Stephen Krashen DRAFT
December, 2016, revised April 2017
There are excellent reasons for the study
of Latin in schools, including the rich culture, literature, and history of Rome, as well as the opportunity to see the results
of language change and borrowing, and the chance to study the litergical use of
But it is not clear that improving
students' English vocabulary is a good reasons for studying Latin. Studies done over the last century confirm
that students who study Latin do better on tests of English vocabulary (1, 2). The effect, however, may be swamped by
another, far more efficient way of building vocabulary.
In every study of the impact of Latin,
the tests of English vocabulary were given soon after Latin study ended, typically
at the end of the academic year. There are no studies confirming the impact of
Latin study in school on adult vocabulary later in life. In fact, there is
reason to suspect that this impact may not exist for dedicated readers or be
There is strong evidence that
self-selected reading for pleasure is a powerful predictor of adult vocabulary,
no matter when it is done during the lifespan (up to age 42 that is; Sullivan
and Brown 2014): Reading for pleasure,
especially fiction, is a significant predictor of adult vocabulary size, and
there is no critical age – more reading as an adult is strongly related to
vocabulary size, controlling for earlier reading.
In other words, older readers who
continue to read for pleasure and interest continue to increase their vocabularies.
There are no studies I know of directly comparing the long-term effect of Latin
study in school on vocabulary with the development of a reading habit. (But see appendix 1.)
What needs to be investigated: Does Latin
study in school have a positive impact
on adult vocabulary size independent of pleasure reading? In other words, given two equally well-read
adults, one who has studied Latin and one who has not not, will the former
Latin student have a larger vocabulary?
My prediction is that the amount of reading will predict vocabulary
size, but that Latin study will make no significant additional
I am suggesting, in other words, that the
power of reading alone is so strong that Latin study will make little
difference. This is, of course, only a
prediction that needs to be confirmed by research, but if it is true, the
argument that Latin is a vocabulary enhancer is only valid in the short run.
Starch (1930) presents data that supports
the prediction that the effect of Latin study wears off, but only compares high
school juniors and college students, the sample is small, and no significance
testing was done. Nevertheless, the trend is clear: University students have
larger vocabulary test scores than high school students, and the "Latin
advantage" is smaller for them (2% versus 4.5%). In fact, university students with no Latin did
better than high schools who had Latin.
From Starch, 1930, table 61.
Bowker (1975), also discussed in footnote one, reported that high-school students with two years of Latin study outscored those without Latin study, but the difference was modest, only 6% on a 150 item test: On a subset of the test contained only Latin-derived words, the Latin student advantage was only 2%.
My hypothesis is that
a reading habit will eventually wipe out any advantage for Latin study on
English vocabulary. This hypothesis also
predicts that there will be less of a Latin advantage if students have
developed a pleasure reading habit. These privledged private school students
most likely had the advantage of a print-rich environment, and most likely many
were dedicated pleasure readers, which explains the weak results for Latin
1: The earliest studies I could find were
Harris (1915) and Otis (1922). Some additional details about the studies: Bassman and Ironsmith (1984) claim that Latin
study resulted in "significantly greater gains in vocabulary than did the
control students" (p. 41). But a close look at the data shows that the
control students made no gain at all over the academic year, and the gain for
the Latin students was about a normal year's growth (effect size = .31 on the
Vocabulary Portion of Stanford Achievement Test).
In some studies, vocabulary tests
contained only words of Latin origin (e.g.
Otis, 1922) and in some studies the list was not restricted, but Latin
origin words make up about half of
English vocabulary (Fromchuck, 1984; Barber, 1985). Strangely, the Latin advantage for Bowker's subjects, high school students from a private school, was larger for a test containing both Latin origin and non-Latin origin words than for only the Latin origin subset. Latin students had only a 2% advantage over non-Latin students on the subset.
2. Could the Latin
advantage be due to pre-existing differences between those who take Latin and
those who do not? If this were the case,
we would expect that those taking Latin
as an elective would score better on English vocabulary tests than those who
did not on tests given before Latin instruction begins.. There is evidence that
this is true (Carr, 1921; Wilcox, 1917), but it was also the case that the gap
between the groups increased with Latin study (Carr, 1921; Wilcox, 1917).
Also, when students
are matched for initial competence, some researchers report that Latin still
has a positive effect on vocabulary test scores (Carr, 1919; Perkins, 1914, who matched for English and
foreign language grades; Paroughian, 1942, matched for IQ).
Douglass and Kittelson (1935) matched students for SES, years of foreign
language study, and English grades, and reported that Latin made only a small,
but positive difference on an English vocabulary test, and Pond (1938), who
also matched students on a variety of factors, reported similar results.
In summary, Latin
students could indeed have an initial advantage, but Latin study still appears
to have a positive effect on scores of English vocabulary tests.
3: It has been argued that the Latin
advantage comes from knowledge of Latin roots and affixes. But the effect of
context on comprehension (clues to meaning outside the word) may be more
powerful than knowledge of roots and
suffixes (clues to meaning inside the word).
Even if knowledge of roots and affixes
are of significant help to young readers, it is likely that well-read adults
who have never studied Latin will have acquired much of this knowledge through reading.
My guess is that you have acquired the
meaning of least half of them.
Bassman, M. and Ironsmith, M. 1984. An
experimental FLES program in Latin. ADFL Bulletin 14, 39-41
Barber, G. 1986. Latin as a practical study. The Classical Journal
Bowker, R. 1975. English
vocabulary comparison of Latin and non-Latin students. Technical Report 831 Johnson O'Connor
Research Foundation, Human Engineering
W. L. Carr, W. L. 1919. English vocabulary of high school freshmen. Classical
Douglass, H. and Kittelson, C. 1935. The
transfer of training in high school Latin to English grammar, spelling, and
vocabulary. Journal of Experimental Education 4(1): 26-33
A. 1984. The measurable benefits of teaching
English through Latin in elementary School. Classical World 78: 25-29.
Harris, J. 1915. A study in
the relation of Latin to English composition. School and Society 2: 251-52.
Otis, A. 1922. The relation of Latin to the study of
English vocabulary and composition.
School Review 30: 45-50.
Parounagian, M. 1942. The Portland derivatives test.
Classical Outlook, 19: 84-95.
Perkins, A, 1914, Latin as a vocational study in the
commercial course. Classical Journal 10: 716.
Pond F. 1938. Influence of the study of Latin on word
knowledge. The School Review, 46: 611-618
Starch, D. 1915. Some experimental data on the value of studying foreign
languages. School Review 23: 697-703.
Starch, D. 1923. Educational Psychology. New York:
Sullivan, A. & Brown, M. (2014). Vocabulary from
Adolescence to Middle Age.
London: Centre for Longitudinal Studies, Institute
of Education, University of London.
Wilcox, M. 1917. Does the study of high-school Latin
improve high-school English? School and Society, 6 (132): 58-60.