[Ads-l] OT - Not the OED

Tue Feb 2 11:30:33 UTC 2016

Ben,

Thanks for pulling me up that it's three (free) example uses per *sense* 
rather than per word -- an important distinction.  Though in the course of 
their response to the issue, ODO didn't draw attention to this, or even note 
that there were supplementary examples included in the various entries, 
albeit only appearing if you click on a drop-down menu.  Odd, as it would 
have made their position stronger in their response to the twitterfury which 
recently occurred.

The relative wealth of example usages (whatever reservations one has about 
them) is one of the strengths of the ODO, as is the ability to shift between 
the British and US English dictionaries, to note different nuances of usage 
between the two registers.  Though that does raise the point that it would 
be useful if the example sentences were at least tagged as, perhaps, "US" vs 
"UK".  Sometimes usages are tagged as "US" or "UK", but not the examples 
themselves. The question here, I suppose, is just how much metadata is 
useful, and to which enquirer, as well as how (and when) such metadata 
should be signalled.  Even the bare date of the origin of the example would 
be useful.

When it comes to 20 examples for the Premium version as against only 3 for 
the free version, I wonder how useful the further examples would be?  My 
feeling would be that, as things stand, an enquirer would move from the 
three uncontextualised examples in the free version, to google, for further 
contextualised examples, rather than simply adding more bare instances of 
use, which is presumably (correct me if I'm wrong) all that the Premium 
subscription adds in this area.  (Access to further dictionaries, which a 
Premium subscription also includes, is another matter.)

In the course of following the newspaper reports of twitterstorm over "rabid 
feminists", I became intrigued over the exact relation of the Oxford 
Dictionary of English, originally published in physical form in 1998, with 
later revisions, and the on-going version which appears in Oxford 
Dictionaries Online.  The substance of the current entry on "rabid" in ODO 
is taken verbatim from the ODE, and as it stands is identical to the version 
which had originally appeared in 1998.  The ODO adds three further example 
sentences.  Presumably the original ODE definition (1998) of "rabid", and 
the illustration chosen -- "rabid feminists" -- was based on the entire 
corpus as it then stood.  But the ODO reflects a *different* corpus, 
containing further material from 1998 to the present.  As an inspection of 
google Ngrams suggests that the use of the term "rabid feminist" peaked in 
the 1990s, then showed a marked decline in frequency of use, this would seem 
to be relevant.  It raises (in my mind) two separate questions.  One would 
be how far the on-going revisions to the ODO are taking this into account, 
while the other would be that it would be useful if such significant changes 
in use were signalled in some way in the ODO, perhaps with something 
analogous to Ngram, but drawing on the corpus of material which the ODO was 
based on rather than the material available to google.

Two constraints are obviously at play here -- the original ODE was 
constrained in the amount of material it could present by the nature of its 
existence as a physical artefact, while just how much scrutiny can be 
carried out is obviously limited to the amount of time and attention ODO can 
devote to this matter.

You say:

"As the ODO is not a historical dictionary like the OED,
there's less of an imperative to supply the date and provenance for
each sentence (though all that information is stored in the back-end
metadata and could conceivably be shared on the front end)."

Indeed, and a failure to notice this crucial distinction, both on the part 
of several journalists writing on the matter and readers commenting on the 
published pieces, was characteristic of the entire furore over "rabid 
feminists".  But if we accept that the OED is an historical dictionary in 
contrast to the ODO as a dictionary of current usage, how adequate is the 
ODO as it stands, and could it be improved?  As it stands, it seems to me 
that the absence of *any* (well, virtually any) metadata undermines the 
usefulness of the ODO, both for casual enquirers and serious researchers.

Further, what metadata is held?  The corpus of material on which the ODO 
draws contains (usefully) on both phonographic and orthographic material, I 
think in the ratio of 1/10, and the orthographic material ranges from 
originally printed material though all kinds of on-line communication. 
Though I don't know whether Twitter feeds are, or should be, included, or 
whether the metadata signals the relation of addressor/addressee -- whether, 
for example, the illustration reflects a situation where a man is speaking 
to a woman, or vice versa.

The fundamental question is perhaps what metadata should be included in the 
ODO presentation of the words it deals with, and how such material could be 
usefully presented.  One possibility might be whereby a Premium subscription 
gives access to a range of metadata not included in the freely available 
ODO.

[As an aside, if the OED corresponds (in Saussurean terms) to a diachronic 
perspective, while the ODO represents a synchronic moment, what range of 
time does the synchronic moment encompass?  I once woke up in the middle of 
the night realising to my horror that I couldn't think of an adequate answer 
to the question of how long a synchronic moment lasted.]

Robin Hamilton

--------------------------------------------------------------------------------

-----Original Message----- 
From: Ben Zimmer
Sent: Monday, February 01, 2016 8:15 AM
To: ADS-L at LISTSERV.UGA.EDU
Subject: Re: OT - Not the OED

---------------------- Information from the mail 
header -----------------------
Sender:       American Dialect Society <ADS-L at LISTSERV.UGA.EDU>
Poster:       Ben Zimmer <bgzimmer at GMAIL.COM>
Subject:      Re: OT - Not the OED
-------------------------------------------------------------------------------

On Sun, Jan 31, 2016 at 9:10 PM, Robin Hamilton wrote:
>
> Does anyone have any strong opinions, either positive or negative, about
> the Oxford Dictionary Online?
>
> In the wake of the recent Rabid Twitterstorm, which finally Made It Big
> in America via the Washington Post on 27/1/16, but had already been
> raging in the MSM in the UK, I've been thinking thots about the
> ODO.
>
> Specifically, how useful do listmembers feel it is to provide three
> citations to illustrate a dictionary entry, none of them with any
> indication of date or provenance?
>
> I have other issues, but that for starters.

You actually get three example sentences per *sense* for each entry.
And if you subscribe to ODO premium, you'll get up to 20 example
sentences per sense. Altogether something like two million example
sentences have been incorporated into the entries under the
appropriate senses, a huge effort in word-sense disambiguation (WSD).

Full disclosure: when I was at OUP in the mid-aughts, I oversaw some
of this work. As the ODO is not a historical dictionary like the OED,
there's less of an imperative to supply the date and provenance for
each sentence (though all that information is stored in the back-end
metadata and could conceivably be shared on the front end).
Dictionaries take differing approaches to this, of course.
Merriam-Webster Unabridged, for instance, credits the authors of many
example sentences, but generally only notable writers (and without
providing dates). The Shorter Oxford also gives authors but not dates.

--bgz

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org 

------------------------------------------------------------
The American Dialect Society - http://www.americandialect.org