Corpus Introduction

Using Corpora in the Language Learning Classroom: Corpus Linguistics for Teachers
Gena R. Bennett
http://www.press.umich.edu/titleDetailDesc.do?id=371534
Michigan ELT, 2010
Part 1
An Introduction
to Corpus Linguistics
Gena R. Bennett
2 Michigan
Using ELT, 2010
Corpora in the Language Learning Classroom
The principles of corpus linguistics have been around for

almost a century. Lexicographers, or dictionary makers, have been collecting exam-
ples of language in use to help accurately define words since at least the late 19th
century. Before computers, these examples of language were essentially collected
on small slips of paper and organized in pigeon holes. The advent of computers led
to the creation of what we consider to be modern-day corpora. The first computer-
based corpus, the Brown corpus, was created in 1961 and comprised about 1 million
words. Today, generalized corpora are hundreds of millions of words in size, and cor-
pus linguistics is making outstanding contributions to the fields of second language
research and teaching.
What Is Corpus Linguistics?

So what exactly is corpus linguistics? Corpus linguistics approaches the study of
language in use through corpora (singular: corpus). A corpus is a large, principled
collection of naturally occurring examples of language stored electronically. In short,
corpus linguistics serves to answer two fundamental research questions:
1. What particular patterns are associated with lexical or grammatical

features?
2. How do these patterns differ within varieties and registers?
Many notable scholars, have, of course, contributed to the development of mod-

ern-day corpus linguistics: Leech, Biber, Johansson, Francis, Hunston, Conrad, and
McCarthy, to name just a few. These scholars have made substantial contributions to
corpus linguistics, both past and present. Many corpus linguists, however, consider
John Sinclair to be one of, if not the most, influential scholar of modern-day corpus
linguistics. Sinclair detected that a word in and of itself does not carry meaning, but
that meaning is often made through several words in a sequence (Sinclair, 1991).
This is the idea that forms the backbone of corpus linguistics.
What Corpus Linguistics Is Not

It’s important to not only understand what corpus linguistics is, but also what cor-
pus linguistics is not. Corpus linguistics is not
able to provide negative evidence

able to explain why
able to provide all possible language at one time.
Gena R. Bennett
Michigan ELT, 2010 An Introduction to Corpus Linguistics 3
Corpus linguistics is not able to provide negative evidence. This means a corpus
can’t tell us what’s possible or correct or not possible or incorrect in language; it can
only tell us what is or is not present in the corpus. Many instructors mistakenly
believe that if a corpus does not present all manners to express a certain idea, then
the corpus is altogether faulty. Instead, instructors should believe that if a corpus
does not present a particular manner to express a certain idea, then perhaps that
manner is not very common in the register represented by the corpus.
Corpus linguistics is not able to explain why something is the way it is, only tell us
what is. To find out why, we, as users of language, use our intuition.
Corpus linguistics is not able to provide all possible language at one time. By
definition, a corpus should be principled: “a large, principled collection of naturally
occurring texts. . .,” meaning that the language that goes into a corpus isn’t random,
but planned. However, no matter how planned, principled, or large a corpus is, it can-
not be a representative of all language. In other words, even in a corpus that contains
one billon words, such as the Cambridge International Corpus (CIC), all instances of
use of a language may not be present.
Gena R. Bennett
Michigan ELT, 2010
Chapter 1
Principles of Corpus Linguistics
Questions We Can Answer with Corpora

Broadly, corpus linguistics looks to see what patterns are associated with lexical and
grammatical features. Searching corpora provides answers to questions like these:
What are the most frequent words and phrases in English?

What are the differences between spoken and written English?
What tenses do people use most frequently?
What prepositions follow particular verbs?
How do people use words like can, may, and might?
Which words are used in more formal situations and which are used in more
informal ones?
How often do people use idiomatic expressions?
How many words must a learner know to participate in everyday conversa-
tion?
How many different words do native speakers generally use in conversa-
tion? (McCarthy, 2004, pp. 1–2)
For the most part, these questions don’t look particularly revolutionary. We already
know the answers to a lot of them. We teach the ideas contained within many of these
questions every day. We can open up almost any grammar, vocabulary, conversation,
or writing textbook and find the answers. Even better, we can apply our expert-user
intuition to find the answers. We’re intimately connected
A frequency list displays to the language; after all, we speak it every day, right? An
the words occurring in exercise may help here. For example, O’Keeffe, McCarthy,
a corpus along with the
and Carter (2007, p. 32) studied a frequency list from a 10
number of times each word
appears. million–word corpus and discovered that the 2,000 most fre-
quent words in the corpus accounted for 80 percent of all the
4
Gena R. Bennett
Michigan ELT, 2010 1: Principles of Corpus Linguistics 5
words present. A mere 2 percent of the words were used repeatedly to account for 8
million words.
For example, degree adverbs demonstrate the extent of a particular feature, such
as thoroughly in the sentence, Her chocolate cake is thoroughly delicious. Keep this
in mind, and think for a moment about these questions.
XX What are some common adverbs of degree? Think of at least four.

