Language Learning 48:2, June 1998, pp.

281–317

Tracking the Incremental Acquisition of Second

Language Vocabulary: A Longitudinal Study

Norbert Schmitt
University of Nottingham

Previous vocabulary research has focused on the size of

lexicons and the number of words learned through various
activities. To date, this type of research has generated little
understanding of how individual words are acquired. To
explore this issue, I tracked the acquisition of 11 words over
the course of a year for 3 adult learners with advanced profi-
ciency in English. I measured 4 kinds of word knowledge:
spelling, associations, grammatical information, and mean-
ing. The participants had little problem with spelling, but
rarely knew all of a target word’s meaning senses or deriva-
tional word forms. Knowledge of the meaning senses of the
target words improved about 2.5 times more than it was
forgotten, and some of the word knowledge types appear to
be interrelated. However, the study did not show evidence
of a developmental hierarchy for word knowledge types.

The mechanics of vocabulary acquisition is one of the

more intriguing puzzles in second language acquisition (SLA).

I would like to thank the participants for their patience and help. Thanks
also to Della Summers and Keith Mardell of Longman Publishing for access
to the BNC, and Gwyneth Fox and Rosamund Moon of COBUILD for access
to the Bank of English corpus. Ron Carter, Mike McCarthy, Paul Meara, and
Diane Schmitt provided valuable comments on earlier versions of this paper.
Correspondence concerning this paper may be sent to the author
at The University of Nottingham, Department of English Studies,
University Park, Nottingham NG7 2RD, United Kingdom. Internet:
[Link]@[Link]

281
282 Language Learning Vol. 48, No. 2

Unfortunately, researchers still can say little with confidence. As

Paul Nation has noted:
There isn’t an overall theory of how vocabulary is acquired.
Our knowledge has mainly been built up from fragmentary
studies, and at the moment we have only the broadest idea
of how acquisition might occur. We certainly have no knowl-
edge of the acquisition stages that particular words might
move through. (Schmitt, 1995b, p. 5)
This lack of knowledge stems not from lack of effort; as Meara
(1987, 1992) and many others have pointed out, there is now a
virtual explosion of vocabulary studies. One reason this increased
research has not yet translated into an adequate understanding
of vocabulary acquisition is because nearly all of it has focused on
the size and growth of lexicons, on how many words are gained (or
forgotten) over time. Among the myriad examples of this research
perspective are studies that have attempted to quantify the
number of words average native speakers (NSs) know (D’Anna,
Zechmeister, & Hall, 1991; Goulden, Nation, & Read, 1990), the
number of words non-native speakers need to know (Hazenberg
& Hulstijn, 1996; Laufer, 1992), the numbers of words learned from
incidental exposure while reading (Huckin, Haynes, & Coady,
1993; Nagy, Herman, & Anderson, 1985), and the number of words
which can be learned by using various exercises, techniques, and
strategies (Avila & Sadoski, 1996; Cohen & Aphek, 1980).
This line of research, though obviously important, does not
explain how individual words are acquired diachronically. Logic
suggests that this requires research which focuses on the individ-
ual words themselves, rather than on the overall growth of lexi-
cons. Such research needs to incorporate a number of key features.
First, the studies have to chart vocabulary development within
individual persons, because learners probably know somewhat
different vocabulary even if they are members of rather homoge-
neous groups; this makes generalizations about the development
of any individual word problematic in a group study. Second, the
studies need be longitudinal in nature: This stems from the
commonsense fact that describing acquisition or loss within an
 Schmitt 283

individual person requires at least two separate measurements.

Third, because words are learned incrementally, and not in a
dichotomous “not acquired/acquired” manner (Meara, 1984; Shar-
wood Smith, 1984), acquisition studies need measurement proce-
dures that can capture degrees of lexical knowledge. Fourth, full
mastery of a word requires more than just a knowledge of its
meaning and form (Nation, 1990, pp. 30–33; Richards, 1976).
Nevertheless, most previous studies have accepted recogni-
tion or production of a single meaning sense as evidence that a
word is “known,” even though this type of response demonstrates
only partial knowledge at best. A truly adequate test of acquisition
would have to capture a much wider range of lexical knowledge.
For example, many teachers assume that grammatical aspects
like word class and derivational morphology are easily learned,
but research has shown that one cannot assume that learners
have mastered these elements. Alderson, Clapham, and Steel
(1997) found that their participants often did not have clear
metalinguistic knowledge about which word class different words
belonged to. Similarly, Schmitt and Meara (1997) found that their
Japanese intermediate students typically had trouble changing
root word forms into their derivatives. These two grammatical
aspects are obvious topics for investigation.
Studies that inform about individual word acquisition have
been either very specific or are thin on the ground. Psychology
journals are filled with studies that have measured the acquisition
of various features of individual words (often in laboratory condi-
tions), such as intuitions of frequency (Hintzman, Block, & Sum-
mers, 1973), a word’s orthographical form (Thomas & Dieter,
1987), and word associations (McNeill, 1963). Other studies have
similarly focused on single elements of lexical knowledge, such as
meaning (Carey, 1978). The problem lies in the fact that these
specifically focused studies tell us little about what else is happen-
ing in the acquisition process at the same time. Surely acquisition
of at least some word knowledge components occurs concurrently;
so researchers need to investigate them together at the same time
in order to establish their interrelationships. Studies that have
284 Language Learning Vol. 48, No. 2

done this are rare. Two of the few that have tackled concurrent
tracks of acquisition were by Nagy, Diakidoy, and Anderson (1993;
meaning and derivative suffixes) and Schmitt and Meara (1997;
word association and suffixes, along with vocabulary size).
To study the incremental acquisition of individual words, one
must be able to measure the degree or depth of knowledge for each
of those words. There are two main approaches for doing this: a
developmental approach and a dimension approach (Read, 1997).
The developmental approach uses scales to describe the stages of
acquisition of a word. Many of these describe 4 levels of knowledge,
including scales by Dale (1965, p. 898), Drum and Konopak (1987,
pp. 79–80), and Schmitt and Meara (1997). One scale that has
received some attention lately is the Vocabulary Knowledge Scale
(Paribakht & Wesche, 1993), which has 5 levels:
Stage 1: The word is not familiar at all.
Stage 2: The word is familiar but the meaning is not known.
Stage 3: A correct synonym or translation is given.
Stage 4: The word is used with semantic appropriateness in a
sentence.
Stage 5: The word is used with semantic appropriateness and
grammatical accuracy in a sentence.
Perhaps the main advantage of using scales is that they
promote an incremental notion of vocabulary acquisition, rather
than the dichotomous not acquired/acquired view. But they pose
serious problems as well. Scales attempt to measure stages of
knowledge in vocabulary acquisition, but defining the stage
boundaries may be problematic if acquisition is in fact a contin-
uum. Scales also tend to suffer from uneven intervals between the
stages. In addition, there is at present no principled way of deter-
mining the number of stages necessary to accurately describe the
acquisition process. Finally, it is difficult to give balanced attention
to both receptive and productive knowledge within a scale; the
 Schmitt 285

