cfiWtSf.
*
M.I.T.
LIBRARIES
-
DEWEY
Dewey
hd28
.M414
00'A-
A GRAMMATICAL MODEL OF ORGANIZATIONAL
ROUTINES IN A TECHNICAL SERVICE ORGANIZATION
by
Brian T. Pentland
CCS TR#
177, Sloan
^
He^fu
WP # 3721-94
August 1994
fj
.
EuejU^-
MASSACHUSETTS INSTI rUTE
OF TECHNOLOGY
DEC 01
1994
LIBRARIES
A GRAMMATICAL MODEL OF ORGANIZATIONAL ROUTINES
IN
A TECHNICAL SERVICE ORGANIZATION
DRAFT
Accepted
for Publication in
Administrative Science Quarterly
Brian T. Pentland
University of California. Los Angeles
Henry H. Rueter
University of Michigan
The authors wish to thank Stephen Barley, Michael Cohen, Connie Gersick, Barbara Lawrence,
Lance Sandelands, Robert Sutton, Burt Swanson, Karl Weick, Elaine Yakura and the anonymous
ASQ reviewers for their comments on earlier drafts of this paper. We would also like to thank
Linda Pike for her expert editing, and Louise Soe and Sidne Ward for their help with the data
analysis. An earlier version of this paper was presented at the Academy of Management, Las
Vegas, NV, August 1992.
Organizational Routines
A GRAMMATICAL MODEL OF ORGANIZATIONAL ROUTINES
IN
A TECHNICAL SERVICE ORGANIZATION
ABSTRACT
This paper explores the sequential structure of work processes
involves high
numbe
..
in
a task unit
whose work
of exceptions, low analyzability of search, frequent interruptions
and
extensive deliberation, and cannot be characterized as routine under any traditional definition.
Yet a detailed analysis of the sequential pattern of action
in
a sample of 168 service interactions
reveals that most calls follow a repetitive, functionally similar pattern.
This apparent
contradiction presents a challenge to our theoretical understanding of routines:
apparently non-routine work display such a high degree of regularity?
question,
patterns
we propose a new
and a
illustrate
a
definition of organizational routines
new methodology
for
how can
To answer
this
as a set of functionally similar
studying the sequential structure of work
processes using rule-based grammatical models.
This approach to organizational routines
juxtaposes the structural features of the organization against the reflective agency of
organizational
members.
Members enact
potentially large) set of possibilities that
regular patterns of action
we
specific
performances from among a constrained (but
can be described by a grammar, giving
label routines.
rise to the
*
.
i
Organizational Routines
Routines occupy a
used
critical position in
to explain the
organization theory.
As genetic
medial quality of organizational structure
Hannan & Freeman,
in
evolutionary theories (Nelson
have become a cornerstone
&
Winter, 1982;
in
theories of organizational learning and adaptation, as well (Levitt
1991; Walsh
action,
& Ungson,
1983).
As memory,
& March, 1988; Cohen,
Routines occupy the crucial nexus between structure and
1991).
to organizational theory, there
have been very few attempts
the sequential structure of organizational routines.
summary measures
collected
difficult to
In
spite of their
to deal empirically with
Instead, existing research has usually
that ignore the sequential structure of the action (Lynch, 1974;
and Macintosh, 1981; Withey,
Routines are
Our
routines
between the organization as an object and organizing as a process.
importance
Daft
material, routines are
Daft,
and Cooper, 1983).
study because they are essentially complex patterns of social action.
tools for characterizing the variable properties of static objects are well developed, but our
tools for characterizing the sequential structure of patterns of action are not.
difficulty lies in
objectify
two
different
senses of the word
"routine."
As a noun,
An
"routine"
additional
is
used
to
a collective capacity to perform recognizable patterns of action (Nelson and Winter,
1982).
The
needed
to
patterns themselves
summarize them as a
may
vary considerably from instance to instance, so a label
fixed entity.
As an
adjective, "routine" indicates a
is
judgment
about a variable property of a pattern of action, such as repetitiveness (Gersick and Hackman,
1990), analyzability (Perrow,
1967), variety (Daft and Macintosh,
1981), or even
"mindlessness" (Ashforth and Fried, 1988). These two senses of the term
considerable amount of confusion about the underlying construct.
are usually spoken of vaguely,
The purpose
of this
paper
is
in
Perhaps
the plural, rather than concretely,
to present a theoretical
model
seem
in
to lead to
this is
why
a
routines
the singular.
that juxtaposes our
concept of
routines as patterns of action with our concept of routineness as a property that such patterns
Organizational Routines
might possess.
function of a
We
illustrate this
medium
model with an example, taken from the software support
sized software vendor, that presents us with an apparent contradiction.
This task unit exhibits repetitive, functionally similar patterns of action
stimuli (classic features of organizational routines), yet
and a highly deliberative search process
analyzability,
has high task
in
response
to defined
low task
variety,
with interruptions (classic
filled
features of non-routine work).
To resolve
routines
this
apparent contradiction, we believe
and the methods
between routines as patterns
we
believe
so,
we suggest
it
is
necessary
we
with which
to
of action
it
is
study them.
necessary
and routineness as a variable property
of
such patterns,
examine the sequential structure of the patterns themselves. To do
a novel methodology
for
representing routines using grammatical models
work by applying the grammatical modeling technique
organization.
the distinction
In addition to clarifying
(Pentland, forthcoming; Olson, Herbsleb and Rueter, forthcoming).
prior
concept of
to rethink the
There are many ways
to
This paper builds on our
to sequential
data from a specific
model sequences of events (Abbott, 1991), but
grammatical models have distinct advantages when applied to organizational routines and other
repetitive organizational processes.
For
that
is
this
a grammar does not specify a fixed outcome;
among which members accomplish
defines a set of possibilities from
action.
First,
highly congenial to the theoretical conceptions of routines
have "sub-routines"
and Simon, 1958).
between events
are
that
in
may be
follows.
models are
widely separated
organizational routines.
in
Each
in
a way
mentioned above. Second,
organizations,
to
form
new
where routines
routines (March
well suited to representing
dependencies
an observed sequence (Chomsky, 1956), which
Finally,
because they provide a clear way
a routine from other instances.
in
can be rearranged and combined
Third, grammatical
that
commonplace
practical value
sequences of
reason, grammatical models acknowledge both structure and agency
grammatical models can capture the layered quality of action
often
specific
it
grammatical models have potential
to distinguish normatively correct instances of
of these features will
be demonstrated
in
the analysis that
Organizational Routines
We
begin by examining various concepts of routines
Drawing on these perspectives,
literature.
in
we argue
that
an organizational routine can be seen
as a set of possible patterns that need not be fixed or automatic.
Simon (1969)
some
in
complex patterns
respects,
of behavior.
in that
we argue
and sociological
the organizational
Our argument
can generate
that situational structures
To demonstrate our argument, we use
the
parallels that of
example
of
a task
that violates the traditional definitions of routineness from the organizational literature.
then use a
action
degree
new grammatical method
in this
task unit to
show
(Olson et
al, in
that in spite of
of sequential structure as defined
its
unit
We
press) to analyze the sequential patterns of
apparent non-routineness,
it
exhibits a high
by the grammar. These findings reinforce the need
a new, broader formulation of the concept of organizational routines, and a
new
for
set of tools for
analyzing their sequential structure.
DEFINITIONS OF ORGANIZATIONAL ROUTINES
At the risk of oversimplifying a large, diverse body of work, one can discern
perspectives
in
the literature on routinized behavior.
There
is,
for
some broad
example, a micro-level
perspective that emphasizes the cognitive processes of individuals (Ashforth and Fried, 1988),
and a more macro-level perspective
(Nelson and Winter, 1982).
that
emphasizes
structural
and
institutional constraints
Levels of analysis are blurred, of course, because situational
factors influence script selection
and performance (Gioia and Poole, 1984), and the cognitive
processes of individuals can enact situational structures (Weick, 1979).
There
is
also a
difference of opinion as to whether routines are fixed entities that exist independently of their
accomplishment (March and Simon, 1958) or as the performances
unfold
anew each
distinction
time (Giddens, 1984).
The
difference
is
of acting subjects that
similar to Weick's (1979)
between organization and organizing; the noun form encourages us
phenomenon as a
static thing, while the
verb form encourages us to see the
dynamic process. And as mentioned above, "routineness"
possess, whether they are fixed or dynamic.
In
is
to
see the
phenomenon as a
a property that such patterns might
the discussion that follows,
we
distill
some
Organizational Routines
insights from
each of these themes and introduce a novel, grammatical perspective on
organizational routines.
Routines as Automatic Responses
Early concepts of routines
defined stimulus.
emphasized a
fixed pattern of individual behavior
governed by performance programs."
March & Simon (1958) went on
in
organizations,
to consider
is
when an
could be considered "routinized":
We
will
been
regard a set of activities as routinized, then, to the degree that choice has
development
simplified by the
of a fixed
response
search has been eliminated, but a choice remains
and systematic computing
The idea
of a fixed
the foundation
Throughout
routine,
we
will
in
to defined stimuli.
the form of a clearly defined
say that the
still
If
activities
are
(March & Simon, 1958: 142)
routinized.
response
to
a given stimulus, accompanied by the absence of search, provides
upon which much subsequent theorizing about routines has been based.
their discussion,
and the response
More
response to a
Using the newly popular computational metaphor, March & Simon (1958:
142) asserted that "[m]ost behavior, and particularly most behavior
activity
in
lies in
to
that the link
the perceptions and reactions of the individuals
and Fried (1988) use the concept
recently, Ashforth
and Abelson, 1977)
March and Simon (1958) argue
of
in
between the stimulus
the organization.
an event schema or
script
(Schank
argue that routines are sustained by the cognitive structures of
individual organizational
members.