XX Give examples of ways you would use these adverbs.
XX Which adverbs do you think are used more often in speaking?
XX Which adverbs do you think are used more often in writing?
XX Which adverbs do you think are used more often overall?
You may have thought of these, among others:
very—My sister is very intelligent.

really—Listening to an in-class lecture can be really difficult.
exactly—Sue always knows exactly what I’m thinking.
quite—Frederick appeared quite surprised by the low mark on his project.
completely—The surprise birthday party was completely unexpected.
too—Working full time and going to school full time is too demanding for
my schedule.
Because corpora don’t contain the same number

From this list of adverbs, we might think
of words, we can’t use a simple frequency count
that really is used more in speaking and
to see in which corpus a word is more common.
quite is used more in writing. Perhaps very For example, very occurs in the spoken portion of
is used most frequently overall. the Corpus of Contemporary American English
The exercise used multiple adverbs of (COCA) 195,000 times and in the written portion
degree: where they’re used, the frequency of the COCA 198,000 times; from looking only at
of use, and some examples of use. This the simple frequency count, we might conclude
that very is used only slightly more in written
information seems like sufficient material
language. But, because the written portion of the
for a lesson, and most teachers would feel
COCA is much larger than the spoken portion, we
comfortable presenting this information in can only get an accurate comparison by calcu-
class. lating how many times very occurs per million
Corpora can give us information like fre- words. This is the normed count. The normed
quency, register, and how language is used, counts in Table 1.1 show that for every million
ideas identified in the adverbs of degree words in the spoken portion of the COCA, very
appears 2,543 times; for every million words in
exercise.
the written portion, very only appears 673 times.
Table 1.1 shows the frequency results This allows us to see that, in fact, very is used sig-
per million (rounded to the nearest one) nificantly more frequently in the spoken portion
from the Corpus of Contemporary Ameri- of the corpus than in the written portion of the
can English (COCA). (See Appendix 1 for corpus.
Gena R. Bennett
6 Michigan
Using ELT, 2010
more information on this and other corpora. COCA will also be discussed in Chapter
4.) The numbers in the Speak column indicate how many times the adverbs very,
really, exactly, quite, completely, too, and thoroughly are used in the spoken portion of
the COCA. The numbers in the Write column indicate how many times the adverbs
are used in the written portion of the COCA, and the numbers in the TOTAL col-
umns indicate how many times the adverbs are used overall.
Very is the most frequently used adverb overall in the COCA, and is especially
frequent in spoken language. Really is the second most frequent adverb in speaking
and overall, while too is the most frequent adverb used in writing. Note that too and
completely are used almost the same in speaking and as in writing. With the excep-
tion of thoroughly, and to a significantly lesser degree too, these adverbs of degree
are used more frequently in spoken language than written language.
So what does all this mean? These data present us
Visit www.americancorpus.org/ with opportunities to show students more accurately
to complete your own search on how to use language. When teaching adverbs of degree,
adverbs of degree. based on the information from this corpus, it would be
prudent to emphasize the following:
XX Focus attention on really and very because they are the most commonly
used adverbs of degree, and students will likely encounter them often.
XX Point out that too and completely are different from really and very because
they are used almost equally in speaking and writing.
XX Show how thoroughly is used differently because it appears more frequently
in writing than in speaking (the only adverb here to significantly do so).
Table 1.1
Frequency Results Per Million of Adverbs of Degree in COCA
Word Speak Write Total
very 2,543 673 3,216
really 1,637 392 2,029
exactly 271 93 364
quite 267 150 417
completely 87 78 165
too 656 699 1,355
thoroughly 7 18 25
Total 5,468 2,103 7571

Source: Corpus of Contemporary American English
Gena R. Bennett
XX Use more listening and speaking activities to teach adverbs of degree

because these particular adverbs are used more than two times more in
speaking than writing.
In a nutshell, corpus linguistics allows us to see how language is used today and
how that language is used in different contexts, enabling us to teach language more
effectively.
The Corpus Approach

Is corpus linguistics a methodology? Is it theory? Most corpus linguists are not will-
ing to answer that question in such terms, but when analyzing language using cor-
pora, there is a “method” to employ.
The Corpus Approach (Biber, Conrad, & Reppen, 1998, p. 4) is comprised of four
major characteristics:
1 It is empirical, analyzing the actual patterns of language use in natural


texts.
2 It utilizes a large and principled collection of natural texts as the basis for

analysis.
3 It makes extensive use of computers for analysis.

4 It depends on both quantitative and qualitative analytical techniques.

1. The Corpus Approach is empirical, analyzing the actual patterns of language use in
natural texts.
The key to this characteristic of the Corpus Approach is authentic language. The
idea that corpora are principled has been mentioned but not what language a corpus
is comprised of. Corpora are composed from textbooks, fiction, nonfiction, magazines,
academic papers, world literature, newspapers, telephone conversations at home or
work, cell phone conversations, business meetings, class lectures, radio broadcasts,
and TV shows, among other communication acts. In short, any real-life situation in
which any linguistic communication takes place can form a corpus.
2. The Corpus Approach utilizes a large and principled collection of naturally occurring
texts as the basis for analysis.
This characteristic of the Corpus Approach refers to the corpus itself. You may work
with a written corpus, a spoken corpus, an academic spoken corpus, etc.
Gena R. Bennett
8 Michigan
Using ELT, 2010
3. The Corpus Approach makes extensive use of computers for analysis.
Not only do computers hold corpora, they help analyze the language in a corpus.
A corpus is accessed and analyzed by a concordancing program. In short, you can’t
effectively utilize corpora, or employ the corpus approach, without a computer.
4. The Corpus Approach depends on both quantitative and qualitative analytical

