Readme.V1

The V1 release captures the state of 4lang circa 2017. The raw list was
published as Appendix 4.8 of

@Book{		  Kornai:2019,
  title		= {Semantics},
  author	= {Kornai, Andr\'as},
  url		= {http://kornai.com/Drafts/sem.pdf},
  publisher	= {Springer Verlag},
  year		= {2019},
  isbn		= {978-3-319-65644-1}
}

and the (partial) implementation of the formal semantics that was at that time
associated to the model, based on Eilenberg machines, is still available in the 
https://github.com/kornai/pymachine repository.

Release 2.5 of Concepticon links to this release by means of the conc_4lang.tsv
file. The format of this file is CNUM CC 4NUM where CNUM is the Concepticon
concept number, CC is some correspondence code, and 4NUM is the 4lang concept number.
An easy example is

1292 p 2043

where 1992 is the numerical code for "BAD" in Concepticon, 2043 is the 4lang
number for "bad" in 4lang, and p means Perfect match. Case differences and
minor spelling differences are considered irrelevant, as would be dialectal
differences as between British "colour" and American "color" or even between
B. "lorry" and A. "truck" which we consider a perfect conceptual match.

Other correspondence codes are as follows.

m (category Mismatch) is used wherever the concepts match up well, except for
lexical category. Some typical examples:

2004 m 1112 BEND/bend. Concepticon takes the verbal sense as basic `Movement
that causes the formation of a curve' whereas 4lang uses the nominal sense
`line, LACK straight/563'.

1786 m 2145 DRY UP/dry `To lose moisture, usually through evaporation or
absorption' verbal, versus dry `LACK wet' adjectival.

s (Subset) is used when the Concepticon entry is a subset of the 4lang entry:

2106 s 1039 ("MALE PERSON" is a subset of "male")

2261 s 1039 "MALE (OF PERSON)" is also a subset of 4lang "male", which is neutral
between adjectival and nominal meanings, but see 1553 p 1039.

In general, the conc_4lang file provides s links only where no p link has been
found. While on many occasions the s correspondence code could be easily fixed
by generalizing the Concepticon definition (e.g. 1040 s 3359 the "The hair on
humans, usually on the head but also including the hair on the body" could be
amended "The hair on humans or animals") we do not advocate for this, since
the choice reflects different goals: Concepticon aims at directly facilitating
cross-linguistic comparisons, so it makes a great deal of sense to distinguish
'capillus' (human hair) from 'saeta' (animal hair), whereas 4lang aims at
maximally reusable conceptual units, for which lumping the two senses together
makes more sense.

u (sUperset) is reserved for when the Concepticon entry is a superset of the
4lang entry. This is rare, so far only one example: 

695 u 1654 GO/go Concepticon has `get from one place to another by any means'
and 4lang has `move, INSTRUMENT leg, =AGT HAS leg' 

n (No entry) is used when a 4lang entry simply lacks an equivalent, or near
equivalent Concepticon entry. some examples are

- n 1407 public `LACK owner' (as in public garden)

- n 172 substance (also no Concepticon entry for "material" or "stuff").

These are best treated as extension requests for Concepticon. The converse
situation, which also occurs, will be resolved gradually: for example 1931
BEESWAX has no 4lang equivalent (where 1797 "wax" refers to growth, as in the
waxing moon). This means that to obtain the 4lang definition, we need to (a)
match up the desired sense in LDOCE and (b) follow the method of Recski:2016.
This is left for later releases.

e error (machine aligned senses don't pass human verification). These will not
appear in published mappings, they are just for internal maintenance.
Example:

1132 e 2722 BEAM beam `A structural member loaded on its narrow face, and
typically used in a horizontal or sloping position to span between bearing
points.' versus beam line, light/739, FROM <sun>