Minimal pairs:

minimal importance?
Adam Brown

Minimal pairs spring to many teachers' minds when the topic of pro-
nunciation teaching is raised. They also form the focus of many course-
books on pronunciation. This article argues that minimal pairs do not merit

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

this attention. There are other aspects of pronunciation which are of
greater importance, and there are other ways of teaching vowel and
consonant pronunciation.

Introduction Minimal pairs are pairs of words which differ in the pronunciation of one
sound only. An example for English is the pair ship and sheep, where
the distinguishing sounds are hi and l\:l. Indeed, this pair has become
famous as it is the title of a pronunciation coursebook (Baker 1981)
which many ELT schools have in their resources library.
Minimal pairs are often used as a short cut in taxonomic phonemic
theory (the analysis of phonology generally used in the 'British school'
and associated with names such as Daniel Jones and A. C. Gimson). If
you can find a minimal pair for two sounds, then they are distinctive
units (phonemes) in the sound system of the language.
After coming to Singapore many years ago, and being given ship vs.
sheep type of drills to perform with students, it occurred to me that this
was not a very meaningful exercise. Singapore is one of the busiest ports
in the world. However, it is a tiny island (the size of the Isle of Man) with
a population of three million. Consequently, land is at a premium, and
there are no animal farms. The nearest most Singaporeans come to
sheep is mutton curry. In short, if Singaporeans don't pronounce the
distinction between ship and sheep clearly, the chances of misunder-
standing are minimal: they are almost certain to mean ship.
Contextual and communicative considerations like this therefore limit
the value of minimal pair exercises. In this article, I look at various
factors which have a bearing on their importance.

Numbers of Although the existence of just one minimal pair is sufficient for us to
minimal pairs conclude that two sounds are different phonemes, we must acknowledge
that some pairs of phonemes have very few minimal pairs. For example,
the few minimal pairs for /J7 and /$/ Confucian, confusion; Aleutian,
allusion; mesher, measure; Asher, azure; glacier, glazier. Reading
through these, you will appreciate that some of them are rather
contrived. Confucian is rare outside the Chinese world. One seldom
ELT Journal Volume 49/2 April 1995 © Oxford University Press 1995 169
 refers to the Aleutian Islands. A mesher ('person or thing that meshes')
is not to be found in most dictionaries. Asher is a proper name. Rare
pronunciations of glacier (with ///) and azure, glazier (with A3/) are
necessary for them to be minimal pairs.
A further limitation with this pair of sounds is that the phoneme A3/
never occurs in word-initial or word-final position in English. The only
exceptions to this are French loanwords such as genre, beige, sabotage,
which in any case are pronounced with Id^l by many speakers. All the
above minimal pairs therefore include the sounds in word-medial
position. The distinction is also rather unstable, in that there are words
for which the pronunciation varies between these sounds, e.g. Asian:
[eijn] or [ei3n].

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

Similarly, you will find that there are only seven minimal pairs for the
distinction /u/ and /u:/, and these are therefore regularly quoted in
coursebooks: pull,pool; full, fool; look, Luke; hood, who'd; would/wood,
wooed; could, cooed; should, shoedlshooed. Again, many sound rare or
contrived.
In contrast, numerous pairs can easily be found for other distinctions,
e.g. Id and /as/, /p/ and Ibl. This factor ought to be reflected in any
statement of the importance of phonemic contrasts in English. So ought
the fact that, for example, the phoneme /o/ occurs as the stressed vowel
in only about forty words in English. I have given an indication of this
importance (called 'functional load') of often confused phoneme pairs in
English elsewhere (Brown 1991a).
The usefulness of functional load, as opposed to the simpler notion of
minimal pairs, is illustrated by the following personal experience (Brown
1988). I once went on a camping trip to Thailand with a group of
Singaporeans. During the trip, my companions referred on several
occasions to my [pe?], e.g. 'Pass me a [pe?]', 'There's a butterfly on your
[pe?]'. I had to keep asking for clarification, as I was never sure whether
they were talking about my bag, back, pack (meaning 'rucksack') or peg
(either 'tent-peg' or 'clothes-peg'). The fact that the contrasts /p/ and Ibl,
Id and /ae/, /k/ and /g/ are all important distinctions in the sound system
of English, with high functional loads, is shown by the number of
confusable words here. Consequently, there is a high risk to intelligibility
for speakers such as my Singaporean companions, who do not make
these distinctions. The interesting thing to note is that there are only two
minimal pairs (bag, back; back, pack). The word peg does not form a
minimal pair with any of the other three. In the circumstances, all four
words were plausible, and in different situations the words beg and beck
could be added to the list of alternatives.