beginning levels of a scale usually focus on receptive knowledge

and the more advanced levels on production.
The dimension approach, on the other hand, describes the
level of mastery of the various component types of word knowl-
edge. This approach has its roots in a seminal paper by Richards
(1976) which set out a number of competencies necessary for
mastery of a word. Later scholars have taken up the idea, suggest-
ing their own lists of word knowledge types (Alexander, 1982;
Blum-Kulka, 1981; Laufer, 1997). Perhaps the most complete and
balanced description of word knowledge to date is that proposed
by Nation (1990, pp. 30–33). It consists of 8 word knowledge
categories, each of which has receptive and productive aspects:
1. The spoken form of a word;
2. The written form of the word;
3. The grammatical behavior of the word;
4. The collocational behavior of the word;
5. The frequency of the word;
6. The stylistic register constraints of the word;
7. The conceptual meaning of the word;
8. The associations the word has with other related words.
Both the weakness and the strength of such a word knowl-
edge list lie in its comprehensiveness. It is hard to imagine how
addressing each of these word knowledge types for each individual
target word could be practical in any normal classroom teaching
or testing situation. Even though the pedagogical applications of
such a list may be limited, Schmitt (1995a) suggested that it could
still be useful to classroom teachers as a conceptual framework
with which to evaluate various vocabulary tests and vocabulary
learning activities. By viewing the tests and activities through
such a framework, teachers can come to a clearer understanding
of which word knowledge types the tests and activities are and are
not addressing.
286 Language Learning Vol. 48, No. 2

It is in a research context that the dimension approach may

prove to be of more value. True, measuring knowledge of several
of the above dimensions would be time-consuming and would
severely limit the number of words that could be studied. However,
such research could produce a very rich description of vocabulary
knowledge, making it well worth the effort. Schmitt and Meara
(1997) noted that the present word knowledge lists are purely
descriptive and have no explanatory power. However, they also
suggest that such descriptive summaries can be used as frame-
works for explanatory research. If research based on a word
knowledge framework could lead to a better understanding of how
each lexical dimension is acquired, researchers would surely be
closer to piecing together the overall lexical acquisition process,
something that remains a major gap in the field. This knowledge
may later make it possible to isolate a few key dimensions of
lexical knowledge, which may be limited enough to have pedagogi-
cal practicality. Meara (1996a) suggested that the size and organi-
zation of the lexicon might be two of these dimensions.
The dimension approach has several advantages beyond its
comprehensiveness. One is the simplifying effect of breaking com-
plex behavior (vocabulary acquisition) into its more manageable
components for analysis. Also, analyzing the components sepa-
rately allows the possibility of discerning their relationships. A
number of these relationships have long been obvious (e.g., be-
tween frequency of occurrence and formality of register; between
word class and derivational suffixes); Schmitt and Meara (1997)
recently demonstrated some of these interrelationships correla-
tionally. An intriguing possibility is that some of these relation-
ships are hierarchal; that is, learned in some type of
developmental order. Developmental sequencing has been posited
in other areas of language, syntactic structures (e.g., Pienemann
& Johnston, 1987) and morphemes (e.g., Larsen-Freeman, 1975),
so it would not be surprising if the principle obtained in the area
of lexical acquisition as well.
In fact, it seems counterintuitive that word knowledge is not
at least partially hierarchal. It is unlikely that the initial exposure
 Schmitt 287

to a word yields much more than some partial impression of its

written or phonological form and one of its meaning senses,
possibly in an idiomatic usage. After more exposures (or some
explicit study), a learner would gradually learn the other kinds of
word knowledge, with perhaps collocational and stylistic knowl-
edge being the last. Indeed, it doesn’t seem reasonable that a
learner would have a rich associative and collocational network
built up without a knowledge of the word’s form, for instance.
Bahns and Eldaw (1993) found some evidence for this intuition;
their participants’ collocational knowledge lagged behind general
vocabulary knowledge. Research designs based on a word knowl-
edge framework would allow investigation into whether some
kinds of word knowledge are acquired before others.
Finally, such word knowledge research may lead to a better
understanding of the movement of vocabulary from receptive to
productive mastery. This movement is still something of a mystery;
researchers are not even sure whether receptive and productive
knowledge forms a continuum, as Melka (1997) argued, or whether
it is subject to a threshold effect, as Meara (1996b) has suggested.
Part of the problem is the typical assumption that the whole word
is either receptively or both receptively and productively known.
The actual situation is probably that, for any individual word, each
of the different types of word knowledge is known to different
receptive and productive degrees. For example, a word’s spelling
might be productively known, some of its meaning senses recep-
tively known, and its register constraints totally unknown. Thus,
research into the underlying receptive/productive word knowl-
edge states should prove informative about learners’ overall abil-
ity to use words in a receptive versus productive manner.
For all of these reasons, I believe that the dimension approach
and the word knowledge framework should be investigated as an
avenue of research. This article reports on an exploratory study
that incorporated the elements discussed above in a longitudinal
study of vocabulary acquisition. It attempted to describe the acqui-
sition of individual words by measuring the development of 4 types
of word knowledge: written form, associations, grammatical
288 Language Learning Vol. 48, No. 2

behavior, and meaning. Ideally, the study would have measured

each of the 8 word knowledge types, but this was not practical due
to time constraints.

Method

Participants

The 4 participants were international students just begin-

ning a course of postgraduate study at a British university. They
came from different countries and were studying in different
departments. They had never previously resided in a country
where English was the first language, except for their respective
university presessional courses. The students voluntarily agreed
to participate; I paid them a token fee of £10 (UK) per session.
Details about the students appear in Table 1. Longitudinal studies
are prone to participant attrition; unfortunately, this one was no
exception. Kor dropped out of her program before the second
session, so there are no longitudinal results for her.

Table 1

Longitudinal Study Participants

Pseudonym ‘Lith’ ‘Kor’ ‘Tai’ ‘Ind’

Native Lithuania Korea Taiwan India
Country
L1 Lithuanian Korean Chinese Tamil
a
Months 2 8 3 5
Age 30 24 30 27
Sex Male Female Male Male
Course of Politics Neural Insurance Life
Study Science Science
TOEFLb 560 530 537 547
a
Length of time spent in English-speaking countries at time of first test
b
Score as submitted as part of admission requirements
 Schmitt 289

Target Words

Because this study investigated the target words in some

detail, I could include only very few, making the selection criteria
even more important than usual. First, I needed a high likelihood
that the participants would be exposed to the words during their
course of study. Because the participants belonged to different
departments in a British university, I considered general academic
vocabulary most suitable. Therefore, I took the majority of the
target words from the University Word List (UWL: Xue & Nation,
1984; reprinted in Nation, 1990, pp. 265–239), a list of academic
words occurring frequently across various academic disciplines.
Second, I chose only polysemous UWL words that had 3 or more
meaning senses, because this would allow examination of the
participants’ growing awareness of the different senses. Third,
because the participants might at least partly know all of these
words (being advanced enough to enter British postgraduate
programs), I required a few lower-frequency, relatively unknown
words in order to allow insights into the beginning stages of lexical
acquisition. I selected 3 words from the 4,000 to 5,000 word level
of the Brown word list (Francis & Kucera, 1982) with the same
criterion of having at least 3 major meaning senses. To confirm
that the eventual target words were likely to fall at various points
on the acquisition continuum for the participants, I piloted 18
candidate words on 12 international students, who were similar
to the eventual participants, to check their familiarity. I reduced
the candidate list to 11 target words, which later piloting indicated
was the maximum that could be addressed in a session approxi-
mately 2 hours long. They included 2 relatively unknown words
(brood, spur), 4 relatively well-known words (abandon, dedicate,
illuminate, suspend), and 5 words in between (circulate, convert,
launch, plot, trace).
290 Language Learning Vol. 48, No. 2

