An abstract is a concise summary of a body of information should as a report, dissertation, thesis or article. Its purpose is to make it possible for potential readers to quickly find out if the work is relevant to their needs and worth reading.

Writers of articles for journals will always be asked to provide an abstract, and these are usually included on CD-ROM and online databases for worldwide reference.

You may be asked to include an abstract at the beginning of a piece of academic work, although confusingly this may also sometimes be called a summary, executive summary or executive abstract. Check with your department which term you should use.

There are in fact two different kinds of abstract: the descriptive abstract and the informative abstract.

These provide a description of the report’s main topic and purpose, together with an overview of the contents. They are usually very short just a sentence or two and may even appear on the title page (or at the beginning of a journal article). All this type of abstract does is give a very general idea of what the writing is about; it like a slightly extended title, or a topic sentence in an introduction. Consequently, it is not necessary to summarise any details or conclusions found in the text.

A descriptive abstract will look something like this:

The research analyzes the different approaches to reducing traffic congestion in the UK and provides conclusions and recommendations for future implementation.

However, if you are asked to provide an abstract here at Birmingham City University, you will probably need to produce an informative type.

As the name implies, informative abstracts should inform the reader: they should make clear what the research is about and give the key information from each section of the report; indicate how it was carried out; and summarise what the main findings and conclusions are.

You should not, however, indicate the structure of the writing this would be included in your introduction.

As suggested by Reading University(personal.rdg.ac.uk/

kcshuwil/cme/abstract.html), the essential parts of an informative abstract are:

• Background. A simple opening sentence or two placing the work in context.
• Aims. One or two sentences giving the purpose of the work.
• Method(s). One or two sentences explaining what was done.
• Results. One or two sentences indicating the main findings.
• Conclusions. One sentence giving the most important consequence of the work.

They also give the following advice:

• Do not commence with this paper&…, this report&… or similar. It is better to write about the research than about the paper. Similarly, do not explain the sections or parts of the paper.
• Avoid sentences that end in &…is described, &…is reported, &…is analysed or similar. These are simply too vague to be informative.
• Do not begin sentences with it is suggested that&…, it is believed that&…, it is felt that&… or similar. In every case, the four words can be omitted without damaging the essential message.

• Do not write in the first person in any form. Thus, not only should you avoid I, but also we, the author, the writer and so on. Again, this is because the abstract should be about the research, not about the act of writing.

Other useful advice:

• Check with your tutor what length of abstract is expected; normally they will be short and concise, with the result that the sentences will be fairly dense and information-heavy.
• Don’t include any secondary information, and take out any unnecessary words, obviously keeping everything grammatical.
• Don’t include any references in the abstract, but do include relevant numerical data.

Here is a sample abstract
(Source: sr.bham.ac.uk/skills/Writing/abstrax.html)

Computerized speech recognition takes advantage of the most
natural form of communication, the human voice. During
speech, sound is generated by the vocal cords and by air
rushing from the lungs. If the vocal cords vibrate, a voiced
sound is produced; otherwise, the sound is unvoiced. The
main problem in speech recognition is that no two voices
produce their sounds alike and that an individual voice va-
ries in different conditions. Because voices do vary and
because words blend together in a continuous stream in
natural speech, most recognition systems require that each
speaker train the machine to his or her voice and that words
have at least one-tenth of a second pause between them. Such
a system is called an isolated word recognition system and
con sists of three major components that process human
speech: (1) the preprocessor which removes irregula rities
from the speech signal and then breaks it up into parts; (2)
the feature extractor which extracts 32 key features from
the signal; and (3) the classification phase which
identifies the spoken word and includes the training mode
and reference pattern memory. Spoken words are identified on
the basis of a certain decision algorithm, some of which
involve dynamic programming, zero crossing rate, linear pre-
dictive coding, and the use of state diagram.

Voice recognition systems offer many applications including
data entry, freedom for mobility, security uses, telephone
access, and helpful devices for the handicapped. However,
these same systems also face problems such as poor
recognition accuracy, loss of privacy among those who use
them, and limited vocabulary sizes. The goal of the
industry is the development of speaker-independent systems
that can recognize continuous human speech regardless of
the speaker and that can continually improve their
vocabulary size and recognition accuracy.

Links to further resources on writing abstracts