Ashforth and Fried (1988) argue that scripted behavior
occurs under the following circumstances:
(1)
presence
of
an event schema;
stimulus cues; (3) presence of action rules; (4) minimal required
unstructured subroutines; and
(6)
llgen (1985) offer a similar
list
that these conditions
to
appear
absence
of interruptions
effort; (5)
(2)
categorizable
absence
and unmet expectations.
of criteria for routinization of individual behavior.
of
Weiss and
Given
be rather common, Ashforth and Fried (1988) argue that
6
Organizational Routines
scripted or mindless behavior
and informal
is
commonplace
and power-based
interactions,
in
operating routines, decision making, formal
interactions.
argument
respect, their
March and Simon (1958).
mirrors that of
At the group level of analysis, Gersick and
Hackman
routines' that retains the notions of a fixed stimulus
A
In this
habitual routine exists
(1990: 69) offer a definition of "habitual
and the absence
when a group repeatedly
of search:
exhibits a functionally similar
pattern of behavior in a given stimulus situation without explicitly selecting
it
over alternative ways of behaving.
Unlike Ashforth
and Fried (1988), Gersick and Hackman (1990) do not
concept of individual scripts as an explanatory construct.
habitual routines
emerge out
of the
complex
interaction of
They argue
rely directly
that at the
group members.
on the
group
Routines are
repetitive patterns of action that are "functionally similar," but not necessarily fixed.
may be some
words, there
variety in the patterns enacted by the group, but
level,
no
In
other
explicit
deliberation.
At the organizational level of analysis, Nelson & Winter (1982) introduced a wide variety of
metaphors
for routines: routines
targets for control, replication,
kind of thing.
The
as genes, routines as memory, routines as truce, routines as
and
imitation.
Each
definition of routine offered
of these
metaphors portrays a routine as a
by Winter (1986) expands on March & Simon's
(1958) notion of a fixed response to include patterns of behavior that function as a recognizable
unit:
...to
the extent to which that reference
is
made
to
a relatively complex pattern of
behavior (or the theoretical representation of such a pattern) triggered by a
relatively
small
number
recognizable unit
in
of initiating signals or choices
and functioning as a
a relatively automatic fashion, to that extent reference
is
Organizational Routines
made
to
something called a routine
in
evolutionary theory.
(Winter,
1986:
165)
Instead of focusing on measurable properties of a task unit's work flow (Perrow, 1967),
Nelson and Winter (1982) shifted the focus of theoretical interest to the patterns themselves,
objectified
The
under the label "routines."
elements of stimulus
definition retains the familiar
and automatic response, but unfortunately, the concept of routine as used
is
unacceptably vague and encompassing, as
Nelson and
It
I
use the word
this
in
evolutionary theory
quote from Winter (1986: 165)
illustrates:
a way of doing things.
"routine" as the generic term for
simultaneously the counterpart of a wide range of terms employed
is
everyday
life
and
in
in
various theoretical languages, including those of orthodox
and behavioral economic theory; among these terms are decision
rule,
technique,
standard operating procedure, management practice, policy, strategy,
skill,
information system,
information
structure,
program, script and organization
form.
Winter's (1986) definition
of constructs across
all
is
both troubling and intriguing because
levels of analysis.
In all
In
as
Effortful
draws upon a wide range
the manifestations Winter (1986) alludes
however, routines are essentially automatic, executed without
Routines
it
to,
explicit deliberation or choice.
Accomplishments
recent social theory, routines are also a central concept, but have received a rather different
theoretical treatment.
of structuration:
Giddens (1984)
"The concept of routinization, as grounded
to the theory of structuration" (p. 60,
The
as a crucial aspect of the theory
identifies routinization
emphasis
in original).
regular or routine features of encounters,
in
in
practical consciousness,
He goes on
to
argue
time as well as
represent institutionalized features of social systems.
Routine
is
in
is vital
that:
space,
founded
in
Organizational Routines
tradition,
custom or
habit,
phenomena need no
but
is
it
a major error to suppose that these
explanation, that they are simply repetitive forms of
behavior carried out 'mindlessly'.
On
the contrary, as
Goffman (together
with
ethnomethodology) has helped to demonstrate, the routinized character of most
social activity
sustain
it
in
something that has
is
their
effortful
is
accomplishment.
"an achievement from
at'
who
continually by those
explicit definition of organizational routines
that routinized social activity
conversation analysis.
be 'worked
day-to-day conduct. (Giddens, 1984: 86).
While Giddens does not offer an
insight here
to
A
similar
in
the literature on discourse and
89) argues, interaction
possibilities," not
critical
.
not mindless or automatic, but rather an
theme can be found
As Schegloff (1982:
among
is
per se the
must always be treated as
something given or pre-planned.
Garfinkel
(1967) and his followers have demonstrated repeatedly that the normal troubles and practical
contingencies of daily
Even some
of the
life
require constant interpretation
most routinized kinds
and buying stamps (Ventola, 1987)
on the part the participants
to
of encounters,
and
such as
exhibit a considerable
repair (Heritage, 1984).
fast food service (Leidner, 1993)
amount
of variety
and require
effort
accomplish successfully. Customers need to be trained and
controlled so that they conform to the expectations of the service provider (Leidner, 1993).
Service interactions are especially prone to variation, of course, because they are interactional
Note that organizational routines frequently require the participation of multiple
products.
individuals;
as complex interactional products, they are unlikely to unfold the same way every
time.
These studies
call into
serious question any concept of routines as fixed or automatic.
potentially resolve the apparent
different
disagreement by suggesting that the authors are
phenomena; perhaps what Ashforth and
than what Giddens (1984) means.
Fried (1988)
mean by a
Unfortunately, insofar as either of
One
talking
routine
is
could
about
different
them uses concrete
Organizational Routines
examples (neither reports any
activity:
appear
to
be
about the
talking
formal and informal encounters and task situations of
The differences
that
data), they
these two broad perspectives on routinized
in
each perspective emerged as a corrective
to
activity
economic dogma
that
all
are not surprising, given
alternative.
organizational behavior literature, the idea that routines are automatic
the
universe of
kinds.
all
an overzealous
same
In
was a
the
reaction against
decision making involves perfect knowledge and optimal choice.
Behavioral economics can also be seen as a continuing reaction against this entrenched orthodoxy
(Winter, 1986).
effortful
In
sociology (and ethnomethodology,
in particular),
accomplishments was a reaction against the Parsonian notion
the idea that routines are
of the social
agents as
judgmental dopes whose behavior was predetermined by social structure (Heritage, 1984).
Seen
in
the context of their antitheses, each of these theses
makes a
There may also be a methodological bias that can help account
The perspective
In contrast,
all
of the minor troubles
the difference
and improvisations
(e. g., Garfinkel,
of daily
life
the perspective that views routines as automatic responses
methods designed
to elicit
Leigh and McGraw, 1989).
and then confirm a
It
in
perspectives.
accomplishments usually draws on detailed
that views routines as effortful
ethnographic observations of people working and interacting
unavoidable that
for
great deal of sense.
specific pattern in the
is
1967), so
it
is
should be apparent.
associated with
minds of subjects
(e. g.,
should not be surprising that these two methodological approaches
lead to differing interpretations of the
phenomenon.
ROUTINES AS GRAMMARS OF ACTION
To help
reconcile these apparently conflicting perspectives on routinized social action,
propose a theoretical model that allows us
model
of organizational routines
to integrate the insights
must acknowledge
that routinized behavior
enabled by the cognitive structures of individuals, such as
1
from each of them.
scripts,
is
we
Such a
constrained and
as well as physical and social
Organizational Routines
At the same,
structures of the organization.
we
that gives rise to the particular patterns
We
must allow
it
define sets of possibilities.
sequence
of
elements or events,
The concept
Kennedy and
Thomason, 1978) and nearly anything else
create
grammars
we can
chromosomes (Gonzalez and
To
set.
Our method
builds
sequences
the
of action
the extent
observed
on Salancik and Leblebici (1988), who used a grammatical
work by using a grammatical model
processes as observed
in
an
in
detect sequential structure and functionally
We
to represent the possible patterns of interaction in food service transactions.
their
for stories
processes (such as routines) can be described as ordered sets of actions,
organizational process using a grammar,
on
applied to any ordered
can be described as an ordered
that
By expressing
model
One can
problem, because
Milanesi, 1990), tasks (Sandelands, 1987),
they can also be described by grammars.
similar patterns.
for this
grammar can be
1987), regular polygons (Miclet, 1986),
political history (Alker,
that organizational
of
good choice
linguistic or otherwise.
(Prince, 1973; Ryan, 1979; Colby,
and agency
observe.
believe that grammatical models are a particularly
grammars
for the individual effort
extend
express the sequential structure of work
to
a particular organization.
Pentland (forthcoming) presents a detailed analogy between organizational routines and
grammar, as shown
the
same way
grammar
in
Figure
that linguists
In that
.
analogy, an organizational routine has a grammar.
use grammar
to describe certain
never the whole routine
1
(or
to
aspects of a routine.