Distinctive An alternative way of analysing (parts of) the sound system of a

features language is in terms of features which sounds share. For example, the
feature [ ± voice] can be used to embrace voiced ([+ voice]) and
voiceless ( [ - voice]), i.e. whether the vocal cords are vibrating or not.
The feature distinguishes several pairs of consonant sounds: /p, b/, /t, d/,
170 Adam Brown
 /k, g/, AT, d3/, It, v/, /G, d /, Is, z/ and /J, 3/. So, if students do not
distinguish /J, 3/, they may also have problems with the other pairs, since
the distinguishing feature is the same.
The importance of thinking in terms of features rather than individual
pairs is shown by the fact that many processes apply to any sound which
has the relevant feature, or group of features:
• Any final [- voice] consonant causes the vowel to be shortened, e.g.
the vowels in tripe and loose are shorter than those in tribe and lose.
• [- voice] fricatives have a louder hissing noise (greater friction) than
their counterpart [+ voice] sounds. There is therefore a louder hiss in
the /f, 0/ of fan, thank than for in the /v, 6/ of van, than.

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

• Initial [— voice] stops have a burst of [- voice] air (aspiration) on
release; [+ voice] stops do not. Compare the /p, t, k/ of pan, ten, con
with the /b, d, g/ of ban, den, gone.
It is therefore more efficient to teach the [± voice] distinction which
subsumes all these consonants, than the eight separate contrasts
(Vaughan-Rees 1986).

Voice quality Voice quality refers to the habitual settings of the vocal apparatus
(tongue, lips, larynx, vocal cords, etc.), which give an overall colouring to
the voice. They characterize the speaker on an individual, social,
geographical, or native language basis. Jenner (1992) describes the
typical voice quality of Southern British English speakers:
Neutral or slightly lowered laryngeal position
Low laryngeal tension
Neutral and relaxed supralaryngeal tract
Active tongue tip
Lax jaw
Slight lip rounding and spreading, but without tension
He further claims that assuming the correct typical voice quality for a
foreign language ('getting into gear' as Honikman (1964) calls it) helps
in the articulation of the individual vowels and consonants of that
language. In other words, an inability to distinguish between III and /0/
may reflect a more deep-seated problem with the overall setting of the
tongue. It may therefore also affect other consonant sounds in perhaps
less obvious ways. He offers a number of exercises of a Suggestopedia
type, aimed at improving voice quality settings for English.

Context As we saw with the ship vs. sheep distinction in Singapore, context
disambiguates in many situations, and a confused pair of sounds may not
lead to confusion of understanding. It is worth quoting here the example
sentences which Baker (1981:8) gives for this pair:
He wants a ship for his birthday.
He wants a sheep for his birthday.
Minimal pairs: minimal importance? 171
 I challenge the reader to supply a context in which both these sentences
are plausible utterances (a G r e e k tycoon? an A r a b sheikh?).

Suprasegmental T h e term 'suprasegmental' is widely used to refer to stress, rhythm,

features intonation, and voice quality. I would like to claim, however, that it
ought to be extended to cover certain features or processes which are
often thought of as segmental phenomena. For example, there is a
phonemic contrast in English between /3:/ and hi. However, /a/ is an
anomalous vowel in English, as it is analysed as occurring only in
unstressed syllables. So, any minimal pair (e.g. foreword, forward)
involves not just a difference between /3:/ and /a/ but also a consequent
difference in stress: forward has an unstressed second syllable, whereas
in foreword it carries secondary stress.