techniques.
This characteristic of the corpus approach highlights the importance of our intu-
ition as expert users of a language. We take the quantitative results generated from
the corpus and then analyze them qualitatively to find significance. Table 1.1 shows
quantitative results. Qualitatively analyzing the results would involve examining
the adverbs of degree in use to understand situations the adverbs are used in. This
is how we answer the question Why?
Target Features
Although intuition may not always be reliable for drawing conclusions about lan-
guage in general, it does often answer the question Why? Intuition is often useful
for helping us form queries for a corpus. Many of the questions that corpora answers
fall into certain areas of language teaching, such as phraseology, lexicogrammar,
registers, English for Specific Purposes (ESP), nuances of language, and appropriate
syllabus design.
Phraseology
Phraseology is the study of phrases. Phraseology is a central element of corpus lin-
guistics: Sinclair (1991) determined that the meaning of a word is found through
several words in a sequence, through phrases. Phraseology includes the study of
collocations, lexical bundles, and language occurring in preferred sequences.
Collocation
The most prominent way of studying phrases is through collocation. Collocation is

the statistical tendency of words to co-occur. This means that when one word is used,
there is a high statistical probability that a certain word or words will occur along-
side of it. For example, look at the noun form of the word deal. The words big, good,
and great are collocations of deal as a noun, meaning that when we use deal as a
noun, we often refer to a big deal, a good deal, and/or a great deal. From studying
collocations, we know that there is a tendency for each collocate of a word to be asso-
Gena R. Bennett
ciated with a single sense of that word. We can see this looking at the phrases using
deal—big deal, good deal, great deal: a big deal is usually an event or situation that
has significant meaning; a good deal generally refers to a bargain; a great deal often
refers to a quantity. Studying collocations provides a deeper understanding of the
meaning and use of a word, such as deal, than simply studying a word alone.
Collocations can also help us better understand particular words used in a certain
phrase. Kennedy (1991) studied between and through, something many language
textbooks have difficulty distinguishing the use of. By studying the collocations of
the two words, Kennedy found that between is usually used after nouns like differ-
ences, distinction, agreement, and meeting, whereas through is more frequently found
after verbs such as go, pass, run, and fall (1991, p. 107). (Chapter 3 of this book pres-
ents an activity dealing with collocation.)
Lexical Bundles
Phraseology also looks at variation in somewhat fixed phrases, which are often
referred to as lexical bundles. Biber, Johansson, Leech, Conrad, and Finegan (1999,
p. 990) define a lexical bundle as a recurring sequence of three or more words. In
conversation, “Do you want me to” and “I don’t know what” are among the most com-
mon lexical bundles (Biber et al., 1999, p. 994). It is important to understand that
lexical bundles are different from idioms. Idioms have a meaning not derivable from
their parts, unlike lexical bundles, which do. Also, lexical bundles are not complete
phrases. Most important, lexical bundles are statistically defined (identified by their
overwhelming co-occurrence), and idioms are not.
One type of lexical bundle is a frame. A frame has set words around a variable
word or words. One example of the use of frames is the expression of future time. In
the Corpus of Contemporary American English, multiple words are used to express
future time using the frame is…to: is going to, is likely to, is expected to, is supposed
to, is about to, is due to. Is and to are the set words of the frame that surround the
variables like going likely, expected, or about.
Preferred Sequences
Phraseology also includes the study of preferred sequences of words. Look at the
adjectives interested and interesting. Hunston (2002, pp. 9–11) explains that learn-
ers often confuse these two words, and explanations of their different meanings do
not usually help students use the words correctly; looking at the phrases someone is
interested in something, an interesting thing, what is interesting, and it is interesting
to see, can give students the ability to use the individual words correctly by providing
an established pattern of use for each word.
Only through corpus study can we find the details of phraseology—collocations,
lexical bundles, and language occurring in preferred sequences.
Gena R. Bennett
10 Michigan ELT, 2010
Using Corpora in the Language Learning Classroom
Lexicogrammar
Another area of language teaching that corpus linguistics addresses is lexicogram-
mar. Lexicogrammar is Sinclair’s (1991) idea that there is no difference between
lexis and grammar, or that lexis and grammar are so closely intertwined that they
cannot be productively studied separately. Certain
To see the most common patterns for lexical items fall in certain patterns and certain
verbs, nouns, or adjectives, visit http:// patterns contain certain lexical items.
candle.cs.nthu.edu.tw/collocation/ An example of the idea of lexicogrammar
webform2.aspx?funcID=9.
includes certain words (lexicon) associated with
certain verbs tenses (grammar): know, matter, and
suppose occur more than 80 percent of the time in the present tense while smile,
reply, and pause occur more than 80 percent in the past tense (Biber et al., 1999, p.
459). We can also find that some verbs are used most frequently in certain clauses:
know and think are associated with that-complement clauses (Biber et al., 1999, p.
661), while the verbs like, want, and seem are associated with to-complement clauses
(Biber et al., 1999, pp. 699).
Hunston and Francis (2000, p. 1–2) offer us a more detailed example.
❶ Philosophy is different from many other disciplines ❷ in that learning about it is as much
a matter of developing skills (in reasoning and argument) as it is a matter of learning a
body of information. ❸ In this sense there are no definitive ‘answers’ to many philosophical
problems: ❹ becoming a philosopher is a matter of becoming able to reason coherently
and relevantly about philosophical issues. ❺ Consequently, valuable contact time with lec-
turers is best spent actually ‘doing philosophy,’ ❻ and that means actively thinking and
talking about it.
The word matter appears three times in the paragraph. What is happening in each
usage?
a matter of developing skills

