marked with focus can be used as focus particles.
Syntactical category: Each connective has one or more word classes of which each has one
or more senses that the word can have when used as this word class. Hover the mouse over a
subtly underlined word class to see the original tagset and word class
which this word has in its source lexicon. The original annotations use wildly different tagsets which are
mapped to the four main word classes used by this app. Non-underlined word classes have not been changed
by the mapping.
Ordering info: For lexicons which include this info, some results contain information about
the grammatical context in which a word can be used. If available, this info is listed next to a word class.
Some prepositions also have an associated set of cases with which they are used.
Discourse relation: The list of discourse relations for each word class lists the possible
semantic senses this word can have when used as this word class. Usually, a word has only one of these semantic
senses at a time. In some entries, however, you will see two senses on one line, separated by a +.
In this case, the word can have these two senses at once. Similar to the word classes, the senses shown here
are mapped from different source tagsets to PDTB3. Hover over an
underlined sense to see the original annotation(s).