What we
a whole language), but only specific instances of
we
it.
same way
allows speakers to produce a variety of sentences
grammar
organizational routine allows
members
natural language like English
can generate an
large lexicon
and a
(Coulmas, 1981).
large
An
number
to
produce variety
of
will call
performances.
infinite variety of
of rules), only
some
of
,
In
1
the
an
Of course, a
sentences (due
to
an extremely
which are ever actually uttered
organizational routine would typically allow a comparatively small
1
is
In linguistics,
them performances.
in
the behavioral world,
use
will
actually observe, empirically,
these are called sentences, but
that English
we
describe certain aspects of a language,
In
t
«
-
.
.
*
Organizational Routines
number
of possible performances, only
may have
language, which
interrogative), a task unit
Employees
rentals,
which are enacted by the participants.
several different categories of sentences
may have
(e. g., declarative,
several different categories of routines
may have
identifiable
pieces
embedded
some
its
within them.
In
organization theory,
repertoire.
In
has long been observed that processes are also
it
constructed from constituents, sometimes called "subroutines" (March
domain
in
sentences are constructed from syntactic constituents such as noun phrases, verb
phrases, and so on.
the
Like a
returns.
Like sentences, performances
linguistics,
of
a video store, for example, would enact different kinds of performances for sales,
in
and
some
of
customer service, a
typical subroutine might
other authorization number.
(Pentland, 1992).
Moves
are
like
Finally, at the
the words
in
be the
most concrete
& Simon,
1958).
In
validation of a credit card or
level,
we have
actions or
moves
the organizational lexicon from which
performances are formed, and can be operationalized through a coding scheme, as described
below.
from
This multi-level analogy allows us to
linguistics,
where the
unit of analysis is the
unit of analysis
is
make
the best use of the corresponding concepts
the sentence.
In
the
method described here, the
performance, and the goal of the analysis to induce a grammar that
describes the language (organizational routine) by analyzing a set of individual sentences
(performances).
(***
This analogy suggests a
INSERT FIGURE
somewhat
1
ABOUT HERE "*
surprising insight: an organizational routine
pattern, but rather a set of possible patterns (enabled
and constrained by a
is
not a single
variety of
organizational, social, physical
and cognitive structures) from which organizational members
enact particular performances.
There are many
activity, for
fall
example, but
within certain
to
bounds.
different variations of a
be recognizable as competent customer
Each instance
is
customer service
service,
each instance must
also, to a greater or lesser extent, the effortful
12
Organizational Routines
accomplishment of the participants and the emergent product of
this definition is quite similar to that of
similar (but not fixed) patterns that
is
emerge through
Hackman
(1990),
What
interaction.
in
that
points to
it
is fixed,
to
some
extent,
the space of possibilities for action (although that too can change, with sufficient effort).
What
are accomplished, with varying degrees of reflective agency and
we
patterns
A
Gersick and
Note that
their interactions.
are the particular
observe.
part of this conception of routines lies
critical
effort,
in
and
the relationship between structure
agency. As Giddens (1984) and others have argued,
rules,
norms, schema,
scripts,
and other
cognitive artifacts are resources for action, but they cannot be understood as determining
action.
Furthermore, individual cognitive structures of
artifacts that
can shape the flow of action
actions called
structure,
moves are shaped by
as well as by the need
among
distributed
individuals
in
to
in
a
Pentland (1992) shows
situation.
the physical and
draw upon the
the organization.
are only one category of
this kind
ritual
how
dimensions of organizational
distributed cognitive resources that are
In
Pentland's framework,
moves are
constrained and enabled by features of the situation, so that a limited repertoire of
be available
to
members
at
any given time. The argument here
beyond
quality of action in organizations extends
The enabling and constraining
hierarchy, division of labor,
unitary
individual
is
moves
moves
will
simply that the situated
to include
sequences
of
moves.
structures that typify organizational work situations (such as
and a myriad
of task-
and
situation-specific particulars)
naturally give rise to regular patterns of action.
Hutchins' (1991) analysis of a bridge crew manually navigating a ship
navigation system failed provides an excellent illustration.
(obtaining sightings, charting the current position, etc.)
constraints
ways
of
and affordances present
accomplishing
while, the
in
to settle
the electronic
distributed nature of the task
combined
with the physical
the situation, created the possibility of several different
After struggling with these possibilities for a
their navigational task.
crew was able
The
when
on one
particular division of labor
1
3
and sequence
of actions that
Organizational Routines
was most
nature of
efficient (Hutchins, 1991).
many
This example highlights the collective and distributed
organizational routines: because no single individual could navigate the ship, the
was a
pattern of action
Defining routines
in this
collective
way
accomplishment.
also helps to clarify and preserve the distinction
between routines
as sequences of action and routineness as a variable property that can be attributed to such a
To
pattern.
for the
reinforce this distinction,
ambiguous and somewhat
repetitive,
more
or less automatic,
one might
substitute the less value-laden term "process"
pejorative term "routine."
embody more
Processes can be more or less
or less variety, search,
perspective, an organizational routine need not be especially "routine," as
in
the
example
From
and so on.
we
this
demonstrate
shall
that follows.
AN EMPIRICAL EXAMPLE
Database International (DBI — a pseudonym) creates and
mainframe computers that we
will call
24 hour-a-day technical assistance
The
staff of
to
customers who are having
1
00 new
software support group on a telephone "hot
of the operators
in
The software support group
SystemTen.
25 specialists received about
answered by one
a software product
sells
line"
calls
a week.
difficulty
Customers
at
for
DBI offers
using their product.
call
the DBI
or by electronic mail, both of which
the "customer service center".
These incoming messages
provided the "stimulus" necessary to invoke the customer service routine at DBI.
After
operators determine whether the caller has a valid service contract (which nearly
do), there
was
essentially
no deliberation about how
to
proceed.
similar response: the operator "opens" a call in DBI's call tracking
problem
to
that follow,
one
of five "functional areas,"
we use data
for
(b) to
all
The stimulus always
callers
elicited
document
In
the sections
the non-routine
develop a grammatical model that describes
of functionally similar patterns.
14
a
database and assigns the
a solution begins.
collected at DBI for two purposes: (a) to
nature of the problem-solving work and
work as a set
where the search
were
this
Organizational Routines
Data
The data reported here were
conducted by the
access
to the
which to work.
informed that:
unit
and provided a passcard needed
Before data collection began,
(1)
no
individual
group was being evaluated
individuals
The management
author (Pentland, 1992).
first
work
collected as part of a larger study of software support hot lines
in
was
all
members
of the
and with management; and
(3) confidentiality
in
level of
office in
DBI support group were
any way;
(2)
no
individual or
would be maintained both with other
(4) the confidentiality of the
involved would be protected with the public.
DBI granted a high
and an unoccupied
to gain entry
obligated to participate
any way;
at
The members
of the
organization and
all
persons
DBI support group were very
cooperative and generous with their time, thereby allowing the collection of the observational
and archival data reported here.
Data collection continued
Observational Data.
observation.
looking things up
in
he was sometimes asked
manuals, or running down the
observe a situation that an informant
the
first
notes.
felt
first
five
days a week, of participant
was predominantly
author's role
to help with simple tasks
hall to
get help.
such as copying,
Occasionally, he
might be especially interesting.
Most
was
invited
of the time,
author sat with a support specialist, observing what he or she did and taking detailed
It
was
usually possible to ask questions to clarify what
have been done, and so on. Unless the
discuss their work.
situations
It
was
situation
was
the day, and the
had happened, what else might
urgent, informants
were generally happy
Throughout the
page number.
text,
of
each day,
quotations from
field
field
notes were typed
For example, "(DBI, 6.3)" would refer to day
six,
page
Archival Data.
With the help of DBI personnel, a random sample of 335 calls was extracted
from DBI's
tracking database, which contained the histories of over 20,000 problems.
1
5
in
notes are referenced by the
three of the field notes.
call
to
also possible to retreat to an office and write up observations about
where note-taking was not possible. At the end
the customary manner.
site,
months,
Given the technical nature of the work, the
that of observer, although
to
for three
Organizational Routines
Colby,
Kennedy and Milanesi (1990) recommend a minimum
creating grammatical models of folktales, so this sample size
we chose one random day
extract this sample,
all
calls received
database
it
would
is
per
of
50 instances as a basis
seemed more
typically involve
many
distinct
is
To
than adequate.
week (Monday through
Friday)
on that day concerning DBI's main product, SystemTen.
assigned a unique number and
for
and selected
Each instance
referred to by the staff at DBI as a
phone conversations. To get 335
calls,
"call,"
a
in this
although
total of
21
days were sampled between March 1988 and February 1989. The average elapsed time on the
calls in the
The
sample was 102 days.
call tracking
database provided convenient access
often stretched out over a period of
weeks
or months.
to the entire history of
The database was used by the DBI
keep contemporaneous notes on what steps had been taken, what needed
Observations by the
first
to
be done
next,
staff to
and so on.
author indicated that the specialists were quite diligent about
maintaining these records.
Since the database served a variety of useful purposes for the
support specialists and their managers, the information recorded
not necessarily complete
problems that
in
every
problem (an essential feature
for
detail.
The records cover
our analysis), but they
was
quite reliable, although
the complete history of each
limit
the
amount
of detail that
can be
recovered about any particular interaction because specialists recorded only the actions and
events that were needed
follows, the
tendency
for their
own purposes. As we
to omit certain actions
shall
demonstrate
in
from the database reduces the
the analysis that
fit
of our
hypothesized grammar, thus understating the extent of the regularity of the actual patterns of
activity.
Coding the archival data
.