a matter of learning a body of information
a matter of becoming able to reason coherently and relevantly
a matter of + -ing.
In this sense or use of matter, it’s much more productive to teach the pattern a matter
of + -ing rather than to focus on the single lexical word matter.
Gena R. Bennett
Register
The third area of language teaching that corpus linguistics addresses is register.
Register is defined as situation of use.1 We use different language with different
audiences—our parents, colleagues, or children—at different times and for different
reasons. Register can be broadly defined—spoken versus written—or more narrowly
defined—conversation versus news or even separate parts of a research paper.
Corpus linguistics addresses language teaching through the study of regis-
ters by illustrating the various phraseology and lexicogrammar used from reg-
ister to register. For example, 90 percent of lexical bundles in conversation are
declarative or interrogative clauses (Biber et al., 1999, p. 999); pronouns are used
slightly more in conversation than nouns, but nouns are used significantly more
than pronouns in fiction, news, and academic writing (Biber et al., 1999, p. 235);
past tense is used more in writing and present tense more in conversation (Biber
et al., 1999, p. 456). In different registers, corpora show us differences of use
in language such as word frequency, word meaning and use, and grammatical
frequency.
ESP (English for Specific Purposes)

ESP is probably one of the most obvious and pointed applications of corpus linguis-
tics. The areas of register, lexicogrammar, and phraseology can all be applied to spe-
cific purposes.
Visit the official AWL website:
The Academic Word List (AWL) (Coxhead, 2000) is
www.victoria.ac.nz/lals/staff/
a well-known example of using corpus linguistics to averil-coxhead/awl/.
address ESP, in this case, academic purposes. By inves-
tigating a corpus comprised of academic language, Coxhead was able to pinpoint the
most frequent vocabulary words used in academic texts; she then made the list avail-
able for instructors to help students focus their vocabulary study. A project is also
underway to further that study by investigating the phraseology and lexicogram-
matical patterns of the top words on the AWL (Byrd, 2007).
Like the corpus created using academic texts to compile the AWL, corpora can
be created and investigated for a myriad of purposes. Right now corpora exist for
nurses and health care professionals, air traf-
The BLC (Business Letters Corpus) is an ESP fic controllers, and switchboard operators, just
corpus. You can search the BLC to explore to name a few. Iowa State University has used
language used by Japanese business people a corpus of research articles for each major of
writing business letters at www.someya-net. its graduate students in order to help them
com/concordancer/. write research articles in their designated field
(Cortes, 2007).
1. The terms register and genre are often interchangeable in corpus studies.
Gena R. Bennett
Nuances of Language
Another area that corpus linguistics addresses in language teaching is nuances of
language; like ESP, nuances of language are also a sort of combination of the areas of
language teaching addressed by corpus linguistics that have already been discussed.
This is the contribution to teaching that, in my experiences, many language instruc-
tors seem to appreciate the most.
Nuances of language refer to questions that students might ask that we just don’t
know the answers to. Often, the questions specifically relate to areas of collocation
and frequency. Our response, as teachers, is usually something like, “There is no dif-
ference/such thing” or “That’s just the way it is in English,” which isn’t particularly
helpful. For example, what if a student says, “I stubbed my large toe.” That doesn’t
sound exactly right. Why? Well, that’s just not the way we say it in English. When do
you use is not (or ’s not) versus isn’t? We’re likely to tell students that there’s prob-
ably not really any difference between the two. But, corpus linguistics can answer
both these questions. McCarthy (2004, p. 4) found that is not is used more with pro-
nouns and isn’t with nouns. And large toe? We use large to describe quantity and big
to describe physical size, so big toe it is.
Syllabus Design
The final area of language teaching that corpus linguistics addresses is syllabus
design. Phraseology, lexicogrammar, register, ESP, nuances of language—all of these
areas can be used to more accurately and effectively design syllabi by helping us see
what students really need to know about language: frequency and collocation for
vocabulary, grammar patterns for different registers, and specific knowledge for spe-
cific purposes. And what are accurate descriptions of it. For example, the present per-
fect appears in almost every grammar textbook. It’s usually defined as “recent past”
or “completed action.” However, a corpus study revealed that more than 80 percent
of the time, present perfect is used to signify “indefinite past” (Mindt, 2000, p. 224).
Statistics of this kind help textbook writers, course designers, and teachers set pri-
orities for the classroom. If you, as a teacher, are armed with this kind of knowledge,
you can supplement course materials with information that is relevant for students.
Tools
Types of Corpora
A corpus is a principled collection of authentic texts stored electronically that can be
used to discover information about language that may not have been noticed through
intuition alone. When you want to consult a corpus, what exactly should you look for?
This is a very important question.
Gena R. Bennett
Because most published materials based on corpora make use of large, general
corpora, many readers may believe this is the type of corpus that can be useful in
the classroom. Actually, there are approximately eight types of corpora—generalized,
specialized, learner, pedagogic, historical, parallel, comparable, and monitor—and
which type should be used depends on the purpose for the corpus; only the four types
of corpora that are most useful for employing the corpus approach directly in the
classroom will be discussed here.
Generalized Corpora
The broadest type of corpus is a generalized corpus. Generalized corpora are often
very large, more than 10 million words, and contain a variety of language so that
findings from it may be somewhat generalized. Although no corpus will ever repre-
sent all possible language, generalized corpora seek to give users as much of a whole
picture of a language as possible. The British National Corpus (BNC) and the Ameri-
can National Corpus (ANC) are examples of large, generalized corpora. The COCA
is also an example of a generalized corpus. These large, generalized corpora contain
written texts such as newspaper and magazine articles, works of fiction and nonfic-
tion, as well as writing from scholarly journals; these corpora also contain spoken
transcripts such as informal conversations, government proceedings, and business
meetings. If generalizations about language as a whole are to be drawn, a large, gen-
eral corpus should be consulted.
Specialized Corpora
A specialized corpus contains texts of a certain type and aims to be representative of

