example Norris and Ortega are cited in an article inApplied Linguisticsby
Lightbown in 2000. Given the production time of articles that means that
she had access to this article in 1998. This is of course not uncommon; earlier
versions of articles are typically circulated and referred to.
Another interesting aspect is the time it takes for an article to“take off”.
The number of citations per year is low in thefirst 5–7 years and the articles
take close to 15 years to reach their full potential. This may also reflect the
overall growth of the number of publications over this period: with a larger
number of publications the chance of being cited increases too.
Related to the impact over time is the problem that some journals built up
backlogs that led to substantial delays in the publication of articles, some-
times up to two years. In recent years more and more journals have made
articles that are still to be published available through their website, which is
useful for thefield but a nightmare to the bibliometrist, who has to define
when a paper was published. Is it the moment the article was available online
or the moment it was published in the paper version of the journal? Long
processing times can be a nuisance, in particular for PhDs whose dissertation
consists of journal articles. Many graduate schools require that at least three
articles in the dissertation have been formally accepted and preferably published.
9.6 Factors influencing citation scores
As mentioned earlier, the data provided answers to a number of questions.
Each question will be dealt with separately below.
How are the citations distributed over time?
The frequency distributions of the citations over time were calculated.
The correlation between the number of citations and the year of publication
is not significant. The distribution shows a peak in the period between 1998
and 2002. As we have seen, publications take time to have an impact, so the
later articles will also be more influential in the years to come. In addition,
the median for the years of publication of this set is 1987, with the oldest
publication from 1972 and the most recent from 2014, which is actually
beyond the time span of this study, but the two cases are recent editions of
older books, like Lightbown and Spada and Ellis. Two factors play a role
here: in order to show that they are up to date, researchers will tend to
include the latest publications on a topic, so there will be a recency effect.
On the other hand, older publications have more time to attract citations.
The former factor should lead to a bias toward recent publications, while the
latter would lead to a bias for older publications. The normal distribution
suggests that the two factors are more or less in balance. For publications
from the 1970/1980s an additional problem is that you could only cite
something if you had access to a physical copy of it, and that was no doubt a
limiting factor in citations (Paul Meara p.c.).
The citation game 113