The Predilex data presents itself in the form of a grid with many named columns, and rows each representing a different entry, dedicated to one semantic predicate, or ‘sememe’.
The two first rows constitute the header, giving out the names of each column: on the first row (hidden in the ODS template file) is a shortened code name, and on the second row a longer, ‘full’ name for the column.
Below is an explanation of the purpose of each column, in order:
The Predilex ID of the entry. It uniquely identify the predicate in Predilex, and is also used in other columns of other entries for referring back to the current entry (for example, for indicating that the other entry's predicate is a hyponym of the one of the entry at hand).
A Predilex ID has the shape CVCVCV where C represents one of 20 consonant letters (bcdfghjklmnpqrstvwyz), and V one of 5 vowel letters (aeiou), making for 100 possible CV syllables, and 1000000 possible CVCVCV ID's. (Also explained here)
The Rank is a single lowercase letter that gives an estimation of the use frequency of the predicate in real language use.
An indicative estimation of the potential of the predicate's word to be used as an etymological component in compound words.
The arity (or ‘valency’, or ‘adicity’) of the predicate, i.e. the number of argument slots it has (excluding the hidden context variable slot).
A list of letters or digits separated by whitespaces, each sign representing the type of the corresponding argument slot of the predicate, in the same order of appearance as in the predicate's definition.
A list of letters separated by whitespaces; each letter corresponds to one of the predicate's argument slots, and each letter indicates its behavior with respect to distributivity: d for “distributive”, c for “collective”, s for exclusively singular references. When a property P is distributive, this means that P(x) entails ∀y. among(y, x) ⇒ P(y).
This column is deprecated. It was intended to indicates the behavior of each argument slot with respect to Kind handling, between the following options: s for “stage-level”, i for “individual-level”, and k for “kind-level”.
Lists of semantic categorization tags separated by whitespaces, in groups separated by semicolons ;, each group being associated with one of the slots of the predicate, in order.
Each group indicates the semantic categories associated with the targed argument slot.
A single list of semantic categorization tags separated by whitespaces, describing the predicate as a whole, and not one of its argument slots in particular.
Indicative list of cherry-picked potential etymologies for words representing this predicate. The potential etymologies are separated by semicolons ;. Each potential etymology consists of a list of Predilex ID's, separated by whitespaces.
List of Predilex ID's separated by whitespaces. Each such ID represents a predicate which describes the same event as the current predicate, but typically with a different argument structure, as exemplified by the following pair:
- 𝜆x. eats(x)
- 𝜆x. 𝜆y. eats(x, y)
List of Predilex ID's separated by whitespaces. Each such ID represents a predicate whose meaning is similar to that of that of the current predicate, i.e. a quasi-synonym thereof.
Same format as the above, but listing antonyms for the predicate.
Same format as the above, but listing hypernyms for the predicate.
Same format as the above, but listing hyponyms for the predicate.
Same format as the above, but listing holonyms for the predicate.
Same format as the above, but listing meronyms for the predicate.
Same format as the above, but listing related predicates, as a sort of ‘see also’.
Same format as the above, but listing predicates that can function as alternatives to the current predicate, usually with a homologuous meaning but a different argument structure. Typically, a single language would need only one of the alternatives, and having several of them at once would be redundant.
Some prominent examples may be the following pairs:
- ⟪➊ is a hand⟫ versus ⟪➊ is a hand part of ➋⟫;
- ⟪➊ surpasses ➋ in the amount to which they satisfy the gradable property ➌⟫ versus ⟪➊ surpasses ➋ in the amount they are in relation ➌ with⟫.
Formal, substitutional definition of the predicate, written in a custom logic notation, with argument slots expressed as lambda expressions. Predilex ID's are used as predicate identifiers in the logic notation, allowing defining the current predicate in terms of other predicates in the Predilex dataset.
All predicates have zero or more argument slots, occurring in a fixed order. The order in which the lambda 𝜆 expressions appear define the order of the argument slots for the predicate.
The hidden context variable is usually not shown explicitly in the logic notation, unless it behaves in an unexpected way (other than being passed down as-is to all the predicates used in the definition). When the context variable is mentioned explicitly, the symbol 🅲 is used.
A list of URLs separated by vertical pipes | surrounded by one whitespace on each side. Each URL points to a media, such as an image or a sound file, representing an instance of the predicate, a physical realization thereof. Such media may serve for example for teaching the meaning of the predicate.
Same as the previous field, but this time with negative sample media: select medias that do not represent instances of the predicate, whereas it might have been expected that they could have been instances thereof. For example, when dealing with a predicate describing a pigeon, positive media like pictures of pigeons might still allow for the supposition that the predicate could have a broader meaning, allowing for other sorts of birds to quality; by adding non-pigeon birds as negative sample media, the boundaries of application of the predicate are made clearer.
Glyphs, symbols, pictograms or ideograms representing the predicate at hand.
An alternative to Predilex ID based on an English expression.
Definition of the predicate in the English language.
The argument slots of the predicates are represented by numbers in black circles, such as ➊, ➋, ➌, ➍; the number represents the index of the argument in the predicate's argument list. This allows for the argument slot marks to occur in any order, albeit it is tried inasmuch as possible as to keep the appearing in their natural order (➊ → ➋ → ➌ → ➍…). Like with definitions in logic notation, the hidden context variable is usually not mentioned explicitly, but when it is, the symbol 🅲 is used.
Additional notes in the English language.
List of semicolon-separated lemmas of the English language which have the predicate at hand as one of their possible meanings. The lemmas may be followed by a subcript number disambiguating the target meaning of the lemma when the lemma is polysemous or has homonyms.
Lemmas may be suffixed with a hash # expression introducing the part of speech and syntactic frame of the lemma. For example, ⟪give#VT_to⟫ means that the lemma ⟪give⟫ is a transitive verb with an additional oblique slot tagged with the adposition ⟪to⟫.
The predicate's argument slots are by default mapped to the same slots in the same order as the English's lemma, and if they map in a different order, a sequence of digits such as 213 indicates how the arguments are reordered: ⟪give#VT_to 213⟫ would mean that the two first argument slots of the lemma are permuted (whereas the default order would have been 123).
These fields have the same purpose as the ⟪English lemmas⟫ field but for languages other than English.
Like the previous fields, this one lists lemmas in natural other natural languages, except this time each lemma is preceded by a bracketed three-letter language code, such as [tgl] for the Tagalog language.
Lemmas in the Toaq constructed logical language.
Deprecated field.
Definition in the Toaq language.
Notes in the Toaq language.
Lemmas in the Lojban constructed logical language.
Notes in the Lojban language.
Lemmas in other constructed languages.
Definition in Eberban.
Lemmas in programming languages; the name of the programming language is shown in square brackets before the each lemma.
Corresponding ID in the WordNet database.
Corresponding ID in the HowNet database.
Corresponding ID in the FrameNet database.
Corresponding ID in the OpenCyc database.