the language of this type. Specialized corpora can be large or small and are often cre-
ated to answer very specific questions. Examples of specialized corpora include the
Michigan Corpus of Academic Spoken English (MICASE), which contains only spo-
ken language from a university setting; the CHILDES Corpus (MacWhinney, 1992),
which contains language used by children; the MICUSP, Michigan Corpus of Upper-
level Student Papers, a collection of papers from a range of university disciplines;
and a medical corpus containing language used by nurses and hospital staff. Special-
ized corpora are often used in ESP settings. The AWL, for example, was generated
from a specialized corpora of academic texts.
Gena R. Bennett
Learner Corpora
A learner corpus is a kind of specialized corpus that contains written texts and/or
spoken transcripts of language used by students who are currently acquiring the
language. Learner corpora are often tagged and can be examined, for example, to
see common errors students made. A well-known learner corpus is the International
Corpus of Learner English (ICLE) (Granger, 2003), which contains essays written by
English language learners with 14 different native languages. While the ICLE is more
generalized, containing writings from learners with 14
When a corpus is tagged, each
different native languages, other learner corpora are word included in the corpus has a
more specialized; for example, the Standard Speaking marker added to it that gives addi-
Test Corpus (SST), comprised of oral interview tests of tional information. Often, tags are
Japanese learners. Targeted instruction can be devel- part of speech markers, enabling
oped for general language teaching or for specific lan- users of corpora to search not only
guage groups depending on the type of learner corpus. for specific words, but also for
specific words used as a particular
Chapter 7 will look at corpus-designed activities created
part of speech.
from a learner corpus.
Pedagogic Corpora
A pedagogic corpus is a corpus that contains language used in classroom settings.

Pedagogic corpora can include academic textbooks, transcripts of classroom interac-
tions, or any other written text or spoken transcript that learners encounter in an
educational setting. Pedagogic corpora can be used to ensure students are learning
useful language, to examine teacher-student dynamics, or as a self-reflective tool for
teacher development.
Creating Corpora
A corpus is a principled collection of authentic texts stored electronically. When cre-
ating a corpus, there must be a focus on three factors: the corpus must be principled,
it must use authentic texts, and it must have the ability to be stored electronically.
A corpus is principled, meaning that the language comprising the corpus cannot be
random but chosen according to specific characteristics. Having a principled corpus
is especially important for more narrow investigations; for example, if you want your
students to look at the use of signal words in academic speech, then it’s important that
the corpus used is comprised of only academic speech. A principled corpus is also nec-
essary for larger, more general corpora, especially in instances where users may want
to make generalizations based on their findings. In creating the Longman Spoken and
Written English Corpus (LSWEC), a 40 million–word corpus created to identify and
understand grammatical patterns in English—the corpus that the information in the
Longman Grammar of Spoken and Written English (Biber et al., 1999) is based on—
Gena R. Bennett
the compilers included a representative sampling from conversation, fiction, news, and
academic prose. Whatever the purpose of the corpus, it must be principled.
A corpus must also include authentic texts. Although there is debate over the defi-
nition of “authentic” texts in second language teaching (see Widdowson, 1990, for
example), for purposes of this discussion, authentic texts are defined as those that
are used for a genuine communicative purpose. In the MICASE, only speech acts that
naturally occurred in the course of routine daily events at a university are included.
The LSWEC includes texts from daily newspapers that were distributed and conver-
sations that took place during participants’ weekly routines. The main idea behind
the authenticity of the corpus is that the language it contains is not made up for the
sole purpose of creating the corpus.
Lastly, a corpus is stored electronically. Corpora can be saved in text format (.txt),
rich text format (.rtf), and/or web-based format (.html), or others, depending on the
concordancing program used to access texts. The electronic storage and easy acces-
sibility of texts is one of the major factors that allows corpus linguistics to be applied
in the classroom.
If you are creating your own corpus, one way to gather principled, authentic texts
that can be stored electronically is through Internet “alerts.” Search engines such
as Yahoo and Google gather email updates of the latest relevant results based on a
topic or specific query generated by the user. Alerts can be used to monitor a devel-
oping news story, keep current on a particular theme, get the latest on a celebrity or
event, and/or keep tabs on a favorite sports team; an example is shown in Figure 1.1.
One way to use corpora created from alerts is to investigate common vocabulary used
in certain topics, such as frequent content words used in articles that discuss the
environment.
Another means of gathering principled, authentic texts that can be stored elec-
tronically is looking at Internet essay sites. Many of the academic essay sites have
a disclaimer that their essays should be used for research purposes only, and should
not to be downloaded or turned in as one’s own work. These sites can be very help-
ful for creating corpora specific for academic writing with term papers, essays, and
reports on subjects such as business, literature, art, history, and science. You can
even access essay sites that aren’t academic. The creation of a corpus using essays
from NPR’s “This I Believe” program can be analyzed for American viewpoints and
language, for example.
Corpora can also be created from resources at hand. Textbooks can be used to cre-
ate a pedagogic corpus to investigate the language of academic textbooks. This would
be especially useful for students enrolled in an Intensive English Program (IEP) or
an English for Academic Purposes (EAP) program. Learner corpora can be created
from the compilation of student work taken from one particular class, for one par-
ticular student, or from a series of students and classes. Students can analyze their
own language use and pinpoint areas that need further instruction or document
progress that has been made.
Gena R. Bennett
Figure 1.1
Google Alerts
FAQ | Sign in
Google Alerts (BETA)

