GENERAL ⎜ ARTICLE

Programming Languages
A Brief Review

V Rajaraman

In this article we review the development of programming

languages and classify them based on their structures
and their applications.

Introduction

Programming languages for computers are developed with the

V Rajaraman is with the
Jawaharlal Nehru Centre
primary objective of facilitating a large number of persons to use
for Advanced Scientific computers without the need to know in detail the internal
Research and the Indian structure of a computer. Languages are matched to the type of
Institute of Science, applications which are to be programmed using the language.
Bangalore. Several
generations of scientists
The ideal language would be one which expresses precisely the
and engineers in India specification of a problem to be solved, and converts it into a
have learnt computer series of instructions for a computer. It is not possible to achieve
science using his lucidly this ideal as a clear specification of a problem is often not
written text books on
programming and
available and developing an algorithm from specifications
computer fundamentals. requires subject knowledge and expertise. In actual practice, a
detailed algorithm to solve a problem is the starting point and it
is expressed as a program in a programming language. A large
number of languages, over a thousand, exist each catering to a
different class of applications. All modern programming
languages (with one exception) are designed to be machine
independent. In other words, the structure of the programming
language would not depend upon the internal structure of a
specified computer; one should be able to execute a program
written in the programming language on any computer regardless
of who manufactured it or what model it is. Such languages are
known as high level machine independent programming languages.

In this article we will briefly review various programming lan-

guages which are currently used. We will look at a classification
of programming languages based on their characteristics and

RESONANCE ⎜ December 1998 43

 GENERAL ⎜ ARTICLE

another classification based on their applications. We will also

point out some of the recent developments in programming
languages.

Assembly Language

The first step in the evolution of programming languages was

the development of what is known as an assembly language. In an
assembly language, mnemonics are used to represent operations
to be performed by the computer and strings of characters to
represent addresses of locations in the computer’s memory
where the operands will be stored. Thus the language is matched
In an assembly
to a particular computer’s processor structure and is thus machine
language,
dependent. A translator called an assembler translates a program
mnemonics are
written in assembly language to a set of machine instructions,
used to represent
which can be executed by a computer. Now-a-days programs are
operations to be
written in assembly language only in applications which are
performed by the
cost sensitive or time critical as efficiency of machine code is of
computer and
paramount importance in these types of applications. A cost
strings of
sensitive application is one in which microprocessors are used
characters to
to enhance the functionality of consumer items such as washing
represent
machines or music systems. In these cases the program is stored
addresses of
in a read only memory and its size is small. Thus code optimisation
locations in the
is important. A time-critical application is use of micro-
computer’s
processors in aircraft controls where real time operation of the
memory where the
system is required. Here again the number of machine
operands will be
instructions executed should be minimised.
stored.
High Level Languages

During the evolution of computers, till about 1955, computers

were slow and had a small memory. Thus programming efficiency
was very important and assembly language was dominant. With
improvements in technology, computers were designed with
larger memory capacity, higher speed and improved reliability.
The tremendous potential of computer applications in diverse
areas was foreseen. It was evident that this potential could be
realised only if a non-expert user could effectively use the
computer to solve problems. It was thus clear that a user should

44 RESONANCE ⎜ December 1998

 GENERAL ⎜ ARTICLE

be concerned primarily with the development of appropriate

algorithms to solve problems and not with the internal logical
structure of a computer. Consequently a good notation to
express algorithms became an essential requirement. For
algorithms to be executed by computers, the notation to express
them should be simple, concise, precise and unambiguous. The
notation should also match the type of algorithm. For example,
programming languages to solve science and engineering
problems should support arithmetic using wide ranging, high
precision real and complex numbers and should have features to
express operations with arrays and matrices. On the other hand,
algorithms for processing business data would have operations
Languages which
to be performed on massive amounts of organised data known as
express step-by-
files. The notation, in this case, must facilitate describing files
step algorithms
and formatting and printing intricate reports. Such notations to
written to solve a
express algorithms are known as high level, machine independent,
problem are
programming languages. High level programming languages are
known as
further classified as procedural and non-procedural. Languages
procedural
which express step-by-step algorithms written to solve a problem
languages
are known as procedural languages whereas those which express
whereas those
specifications of a program to be solved are known as non-
which express
procedural. We will first discuss the common features of
specifications of a
procedural languages.
program to be
Procedural Languages solved are known
as non-procedural.
Procedural languages have facilities to:

i) specify data elements such as real, integer, boolean, characters

and data structures such as arrays, matrices, stacks, records, sets,
strings of characters, lists, trees, etc.,

ii) control structures to sequence operations to be performed.