Word Knowledge Types and Measurement

I focused on knowledge of spelling, associations, grammatical

information, and meaning. I also measured collocation, but the
results indicated that the experimental measurement procedure
needed further development (see Schmitt, in press a). I eliminated
the other word knowledge types because (a) the target words had
no obvious register marking, (b) the words were given to the
participants orally, thus eliminating the possibility of a productive
measure of phonological form, and (c) a previous study had high-
lighted the difficulty of measuring intuitions of frequency (see
Schmitt & Dunham, 1997).
On the assumption that the various word knowledge types
are learned incrementally, the measurement procedures had to be
as sensitive as possible to increments in the different word knowl-
edge aspects, ranging from no knowledge at all to full NS-like
mastery. The test of written form (spelling) consisted of a 4-point
rating system. Zero (0) on the scale indicated that the participant
demonstrated no knowledge of a word’s spelling. One (1) signified
that the participant could give the initial letters of the target word,
but omitted some later letters, added unnecessary letters, or
transposed letters. Two (2) indicated that the word was phonologi-
cally correct, but perhaps some vowels or consonants were re-
placed by similar-sounding but erroneous items (brood—*brud;
illuminate—*elluminate). Three (3) indicated fully correct spell-
ing. This scale almost certainly suffers from the scale-specific
problems discussed above, but I considered it adequate to indicate
a progression in the mastery of written form. In an assessment of
how NSs might perform on this measure, 3 NS pilot respondents
(see below) scored 100% correct spelling of the target words.
The association measurement procedure asked participants
to give 3 responses for each target word stimulus. These responses
I compared to an NS norming list. I scored each of the participants’
responses according to how many of the norming NSs also gave
that response. The 3 scores were totalled and matched against a
4-category profile of NS behavior. In Category 0, none of the 3
 Schmitt 291

responses matched any of those on the norming list, in which case,

no NS-like association behavior was demonstrated. In Category 1,
some responses matched infrequent ones on the norming list,
indicating a minimal amount of NS-like association knowledge. In
Category 2, the responses were similar to those typical of the NS
norming group, indicating NS-like associations. Last, in Category
3, the responses were similar to those in the top half of the NS
norming group, indicating a NS-like rating in which even more
confidence can be put (see Schmitt, in press b, for more detail).
Eighty-two percent of the association responses from the 3 NS
respondents belonged in Categories 2 or 3, while 12% fell into
Category 1 and 6% into Category 0.
I obtained the norming word class and derivational forms
from 3 dictionaries: the Longman Dictionary of English Language
and Culture (1992), the Oxford Advanced Learner’s Dictionary
(1995), and Webster’s Ninth New Collegiate Dictionary (1987). The
participants got 1 point for knowing the word class of the target
word, and 1 point for knowing how to transform it into each of the
3 other word classes. If a form for a word class does not exist,
participants got credit for being able to state that fact. When 2 or
more forms were possible for any word class, only 1 was required
for credit. For example, participants were awarded 1 point for
knowing that illuminate is a verb, and 1 point each for knowing
illumination is the noun form, illuminated or illuminating (only
1 required) is the adjective form, and that no common adverb form
exists. During the development of this section, I noted that the
norming data from the dictionaries sometimes conflicted with the
NS pilot participants’ answers, particularly for adverbials, with
the dictionaries occasionally listing forms that the NSs found
strange. In these cases, I consulted the British National Corpus to
check those forms’ frequency of occurrence. If it was very low, I still
accepted it as a possible form for that word class, but I also
considered acceptable an answer that no form existed. For exam-
ple, the very rare adverb form of circulate, circularly, is so uncom-
mon that I also accepted the answer “No form exists.” Thus, the
possible scores ranged from 0 (knowledge for no word class) up to
292 Language Learning Vol. 48, No. 2

4 (knowledge for all four word classes). The NS pilot participants

knew all 4 word classes for 70% of the target words and 3 word
classes for the remaining 30%.
Because this study attempted to describe vocabulary acqui-
sition up to the level of full mastery, it was important to measure
knowledge of all of the major meaning senses of the target words.
(Knowing only a single meaning sense for a polysemous word must
be considered only partial knowledge.) I consulted the 3 diction-
aries to determine the major meaning senses. For cases in which
they disagreed, I made decisions based on the responses from both
the NS and NNS pilot participants and on corpus data. Whereas
I only measured the other word knowledge types productively, it
was both feasible and desirable to measure both receptive and
productive knowledge of word meaning, because a major part of
the incremental acquisition of word meaning probably involves
the move from receptive to productive mastery of different mean-
ing senses. In addition, meaning has traditionally been the type
of word knowledge given most weight in vocabulary testing; it was
relatively easy to tap into the participants’ receptive knowledge.
I asked the participants to explain all of the meaning senses
they knew for each target word. After the participant could not
think of any additional senses, I gave prompt words designed to
elicit additional senses that the participant might know but could
not recall. The prompts were designed to trigger the related sense
if the participant knew it, but not to give it away if it were
unknown. For example, for the target word spur, the prompt word
horse was designed to suggest the meaning, “metal device worn on
the heel of a boot used to guide or encourage a horse.” If the
participant could not describe the meaning sense after the prompt,
I scored that meaning sense as “unknown.”
With no real guidance in the literature about how to relate
productive and receptive meaning knowledge to each other and
score them, I devised the following scoring system. Because the
whole study was an exploratory look into a new area, I felt that a
simple, transparent system would be best. Therefore, I assumed
unprompted explanations of meaning sense demonstrated
 Schmitt 293

productive knowledge and awarded them 2 points. I considered

prompted explanations to be related to receptive knowledge and
gave them 1 point. Unknown meaning senses received 0 points.
Because the target words had differing numbers of meaning
senses, a convenient way of comparing the different words was to
calculate a meaning proportion by taking the participant’s total
point score for each word and dividing that by the number of
possible points (= number of meaning senses × 2 points each).
Thus, a meaning proportion of .50 could indicate knowing all
meaning senses receptively, half productively, or some combina-
tion of the two.
Of course, even NSs do not necessarily know every meaning
sense for all of the target words, but the 3 NS pilot participants
demonstrated an understanding of most of the 61 meaning senses:
Participant A, productive knowledge 56%, receptive knowledge
20%, unknown 24%, meaning proportion .66; Participant B, 69%,
21%, 10%, .80; Participant C, 61%, 21%, 18%, .71.
Once the entire measurement battery was assembled, I pi-
loted it on the 3 NSs mentioned above, resulting in some fine-
tuning of the meaning prompts. Next, I tested 3 international
postgraduate students (1 Japanese, 1 Brazilian, and 1 Taiwanese)
from the same cohort as the eventual study participants. Their
results indicated that the measurement battery would be suitable
for the main study.