Alerts
Welcome to Google Alerts Create a Google Alert

Enter the topic you wish to monitor.
Google Alerts are email updates of the latest relevant Google results (web,
news, etc) based on your choice of query or topic.
Search terms: environment
Some handy uses of Google Alerts include Type: Comprehensive
How often: once a day

• monitoring a developing news story
• keeping current on a competitor or industry Your email:
• getting the latest on a celebrity or event
• keeping tabs on your favorite sports teams Create Alert
Create an alert with the form on the right. Google will not sell or share your email address.
You can also sign in to manage your alerts
© 2010 Google - Google Home - Google Alerts Help - Terms of Use - Privacy Policy
Google Alerts allow users to collect email updates based on a specific topic or query as shown here. You can
choose register, news, blogs, the web, videos, groups, or all of the above (as shown). Either once a day (as
shown), as it happens, or once a week, text will be emailed that relates to a specified topic. These texts can
be combined to investigate vocabulary or grammar patterns, for example, for particular themes.
An important aspect related to creating corpora is the issue of copyright, especially

if findings from a corpus will be distributed via a handout or published in any form.
Contact Internet sites for their permissions policy, and always get students’ writ-
ten permission before using their work for any purpose. Some institutions may also
require you to complete an Internal Review Board (IRB) application for a corpus study.
An institution’s IRB serves to monitor research involving human subjects. The role
of the IRB is paramount in medical studies during which physical or psychological
damage may be done to research participants, although studies involving research of
normal educational practices—such as those concerning instructional strategies or
the effectiveness or comparions of instructional methods—can often be exempted so
long as a clear demonstration can be made that human subjects will not be identified
(e.g., the use of numbers instead of names or the keeping of research documents in
locked drawers). Copyright is a serious issue and should not be overlooked.
Concordancing Programs
Two tools are needed to effectively apply corpus linguistics in the classroom: the cor-
pus and the concordancing program. Concordancing programs are computer software
used to access and sort data from the corpus. Most large corpora, like the MICASE,
have a built-in concordancing program. When the concordancing program is built in,
Gena R. Bennett
as is the instance with the MICASE, only the search options are important to under-
stand for the program. See Figure 1.2 for search options in the MICASE concordancer.
When you create your own corpora or want to access a corpus that does not have
a built-in concordancing program, it’s important to note that many effective concor-
dancing programs are available. Some concordancing programs are very affordable,
and others are free.
A concordancing program should be chosen based on what information you want
from your corpus. On the basic level, all concordancing programs provide a frequency
list that provides each word used in the corpus as well as the number of times the
word appears in the corpus. Often, the list can be sorted by frequency or alpha-
betically. In addition, most concordancing programs show concordance lines from
the corpus. Concordance lines are all the instances of a word or phrase in the corpus.
Many concordancing programs will allow you to sort the concordance lines according
to what comes before or after a specified word or phrase and will let you see more
extended context of the word as it appears in the corpus.
Figure 1.2
The MICASE Corpus
Michigan Corpus of Academic Spoken English
Home Search Browse Help
Search
Speaker Attributes Transcript Attributes

Enter the exact word or phrase you wish to find in
the box. The wildcard character * may be used at
Academic Position/Role Speech Event Type
the end (but not the beginning) of a search word or
phrase to represent zero or more characters (e.g. All All
typing walk* will give you walk, walks, walked, Junior Faculty Advising Sessions
and walking). If you wish to search the entire Junior Graduate Student Colloquium
corpus, use the default setting on the speaker and
transcript attributes. If you wish to do a more Native speaker status Academic Division
specific search, choose the speaker and transcript All All
level criteria using the menus on the right. When Non-native speaker Biological and Health Sciences
you click the button, utterances by speakers that fit Near-native speaker Humanities and Arts
the speaker-level criteria within transcripts that fit
the transcript-level criteria will be found.
First language Academic Discipline
Find: All All
Arabic Afroamerican and African Studies
Submit Search Armenian American Culture
Participant Level
All
Junior Faculty
Junior Graduate Students
Interactivity Rating
All
Highly Interactive
Highly Monologic
MICASE allows users to search for a word or phrase within the corpus. You can also narrow your search by
specifying various speaker attributes or transcript attributes, so that the concordancing program will only
display results of the word or phrase you are searching that are found within those attributes.
Source: http://micase.elicorpora.info. Used with permission.
Gena R. Bennett
An example of this type of basic concordancing program Download TextSTAT at http://