The coding scheme used here was adapted from Pentland (1992)
take advantage of the call tracking database used at DBI.
set of basic
problems
moves
that support specialists
Pentland's (1992) analysis defined a
use to resolve problems, such as transferring
to other areas, escalating to the engineering staff,
This framework
was extended
to include
to
and
referring to other vendors.
problems that were transferred
1
6
to the
customer
for
Organizational Routines
Additional moves, such as opening the call
additional information or for testing.
call,
were added
for this analysis,
changes the status
as shown
one
of a call to reflect
development), their action
in
Figure 2.
of these
moves
automatically recorded
is
unambiguous way. Thus, the codes could be read
need
little
archival
grammar
INSERT FIGURE
sample was divided
(N = 168).
two
into
= 167) and a
inductively (N
statistically
in
The
calls).
code from the sequence)
Non-routine
As mentioned
Problem
in
test
sample
lack of
that
The evidence supporting
summarized
in
call
96%
of
to test the
grammar
which coded one half of the
unit-by-unit
sequences
agreement (326
calls.
of
338
to minor unitizing errors
off
Hewes and
by one position, resulting
in
Poole, 1984).
Work
this
contention
is
tracking database
based on an analysis
of the training
sample
and from observational data. This evidence
the following paragraphs.
High Numbers of Exceptions
input stream (or
was used
agreement was due
that shifted the
Solving
"*)
the introduction, this task unit's work would be difficult to characterize as
(N = 167) taken from DBI's
is
a very clear,
a training sample that was used to develop the
parts,
apparent disagreement on subsequent codes (Folger,
routine.
call to
from the database records with very
ABOUT HERE
2
Two coders were employed, each
a subset of 50
(omitting a
in
DBI
at
example, escalating the
the database
directly
After a brief training session, the coders achieved
codes
(for
a support specialist
for interpretation.
(***
The
in
When
and closing the
.
There are several indicators of the frequency of exceptions
raw materials)
to the
DBI software support group.
1
7
One
indicator
is
the
in
the
Organizational Routines
frequency of duplicate calls (different customers calling with the
In
many
with duplicate topics
its
in
software support
have not previously been encountered,
To assess
this
frequency,
for
five calls
was unique.
(calls
evidence that the problem was
call for
Seventy three of the
faci that solutions for
new problems
which a solution cannot be looked up or
problems" that had not previously been encountered.
by the
the frequency of
is
we examined each
recognized as one with a known solution.
partly explained
were only
compatibility with DBI's software); every other topic
Another indicator of task variety
retrieved).
the training sample, there
In
publish
(IBM had recently released a new operating system and these customers
had questions about
that
or similar problems).
common and some vendors
technical service hot lines, similar questions are
newsletters of "commonly asked questions."
same
calls
(43.7%) were "new
The high frequency
of
new problems
is
"known" problems are distributed to customers
on a regular basis via maintenance tapes. Thus, customers tend
to call with
new
or exceptional
problems.
Even
if
a customer has a familiar kind of problem, there are frequently details about that
customer's installation that
make
their specific
problem exceptional.
large data processing centers; they frequently have multiple
networks, and databases containing
dependent problems
many
millions of records.
customers have a similar kind of problem, the
of the
large telecommunications
There are a variety of system
that result from unique combinations of processors, storage devices,
networks, operating systems, and other software
assessment
CPUs,
DBI's customers operate
frequency of
in
use
at the
similarity
new problems tends
to
customer
can be
Even when two
site.
difficult to
detect.
Our
understate the variety introduced by these
contingencies.
Low Task
Analvzabilitv
rationalized
.
Although the world of computer software can sometimes be highly
and well-understood, the task environment
many problems
difficult to
analyze.
First,
the product
explained:
18
at
DBI had certain features that made
was very complex, as
this
engineer
Organizational Routines
I
can
tell
way on
you that SystemTen
enough
earth to create
the realm of
human
result,
was
it
the scale
cases
for
terms of complexity. There's no
in
There's no doubt that
it.
beyond
Certain modules you just have to put a skull
as a black box. (DBI, 34.6)
treat
even where
was
This
to look.
especially true of "data dependent" or timing
which could usually only be reproduced on the customer
difficult to
it's
not always possible to be sure of what caused a problem, what a feasible
solution might be, or
errors,
test
understanding.
and crossbones on and
As a
is off
when
construct an explanation for a problem even
Second,
site.
the
it
was sometimes
symptoms were known, as
this
engineer explained:
Even
if
you have the dump,
see that
it
"footprints in the
call
tracking database.
it
was a "one
symptoms
was
really
have
a footprint
is
Out
of these
167
calls, forty
the sand.
You can
in
the training sample taken from DBI's
seven (28.1%) were never solved.
Either
time occurrence" for which no meaningful diagnostic data could be
some aspect
not always possible to obtain
of a problem,
in
happened. (DBI, 5.13)
it
sand" phenomenon can be detected
collected or the customer upgraded
Finally,
you
happened, but you have no idea why
The
the problem
all
much
less to explore
of their
system and the problem went away.
enough information from users
its
causes.
In
to isolate the
these cases, there
was no way
to
analyze what had happened.
Extensive Deliberation and Search.
Given the complexity
of the
problems being analyzed,
should not be surprising that considerable deliberation and search was required.
The
it
first
author often observed technical support specialists working several hours on a single problem,
sifting
through "dumps" and "snaps" (foot-thick stacks of paper
representations of the
file
buffers,
filled
with hexadecimal
main memory, and other diagnostic information), or
stepping through programs one instruction at a time.
Other members of the support
frequently consulted, and calls were frequently transferred
1
9
between
staff
functional areas or
were
Organizational Routines
escalated to the developers
to
bear on the problems.
who wrote
the code so that the appropriate expertise could be brought
Because so many problems were
necessary to transfer responsibility
for
site-specific,
a problem to the customer
who
more information could be gathered. Customers were sometimes asked
to run
In
a test procedure and report the
no sense could
problem was
to
fix
do about
this
In
one
should be issued
new
version.
In
was an
additional
phase
some
frequently
reported
so that
it,
to recreate a
problem or
When
the source of a
of deliberation involved in deciding
particularly difficult case, several
for the current version of the
was
example.
work be described as mindless or automatic.
identified, there
it.
results, for
it
meetings were required
to
decide
product (as a "patch") or delayed
cases, a customer could simply "work around" the problem.
what
until
In
if
the
the
other
cases, a problem could be fixed by supplying a customer with a customized "ZAP" that changed
their
copy of the program.
to the underlying product.
In
still
other cases,
it
might require more extensive design changes
Choosing among these alternatives often required
escalated to higher level technical or managerial personnel.
navigational task, this work required the collaboration of
solve
Taken
that a
problem be
Like Hutchins' (1991)
many
individuals;
no
individual could
of the problems.
all
together, these indicators strongly suggest that this task unit's
terms of definitions based on Perrow (1967),
analyzability of search, at best.
this unit
They operated
work was non-routine.
In
had high task variety and moderate
with considerable deliberation
and search and
frequent interruptions, thus violating Ashforth and Fried's (1988) criteria for routinized
behavior.
routine,
Given these findings,
we
might be tempted to speculate that
no functionally similar pattern that characterized
it
this task unit
work processes.
had no
This conclusion,
however, would be based on an unwarranted inference from the properties of the work to the
sequential structure of the actions required to accomplish
develop a grammatical model and use
it
it.
In
the following section,
we
to test for repetitive, sequential patterns of action in the
work.
20
Organizational Routines
Constructing
Grammatical
the
There are several steps required
They are described and
steps
will
method
1990;
and
be
familiar,
builds
on a
it
The
first
in
to the call.
As
be less
familiar
and
will
In this
.,hich is
one performance
two different times (W
in
row
(1)
first
step
W), transferred
were taken
coded data
can be read as
respect, this
(Gottman and Roy,
creating a
in
to
is
become
grammar
in
familiar with
will
in
(C).
In
to
as "strings" or
was opened
to the user (US), transferred
worked on two more times (W W), and then closed
The
the process of responding
be referred
follows: the call
Each row
Figure 3.
of the software support routine.
(or actions) that
the discussion that follows, the
of these
receive special attention here.
twenty coded calls from the training sample are shown
call,
Some
to allow replication.
The more specialized steps involved
each row represent moves
In
in sufficient detail
with any inductive procedure, the
"sequences." The sequence
(PS),
here
Poole, 1990).
to the data will
represents a single
symbols
illustrated
variety of familiar techniques for handling sequential data
Exploring the data.
the data.
a grammatical model of an organizational process.
to construct
such as the basic coding procedure described above.
Van de Ven and
fitting
Model
back
(O),
worked on
to product support
general, the coding procedure
reduces a great deal of other information down to these strings of symbols.
For example,
we
record the sequence of events but not the interval between them (Abbott, 1991, refers to this as
"event time" as opposed to "clock time") and a wide variety of work activities
is
compressed
a single category (W).
(***
Nonetheless, the variety
constructed from the
INSERT FIGURE 3 ABOUT HERE
among
same
the performances
in
Figure 3
"*)
is striking.
They are
all
basic lexicon of moves, but the sequences vary considerably from
21
into
.
}
Organizational Routines
performance to performance. Of these 20
and
pattern (lines 4, 8, 11, 12
the other calls entail unique
support the conclusion that
The goal
was
the
of action, yet they
responding to a customer problem.
similar" objective of
the data.
18) (none of these
sequences
this
same
only the five simplest calls had the
calls,
same problem, by
is
to
All of
accomplish the "functionally
all
seem
Superficially, these data
task unit has no routine; there
of the next step in the analysis
the way).
seems
summarize
to
be too much variance
to
this sequential
in
variance using
a simple grammar.