Interview Procedure

The first session (T1) was at the beginning of November 1995,

with subsequent sessions at approximately half-year intervals.
The only exception was Tai’s T3, which was administered 3 months
after his T2 because he left his university early due to financial
reasons.
The format was a one-on-one interview, with me working
down the interview instrument (see Appendix) in lockstep order.
The sessions were generally held in my or the students’ office; I
tape-recorded them for future reference. After a brief exchange of
294 Language Learning Vol. 48, No. 2

pleasantries to set the student at ease, I either explained or

reminded the student about the format of the session. (In the T1,
we went through the entire battery for the example word secure.)
Next, I checked to see whether the student was comfortable with
the metalinguistic notions and terms noun, verb, adjective, and
adverb. If so, the interview proceeded. Occasionally some instruc-
tion was required before this point. The students had lists of words
that belong uniquely to a single word class for reference. These
steps insured that lack of metalinguistic knowledge was not a
factor in the grammar task.
At this point, the session proper began. The first type of word
knowledge measured was written form. I asked “How do you spell
_______?” The student had pencil and paper available throughout
the interview, and was allowed to either spell the word out orally
or write it out. In practice, the students normally did both simul-
taneously. If they misspelled the word, I gave the correct spelling
at that point so that they would have it in front of them for the
rest of the questions on that word. I next elicited associations by
using the following instruction, “Please give the first 3 words you
think of when you hear the word ______.” The next lexical aspect
elicited was grammatical knowledge, initiated by the prompt
“What word class (part-of-speech) is _______?” Depending on the
answer, I went on to ask “Is there a (noun, verb, adjective, adverb)
form? If so, what is it?” for the remaining 3 word class forms. I then
asked the students to explain any meaning senses they knew for
the target word, reminding them to use any means necessary to
convey understanding of the meaning senses (give definitions, give
examples, use the word in sentences, draw sketches or diagrams,
use gestures, etc.). After I determined which meaning senses the
student knew productively, I gave the prompts, one-by-one, includ-
ing the word class of that meaning sense, to elicit any receptive
knowledge of meaning. In the T2 and T3 sessions, after the tests
for each word, I went back and explained the different meaning
senses to the student, both to keep the students’ interest up during
the rather long interview sessions and to make sure that they
 Schmitt 295

received at least some input from which they could learn about
the words. After this was finished, I went on to the next word.

Results

The results for all 4 students are summarized in Tables 2

through 5. The study’s design allowed statements about how the
4 kinds of word knowledge developed over time, but not why this
knowledge developed, or what facilitated or inhibited develop-
ment. Lith and Ind both confirmed that they did not study the
words between test sessions, and so the only time they explicitly
focused on the words was in the T2 and T3 sessions. Thus they
must have gained any additional exposure in a naturally-
occurring context. On the other hand, Tai reported explicitly
looking up the words in a dictionary and studying them to a
minimal extent.

Discussion

Meaning Knowledge

In only one case was a meaning proportion of 1.00 attained

in a single session, signifying that all of the meaning senses of that
target word were known productively. Thus, in all but this one case,
the students had only partial meaning knowledge. In addition, the
average meaning proportion was generally less than .50, indicat-
ing that the “partial” knowledge was nowhere near full productive
mastery. In only 5 cases—all from one student, Lith—were all
meaning senses known to some degree (indicated by “Unknown”
= 0). The limited state of meaning knowledge these tables paint is
surprising; one might have assumed that advanced students like
these would know the majority of the target words fairly well. The
upshot is that advanced NNSs may have mastery over only a
rather limited number of the possible meaning senses of a word,
even if they are proficient enough to study in British universities.
Table 2

Longitudinal Study Results for Lith

Spelling Association Grammar Meaning

Session 1 2 3 1 2 3 1 2 3
1 2 3
P R U MP P R U MP P R U MP
Abandon 3 3 3 2 3 2 2 2 3 1 1 4 .25 1 3 2 .42 3 1 2 .58
Brood 2 2 3 0 0 2 2 2 2 0 0 6 .00 0 0 6 .00 1 0 5 .17
Circulate 3 3 3 1 1 2 2 2 3 1 2 1 .50 1 3 0 .63 1 1 2 .38
Convert 3 3 3 1 1 2 2 3 2 1 2 1 .50 3 0 1 .75 3 0 1 .75
Dedicate 3 3 3 2 2 2 3 2 2 1 3 0 .63 3 1 0 .88 2 1 1 .63
Illuminate 2 2 2 3 3 3 3 2 3 2 1 1 .63 2 1 1 .63 4 0 0 1.0
Launch 2 3 3 0 3 3 3 2 4 2 1 5 .31 2 4 2 .50 5 3 0 .81
Plot 3 3 3 0 1 2 3 2 3 0 1 5 .08 0 2 4 .17 3 0 3 .50
Spur 2 2 3 1 1 0 2 2 4 0 0 7 .00 0 0 7 .00 0 0 7 .00
Suspend 3 3 3 2 2 3 2 2 3 2 0 3 .40 4 0 1 .80 4 0 1 .80
Trace 3 3 3 0 2 2 3 2 3 1 3 3 .36 4 1 2 .64 4 0 3 .57
Mean 2.64 2.73 2.91 1.09 1.73 2.09 2.45 2.09 2.91 1.00 1.27 3.27 .33 1.82 1.36 2.36 .49 2.73 0.55 2.27 .56
Note. P=productive knowledge, R=receptive knowledge, U=unknown, MP=meaning proportion.
Table 3

Longitudinal Study Results for Ind

Spelling Association Grammar Meaning

Session 1 2 3 1 2 3 1 2 3
1 2 3
P R U MP P R U MP P R U MP
Abandon 3 3 3 1 0 0 2 2 2 1 1 4 .25 1 2 3 .33 1 0 5 .17
Brood 2 2 2 0 0 0 1 1 0 0 0 6 .00 0 0 6 .00 0 0 6 .00
Circulate 3 3 3 1 3 3 2 2 2 2 1 1 .63 2 0 2 .50 1 2 1 .50
Convert 3 3 3 1 1 3 4 3 3 2 0 2 .50 2 0 2 .50 2 0 2 .50
Dedicate 3 3 3 2 2 2 2 2 2 1 1 2 .38 2 1 1 .63 3 0 1 .75
Illuminate 3 3 3 3 3 3 3 3 1 2 1 1 .63 2 0 2 .50 2 0 2 .50
Launch 2 2 2 3 2 3 2 3 1 3 0 5 .38 2 2 4 .38 3 2 3 .50
Plot 3 3 3 2 2 1 2 3 2 2 2 2 .50 3 0 3 .50 2 0 4 .33
Spur 3 2 2 2 2 3 1 2 1 0 0 7 .00 0 0 7 .00 0 0 7 .00
Suspend 3 3 3 1 1 0 3 2 3 2 0 3 .40 2 1 2 .50 2 0 3 .40
Trace 3 3 3 2 1 1 1 3 1 1 1 5 .21 2 1 4 .36 2 1 4 .36
Mean 2.82 2.73 2.73 1.64 1.55 1.73 2.09 2.36 1.64 1.45 0.64 3.45 .35 1.64 0.64 3.27 .38 1.64 0.45 3.45 .36
Note. P=productive knowledge, R=receptive knowledge, U=unknown, MP=meaning proportion.
Table 4