An if then else structure is necessary to allow programs to follow
different sequences of statements based on testing a condition.
For example, the following statement:

if ( a > b ) then
x=y+z;

RESONANCE ⎜ December 1998 45

 GENERAL ⎜ ARTICLE

Procedural p=q+t
else
languages are
x=y–z;
designed using a
p=q*t
set of syntax rules,
endif
which precisely
specify the ‘words’ commands that the statements x = y + z and p = q + t are to
of the language, be executed if ( a > b ) is true. If ( a > b ) is false x = y – z and
and how they may p = q ∗ t are executed.
be combined
legally. iii. Repetition structures which carry out a group of statements
again and again while a condition is true as shown below

while (a > b ) do
x=y–z;
p=q*r
end while

iv. Statements to input and output data.

Procedural languages are designed using a set of syntax rules,

which precisely specify the ‘words’ of the language, and how
they may be combined legally. The rules of syntax are specified
using a notation called Backus–Naur Form (BNF) which
recursively defines various syntactic units of the language. These
rules are similar to the ones used by the great Sanskrit
grammarian Panini. A sample BNF definition of a variable
name is

< variable name > : = < letter >

< variable name > : = < letter > < digit >
< variable name > : = < variable name > < variable name >
where < letter > is any upper case English letter A to Z and
Each syntactically < digit > is any digit between 0 and 9.
correct structure Observe the third line in the above definition, which is a
should have one recursive definition.
and only one
Besides rules of syntax each language has semantic rules. Each
semantic
syntactically correct structure should have one and only one
interpretation.
semantic interpretation.

46 RESONANCE ⎜ December 1998

 GENERAL ⎜ ARTICLE

Associated with each high level language is an elaborate compu- Interpreters are
ter program which translates it into the machine language of the easy to write but
computer in which it is to be executed. There are two types of the translated
translators. One of them takes each statement of the high level programs’
language, translates it and immediately executes it. This is execution is slow.
called an interpreter. Interpreters are easy to write but the
translated programs’ execution is slow. The other approach is to
scan the whole program and translate it into an equivalent
machine language program. Such a translator is called a compiler.
A compiler is a complex program but the compiled machine
code takes lesser time to execute compared to an interpreted
program.

A Classification of Programming Languages

We give in Figure 1 a classification of programming languages.

We have classified high level machine independent languages
into three groups, namely, procedural, non-procedural and Figure 1. Classification of
Programming Languages.
problem-oriented. Procedural languages have as their starting

Examples
Fortran 95
Algorith- COBOL
mic C
Smalltalk
Procedural Object C ++
Oriented JAVA
Visual Basic
Scripting Perl

High Level Non - Functional LISP

(Machine Procedural ML
independent) Logic PROLOG

Numerical – MATLAB
Programming Problem
Languages Oriented Symbolic – MATHEMATICA

Low Level Assembly Publishing – LATEX

(Machine Language
dependent)