Create a
Grammar
that Describes the Pattern
.
towards a formulating a hypothetical grammar.
As we explore the
data,
The grammar
consist of a set of phrase
will
structure rules (sometimes called "rewrite rules," for reasons that
moment)
that specify patterns
in
the coded data.
the usual sense of rules that organizational
Members
of the
the kind
in
Formulating a
grammar
is
top down, by starting with the whole routine
Alternatively,
one can work from the bottom
common sub-sequences. A top-down
of the routine
is
in
members
art.
known
in
rules,
Rather, phrase structure rules of
observed
it
about the sequence of actions observed
Whether one
is
in
These
These are
like
coded data and
priori
is
identifying
overall logic
expectations based on theory.
to
priori
account
A
theory and wants to
for the largest
number
of
rules form a kind of working hypothesis
the routines.
working top-down or bottom-up,
parts of the routine.
the coded data.
most appropriate when the
bottom-up strategy would be more appropriate when one has no a
observations with the simplest set of rules.
in
into constituent parts.
up, by starting with the
case, the goal
any more than they would
begin to formulate a grammar from the
and breaking
strategy would be
In either
will
follow or refer to while doing their work.
One can
advance, or when one has a
work inductively from the data.
become apparent
Note that these rules need not be thought of
for describing patterns
something of an
be working
a
a conventional regression model.
used here are simply devices
will
in
DBI technical support group do not follow these
follow the rules implicit
we
it
may be
helpful to look for the constituent
the "syntactic constituents" (Cook, 1988) that
22
make up a
Organizational Routines
These are sub-patterns
formal grammar.
can be meaningfully isolated from the
that
sequence. For example, each software support
for the solution,
followed by a closing.
Software
Support
"-->"
where the symbol
This
"followed by."
is
Call
be
rule could
rules as describing
into
The next step
its
in
procedures
Solution,
for
and the commas
in
the
list
Closing.
indicate
Each
side.
rule
In
has a single symbol on the left-hand
al.,
side,
and
English grammar, for example, one might have
written
how
as follows: S
-->
NP, VP. One can think of these H ,irase structure
a particular entity (such as a customer service interaction) can be
constituent parts.
In this
respect, the
is
to
enumerate the ways
in
which each constituent can
method shares some common ground
with ethnographic
mapping means-ends semantic domains (Spradley, 1979). The semantic domain
for
here can be thought of as the "ways to accomplish" the constituent
general, there
model
Search
could be expressed as:
a simplified version of how phrase structure grammars (Gazdar et
developing a grammar
be accomplished.
of interest
of"
this
says that a sentence (S) can consist of a noun phrase (NP) and a verb phrase (VP).
rule that
expanded
grammatical form,
can be read as "consists
one or more symbols on the right-hand
Such as
consists of an opening, followed by a search
call
Opening,
-->
1985) are used to represent sentences.
a
In
rest of the
this,
may be
several possible
one simply includes
alternative gets
one
rule.
ways
to
Every
usually several
ways
retail
that
question.
alternative expansions of that constituent;
each possible
Theoretically, these alternatives are functionally similar
satisfy
a
in
the sense
common means-ends
transaction, for example, involves form of payment, but there are
payment can be accomplished
(e.g.,
cash, check, credit card,
Spradley's (1979) terms, these alternatives comprise the domain, "ways of paying."
terms, they would be included as separate rules
forthcoming).
In
accomplish a given constituent of a routine. To
suggested by Gersick & Hackman (1990) because they
relationship.
in
in
Each constituent must be expanded
23
a
grammar
in this
etc.).
In
In
our
of retail transactions (Pentland,
manner
until
it
can be expressed
in
Organizational Routines
terms of the basic lexicon or coding scheme. This
is
what allows the grammar
to
be tested
against the data, as described below.
was used
This procedure
process, as
shown
examining the sequences
carefully
observational data.
As
see how
well
sample
was
of
in
to
The
required.
first
O.
with a single
more
If
more
it
rule in Figure
in
detailed data
detail to include the
The body
of
a solution.
each
a
calls are
new
ways
call, this
moves
include any
there are
of
judgment
some
is trivial:
constituent could have
initial
opening
been coded
conversation with the caller,
In that
case,
we would add
be accomplished.
that represent the process of searching for
sequence
of transferring, escalating, or working
special constituents within the search process, that
For example,
in
the second line of Figure 3 the
that responsibility for the
problem was turned over to
and then taken back by product support (PS). This sequence happens frequently
problems where the only way
to
advance the
information or test something on their system.
solution
When
completed, the specialist assumes responsibility again.
patterns
amount
by a closing. Since the opening was
the tracking database.
that an opening could
sequence "US PS" occurs. This means
in
each
answered, the
call into
labeled "user" and "developer."
the user (US)
for
consists of a sequence of
In addition,
to the training
it
4 states that the overall structure of the pattern consists
for the solution, followed
The search process can
on the problem.
we have
call
insights gained from the
the lexicon, the rule that expresses the opening
way
rules to express the alternative
and applying
were developed by
grammar took shape, we applied
were available
to enter
identified
Like any inductive exercise, a certain
the data.
fit
DBI software support
for the
we have
the training sample
the hypothetical
symbol
and the steps required
particular rules
in
an opening, followed by a search
coded
-->
The
Figure 4.
in
a hypothetical grammar
to induce
is
to
have the customer gather more
the information arrives or the test
Figure 4 also includes a similar set of
between the developer (DV) and product support (PS)
constituents.
24
is
that
we have
labeled "developer"
Organizational Routines
(***
Each
INSERT FIGURE 4 ABOUT HERE
ends with a closing sequence
call
indicating that the call
is
that involves
a tentative
fix,
comment
"done", followed by a trailing
Observations of the work process and an examination of the
call
***)
followed by an action
what was done.
that explains
data indicated that closing a
call
does not necessarily imply solving the problem. Many
of the calls that are handled by
reports of one-time problems that are never duplicated
and never solved. Nonetheless, these
calls are usually either
alternative
sequence
is
codes
indicating that a
involving a test of the
comment
Rewriting
it
that
I
fix
rules,
The "done"
fix).
(inactive).
some
In
is
very similar to what
we would
tentative
fix
internal structure
contains one of three
is
represented by either
C
D
(closed) or
another symbol indicating an additional
calls, there is
call (E).
grammar against
to test the
have an
or T) followed by a work episode (usually
constituent
consists of taking each coded string and
Rewriting
R
A
explains what steps were taken to close the
The procedure used
.
tried (F or
Figure 4.
includes an
and developer episodes,
level constituents that
in
grammar
reason, the
this
Like the user
fix.
as shown
has been
For
inactive.
that omits the tentative
expressed by further
(deferred) or
become
and "done" are intermediate
"tentative fix"
that
closed or
DBI are
literally
rewriting
the data
it
is
called "rewriting", since
using the hypothesized rules.
One searches
normally describe as coding.
data
for
a specified pattern, but instead of simply annotating the data, one actually replaces the original
data with the code (hence the term rewriting).
with the "search
one looks
For simple rules, one could accomplish this
and replace" option available
for patterns in the
found, that part of the data
in
any word processing program.
data that match the right-hand side of a
is
rule.
When
replaced with the left-hand side of the rule (the
way, detailed patterns
constituent).
In this
constituents.
This basic procedure
is
in
the data are replaced by the
repeated
for
25
each of the rules
in
the
In rewriting,
a match
name
is
of the
more abstract
grammar
until
no
Organizational Routines
further rewriting
This can be accomplished manually, or
possible.
is
it
can be accomplished
using a computer program designed for pattern matching and re-writing. 1
The
Figure 5 illustrates the procedure for rewriting a single string, one rule at a time.
row
in
Figure 5
shows
the raw data, consisting of a sequence of seven moves.
row shows the same data
the path of
after
each data element as
indicate that
Thus,
the rule "Opening --> O."
The
it
an element has been
simple sentence diagram.
C."
has been rewritten by a single
it
ability of rewriting to
the rule "Closing --> Done,
to a single
were
If
grammar
is
perfectly
of strings in a
Comment" was used
symbol
grammar
in
sample
fits
is
been
be sequences
in
in
the data that
For the analysis that follows,
I
lei
it
Double
lines
would resemble a
been
rewritten using
lines.
"Done
-->
where
the fourth row,
two constituents with one, as
In
re-written
the final row, three constituents
that represents the overall pattern.
to
in this
example, then
use
do not
it
fits
the
have been "parsed" by the grammar. This would be
fits
this
measure
by the grammar
in
is
one measure
the analysis that follows.
fit
we used
the hypothetical grammar.
a program written
personal computer.
26
in
like
The number
the model with an error term of zero.
observed sequences can always be completely
all
and run on a Macintosh
will
vertical lines trace
rewritten using the rule
demonstrated
to replace
that are completely rewritten
we
inverted,
Each successive
each of the next three rows, two symbols were
In
a regression that
the data;
were
the second row, note that the "O" has
summarize patterns
and can be said
example, however, not
1
this figure
completely rewritten to a single symbol, as
having a data point
the
If
symbol, as indicated by the converging
unified into a single
a string
rewritten.
in
indicated by the converging lines.
The
transformed by the rewriting process.
is
the next row, the "C" has
In
rule.
first
of
how
well
Unlike this
rewritten.