is TextSTAT 2.8 (Hüning, 2008). TextSTAT, Simple Text neon.niederlandistik.fu-berlin.
Analysis Tool, can be downloaded free from the Internet de/textstat/.
and has all the basic features needed to access a corpus:
uploading of files to create a corpus, retrieving of word forms, viewing of concordance
lines, and accessing of extended context. These basic features allow users to retrieve
word frequencies from the corpus and sort alphabetically (or by frequency or by ret-
rograde, alphabetically backward), to establish a minimum/maximum frequency, and
to search for words containing affixes, as well as search and view concordance lines
and extended context. WordSmith Tools (Scott, 2004) and MonoConc Pro (Barlow,
2007) are other concordancing programs that can be downloaded from the Internet.
WordSmith Tools and MonoConc Pro do have a cost, but a personal license (for use
on your personal computer) is affordable.
Procedures
A Framework for Creating Corpus-Designed Activities

Parts 2 and 3 of this book focus on the applications of corpus linguistics to language
teaching, with many of the chapters (Chapters 4–7) focusing specifically on corpus-
designed activities that can be used in your classroom.
As shown in Table 1.2, creating corpus-designed activities involves seven steps.
Table 1.2
A Framework for Creating Corpus-Designed Activities
Ask a research question.
Determine the register on which your students are focused.
Select a corpus appropriate for the register (or compile authentic texts from that
register).
Utilize a concordancing program for quantitative analysis.
Engage in qualitative analysis.
Create exercises for students.
Engage students in a whole-language activity.

Gena R. Bennett
Ask a research question.
The “research question” for a corpus-designed activity could resemble, “What’s the
difference between through and between?” which was discussed earlier in this chap-
ter. Or, “How do you use signal words in academic speaking?” which will be explored
in Chapter 5.
Determine the register on which your students are focused.
As previously discussed, language is used differently in different registers. It’s

important when creating a corpus-designed activity that you know which register is
relevant for your students. If students are practicing informal conversation, looking
at a corpus of academic papers won’t be helpful.
Select a corpus appropriate for the register (or compile authentic texts from that
register).
Whether you create your own corpus or you use a corpus that already exists is not
particularly relevant to creating a corpus-designed activity beyond the appropriate
register and size of the corpus. What is important is that the corpus contains authen-
tic language used for real-life communication.
Utilize a concordancing program for quantitative analysis.
Utilizing a concordancing program is the third step in the corpus approach. To create
a corpus-designed activity, a concordancing program must be used to access the lan-
guage stored in the corpus. If you are using a corpus that has a built-in concordanc-
ing program, be sure to understand all the search functions. If you are employing an
outside concordancing program with a corpus, be sure that the program performs the
functions you need to answer your research question.
Engage in qualitative analysis.
Qualitative analysis is the last step of the corpus approach. Most often, qualitative
analysis will answer the question Why? For qualitative analysis, take the quanti-
tative information given by the corpus’s concordancing program and determine its
significance.
Create exercises for students.
Preparing concordance lines and traditional fill-in-the-blank and gap-fill activities for
students to examine and engage in is a central element of corpus-designed activities.
Gena R. Bennett
Engage students in a whole-language activity.
The grammar methodology of form, meaning, and use is also applicable to apply-
ing corpus linguistics in the classroom. The first and second steps of this methodol-
ogy, form and meaning, are addressed through corpus-designed activities; but when
applying corpus linguistics in the classroom, you should also ensure that students
have an opportunity to use the target feature under investigation. Creating gap-fill
exercises and whole-language activities will provide such opportunities for students
and encourage acquisition of the target features at hand. Examples of these types of
activities are included in Part 3.
Look again at the steps for creating corpus-designed activities listed in Table 1.2.
What do you notice? The steps of the process are not numbered; this is because they
may not take place specifically in this order. You may start with a research question,
a corpus, a register, or even a whole language activity.
As you read Chapters 4–7, refer back to this framework for creating corpus-
designed activities; it will help you make the connection from the reading and exer-
cises to creating your own activities.
Modifying Activities by Language Level

Most of the activities discussed in Part 3 require at least an intermediate to high-
intermediate level of English, but there are ways to modify corpus-designed activi-
ties to make them more accessible to your students. By their nature, corpus activities
are for more advanced levels, but they can be adapted for students at lower levels.
To modify corpus-designed activities for low (beginning to low-intermediate) lan-
guage level:
XX Ask simple research questions.