Longitudinal Study Results for Tai

Spelling Association Grammar Meaning

Session 1 2 3 1 2 3 1 2 3
1 2 3
P R U MP P R U MP P R U MP
Abandon 3 3 3 1 1 1 2 1 4 1 1 4 .25 1 2 3 .33 1 1 4 .25
Brood 2 2 3 0 0 2 2 2 3 0 0 6 .00 0 1 5 .08 1 1 4 .25
Circulate 3 3 3 2 3 2 2 2 4 1 0 3 .25 1 0 3 .25 1 1 2 .38
Convert 3 3 3 1 2 3 2 3 3 1 1 2 .38 1 1 2 .38 2 0 2 .50
Dedicate 3 3 3 2 0 2 2 2 3 1 0 3 .25 2 1 1 .63 1 1 2 .38
Illuminate 3 3 3 1 3 3 2 3 3 2 0 2 .50 2 0 2 .50 3 0 1 .75
Launch 2 3 3 0 3 3 3 2 4 0 1 7 .06 2 2 4 .38 3 1 4 .44
Plot 3 3 3 1 3 3 2 2 3 0 2 4 .17 2 2 2 .50 2 1 3 .42
Spur 0 3 3 0 2 0 2 4 4 0 0 7 .00 1 1 5 .21 4 0 3 .57
Suspend 3 3 3 2 3 3 1 2 2 2 1 2 .50 2 2 1 .60 2 1 2 .50
Trace 3 3 3 2 2 2 0 3 4 2 1 4 .36 2 2 3 .43 2 2 3 .43
Mean 2.55 2.91 3.00 1.09 2.00 2.18 1.82 2.36 3.36 0.91 0.64 4.00 .25 1.45 1.27 2.82 .39 2.00 0.82 2.73 .44
Note. P=productive knowledge, R=receptive knowledge, U=unknown, MP=meaning proportion.
Table 5

T1 Results for Kor

Spelling Association Grammar Meaning

P R U MP
Abandon 3 1 3 0 5 1 .42
Brood 2 2 3 0 1 5 .08
Circulate 3 2 2 1 0 3 .25
Convert 3 2 3 0 3 1 .38
Dedicate 3 1 4 1 0 3 .25
Illuminate 3 0 2 1 0 3 .25
Launch 3 2 3 1 2 5 .25
Plot 1 0 1 0 1 5 .08
Spur 3 2 3 1 2 4 .29
Suspend 3 0 2 0 2 3 .20
Trace 3 1 2 2 2 3 .43
Mean 2.73 1.18 2.55 0.64 1.64 3.27 .26
Note. P=productive knowledge, R=receptive knowledge, U=unknown, MP=meaning proportion.
300 Language Learning Vol. 48, No. 2

Longitudinally, 2 of the students, Lith and Tai, progressed

steadily in their meaning knowledge, while Ind remained rela-
tively static. There is no obvious reason for this; Ind seemed to take
the sessions seriously and certainly had better language profi-
ciency than Tai.
Examining the changes in the state of the students’ meaning
knowledge (Table 6), one first notices that the vast majority of
meaning senses stayed at the same state of knowledge (72%,
263/366). This suggests knowledge of meaning sense has a certain
amount of inertia and does not change easily. This is probably to
be expected, as acquiring a large number of meaning senses
quickly and easily might be too auspicious to hope for, at least in
L2 learning. On the other hand, this stability means there is
not a large amount of forgetting either. In the 103 cases where

Table 6

Changes in State of Knowledge for the Different Meaning Senses

of a Word

Lith Ind Tai Totala

R➚P 14 3 10 27
U➚R 8 9 11 28
U➚P 8 2 9 19
Total ➚ 30 14 30 74
P➘R 2 0 5 7
R➘U 4 8 4 16
P➘U 1 3 2 6
Total ➘ 7 11 11 29
P➙P 28 31 19 78
R➙R 11 3 7 21
U➙U 46 63 55 164
Total ➙ 85 97 81 263
Note. P=Productive Knowledge of Meaning Sense, R=Receptive Knowledge
of Meaning Sense, U=Unknown Meaning Sense, ➚=Improves to,
➘=Deteriorates to, ➙=Remains in Same State.
a
N=122 per student (61 meaning senses × 2 intervals, T1-T2 and T2-T3),
N=366 total (122 × 3 students)
 Schmitt 301

meaning knowledge did change, it improved 2.5 times more than

it deteriorated. In fact, forgetting occurred in only about 8% of the
cases; there was improvement in 20%. Each of the students gained
more than they lost in terms of number of progressions and
regressions.
It is informative to check where the regression and progres-
sion occurred. It was uncommon for meaning senses in a produc-
tive state to slip down to either receptive or unknown states (as
defined here). Forgetting was more likely a shift from a receptive
to an unknown state; Tai alone lost more productive knowledge
than receptive knowledge. However, the overall level of forgetting
was still quite low. These results suggest that once a meaning
sense is known productively, it is not very likely to be forgotten, at
least not over a 6-month period.
In the 3 students’ total of 74 cases of progression, meaning
knowledge moved from receptive to productive and from unknown
to receptive a similar number of times. As might be expected, there
were fewer cases of meaning knowledge making the presumably
larger move from unknown to productive, although the existence
of such cases suggests that a productive level of knowledge can be
attained within 6 months with only natural exposure as input. We
know this because there were unknown to productive improve-
ments from T1 (students were not told the meaning senses of the
words in this session) to T2.
Table 7 compares the degree of meaning knowledge with
association and grammar knowledge, again illustrating the par-
tial nature of the students’ meaning knowledge. There were only
3 instances of a meaning proportion higher than .8; the majority
of instances fell in the .2 to .6 range. In general, association and
grammar knowledge seem to increase in line with meaning knowl-
edge. A meaning proportion level of .4 to .6 appears to match the
NS-like association threshold of 2. The data also suggest that
students can possess other kinds of word knowledge even when
they can demonstrate no meaning knowledge. In this situation,
the participants usually demonstrated knowledge of 2 word
classes, which resulted in a mean grammar score of 1.69; in only
302 Language Learning Vol. 48, No. 2

Table 7

Meaning Scores vs. Association and Grammar Scores

Meaning Proportion

.00 .00– .20– .40– .60– .80–

.20 .40 .60 .80 1.0
Association Score .69 .73 1.51 2.09 2.13 2.67
(Max. 3)
Mean Grammar Score 1.69 2.09 2.49 2.56 2.31 3.00
(Max. 4)
Number 13 11 35 32 16 3

one instance was there a grammar score of zero. They also pro-
duced some NS-like associations, but further analysis showed
these were tied to a memorized idiom and so are suspect (see
below).

Knowledge of Written Form (Spelling)

One of the more noticeable things in Tables 2 through 5 is

that the participants did not seem to have much trouble spelling
the target words. Although these words do not seem particularly
tricky, there are still cases where they cannot be spelled directly
from the phonological rendering; for example, the schwa in the
final syllable of abandon could be virtually any vowel. If the
students demonstrated any meaning knowledge of the words at
all, they were almost always able to spell them too. In the 13 cases
where they could not demonstrate any meaning knowledge, they
could still produce a phonologically correct spelling 10 times and
a completely correct spelling twice; only once did the student have
no idea. These students had apparently reached a level of profi-
ciency where they were able to use sound-symbol correspondences
to produce at least phonologically correct spellings of unknown
words.
 Schmitt 303

There were exceptions, however. For example, by T3, Lith

knew all of the meaning senses for the word illuminate produc-
tively (the only full meaning marks in the study) and also scored
high on the other word knowledge measures. Still, Lith persisted
in spelling illuminate with an “e”: T1, “elluminate”; T2, “elumi-
nate”; T3, “eluminate.” Having a good understanding of other types
of word knowledge does not necessarily mean that one will know
how to spell a word correctly.
In most cases, the students improved their spelling scores
over the course of time. Only twice was there any backsliding (Ind,
launch and spur). One might hence infer that once the spelling of
a word is mastered it is not usually forgotten. However, the
students had probably been spelling the vast majority of these
words correctly for quite some time. Therefore, I probably do not
have enough data about words that have been recently learned
and consolidated (e.g., brood and spur for Tai, or plot for Lith) to
make strong claims about words just over the threshold of spelling
control.