RESONANCE ⎜ December 1998 47

 GENERAL ⎜ ARTICLE

point an algorithm to solve the problem. Languages such as

An object models
FORTRAN, COBOL and C are purely algorithmic. These
a complex real
languages provide a methodology to break up a large job into a
world or an
number of tasks and programming the tasks independently as
abstract object.
functions or subroutines. These functions or subroutines are
then combined to form a program. The general idea is to
simplify debugging a program and to reuse the procedures in
other programs which may need them. Over the years it was
realised that this was not sufficient to enable re-use of programs.
Subroutines and functions are too rigid. They require the
specification of the type of data to be used a priori and the data
to be passed to them in a pre-specific order. As the cost of
programming continually increased it was realised that ‘building’
programs using a library of reusable ‘components’ was imperative.
This led to the emergence of the so-called object-oriented languages.
In these languages the concept of subroutine/functions is
extended to that of an object. An object models a complex real
world or an abstract object. A real world object, for example, is
a student whereas an abstract object is a course taken by a
student. In an object oriented (OOP) program an object is
modelled by a collection of data structures and a set of procedures
that can be performed on this data structure. A program consists
The advantages of of a collection of objects, each object providing a service when it
object oriented is invoked and all the objects co-operating to get the job done.
programming Objects are invoked by sending messages to them and objects
(OOP) accrue only return messages when the job is done.
when a large
software project is The action performed in response to a message can vary
undertaken. The depending on the data and type of parameters. This is called
methodology of polymorphism. Objects form class hierarchy with super class
OOP enables a (parent) and subclass (child) relationship. An object can use
programmer to procedures and data defined on objects in its superclass through
remain close to inheritance.
the conceptual
The advantages of object oriented programming (OOP) accrue
higher level model
only when a large software project is undertaken – also known as
of the real world
‘programming in the large’. The methodology of OOP enables
problem.
a programmer to remain close to the conceptual higher level

48 RESONANCE ⎜ December 1998

 GENERAL ⎜ ARTICLE

model of the real world problem. One of the earliest OOP JAVA is getting
languages to be developed was Smalltalk. It, however, did not wide acceptance
become popular. Currently an object oriented version of C now as a
known as C++ is the most popular OOP language. programming
language to write
Another development, which has taken place in the last few
applications for a
years, is the internet – an international network of a very large
network of
number of national computer networks. The technology
heterogeneous
developed in creating the internet has been adapted for
computers.
networking computers within an organization. A computer
network within an organization using protocols and providing
services similar to an internet is called an intranet. In both inter
and intranet small application programs (agents or objects to
perform some services – known as applets) may be developed at
any one of the computers connected to the network. One would
like to create a new application by using these applets by either
importing them to one’s own computer or using them via the
network. A language known as JAVA, which is an object
oriented language achieves this. This language achieves machine
independence by defining a JAVA virtual machine for which
the compiler is written. The JAVA code compiled for the virtual
machine is then executed on any machine by an interpreter
which generates machine code from the compiled code. This
technique makes it easy to port JAVA language to any machine Figure 2. Illustrating
quickly (see Figure 2). JAVA is getting wide acceptance now as portability of JAVA.

COMPILER
Machine code
Java Written in a high for JAVA Virtual
Source Program level language Machine (called
(e.g. C) Java Byte Code)

INTERPRETER

Machine Code Written for

of Target Machine Target Machine

RESONANCE ⎜ December 1998 49

 GENERAL ⎜ ARTICLE

Scripting a programming language to write applications for a network of

languages assume heterogeneous computers.
that a collection of
Scripting Languages: Programming languages such as C and
useful programs,
JAVA are also known as system programming languages as they
each performing a
have been used to develop large systems. For example C has
task, already
been used to write the Unix operating system. System
exists in other
programming languages are strongly typed, that is, each variable
languages. It has
must be declared as a particular type – real, integer, pointer etc.
facilities to
Typing is used both for easy readability and enabling more
combine these
efficient compilation and error detection. Another class of
components to
languages, which are gaining wider acceptance is called scripting
perform a complex
language [3]. Scripting languages assume that a collection of
task.
useful programs, each performing a task, already exists in other
languages. It has facilities to combine these components to
perform a complex task. A scripting language may be thus
thought of as a gluing language, which glues together components.
One of the earliest scripting languages is Unix Shell. Unix shell
filter programs, read a stream of bytes from an input and write a
stream of bytes on to an output. Any two programs can be
connected by attaching the output of one program to the input of
the other. The following shell commands stack three filters to
count the number of lines in the selection that contains the word
‘language’.

