198 Forms of Say
that (having been) said’.”^3 Likewise, Poutsma ( 1905 : 725) notes that “being
or having been is often suppressed” in absolutes. The suggestion here is the
intuitively plausible one that the full clause (with) that having been said is
the (historical) source of the reduced clauses (with) that being said and (with)
that said.^4
However, grammarians are in agreement that the opposite course of devel-
opment should be postulated. Jespersen (1946: 55– 57) sees the simple pas-
sive participle as original, with the “more clumsy construction” with being
appearing in the sixteenth century, perhaps as a means of preventing ambi-
guity (in cases where the participle might be interpreted as a past tense).
Jespersen ’s examples of the fuller, again “clumsier,” construction with hav-
ing been date from the eighteenth century. Visser ( 1972 : 1259ff.) too consid-
ers that absolute constructions with the simple past participle are original;
forms with having/ being/ having been were introduced in the modern period
and are now often preferred (1266– 1267). Finally, for Curme ( 1931 : 153),
absolute constructions without the copula are original, with the more modern
trend being to take tense ( having ) and voice ( having been ), especially when
active.
In order to resolve the question of the origin and history of the “that said”
construction, the remainder of this section takes a corpus- based approach.
In addition to the quotation databases of the OED and MED, the corpora
used in the following study included EEBO, ED, CLMET3.0, CEN, UofV,
and COCA.^5
The OED entry (s.v. say , v.^1 and int., def. P11) suggests a recent origin: The
fi rst example of having said that is 1820, of that being said is 1908, and that
said 1992.^6 We make the starting assumption that the full clause is the source of
the reduced clause, as might also be expected in terms of grammaticalization.
Do we fi nd historical evidence for this assumption?
3 In a response to the Language Log entry (see http:// languagelog.ldc.upenn.edu/ nll/ ?p=1807 ),
Lawler points out that he did not intend this as an historical explanation but rather as an “indi-
cation of how the phrase could be unpacked by adding stuff to make it unambiguous (if also
uncolloquial).” I retain the argument here because of its intuitive appeal.
4 An early statement of this view is found in Davies ( 1869 : 115): “sometimes being and hav-
ing been are omitted; as, ... This said, that is, This having been said.” In contrast, Whitney
(1886: 224) presents the opposite view. After introducing this said , he says, “Instead of a simple
passive participle ... being taken directly with the noun or pronoun in absolute construction, an
auxiliary being or having been is very often introduced ... for example this having been said .”
5 No examples of the constructions were found in the Modern English section of HC, Lampeter,
or CED. The “that said” construction is a relatively low- frequency item. Even the OED quota-
tion database (59 million+ words) yields only 4 parenthetical uses of that said , 24 of this said , 2
of that being said , 1 of this being said , 1 (invented example) of that having been said , 0 of this
having been said , 7 of having said that , and 1 of having said this.
6 The only example of having been said in the OED is the concocted example from Safi re ( 2002 )
(OED, s.v. participial, n. and adj., def. B).