Association Knowledge

The associations given by both Ind and Tai became more

NS-like over time; Lith remained at about the same level. This is
difficult to explain; from my experience, Lith had a similar ESL
proficiency to Ind, and noticeably better than Tai. Of the 33
association cases (3 students × 11 words), 23 showed stability or
progress, while 10 indicated some backsliding. However, if we
count the times that a student backslid from NS-like knowledge
(2 or 3) down to not NS-like (0 or 1), then the total number of cases
is only 4. Three of the remaining 6 instances of backsliding dropped
from 3 to 2, meaning these performances were all NS-like, but with
slightly less typical associates. In the other 3, the students had
dropped from a level of producing only minimally NS-like associ-
ates (1), usually idiosyncratic, to not producing any norm-list
matches at all (0). In general, students’ association knowledge
progressed in an ever-improving direction; and where there were
304 Language Learning Vol. 48, No. 2

downward fluctuations, they were seldom across the NS-like

threshold.
Schmitt and Meara (1997) found that their participants,
having judged a word as unknown, could not produce NS-like
associations that were on a norming list. The present study re-
quired demonstrations of knowledge rather than self-evaluation,
but its results appear to suggest that learners can produce NS-like
associations with no corresponding meaning knowledge. However,
the associations themselves make the reasons for this clearer.
(1) Ind T1 - moment suddenly [no answer]
(2) Ind T2 - moment events suddenly
(3) Ind T3 - moment horse [no answer]
(4) Lith T1 - spin movement beginning
(5) Lith T2 - water around quick
Ind’s NS-like association scores were mainly achieved on the basis
of knowing the idiom spur of the moment. Because moment was on
the norming list at a high enough value to put a student over the
NS-like threshold by itself, Ind produced 3 NS-like scores by
knowing only one fixed phrase. Even more interesting, at T3 Ind
could produce the primary association horse even though having
no idea of the meaning of spur. Ind was either a lucky guesser or,
more likely, had some small sense of at least one lexical field with
which the word spur connected. If so, this might be one of the
earliest indications that Ind’s acquisition of spur had begun. Lith
gave movement and quick very rare NS-like associations (1 and 2
responses respectively). Taken together, they suggest that Lith
had some idea of “movement” in connection with spur. Again, this
may hint at the very start of acquisition, or, considering that Lith
gave no NS-like associations at T3, might have been a fluke.
Combining these results with those in Schmitt and Meara (1997)
leads to the revised conclusion that learners are very unlikely to
be able to give NS-like associations unless they know at least one
 Schmitt 305

meaning sense of a word—unless that word is part of a known

phrase.
Table 8 illustrates how association knowledge relates to
grammar and meaning knowledge. Unsurprisingly, as mean asso-
ciation scores increased, so did average meaning scores. With only
4 participants, I thought it inappropriate to use statistical proce-
dures to verify this; however, the trend in the table seems clear
enough. The acquisition of grammatical knowledge, on the other
hand, did not seem strongly related to the development of more
NS-like associations. Schmitt and Meara (1997) suggested that
different kinds of word knowledge are interrelated; however, the
data in the present study suggest that the strength of the relation-
ship may vary between different word knowledge types, some
perhaps being unrelated. It seems meaning knowledge has closer
links with association and grammar knowledge than the latter
two do with each other.

Grammar Knowledge

One reason for the lack of parallel progression between

association and grammatical knowledge is that the students’
grammatical knowledge seemed already somewhat advanced
when their association knowledge was still at the point where they

Table 8

Association Scores vs. Grammar and Meaning Scores

Association Category

0 1 2 3
Mean Grammar Score 2.19 2.36 2.42 2.42
(Max. 4)
Average
Meaning .17 .37 .41 .52
Proportion
Number 21 25 38 26
306 Language Learning Vol. 48, No. 2

could give no NS-like associations. They seemed to have a rela-

tively strong sense of the grammatical aspects of words, even when
not knowing them according to meaning (Table 7). In most cases,
the students could readily give the word’s class and at least one
derivative form.
Grammatical acquisition longitudinally was erratic for Lith
and Ind; only Tai’s mean scores consistently improved. Only Tai
reported explicitly looking up the target words in a dictionary and
studying them, which might have helped; grammar information
like that required in the grammar task is available in the typical
dictionary. Association information is not given in dictionaries, but
Tai’s improved association scores could be attributed to better
mastery of the words’ meaning gained from explicit study. Perhaps
these results illustrate the value of explicit study in addition to
implicit learning; however, to test this empirically would require
a redesigned, future study.
Table 9 compares the longitudinal grammar scores to those
of the other types of knowledge tested. Both types of knowledge
increase more or less steadily through Grammar Categories 0 to
3 but then drop unexpectedly at Category 4. I could only deduce
that the students had moved from Category 3 to Category 4 by
stating there was no form for a certain word class, because that

Table 9

Grammar Scores vs. Association and Meaning Scores

Grammar Category

0 1 2 3 4
Mean Association Score 1.00 1.55 1.54 1.86 1.50
(Max. 3)
Average
Meaning .18 .23 .38 .45 .38
Proportion
Number 2 11 52 35 10
 Schmitt 307

would have been somewhat easier than actually having to give the
form if it existed. However, that turned out not to be the case; most
of the students reached Category 4 by supplying a word class form.
Schmitt and Meara (1997) reported that their beginning/
intermediate Japanese EFL students did not have very good
mastery of the different derivative forms of a word. The present
study has found the same situation even with advanced NNSs who
are capable of pursuing postgraduate studies in an English-
speaking university. In only 10 cases out of the 110 possibilities
did the participants know all 4 word classes (9%); in almost 60%
of the cases they knew 2 or fewer. Even when they knew words
rather well, their average number of word class forms was usually
3 at best (Tables 2–5). This definite gap in these advanced learners’
morphological knowledge calls into question the assumption that
a learner who knows one member of a word family can easily learn
the others.
Breaking the results down into the individual word classes
shows that the students knew some better than others (Table 10).
However, the target words were nouns and verbs, and so these
word classes could be answered through recognition alone; adjec-
tive and adverb word forms had to be produced. Still, the figures
show such a disparity that it seems nouns and verbs were the best
mastered word classes; adjectives and adverbs appear to be
learned at a later stage, and still represented knowledge gaps even
for these advanced students. This suggests that adjective and
adverb forms are not so readily learned from general exposure
(perhaps due to their lower frequency of occurrence) and hence
might be good candidates for explicit instruction.