select | grep language | wc

The program select reads the given text that is currently on the
display and prints the text on its output; the grep program reads
its input and prints as its output the lines containing the word
‘language’; the wc program counts the number of lines on its
input. Each of these programs select, grep and wc are indepen-
dent programs which could be combined with other programs
also in many ways. Another popular scripting language is Visual
Basic, which is used to develop Graphical User Interfaces (GUI) on
the screen of a Visual Display Unit. It is expected that with
increasing complexity of applications it will be more cost effective

50 RESONANCE ⎜ December 1998

 GENERAL ⎜ ARTICLE

Assembly System Programming Scripting

(e.g. C) (e.g. PERL)

No.of instructions per

statement of language 1 5 100

Degree of typing None Strong Weak

Applications Time Critical, Routine GUI, Gluing

Cost Critical applications components

to glue together existing ‘program components’ using scripting Table 1. Comparison of

languages [3]. In Table 1 we give a comparison of some of the Languages.

languages.

Non-procedural Languages: In procedural languages (also

known as imperative languages) each statement causes the values
stored in one or more memory locations to change. Program
design consists of writing a sequence of statements, which
transform the ‘state’ of the memory from an initial state to a final
state which is the solution to the problem.

Non-procedural functional languages solve a problem by applying

a set of functions to the initial variables in specific ways to get
the answer. The syntax of such languages is similar to

fn (fn–1 (fn–2 ..........f1 (data))......)

where f’s are the successive function applications which transform

their arguments which, at the start, is the initial data. LISP and
ML are two languages in use which support this model. LISP
has been widely used to program artificial intelligence Non-procedural
applications. functional
Another non-procedural class of languages is called rule based languages solve a
languages or logic programming languages. A logic program is problem by
expressed as a set of atomic sentences (known as facts) and Horn applying a set of
clauses (if then rules). A query is then posed. Execution of the functions to the
program now begins and the system tries to find out if the initial variables in
answer to the query is true or false for the given facts and rules. specific ways to
PROLOG is the best known language of this type. get the answer.

RESONANCE ⎜ December 1998 51

 GENERAL ⎜ ARTICLE

Languages known Problem Oriented Languages: Problem oriented languages are

as 4GLs (Fourth designed to solve a narrow class of problems. A user of such a
Generation language need not express in detail the procedure used to solve
Languages) are a problem. Readymade procedures are pre-programmed. The
also used which user merely presents the data in a flexible ‘language’. MATLAB
provide query is a very popular language among scientists and engineers to
languages to solve a wide class of problems in digital signal processing,
access data from control systems, modelling systems described by differential
data bases and equations, matrix computations etc.
manipulate them.
Another class of problem oriented languages is for symbolic
manipulation, for example, simplifying a complex algebraic
expression or getting the indefinite integral of a complex expre-
ssion. MATHEMATICA is a popular language of this type.

Classification Based on Applications

Another method of classifying computer languages is by

applications. The major applications of computers are in the
following areas:

i. Business Data Processing where large files are to be processed.

COBOL has been the dominant language in this area. We have
seen, however, the emergence of spreadsheet based ‘languages’
for answering ‘what if’ type questions. Languages known as
4GLs (Fourth Generation Languages) are also used which
provide query languages to access data from data bases and
manipulate them. 4 GLs also have special features like ‘fill in
the blanks’ to obtain answers to queries and for designing good
looking forms.

ii. Scientific applications require numeric intensive computing

such as those used to solve problems in science and engineering.
Fortran 90 is the dominant language in this area. C is making
inroads. Recently Fortran 95 standard has been published
which incorporates features to write Fortran programs for parallel
computers.