There may
Since rewriting tends to
MacPROLOG
4.0 (Quintus Corp.),
Organizational Routines
reduce the length of the string being rewritten, the percentage reduction
indicator of the
degree of
(Olson et
fit
Our hypothesis
Testing the grammar.
is
length
another
is
forthcoming).
al,
INSERT FIGURE 5 ABOUT HERE
(***
in
"*)
that in spite of frequent interruptions
and extensive
search, the support process embodied sequential patterns that could be represented by a
grammar.
A
convenient
hypothesis that allows us to formulate a
null
statistical test is that the
observed sequences of action are random and do not follow the hypothesized pattern. To
whether the service process
rewriting
procedure on second
data were not included
One hundred
strings,
in
DBI conformed
half of our
to
our hypothesized grammar,
coded data, the
the training sample that
test
was used
most were
rewritten to
some
grammar
to
we needed
reduction
the remaining
in
the length
the kind described by Olson et
We
random permutations
al
of
(forthcoming).
each
same
length
grammar
to
each
of the
Thus, each randomized dataset
and composition
(To visualize this procedure, imagine taking the sentences
each sentence 100 times
simple bootstrap procedure of
generated 100 sets of random data by
string in the real dataset.
consisted of 168 strings, each with the
it
a set of random sequences. Since no probability
we performed a
the hypothesized
much
In
to estimate the probability that
distributions are available for this kind of data,
in
that these
These ungrammatical cases are discussed below.
could be obtained by applying the
rearranging the words
the
to formulate the hypothetical model.
extent, but not always with
test the statistical significance of this result,
creating
we performed
sample (N = 168). Note
nine of the strings, or about 64.9%, were rewritten completely.
of the string of symbols.
To
at
test
in this
to create
of
original.
paragraph and randomly
100 new paragraphs.)
100 randomized datasets; the
27
elements as the
We
mean number
applied
of strings
*
-
-
*
-
.
Organizational Routines
was 9.57
completely rewritten
result
The observed
(out of 168), with a standard deviation of 2.77.
from the real data (109 out of 168) was over 35 standard deviations higher than the
mean.
Analysis of unarammatical patterns.
grammar might be
answer
better or
this question,
it
is
useful to analyze the residual patterns that
incompletely rewritten calls)
disparity in the test
was
sample (appearing
never taken back by DBI's product support group.
had
implicitly
whether some other
do not
in
fit
the hypothesized
twenty four of the
the presence of a single "US" without any corresponding
This pattern would imply that responsibility for the
specialist
of course,
more parsimonious than our hypothesized grammar. To help
grammar. The most common
"PS."
The question remains,
In
was
call
nearly
all
transferred to the user and
of these instances,
taken responsibility by continuing to work on the
other action, but had neglected to indicate this
in
the database.
a DBI
The existence
of this pattern
makes sense, because
the high workload encourages support people to give problems
take them back only
necessary.
if
reduces the apparent
fit
of the
of calls completely rewritten
The remaining few
seemed
to
officially
Many
people simply attended
more fundamental departures from the grammar.
In
to require additional work.
had been transferred
to
For
these cases, problems that
Thirteen of the calls had
to the
customer
for further
data which
Without customer pressure on these problems, the support
more pressing matters.
These exceptional patterns are
in
the data had reflected this move, then the percentage
closed (or deferred or declared inactive), although work had apparently
apparently never materialized.
ways
If
had been closed and then reopened.
of these calls
simple structure that
to
would exceed 79%.
have been solved turned out
never been
ceased.
calls
away and
This omission on the part of the DBI staff significantly
grammar.
calls reflected
example, seven of the
some
call or taking
theoretically important
we have hypothesized
is
because they indicate
not always attained
which the hypothesized grammar might be improved.
28
in
that
practice.
One might want
even the
They also suggest
to
add a
rule, for
Organizational Routines
example, that allows
support.
One
user episodes that do not include returning responsibility
for
based on
to
over
90%
and then re-opened. These
of the test sample, but they
insights from the fieldwork, of
Furthermore, there
keep adding rules
which to exclude
is
until
every possibility
essentially a
is
we do
rules
would improve the coverage
is
judgment
model;
overfitting the
covered.
call.
of the
would also contradict our normative sense,
what constitutes a complete, competent
always the danger of
against a null hypothesis,
different
product
might also want to add a rule that allows patterns that have no closing constituent,
or patterns that are closed
grammar
to
The decision
Although
not yet have tests to
we can
compare
in principle,
call.
one can simply
of which rules to include
test
and
a hypothetical grammar
the relative performance of
grammars (Simon, 1992).
DISCUSSION
This grammatical analysis reveals a great deal of regularity
superficially to
shown
in
captured
be quite unstructured.
figure 3)
in
appear
to
The observed
it
Even though the software support performances (as
We
embodies a
regularity arises from the
structural features of the situation
transferred
between
in
way
specialists, information
on.
the work
must be
If
in
routines are often associated with specialized
repetitive
is
organized
These assets define a
is
(e. g., division of labor),
of the participants.
solicited
Calls
must be
from the user, calls must be
our dataset allowed us to code the process at a
would emerge.
which routines are embedded
specific assets."
can be
repetitive, functionally similar pattern of action.
finer level of detail, additional regularities
way
that
are forced to conclude that although this work
and the cognitive models
opened before they can be closed, and so
of the
data that would appear
be quite diverse, they share an underlying structure
the rules of a grammar.
apparently non-routine,
in
It
organizations.
human
could hardly be otherwise, because
As Winter (1986: 168)
or physical capital that he calls "routine-
set of possible actions which give rise,
patterns.
29
notes,
in
practice, to
Organizational Routines
In
the DBI software support group, for example, specialists have a limited lexicon of
At any point
with which to accomplish their work.
how
to
proceed with solving a problem, as reflected
creativity
and diligence applied toward
next are quite limited.
example, once a
Further, there
in
the lexicon.
this goal, but the
a user,
it
variety;
to
be
identical.
What we
these patterns
They
present.
will
emerge whether or not
arise
'
jm
tue
way
affordances of the situation.
the work
of
among some
of these
moves. For
fall
far short of constraining
all
can best be described as constrained
find, instead,
performances that are functionally
There may be a great deal
must be transferred back. These dependencies
give additional structure to the performances, but they
performances
few choices about
generic choices about what can be done
a logical dependence
is
call is transferred to
time, there are only a
in
moves
similar, but not necessarily the
same.
Finally,
the preconditions for purely scripted behavior are
is
organized, as reflected
This interpretation
in
the constraints
quite consistent with Nelson
is
and
and Winter's
(1982) ideas about routine-specific assets, routines as truce, and other contextual features
that enable
and constrain
they allow the agents
in
action.
Cognitive structures are also important, of course, insofar as
the situation to employ the appropriate assets, observe the truce, and
thereby produce the desired pattern of action.
Conceptualizing routines grammatically draws attention to the relationship between an
organization's structure and the
sequences
of action that occur there.
structure defines the set of possibilities, but not the particular
becomes
action.
natural to ask
For example, might a
developers
To even begin
make up
to
How
in
new way
alter the pattern of
patterns of work?
that
how changes
work?
In this
sequences we observe.
structural features might alter the
of
framework,
Thus,
observed pattern
it
of
communicating and sharing information with users and
How
would a
persistent are routines
in
different division of labor affect the
the face of changing institutional structures?
address such questions, one must be able to summarize the patterns of action
the hypothesized routines.
30
Organizational Routines
Grammatical models have
for this
purpose.
distinct
advantages over other kinds of models
(e.g.,
Markov models)
they describe a set of possible patterns of action from which
First,
This formulation has theoretical value, because
enact particular patterns.
it
accommodates
apparently conflicting perspectives on routines as fixed patterns and routines as
accomplishments.
It
has methodological value, as
well,
because
members
the
effortful
real routines are unlikely to
Rather, they display a variety of functionally similar patterns that
display "fixed" patterns.
can be concisely summarized using the kind of formalisms described here. Second, phrase
way
structure rules provide a convenient
routines") that
expanded
into
to describe the layers of constituents ("sub-
can be rearranged and combined
to
form
new
Each constituent can be
patterns.
a family of functionally similar patterns that represent alternatives ways of
accomplishing that part of the routine.
Third, as illustrated in the
case of the user and
developer episodes, grammatical models are well suited to representing connections between
events that
way
may be
widely separated
in
a sequence.
to distinguish normatively correct instances of
illustrated in the
tendency of the DBI support
Fourth, grammatical
models provide a clear
a routine from other instances, as
staff to
avoid claiming responsibility for certain
calls.
Last but not least, the use of a grammatical model allows us to operationalize the definition of an
organizational routine as a set of repetitive, functionally similar patterns of action with a
degree
is
of precision that
a "pattern of action"
grammar
has not been possible before.
if
it
In particular,
a sequence of coded symbols
can be parsed by a grammar. The patterns
of action within the
are established as "functionally similar" by their inclusion within a
ends semantic domain (Spradley, 1979). That
particular "end."
instances of
it
And
within
to interpretation
finally,
some
is,
a pattern of action
each pattern
is
"repetitive"
in
if
the
grammar
one can
time period of practical or theoretical interest.
and refinement
in
common meansis
a "means"
collect multiple
While
subject
still
oarticular settings, these operationalizations help to
the concept of organizational routines
more
explicit
counts as a routine and what does not.
31
by defining empirical
to a
criteria for
make
what
Organizational Routines
A New Comparative Framework
If
we
carry the analogy to language forward,
comparative framework
similarities
between
patterns.
in
many
and so on.
