Skip to content

edge feature, indicating linguistic dependency between elements that do not necessarily have an embbedding relationship.

An ETCBC interactive encoding program sy04types presents for each clause one or more mother clauses in which it could be anchored. A clause’s possible mother clauses are ranked in order of probability. This ranking is based on multiple types of observations:

  • grammatical and lexical correspondences between a clause and its possible mother clause;
  • the number of earlier occurrences of a similar clause connection;
  • the distance (in clause atoms) between the two clauses.

Eventually, it is the analyst who has to decide which of the proposed clause connections is the most adequate one. His or her decisions are tabulated in different levels of indentations reflecting parallel or dependent relations between connected clauses. (Kalkman, Verbal Forms in Biblical Hebrew Poetry, 2015, p. 118)

The mother relation exists between objects of many different kinds. The feature code refers to it a lot.

See the AtomsAndMothers notebook which makes some basic explorations into these matters.


This description needs more body: what objects can be mothers, what daughters? What is the linguistic meaning of mother in all those cases? Are there formal characteristics indicating a mother relationship? Examples needed!


MQL Implementation

In MQL, mother is not a feature, but it is possible to query with conditions involving the mother.

Here is a query that looks up occurrences of the word swear (CB<) if it occurs in a clause atom that is the mother of a following clause which is a quotation (txt = 'Q'):

select all objects where
    [clause_atom as c1
        [word focus lex = "CB<["
    [clause txt = 'Q'
        [clause_atom mother has c1.self

N.B.: Note the usage of has here. In previous versions (up to 4b) the MQL data has been modeled in such a way that every object can have it most one mother. Users of that version of the data base have learned to write

clause_atom mother = c1.self

which will lead to an error in the current version.

Text-Fabric implementation

In the Text-Fabric representation of the BHSA dataset, mother is an edge feature. The nodes correspond to the objects, and the edges to relationships between nodes. The edges that belong to the mother feature, correpond to the mother relationship.

We count how many mothers nodes can have (it turns to be 0 or 1). We walk through all nodes (for c in N()) and per node we retrieve the mother nodes (E.mother.v(c)), and we store the lengths (if non-zero) in a dictionary (mother_len):

mother_len = {}
for c in N():
    lms = E.mother.f(c) or []
    nms = len(lms)
    if nms: mother_len[c] = nms
See also