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

Similarly, the difference between so-called strong and weak forms of
grammatical words, or the pronunciation of compound nouns, has
consequences for stress and rhythm as well as vowel quality. For
instance, in the following sentence pairs, the words in the second
sentence of each pair not only have strong vowel forms but also attract
some degree of stress. They therefore affect the rhythm of the sentence
as well as its meaning. In other words, the occurrence of /a/ is a
suprasegmental feature as much as a segmental one.

1 a. H e has some [sam] good ideas, (a genuine compliment)

b. H e has some [SAITI] good ideas. ( . . . but also some ridiculous ones),
(a back-handed compliment)
2 a. She said that [dat] coffee had risen in price, (all coffee)
b. She said that [diet] coffee had risen in price, (that particular type
of coffee)
3 a. H e is a gentleman [man], (compound noun)
b. H e is a gentle man [maen]. (adjective + noun)

It is unfortunate that vowel and consonant segments constitute the main

focus of pronunciation in the minds of many students and teachers.
Writers are nowadays convinced of the importance of suprasegmentals
in pronunciation, although the priority given to segmentals in course-
books may lead readers to overemphasize their importance.

If the child could paint the picture, [intonation and rhythm] would b e
the wave on which the other components ride up and down; but the
linguist is older and stronger, and has his way—he calls them
suprasegmentals, and makes the wave ride on top of the ship.
(Bolinger 1961, quoted by Gilbert 1984:1-2)

Communicative Perhaps the greatest criticism against minimal pair drills is that they are
teaching not communicative, and therefore lack interest for the student. T h e r e is
a saying in E L T circles: 'a drill is something used for boring'. Drills of
course have their uses, as means of checking whether students are
capable of making sound distinctions. However, since they do not
involve the students in meaningful exchanges, they can be demotivating.

172 Adam Brown

 Teachers whose pronunciation work consists of little more than drills
and quasi-drill passages, are therefore unlikely to foster any pronuncia-
tion skills or interest in the language in their students.

Preparing materials Very little has been written on communicative activities for pronuncia-
tion teaching. Celce-Murcia (1987) and Pica (1984) both include a
number of techniques for English phonemes. Celce-Murcia (1987:10)
outlines the steps in preparing communicative pronunciation materials:
1 Identify your students' problem areas (different groups of students
may have different problems)
2 Find lexical/grammatical contexts with many natural occurrences of

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

the problem sound(s).
3 Develop communicative tasks that incorporate the word.
4 Develop at least three or four exercises so that you can recycle the
problem and keep practicing the target sound(s) with new contexts.
In other words, the same types of activities used to teach other language
areas communicatively can also be used to teach pronunciation.

Sample activities As an illustration, Celce-Murcia describes the following activities to be

introduced at intervals to practise the th sounds, /9, 61—problems for
most learners of English:
1 Doctor examining patient. This involves parts of the body such as
mouth, teeth, throat, thumb, and thigh.
2 Calendar. The th sounds also occur in Thursday, month, all numbers
with the number 3 in them {third, thirtieth, etc.), and almost all
ordinal numbers (fourth, twentieth, etc.).
3 Family members. The three words father, mother, and brother (and
grandfather, brother-in-law, etc.) are important here.
4 The Thorpe family. The family has members such as Arthur,
Dorothy, and Keith. (It is also worth pointing out here that the
names Anthony, Thomas, and Esther do not normally have a th
sound in British English, despite their spelling.)
These activities practise the th sounds in addition to points of grammar
(e.g. questions), vocabulary building (e.g. other words for relatives) and
functional-notional categories (e.g. 'at the doctor's'). You will notice
that none of Celce-Murcia's exercises involve minimal pairs as such. This
is partly due to the very limited number of pairs involving the th sounds.
Rogerson and Gilbert (1990) give exercises which involve minimal pairs,
but improve on the traditional practice of requiring students to simply
identify which of a pair of sentences is being read out. Instead, students
have to respond with an appropriate reply:
Q: Where did you sleep? A: In a hotel.
Q: Where did you slip? A: On the ice.
Minimal pairs: minimal importance? 173
 Q: When did he leave? A: At 2 o'clock.
Q: When did he live? A: In the 19th century.
Q: Are you ready, team? A: Yes, we're all here.
Q: Are you ready, Tim? A: Yes, I'm coming.

