OPERATIONopr.eng.suppletive-comparison
suppletive comparison
The replacement of a graded base by an unrelated word, so that the graded step of a very common word has to be learned rather than built.
Suppletion is the limiting case of grading: not a base altered by a rule but a second word standing in for it. The set is small, and its members are among the commonest gradable words in the language, which is the ordinary distribution for suppletion wherever it is found. Two consequences follow that no regular grading produces. A suppletive graded shape cannot be generated and has to be stored step by step, and one such shape can belong to two words at once, so the relation between a shape and a word is not a function in either direction.
The rule declares itself irreducible and names the rule it overrides, because the honest content of a suppletion statement is which words are exempt from the regular route, never what the exempt shape turns out to be. A coverage of zero is what that honesty looks like as a number. The set filed here is smaller than the set the language has: the comparison record also names bad and far, and neither is filed as a word in this graph yet, so the rule reaches only the three words that are.
- status
status documentedasserted, and the sources agree- provenance
provenance authoredwritten for Logos, from first principles- derivation status
derivation_status irreducibleit follows from nothing: an arbitrary fact of the language- salience
salience - No salience. Where no frequency source applies the object carries none, rather than a default or a guess (03-schema.md §3.5).
- aliases
aliases - None recorded.
- sources
sources - An empty list, which is an assertion: this rests on nothing external, and somebody checked. It is not the same as nobody having looked.
What it does, and when it fires
operation.ruleThe domain it works in, and the firing condition as a sentence. The checkable form of that condition is the selects group below.
- domain
domain morphologicalit changes the shape of words- trigger
trigger A word of the small stored set is graded, so the step that would have been built from its base is served instead by a word sharing no material with that base.
How much it covers
operation.coverageHow many cases the operation actually derives, out of how many it claims, and how that was counted.
derives 0 of 1enumerated
All of it. This rule is the statement that a set of graded shapes follows from nothing, so a coverage of zero is the accurate number rather than a gap: it names which words hand their graded steps to an unrelated word, and it never says what that word looks like.
every case was counted
Everything joined to this object
One group per typed edge, in both directions. A1b guarantees no object node is an orphan, so this is the whole navigation: an object with no edges would have failed the build rather than reached this page.
What this object is a kind of
Walking this object's own `is-a` edges: what it is a kind of. Several branches from one object are several classification axes, not competing supertypes.
- objects
- 1
- places drawn
- 1
- multi-parent
- None in this view. This subgraph happens to be a tree, which the explorer draws as one without assuming it is one.
The replacement of a graded base by an unrelated word, so that the graded step of a very common word has to be learned rather than built.
Siblings are listed by name, then by id to break ties. That is a display choice made here and not an order the graph asserts.
Fan-out per dimension
A4One row per dimension, largest first. That is a display order, not a verdict: 03-schema.md declares no threshold above which fan-out is too much, so this reports the count and grades nothing.
- case13 values
- mood8 values
- voice8 values
- aspect7 values
- aktionsart6 values
- number6 values
- person6 values
- tense6 values
- comparison4 values
- evidentiality4 values
- gender4 values
- politeness4 values
- valency4 values
- animacy3 values
- definiteness3 values
- deixis3 values
- finiteness2 values
- information structure2 values
- interrogativity2 values
- polarity2 values
- switch-reference2 values
What kinds of this object there are
Walking the `is-a` edges that point at this object: what kinds of it there are. An object reachable two ways is drawn in both places and is still one object.
- objects
- 1
- places drawn
- 1
- multi-parent
- None in this view. This subgraph happens to be a tree, which the explorer draws as one without assuming it is one.
The replacement of a graded base by an unrelated word, so that the graded step of a very common word has to be learned rather than built.
Siblings are listed by name, then by id to break ties. That is a display choice made here and not an order the graph asserts.
applies-to
edges.applies-tothe operation's input
this operation applies to
- LEXEMEgood
lex.eng.goodno page in v1documentedThis edge type carries no payload. The assertion is the edge itself.
- LEXEMEmany
lex.eng.manyno page in v1documentedThis edge type carries no payload. The assertion is the edge itself.
- LEXEMEmuch
lex.eng.muchno page in v1documentedThis edge type carries no payload. The assertion is the edge itself.
overrides
edges.overridesdeclared arbitrariness (A7)
this object overrides
reasonA replaced base leaves the suffix nothing to attach to, so in the graded steps this rule serves, suffixation has no input and does not fire. The two cannot both apply to one step, and the replacement wins, because once it has applied there is no base left for the suffix to read.
these objects override this one
- FORMbetter
frm.eng.betterdocumentedreasonThe suppletion rule names which common words hand their graded steps to an unrelated word, and declares a coverage of zero for them because it never predicts what that word looks like. This shape is exactly that residue: the graded step of good, drawn from a base the plain form never uses. It is the case the rule was written to name, not an exception left over from it.
- FORMmore
frm.eng.moredocumentedreasonThe suppletion rule names which common words hand their graded steps to an unrelated word, and declares a coverage of zero for them because it never predicts what that word looks like. This shape is that residue twice over: it is the graded step of two separate quantifiers at once, drawn from neither plain shape, so one form belongs to two words and nothing connects it to either. The work it does standing in front of a long base is a second job and is not what makes it arbitrary.