A
the
there are grammatical
may have
the
is
of activity into concise, high-level
allows linguists to identify families of
that share certain syntactic characteristics
recruiting,
a grammatical approach
of
The
different.
such as sales, purchasing,
advantage
different.
structural
each organization may be quite
in
potentially powerful
same way grammar
kind of
words are completely
organizations
summarize innumerable diverse patterns
In
new
human language,
true of other kinds of processes, as well,
auditing, planning,
ability to
In
different languages, although the actual
although the specific actions taken
same may be
suggests the possibility of
for organizational analysis.
By analogy, the customer service process
similarities,
it
human languages
(Newmeyer, 1983), a grammar
of routines should
allow organizational scholars to identify families of organizational processes and to chart their
changes over
time.
The grammatical framework becomes
profusion of lexical items, such as nouns.
grammar
As a
facilitates
when
especially valuable
By creating categories
there
is
a
of functional equivalence,
comparisons across sets of apparently diverse instances.
practical matter, grammatical
models seem most appropriate
sequential records are kept or can be easily generated.
especially for processes that are distributed
in
for situations
Observational data
time and space.
is
Fortunately,
where
expensive,
many
kinds of
business processes generate archival data as a byproduct of normal operations which could be
analyzed using these methods.
Bill
collectors, for
could potentially be useful (Sutton, 1991).
all
example, keep records of
With the increased use of information systems
kinds of service and manufacturing organizations, the necessary data
more and more
Assuming
their activities that
is likely
to
in
become
available.
that data are available, the
success of
this
approach
formulate an appropriate lexicon and syntactic constituents.
abstraction are analogous to the constructs
one
finds
32
in
will
depend on our
These intermediate
traditional
ability to
levels of
nomological research, except
Organizational Routines
that instead of referring to the variable properties of objects, they refer to actions or
By introducing abstract constituents
collections of actions.
"closing") to represent parts of a routine,
compared
in
terms of
their constituents.
it
allows routines that differ
In linguistics,
it
is
in
In
is
the existence of syntactic
all
subsequent
the arrangement of these constituents that characterizes the
much more so
structure of a language,
surface structure.
It
than the details of the vocabulary or other aspects of
the case of software support, the simple three-part structure for calls
(opening, search for solution, closing) suggests the possibility of comparing the
different organizations
be
their details to
constituents such as noun phrases and verb phrases that forms the basis for
theorizing (Cook, 1988).
and
"opening", "search"
(e.g.,
ways
accomplish openings, searches, or closings, independent of
that
how
the other
parts of the overall .out me are accomplished.
Assessing
Models
Alternative
Although organizational routines might be expected to have a great deal of regularity, there
always the
possibility of exception,
producing contingencies.
fit
an hypothesized model.
It
breakdown,
seems unreasonable
It
creativity, shirking, error,
to
expect that
would be highly desirable
to
all
and other
variety-
instances of a routine would
have a systematic way of ascertaining
whether an hypothesized model could be improved upon by adding or deleting
rules.
(1992) notes that the best available test for the adequacy of rule based models
heuristic that the
is
model should explain many more cases than the number
is
Simon
the general
of rules.
Using
this
rule of
thumb, our hypothesized model fares reasonably well by explaining 109 cases (plus a
similar
number
in
the training sample) with only 13 rules.
between the complexity
currently
have no
of the
grammar and
statistical test for
the
number
of
There
cases
it
is
clearly a trade-off here
can describe, but we
assessing the relative contribution of individual rules.
33
t
Organizational Routines
Some
Assumptions
Issues and
The key assumption
of this
is
that organizational
This assumption
syntactic constituents.
survey responses
mean
and aggregated.
While
is
method
same
the
is
no
thing to different respondents
The advantage
best employ
routine
strings of
summary measures
approach
is
is
in
we observe
us from the object
to distance
that unlike traditional linear models,
assumes
that
it
is
meaningful
are always just the
tip
of
to isolate the
which
at
reduced
to
synchronic structure of routines
As Fabian (1990)
they are created by
diachronic
in
it.
We
would argue, however,
order to achieve a degree of
moment
in
time.
Finally, there is
always the
that
in
of existing
at
any
moves), but the source of these
to the model.
risk that
way
Rules are resources
in
a
determine action (Zimmerman, 1970; Giddens, 1984).
can embody physical or
grammar
describing patterns of social action with
understanding of the nature of the social world.
such a
we observe
This framework can be used to describe lexical evolution (new
life
of
helpful to bracket off the
of action
misinterpreted as an attempt to objectify social
An example
repetition of
which they occur just as
is
it
about the patterns
clarity
moves) and syntactic evolution (new patterns
changes would be exogenous
notes, the performances that
an iceberg that consists of the rehearsal and
these patterns over time; these performances create the context
rules
it
any methodology that involves a
that the experiences of the participants are
from their diachronic context (Barley, 1990).
particular
Blumer, 1968),
coded symbols.
This method also
much as
than the assumption that
of the overall routine, the essential sequential structure of the
The disadvantage
preserved.
is
of this
meaningful
and can therefore be compared
phenomenon, the representation tends
representation of a complex
into
criticized (Cicourel, 1964;
more commonly treated as completely unproblematic. As
of our inquiry.
can be coded
different, in principle,
assumption has been
this
life
logical constraints
rule (from Salancik
&
on the
is
for action,
is
but they do not
a sense
possibilities for action in
is
that
be
inconsistent with our best
Nonetheless, there
Leblebici, 1988),
34
that
will
in
which
a situation.
one cannot eat a meal
Organizational Routines
before
it
has been prepared. Note that
has been prepared,
deterministic, but
it
it
this rule
does not force one
to eat
a meal
just
simply rules out the possibility of reversing this sequence.
does constrain the set
methodology makes no attempt
to
of possibilities.
In
is
not
is
It
it
recognition of these issues, our
uncover any rules that people
rules to describe patterns of action, this
because
Although
"really use."
we use
merely a convenient way of summarizing sequential
patterns of action that allows comparison and generalization across settings.
For
this
reason, the models presented here are rather different
symbolic or computational models that have been used
theory.
Unlike
in
processes
when confronted
that,
reproduces a particular pattern of behavior.
an ant carrying a crumb,
for
example,
is
members
and organization
a reductionist model of
that the circuitous path of
While people have more extensive cognitive
do work
in
constrained environments that
Grammatical models of organizational routines provide a
convenient way to summarize those possibilities and render them available
analysis, but they leave the
this respect,
enactment of
particular patterns
open
to unfold
for inspection
anew each
and
time.
In
grammatical models capture the essence of the phenomenon, as eloquently
expressed by one of the anonymous
road.
of the
with a set of situational constraints,
Simon (1969) argues
of organizations
limit their possibilities for action.
to create
many
the product of the ant's simple rules for navigation
applied to the constraints of the environment.
capabilities than ants,
than
cognitive science
Simon (1969, 1992), we are not attempting
individual cognitive
in spirit
ASQ
reviewers: "Routines are like ruts
They do not exactly determine where the next wagon
describe where past
will
in
go, but neither
a well-traveled
do they merely
wagons have gone."
CONCLUSION
Imagine a community of scholars
color at the center of
some
yellow, green or blue
and no way
for
of their
whom
color
is
a
critical
concern.
These scholars place
most important theories, yet they have no words
of distinguishing
35
among them.
for red,
Organizational scholars are not
Organizational Routines
quite this
weak,
bad
at best.
off,
of course, but our tools for describing
Our theories
refer to routines
and generalizing about routines are
as though they were concrete objects, yet
no systematic means of describing these objects. By providing a way
structure of routines, grammatical
organization theory
in
the patterns of action
ways
we
that
call
models create the
to represent the sequential
possibility of investigating
were not previously possible.
It
36
we
core issues
in
remains to be seen whether
routines merit the theoretical importance
without a methodological lens to help us see these patterns,
we have
we have
given them, but
might never know.
Organizational Routines
REFERENCES
Andrew
Abbott,
1991 "A primer on sequence methods." Organization Science,
Alker,
Hayward
375-392.
R., Jr.
1987 "Fairy Tales, Tragedies, and World
Possibilist
Ashforth, Blake E.
Histories:
Towards
Interpretive Story
Grammars as
World Models." Behaviormetrika, 21: 1-28.
and Yitzhak Fried
1988 "The Mindlessness
Barley,
1:
of Organizational Behaviors."
Human
Relations, 41: 305-329.
Stephen R.
1990 "Images of Imaging: Notes on Doing Longitudinal Fieldwork." Organization Science,
1:
220-247.
Blumer, Herbert
1969 Symbolic Interactionism: Perspective and Methods, Englewood
Cliffs,
NJ: Prentice-Hall.
Chomsky, Noam
1956 "Three models
IT-2:
Cicouiel,
for the description of
language." IRE Transactions on Information Theory,
113-124.
Aaron
1964 Method and Measurement
in
Sociology,
New
York: Free Press.
Cohen, Michael D.
1991 "Individual Learning and Organizational Routine: Emerging Connections." Organization
Science,
2:
135-139.
Colby, Benjamin N., Sayuri Kennedy and Louis Milanesi
1991 "Content Analysis, Cultural Grammars, and Computers." Qualitative Sociology, 14: 373-
387.
Cook, V.
J.
1988 Chomsky's Universal Grammar, Oxford, England: Basil Blackwell.
37
Organizational Routines
Coulmas, Florian
1981 Conversational Routine: Explorations
in
Standardized Communication Situations and
Prepatterned Speech, The Hague: Mouton.