iii. System programs such as those used to write compilers and

52 RESONANCE ⎜ December 1998

 GENERAL ⎜ ARTICLE

Box 1. Ada Language

Most computer languages evolve from the work of a small group and take years to get standardised.
Through an initiative from the United States Department of Defence in early 1970s a standard for a
programming language called Ada was approved in 1983 before a working compiler was written. Ada
was designed based on world wide competition, where a French entry by Jean Ichbiah won in 1979.
Initially the language was named DOD-1 but the name was changed later to Ada in honour of Lady Ada
Augusta Lovelace who is reputed to have programmed an early mechanical computer designed by Charles
Babbage in UK in the 1850s. Ada is a large complex language, which includes the concept of tasks,
concurrent execution, real-time execution of tasks, exception handling and abstract data types. Due to
its complexity, compilers did not appear till 1987 in spite of support and funding by US Department of
Defence. It was revised in 1995 to include better object orientation and better tasking models for
processes. It is, however, more or less dead today probably due to its complexity and strong competition
from C and later C++ and Java.

operating systems. In this area C and more recently C++

dominate. A language known Ada was specially designed to
write programs for these applications but did not become popular.
(See Box 1)

iv. Scripting programs: Another class of applications is to combine

‘program components’ to build large programs. Examples of
these are: commands to ‘back up’ files at specified times, sending
replies automatically to email messages and invoking certain
processes automatically when some conditions are satisfied.
Languages have been developed to specify such tasks and
sequence them to execute automatically. In UNIX operating
system the user command language is called the shell and
command programs as shell scripts. This class of languages is
called scripting languages. One such language is called PERL
(Practical Extraction and Report Language). Visual Basic is
used to develop graphical user interfaces.

v. Artificial intelligence applications are characterised by algo-

rithms, which search large data spaces for specific patterns.
Typical examples are chess playing programs which generate
many potential moves and search for the ‘best’ move within a
given time using heuristic rules. LISP and Prolog are preferred
languages in this area.

RESONANCE ⎜ December 1998 53

 GENERAL ⎜ ARTICLE

Box 2. Survival of Programming Languages

A language does not seem to survive even if it is very good and has strong Government support. Complex
interplay of vendor support, committed user groups, corpus of existing application programs, ease of
learning and use, good compilers giving efficient object code and popular hype seem to prop up some
languages at the expense of others. Except for perennials such as FORTRAN and COBOL most other
languages have a life of less than a decade. Even in these cases only their names have survived! The
modern versions of FORTRAN and COBOL are quite different from the versions of the 60s.

vi. Publishing has become an important application of compu-

ters. Languages for word-processing are proliferating and have
special formatting commands, print commands etc. TEX is a
popular language used to typeset material with complicated
mathematical equations. The TEX translator produces a program
in the Postscript page description language for printing the
material using a laser printer.

Conclusions

The area of programming languages is dynamic, even somewhat

chaotic (See Box 2). As more sophisticated hardware systems
appear in the market new computer applications emerge. These
applications spawn new languages to solve such applications.
Another trend is the continuous increase in complexity of
applications as hardware become more sophisticated and cheaper.
The increase in size of programs needs new methods of tackling
complexity while keeping the cost of program development low
and ensuring correctness of programs.
Address for Correspondence
V Rajaraman
Suggested Reading
IBM Professor of Information
[1] Pratt T W and Zelkowitz M V. Programming Languages. (3rd edition),
Technology, Prentice Hall of India, New Delhi, 1996.
Jawaharlal Nehru Centre for [2] Bird R and Wadler P. Introduction to Functional Programming. Prentice
Advanced Scientific Research Hall Inc., Englewood–Cliffs, N J, USA, 1988.
and [3] Ousterhout J K. Scripting: High level Programming for the 21st Century.
Hon.Professor, Supercomputer Computer. (IEEE, U.S.A), Vol.31, No.3, 23–30,March 1998.
Education & Research Centre [4] Wilkes M V. A Revisionist Account of Early Language Development.
Indian Institute of Science, Computer. (IEEE, U.S.A.), Vol.31, No.4, 22–26, April 1998.
Bangalore 560 012, India [5] Mohan T S. The Java Internet. Resonance. Vol.1.No.5,1996.
[6] Shyamsundar R K. Introduction to Algorithms–4 Turtle graphics.
email: [email protected]
Resonance. Vol.1.No.9,1996.

54 RESONANCE ⎜ December 1998