Different As Celce-Murcia notes, different groups of students may have different

students, problems as far as phoneme confusion is concerned. This is one of the
different major drawbacks of minimal pair drill books published (usually in
problems Britain or the USA) for worldwide distribution. For example, the vowels
/ae/ and lal are problems for many foreigners, and drill books therefore
contain exercises practising the distinction. However, they are not a
problem in Singapore. Instead, Singaporeans tend to confuse /ae/ with Id,

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

and la:l with /A/. SO, bad and bed are pronounced the same (both
sounding like bed), and bard and bud are also the same (both sounding
like bud. Since no Singaporean confuses bad and bard, exercises drilling
this distinction are therefore totally irrelevant in Singapore.
In classes of students of mixed nationalities and LI backgrounds, such
all-purpose drillbooks may be useful. However, classes, especially those
in foreign countries, are usually composed of homogeneous students. In
such circumstances, tailored pronunciation materials should be used
practising only those features which are problems for the specific
students. These are often the product of local writers and publishing
houses.

Conclusion Minimal pairs immediately spring to most teachers' (and students')

minds when the topic of pronunciation is raised. However, important
though they are, they should not be overemphasized at the expense of
other aspects of pronunciation, such as stress, rhythm, intonation, and
voice quality.
Where pronunciation is concerned, locally-produced materials covering
problem areas for specific students are generally more relevant than all-
purpose international drill books.
Received September 1993

174 Adam Brown

References Jenner, B. 1992. 'The English voice' in Brown
Baker, A. 1981. Ship or Sheepl (2nd edn.). 1992.
Cambridge: Cambridge University Press. Pica, T. 1984. 'Pronunciation activities with an
Bolinger, D. L. 1961. Forms of English. Cam- accent on communication.' English Teaching
bridge, MA.: Harvard University Press. Forum 22/3: 2-6. (Also in Brown 1991).
Brown, A. 1988. 'Functional load and the teaching Rogerson, P. and J. B. Gilbert. 1990. Speaking
of pronunciation.' TESOL Quarterly 22: 593- Clearly. Cambridge: Cambridge University
606. (Also in Brown 1991b). Press.
Brown, A. 1991a. Pronunciation Models. Singa- Vaughan-Rees, M. 1986. 'Sink the ship and shoot
pore: Singapore University Press. the stupid sheep!' IATEFL Phonology Special
Brown, A. (ed.). 1991b. Teaching English Pro- Interest Group Newsletter 1: 7.
nunciation: A Book of Readings. London:
Routledge. The author
Brown, A. (ed.). 1992. Approaches to Pronuncia- Adam Brown is a training consultant in the
tion Teaching. London: Macmillan and The

Downloaded from http://eltj.oxfordjournals.org/ at Freie Universitaet Berlin on May 3, 2015

Business Communication Unit of The British
British Council. Council, Singapore. He has an MA and a PhD
Celce-Murcia, M. 1987. 'Teaching pronunciation in phonetics, from the University of Edinburgh.
as communication' in D. Morley (ed.). 1987. He has taught at universities in Singapore,
Current Perspectives on Pronunciation. Malaysia, Thailand, and the UK. His specific
Washington, DC: TESOL. interests are in business communication, spelling,
Gilbert, J. B. 1984. Clear Speech. (Teacher's and pronunciation teaching, and he has written a
manual) Cambridge: Cambridge University number of books and articles, including Teaching
Press. English Pronunciation: A Book of Readings
Honikman, B. 1964. 'Articulatory settings' in D. (Routledge 1991), Pronunciation Models (Singa-
Abercrombie et al. (eds.). 1964. In Honour of pore University Press 1991), and Approaches to
Daniel Jones. (Also in Brown 1991b) London: Pronunciation Teaching (Macmillan and The
Longman. British Council 1992).

Minimal pairs: minimal importance? 175