Checking for the Developmental Sequencing of Word Knowledge

In my introduction, I raised the possibility of a developmental

hierarchy of types of word knowledge. If this were the case,
acquisition of the four word knowledge types should fall into an
implicational scale. I explored the possibility of such a scale by
means of the Guttman procedure (see Hatch & Lazaraton, 1991,
308 Language Learning Vol. 48, No. 2

Table 10

Number of Words Known in Each Word Class

Noun Verb Adjective Adverb

n % n % n % n %
Litha 30 91 33 100 11 33 8 24
Inda 21 64 28 85 6 18 13 39
Taia 30 91 31 94 10 30 11 33
Korb 10 91 10 91 7 64 1 9
Totalc 91 83 102 93 34 31 33 30
a
Total N=33
b
Total N=11
c
Total N=110

for details and formulas). This required making decisions about

when each type of word knowledge could be considered “acquired”
and operationalizing these decisions in a series of cut-points. As
Anderson (1978) showed with morpheme studies, the placement
of the cut-point can seriously affect the scaling results. The crite-
rion of 80% accuracy typically used in grammatical studies did
little good in this study. However, I could fortunately still generally
make principled decisions. The association criterion I had already
experimentally determined as being Category 2 (Schmitt, in press
b). Judging from the data here, the strength of the students’
performance on the written form task suggests that fully correct
spelling is the logical criterion; the vast majority of spelling per-
formances were at that level.
As for grammar knowledge, the students seemed to know a
target word’s word class plus one derivative form (2 word classes)
even if they had little or no idea what the word meant. On the
other hand, they rarely knew all 4 word class forms. Hence,
knowledge of 3 word class forms seems the most reasonable
criterion. Criteria for meaning were not so easily resolved. Because
there seems to be no principled way to set the meaning proportion
criterion, I explored three proportions: .25, .38, and .50. No
combination of parameters led to an indication that implicational
 Schmitt 309

scaling was present. There were simply too many “errors” for the
scaling to be valid. Therefore, at this early point I found no
evidence of a developmental hierarchy for word knowledge types.

Limitations

The intensive and detailed one-to-one interview employed in

this study should have produced as valid a measurement of the
various word knowledge types as is now possible. However, there
are some possible weaknesses in the procedure. First, even with
repeated probing, I sometimes found it difficult to determine the
students’ knowledge of the subtle differentiation between similar
meaning senses without actually giving away those differences
(e.g., abandon = “leave” or “desert and not return” vs. “leave
because of danger”). Future studies of this type should probably
use only clearly distinguishable meaning senses. Second, I was the
only rater; this inevitably involved a certain amount of subjectivity
in scoring. Third, this study has given some indication of the
manner in which 4 types of word knowledge were acquired con-
currently. It was not designed to isolate the factors affecting this
acquisition; that remains for future studies.
Revised version accepted 10 November 1997

References

Alderson, J.C., Clapham, C.M., & Steel, D. (1997). Metalinguistic knowledge,

language aptitude, and language proficiency. Language Teaching Re-
search, 1, 93–121.
Alexander, R. (1982). What’s in a four-letter word? Word meaning in English
and second language learning. Die Neueren Sprachen, 81, 219–224.
Anderson, R. (1978). An implicational model for second language research.
Language Learning, 28, 221–282.
Avila, E., & Sadoski, M. (1996). Exploring new applications of the keyword
method to acquire English vocabulary. Language Learning, 46, 379–395.
Bahns, J., & Eldaw, M. (1993). Should we teach EFL students collocations?
System, 21, 101–114.
310 Language Learning Vol. 48, No. 2

Blum-Kulka, S. (1981). Learning to use words: Acquiring semantic compe-

tence in a second language. In M. Nahir (Ed.), Hebrew teaching and applied
linguistics (pp. 229–243). Washington, DC: University Press of America.
Carey, S. (1978). The child as word learner. In M. Halle, J. Bresnan, & G. A.
Miller (Eds.), Linguistic theory and psychological reality (pp. 264–293).
Cambridge, MA: MIT Press.
Cohen, A. D., & Aphek, E. (1980). Retention of second-language vocabulary
over time: Investigating the role of mnemonic association. System, 8,
221–235.
Dale, E. (1965). Vocabulary measurement: Techniques and major findings.
Elementary English, 42, 895–901, 948.
D’Anna, C.A., Zechmeister, E.B., & Hall, J.W. (1991). Toward a meaningful
definition of vocabulary size. Journal of Reading Behavior, 23, 109–122.
Drum, P., & Konopak, B.C. (1987). Learning word meanings from written
context. In M.G. McKeown & M.E. Curtis (Eds.), The nature of vocabulary
acquisition (pp. 73–87). Hillsdale, NJ: Lawrence Erlbaum.
Francis, W. N., & Kucera, H. (1982). Frequency analysis of English usage.
Boston: Houghton Mifflin.
Goulden, R., Nation, P., & Read, J. (1990). How large can a receptive vocabu-
lary be? Applied Linguistics, 11, 341–363.
Hatch, E., & Lazaraton, A. (1991). The Research manual. New York: Newbury
House.
Hazenberg, S., & Hulstijn, J.H. (1996). Defining a minimal receptive second-
language vocabulary for non-native university students: An empirical
investigation. Applied Linguistics, 17, 145–163.
Hintzman, D.L., Block, R.A., & Summers, J.J. (1973). Modality tags and
memory for repetitions: Locus of the spacing effect. Journal of Verbal
Learning and Verbal Behavior, 12, 229–238.
Huckin, T., Haynes, M., & Coady, J. (1993). Second language reading and
vocabulary learning. Norwood, NJ: Ablex.
Larsen-Freeman, D. (1975). The acquisition of grammatical morphemes by
adult ESL students. TESOL Quarterly, 9, 409–430.
Laufer, B. (1992). Reading in a foreign language: How does L2 lexical knowl-
edge interact with the reader’s general academic ability? Journal of
Research in Reading, 15, 95–103.
Laufer, B. (1997). What’s in a word that makes it hard or easy: Some
intralexical factors that affect the learning of words. In N. Schmitt & M.
McCarthy (Eds.), Vocabulary: Description, acquisition, and pedagogy
(pp. 140–155). Cambridge, UK: Cambridge University Press.
Longman dictionary of English language and culture. (1992). Harlow, UK:
Longman.
 Schmitt 311