XX Find your own concordance lines.
XX Adapt lines for students’ level.
XX Present students with fewer lines.
XX Encourage group or whole class work.
Ask simple research questions.
General questions that have to do more with language as a whole are usually less
ambiguous and more significant for lower-level students than specific questions like,
“What’s the difference between through and between?” For example, if you are teach-
ing a group of Hispanic students who consistently use have to tell how old they are,
a simple activity with have versus am when giving age can be very enlightening for
learners.
Gena R. Bennett
Find your own concordance lines.
Instead of utilizing a whole corpus of authentic English for lower-level students to

navigate, look for instances of the pattern you would like to investigate in materials
around you, like the textbook, a newspaper, or magazine.
Adapt lines for students’ level.
Materials that you intend to use in your corpus-designed activity can be modified for
sentence structure and vocabulary so long as the feature under investigation remains
intact. For example, these following sentences are used in the Chapter 5 materials
on using though in academic speaking and exemplify the notion that though can be
used to show a contrast in ideas.
The two disciplines do not appear on the surface to have very much in common. His-
torically, though, anthropologists and epidemiologists have worked together for a very
long period of time.
The sentence structure and vocabulary in these two sentences is quite advanced for
low-intermediate learners, but they can be modified and still demonstrate though
showing a contrast in ideas:
The study of culture and the study of people’s health do not seem similar, though they
have been studied together for a very long time.
While some corpus linguists may argue against this method, I believe it’s a realis-
tic way for lower-level students to take advantage of the benefits of learning from
corpus-designed activities and outweighs any drawbacks of working with adapted
language.
Ask students to work with fewer lines.
Students don’t necessarily have to study 20–50 lines in a corpus-designed activity;

10 lines may be enough for lower-level students to study, especially in investigations
dealing with more general ideas, such as have versus am, as previously discussed.
Encourage group or whole class work.
Completing a corpus-designed activity as a whole class is another great way to give

lower-level students exposure to the benefits of learning from corpus-designed activi-
ties without being too challenging for their language level.
Gena R. Bennett
On your Own
For many instructors who want to provide the best possible instruction for their
students but are pressed for time and resources, theoretical principles are often a
luxury. What makes corpus linguistics work, so to speak, are the practical activities
that can be used with students in an everyday classroom. Knowledge of what corpus
linguistics is and is not, questions that corpora can answer, the corpus approach,
types of corpora and concordancing programs, and how to create corpus-designed
activities all help to provide a solid foundation for understanding the applications of
corpus linguistics. Try these activities to get an idea of how you can apply corpora to
your language learning classroom.
1. Go to www.collins.co.uk/corpus/CorpusSearch.aspx to view the top 100 col-

locations of a word or words.
2. Visit http://corpus.byu.edu/bnc/x.asp to view a side-by-side comparison of fre-
quency and collocation in spoken versus written language.

Corpus Introduction

Uploaded by

Corpus Introduction

Uploaded by

Using Corpora in the Language Learning Classroom: Corpus Linguistics for Teachers

The principles of corpus linguistics have been around for

What Is Corpus Linguistics?

1. What particular patterns are associated with lexical or grammatical

Many notable scholars, have, of course, contributed to the development of mod-

What Corpus Linguistics Is Not

 able to provide negative evidence

Principles of Corpus Linguistics

Questions We Can Answer with Corpora

 What are the most frequent words and phrases in English?

XX What are some common adverbs of degree? Think of at least four.

You may have thought of these, among others:

 very—My sister is very intelligent.

Because corpora don’t contain the same number

very 2,543 673 3,216

really 1,637 392 2,029

exactly 271 93 364

quite 267 150 417

completely 87 78 165

too 656 699 1,355

thoroughly 7 18 25

Total 5,468 2,103 7571

XX Use more listening and speaking activities to teach adverbs of degree

The Corpus Approach

1 It is empirical, analyzing the actual patterns of language use in natural

3. The Corpus Approach makes extensive use of computers for analysis.

4. The Corpus Approach depends on both quantitative and qualitative analytical

The most prominent way of studying phrases is through collocation. Collocation is

a matter of developing skills

ESP (English for Specific Purposes)

A specialized corpus contains texts of a certain type and aims to be representative of

A pedagogic corpus is a corpus that contains language used in classroom settings.

Google Alerts (BETA)

Welcome to Google Alerts Create a Google Alert

How often: once a day

You can also sign in to manage your alerts

An important aspect related to creating corpora is the issue of copyright, especially

Michigan Corpus of Academic Spoken English

Home Search Browse Help

Speaker Attributes Transcript Attributes

An example of this type of basic concordancing program Download TextSTAT at http://

A Framework for Creating Corpus-Designed Activities

 Determine the register on which your students are focused.

 Utilize a concordancing program for quantitative analysis.

 Engage in qualitative analysis.

 Create exercises for students.

 Engage students in a whole-language activity.

Ask a research question.

Determine the register on which your students are focused.

As previously discussed, language is used differently in different registers. It’s

Utilize a concordancing program for quantitative analysis.

Engage in qualitative analysis.

Create exercises for students.

Engage students in a whole-language activity.

Modifying Activities by Language Level

XX Ask simple research questions.

Ask simple research questions.

Find your own concordance lines.

Instead of utilizing a whole corpus of authentic English for lower-level students to

Adapt lines for students’ level.

Ask students to work with fewer lines.

Students don’t necessarily have to study 20–50 lines in a corpus-designed activity;

Encourage group or whole class work.

Completing a corpus-designed activity as a whole class is another great way to give

1. Go to www.collins.co.uk/corpus/CorpusSearch.aspx to view the top 100 col-

You might also like

able to provide negative evidence

What are the most frequent words and phrases in English?

very—My sister is very intelligent.

Determine the register on which your students are focused.

Utilize a concordancing program for quantitative analysis.

Engage in qualitative analysis.

Create exercises for students.

Engage students in a whole-language activity.