Daft,
Richard
L.
and Norman B. Macintosh
1981 "A Tentative Exploration into the Amount and Equivocality of Information Processing
Work
Organizational
Units." Administrative
in
Science Quarterly, 26: 207-224.
Fabian, Johannes
1990 Power and Performance: Ethnographic Explorations through Proverbial Wisdom and
Theatre
Folger,
Joseph
in
P.,
Shaba, Zaire, Madison, Wl: University
Dean
1984 "Coding Social
E.
Hewes and
Interaction." In
of
Wisconsin Press.
Marshall Scott Poole
Brenda Dervin and Melvin
Communication Science, Volume
IV:
J.
Voight, (eds.) Progress
in
115-161. Norwood, NJ: Ablex.
Garfinkel, Harold
1967 Studies
Ethnomethodology, Englewood
in
Ewan
Gazdar, Gerald,
Klein, Geoffrey
Cliffs,
NJ: Prentice-Hall.
Pullum and Ivan Sag
1985 Generalized Phrase Structure Grammar, Cambridge, MA: Harvard University Press.
Gersick, Connie
J.
and
J.
1990 "Habitual Routines
Richard
in
Hackman
Task-Performing Groups." Organizational Behavior and
Human
Decision Processes, 47: 65-97.
Giddens, Anthony
1984 The Constitution of Society, Berkeley, CA: University of California Press.
Gioia, D.
1984
and
P.
"Scripts
Poole
in
Organizational Behavior,"
Academy Management Review,
9,
449-459.
Goffman, Erving
1981 Forms of Talk, Philadelphia: University of Pennsylvania Press.
Gonzalez, Rafael C. and Michael G. Thomason
1978 Syntactic Pattern Recognition: An Introduction, Reading, MA: Addison-Wesley.
38
Organizational Routines
Gottman, John M. and Anup K. Roy
1990 Sequential Analysis:
T.
1983 "Structural
for
Behavioral Researchers, Cambridge: Cambridge
Press.
University
Hannan, Michael
A Guide
and John R. Freeman
and organizational change." American Sociological Review, 29: 149-
inertia
164.
Heritage, John
1984 Garfinkel and Ethnomethodology, Cambridge, England:
Polity Press.
Hutchins, Edwin
1991 "Organizing Work by Adaptation." Organization Science,
2:
14-39.
Leidner, Robin
1993 Fast Food, Fast Talk: Service Work and the Routinization
of
Everyday
Life,
Berkeley:
University of California Press.
Leigh,
Thomas W. and
Patrick F.
McGraw
1989 "Mapping the procedural knowledge
of industrial sales personnel:
investigation." Journal of Marketing,
Levitt,
53:
A
script-theoretic
16-34.
Barbara and James G. March
1988 "Organizational Learning."
Annual Review of Sociology: 319-340.
Lynch, Beverly P.
1974 "An Empirical Assessment of Perrow's Technology Construct." Administrative Science
Quarterly,
March, James
G
and Herbert A. Simon
1958 Organizations,
Miclet,
338-356.
19:
New
York: John Wiley and Sons.
Laurent
1986 Structural Methods
in
Pattern Recognition,
New
York: Springer- Verlag.
Nelson, Sidney G. and Richard R. Winter
1982 An Evolutionary Theory
Newmeyer, Frederick
of
Economic Change, Cambridge, MA: Harvard University Press.
J.
39
Organizational Routines
1983 Grammatical Theory:
Its
Limits
and
Its
Possibilities,
Chicago: University of Chicago
Press.
Olson, Gary M.,
James
D. Herbsleb
and Henry H. Rueter
Forthcoming "Characterizing the Sequential Structure of Interactive Behaviors through
Statistical
and Grammatical Techniques." Human Computer
Interaction.
Pentland, Brian T.
1992 "Organizing Moves
37:
in
Software Support Hot Lines." Administrative Science Quarterly,
527-548.
Pentland, Brian T.
Forthcoming "Grammatical Models of Organizational Processes." Organization Science.
Perrow, Charles
1967 "A Framework
for the
Review, 32:
Comparative Analysis of Organizations." American Sociological
194-208.
Prince, Gerald
1973 A Grammar of
Stories,
The Hague: Mouton.
Ryan, Marie-Laure
1979 "Linguistic models
Semiotica, 28:
in
Narratology:
From
structuralism to Generative Semantics."
127-155.
and Husseyin Leblebici
Salancik, Gerald R.
1988 "Variety and Form
in
Organizing Transactions:
Nancy DiTomaso and Samuel
A
Generative
B. Bacharach, (eds.)
Research
Grammar
in
of Organization." In
the Sociology of
Organizations: 1-32. Greenwich, CT: JAI Press.
Sandelands, Lloyd
E.
1987 "Task Grammar and
Attitude." Motivation
and Emotion,
11: 121-143.
Schank, Roger C. and Robert P. Abelson
1977
Scripts, Plans,
Goals and Understanding: an Inquiry
Hillsdale, NJ:
Schegloff,
Emanuel
Lawrence Erlbaum.
A.
40
into
Human Knowledge
Structures,
Organizational Routines
1982 "Discourse as an
come between
interactional achievement:
sentences."
on Languages and
In
Some uses
Deborah Tannen,
(eds.)
of 'uh-huh'
and other things
that
Georgetown University Round Table
Linguistics 1981: 71-93. Washington,
DC: Georgetown University
Press.
Simon, Herbert A.
New
1957 Administrative Behavior,
York: Macmillan.
Simon, Herbert A.
1969 The sciences of the
artificial,
Cambridge, MA: MIT Press.
Simon, Herbert A.
1992 "What
Spradley,
is
an "explanation"
James
of behavior?" Psychological Science, 3:
P.
1979 The Ethnographic Interview,
Sutton, Robert
150-161.
New
York: Holt, Rinehart, Winston.
I.
1991 "Maintaining norms about expressed emotions
-
The case
of
bill
collectors."
Administrative Science Quarterly, 36: 245-268.
Van de Ven, Andrew H. and Marshall
1990 "Methods
for
Scott Poole
Studying Innovation Development
Program." Organization Science,
1:
in
the Minnesota Innovations Research
313-335.
Ventola, Eija
1987 The Structure
of Social Interaction:
A
Systematic Approach to the Semiotics of Service
Encounters, London: Frances Pinter.
Walsh, James P. and Gerardo R. Ungson
1991 "Organizational Memory."
Academy
of
Management Review,16:
57-91.
Weick, Karl E.
1979 The Social Psychology
of Organizing,
Second
Edition,
New
York:
Random House.
Weiss, Howard M. and Daniel R. Ilgen
1985 "Routinized behavior
in
organizations." Journal of Behavioral Economics, 14:
Winter, Sidney G.
41
57-67.
Organizational Routines
1986 "The research program
of the behavioral theory of the firm:
evolutionary perspective."
In
B.
Gilad and S. Kaish, (eds.)
Orthodox
Handbook
critique
and
of Behavioral
Economics, Volume A: Behavioral Microeconomics: 151-188. Greenwich, CT: JAI Press.
Withey, Michael, Richard
1983 "Measures
Scale."
L. Daft
of Perrow's
Academy
of
and William H. Cooper
Work
Unit Technology:
Management
An
Empirical
Assessment and a New
Journal, 26: 45-63.
Zimmerman, Donald H.
1970 "The
Practicalities of
Rule Use."
Toward the Reconstruction
In
Jack Douglas,
of Sociological
42
(ed.)
Understanding Everyday
Life:
Knowledge: 221-238. Chicago: Aldine.
Organizational Routines
Figure
1:
Analogy
Behavioral
between
behaviorlal
Routine: the set
of
possible performances for a
particular task, described
by a grammar.
Example: customer service
grammatical
Grammatical
Entity
Organizational
and
in
part
entitles
Counterpart
Organizational Routines
Figure
2:
Coding scheme
Code
for
software
support
process
Organizational Routines
Figure
3:
1
Sequence
of
moves
in
subset of the
call
sample
Organizational Routines
Figure
A grammar
4:
Grammatical
CALL
for
DBI's
software
Rule
A whole
closing.
search
call,
consisting of a opening, search
closing.
and
for solution
-->
opening
process
Explanation
opening, search for solution,
-->
support
Opening move
O.
for solution-->
any order( work,
transfer, user, developer
Search
of
a
call.
for solution, consisting of
any sequence
and
of working, transferring, as well as user-
).
developer- episodes.
closing --> tentative
done, comment,
fix,
Final stage of a call,
closing --> done,
fix
--> (F or
done
(C
or
-->
comment
work
comment.
tentative
-->
(W,
-->
I
or
(E or
W,
)•
which
may
include a
one of several symbols
indicating that work is done, and a trailing
comment that explains what was done.
tentative
R
or
D).
blank).
T
),
fix,
work.
Mote that there are two kinds of closings (with
"tentative fix", and that each
constituent in the closing (tentative fix, done,
and comment) has its own rule.
and without a
An episode
of
work on the call, consisting of a
any length.
string of "w"s of
user
user
-->
-->
US, PS.
Change
US, work, PS.
and back to product support, with an optional
work episode intervening.
responsibility for a call to the user
developer ->
DV, PS.
Escalate a call to a development and then back
to product support, with an optional work
developer
DV, work, PS
episode intervening.
-->
46
MIT LIBRARIES
Organizational Routines
3
Figure
Rewrite
5:
Rewriting
Rule
a
Applied
sample
call
TOAD ODflEMQfiM
S
?!76
Date Due
fi£C
S
Lib-26-67