McNeill, D. (1963). The origin of associations within the same grammatical

class. Journal of Verbal Learning and Verbal Behavior, 2, 250–262.
Meara, P. (1984). The study of lexis in interlanguage. In A. Davies, C. Criper,
& A.P.R. Howatt (Eds.), Interlanguage (pp. 225–235). Edinburgh, Scotland:
Edinburgh University Press.
Meara, P. (1987). Vocabulary in a second language (Vol. 2). London: CILT.
Meara, P. (Ed.) (1992). Vocabulary in a second language (Vol. 3) [Special
issue]. Reading in a Foreign Language, 9.
Meara, P. (1996a). The dimensions of lexical competence. In G. Brown, K.
Malmkjaer, & J. Williams (Eds.), Competence and performance in language
learning (pp. 35–53). Cambridge, UK: Cambridge University Press.
Meara, P. (1996b). The vocabulary knowledge framework [On-line]. Available
Internet: [Link]
Melka, F. (1997). Receptive vs. productive aspects of vocabulary. In N. Schmitt
& M. McCarthy (Eds.), Vocabulary: Description, acquisition, and pedagogy
(pp. 84–102). Cambridge, UK: Cambridge University Press.
Nagy, W.E., Diakidoy, I.N., & Anderson, R.C. (1993). The acquisition of mor-
phology: Learning the contribution of suffixes to the meanings of deriva-
tives. Journal of Reading Behavior, 25, 155–170.
Nagy, W.E., Herman, P.A., & Anderson, R.C. (1985). Learning words from
context. Reading Research Quarterly, 20, 233–253.
Nation, I.S.P. (1990). Teaching and learning vocabulary. New York: Newbury
House.
Oxford advanced learner’s dictionary. (1995). Oxford, UK: Oxford University
Press.
Paribakht, T.S., & Wesche, M.B. (1993). The relationship between reading
comprehension and second language development in a comprehension-
based ESL program. TESL Canada Journal, 11, 9–29.
Pienemann, M., & Johnston, M. (1987). Factors influencing the development
of language proficiency. In D. Nunan (Ed.), Applying second language
acquisition research (pp. 45–141). Adelaide, Australia: National Curricu-
lum Resource Center, Adult Migrant Education Program.
Read, J. (1997). Vocabulary and testing. In N. Schmitt & M. McCarthy (Eds.),
Vocabulary: Description, acquisition, and pedagogy (pp. 303–320). Cam-
bridge, UK: Cambridge University Press.
Richards, J.C. (1976). The role of vocabulary teaching. TESOL Quarterly, 10,
77–89.
Schmitt, N. (1995a). A fresh approach to vocabulary using a word knowledge
framework. RELC Journal, 26, 86–94.
Schmitt, N. (1995b). The word on words: An interview with Paul Nation. The
Language Teacher, 19, 5–7.
312 Language Learning Vol. 48, No. 2

Schmitt, N. (in press a). Measuring collocational knowledge: Key issues and
an experimental assessment procedure. ITL Review of Applied Linguistics.
Schmitt, N. (in press b). Quantifying word association responses: What is
nativelike? System.
Schmitt, N., & Dunham, B. (1997). Native and nonnative intuitions of word
frequency. Manuscript submitted for publication.
Schmitt, N., & Meara, P. (1997). Researching vocabulary through a word
knowledge framework: Word associations and verbal suffixes. Studies in
Second Language Acquisition, 19, 17–36.
Sharwood Smith, M. (1984). Discussion of P. Meara’s “The study of lexis in
interlanguage.” In A. Davies, C. Criper, & A.P.R. Howatt (Eds.), Interlan-
guage (pp. 236–240). Edinburgh, Scotland: Edinburgh University Press.
Thomas, M.H., & Dieter, J.N. (1987). The positive effects of writing practice
on integration of foreign words in memory. Journal of Educational Psy-
chology, 79, 249–253.
Webster’s ninth new collegiate dictionary. (1987). Springfield, MA: Merriam-
Webster.
Xue, G., & Nation, I.S.P. (1984). A university word list. Language Learning
and Communication, 3, 215–229.

Appendix

The following word class forms were accepted in this study.

In addition, the various meaning senses are illustrated here, along
with their receptive prompts, including word class, in parentheses.

Abandon

(Noun) (Verb) (Adjective) (Adverb)

abandon/ abandon abandoned [no form]
abandonment
1. (abandon a baby - verb) leave or desert and not return
2. (abandon a ship - verb) leave because of danger
3. (abandon a project - verb) give up before finishing
4. (abandon a political leader - verb) withdraw help or support
from somebody
 Schmitt 313

5. (abandon yourself to despair - verb) allow oneself to be

completely controlled by something
6. (gay abandon - noun) state where feelings and actions are
uncontrolled or uninhibited

Brood

broodiness/ brood broody/ broodily/

brood brooding broodingly
1. (problem - v) spend time anxiously thinking about some-
thing
2. (clouds - v) to hang closely; menacing, threatening
3. (ducks - n) a family of young creatures, esp. birds
4. (situation in a household - n) the children of one family
5. (horse - adj) animal used for breeding
6. (hen - v) sit on eggs in order to hatch them

Circulate

circulation circulate circular/ [no form]/

circulatory circularly
1. (liquid - v) to move around in a closed system
2. (information - v) disseminate or spread widely
3. (air - v) to move about freely
4. (party - v) move from one person to the next

Convert

converter/ convert converted [no form]

convert/ convertible
conversion
314 Language Learning Vol. 48, No. 2

1. (sofa - v) change [be able to be changed] from one form or

use into another
2. (religion - v) change one’s beliefs
3. (religion - n) person who has changed beliefs
4. (football - v) gain extra points after scoring a goal or touch-
down

Dedicate

dedication dedicate dedicated [no form]/

dedicatedly
1. (cause - v) devote oneself to a good cause
2. (publication - v) address one’s publication to someone
3. (church - v) devote something to a sacred purpose
4. (money - v) to set aside something for a particular reason

Illuminate

illumination illuminate illuminated/ [no form]

illuminating
1. (candle - v) to give light or cast light on something
2. (festival - v) to decorate a street or building for a special
occasion
3. (difficult idea - v) cause to understand, make clear
4. (manuscript - v) decorate a book with gold paint and colors

Launch

launch/ launch [no form]/ [no form]

launcher (newly) launched
 Schmitt 315

1. (attack - v) put something or somebody into action, begin

2. (merchandise - v) make a new product publicly available
3. (navy - v) put a new ship into the water
4. (missile - v) put something in motion or on its course
5. (oneself into something - v) begin enthusiastically some-
thing important or something that will take a long time
6. (out into something [career] - v) to do something new or
more exciting or profitable
7. (yacht - n) a large motor boat
8. (baseball - v) to throw long, high, or very hard

Plot

plot plot plotted [no form]

1. (spy - v) to make a secret plan to do something
2. (spy - n) the secret plan
3. (graph - v) to connect a series of points into a curve
4. (navigation - v) to mark something on a map
5. (real estate - n) a small piece of land marked for a special
purpose
6. (novel - n) outline of events in a play or novel

Spur

spur spur [no form]/ [no form]

spurred
1. (horse - n) metal attachment on the heel of a boot
2. (horse - v) to use these spurs
3. (sports coach - v) to urge or encourage forcefully
316 Language Learning Vol. 48, No. 2

4. (complaints - n) event or influence that encourages action

5. (mountains - n) length of high ground coming out of moun-
tains
6. (trains - n) track or road that goes away from the main line
7. (rooster - n) back part of a bird’s foot

Suspend

suspension/ suspend suspended [no form]

suspender(s)
1. (gravity - v) to hang something from above
2. (rail services - v) to temporarily stop or prevent from being
in effect
3. (prison sentence - v) not enforce, delay, or happen at a later
time
4. (policeman - v) prevent someone from holding usual position
[because of misbehavior]
5. (dust - passive verb) hold still in liquid or air

Trace

trace/ trace traceable/ [no form]

tracing tracing/
traced
1. (detective - v) find something or someone by following their
course
2. (clue - n) mark or sign of the presence or someone or
something
3. (family tree - v) going back in time to find the origins or proof
of something
 Schmitt 317

4. (book about the monarchy - v) follow the course, develop-

ment, or history of something
5. (picture - v) to copy something by drawing its lines on
transparent paper
6. (poison - n) very small amount of something
7. (horse - n) part of the harness which pulls a cart