Human Resource Management, Fall / Winter 1998, Vol. 37, No. 3 & 4, Pp. 235–248© 1998 John Wiley & Sons, Inc.
CCC 0090-4848/98/030235-14
Jacqueline Rowley Mayfield, Milton Ray Mayfield, and Jerry Kopf
This article bridges theory and practice to show that superiors’ u of Sullivan’s (1988) “motivat-ing language theory” correlates significantly with subordinates’ performance and job satisfaction.In brief, Sullivan hypothesized that superiors’ u of motivating language (including (1)perlocutionary or direction-giving, (2) illocutionary or sharing feelings, and (3) locutionary or explaining culture) would h
ave positive impact on key worker productivity and process outcomes including performance and job satisfaction. This theory was tested by the authors using a LISREL analysis and found to be true.
Introduction
Extensive rearch has shown the importance of a leader’s oral communication skills to suc-cessful outcomes (Graen & Scandura, 1987;Graen, Scandura, & Graen, 1986; Robbins,1993). For decades, implicitly or explicitly , man-agement theorists have identified leader com-munication as a key means for improving worker motivation. For example, the Ohio State stud-ies emphasized two key leader dimensions, con-sideration and initiating structure, which are often expresd in verbal language (Robbins,1993). Furthermore, clear expression of the link between performance and rewards has been the basis for the expectancy and path-goal models of motivation and leadership models (Arnold,1981; Danreau, Cashman, & Graen, 1973;Keller, 1989). Equally important, recent stud-ies have begun to explicitly investigate the mo-tivational impact of a leader’s spoken language on employee performance, affect, and career advancement (Conger, 1991; Fairhurst & Chan-dler, 1989; & Waldron, 1991).The relationship between leader verbal skills and outcomes is clearly embodied within the motivating language model (Sullivan,1988). In brief, this theory predicts that stra-tegic applications of leader oral communica-tion have positive measurable effects on subordinat
e performance and job satisfaction.
The suppositions are credible and could pave the way to important insights about leader behavior. Our test of Sullivan’s predictions also satisfies the dual criteria for relevant rearch in organizational communication—specifically ,relating current theory to pragmatic organiza-tional issues and responding to the needs of business practitioners (Smeltzer, 1993). Unfor-tunately , the model has not yet been satisfacto-rily operationalized. For the reasons, this article prents a study conducted to test the motivating language model. First, we define mo-tivating language theory more explicitly . Then we discuss measures ud to collect relevant data. Next, a model is introduced to test the theory’s core hypothes. Finally , results and future implications are discusd.
In brief, this theory predicts that strategic applications of leader oral
communication have positive measurable effects on subordinate
performance and job satisfaction.
Theory and Hypothes Management and social sciences literatures have often discusd the hypothesized link between a leader’s language and key outcomes. To date, the majority of this work i
s theoreti-cal. Within this stream, Daft and Wiginton (1979) obrved that high variety, verbal lan-guage is a tool for managerial control. In ad-dition, Gronn (1983) prented an instructive ethnographic language analysis of how a school principal maintained dominance through the deliberate u of talk.
While Gronn’s study tested more than communication theory, it also succeeded in operationalizing a number of underlying pre-mis about the language of leaders. Bad on a coding system for recorded talk, Gronn obrved that educational administrators de-liberately cho words to both tighten and loon the grip of control on subordinates—a direct reference to Weick’s looly coupled systems (Weick, 1979a; 1979b). Furthermore, Gronn speculated that leaders could improve their performance by analyzing recorded con-versations with subordinates.
Language has not been described as merely a mechanism for leadership control. Conger (1991) saw language as a means of motivating and conveying strategic vision to subordinates. In addition, oral communication has been modeled as both a form of manage-rial influence and mitigation (Drake & Moberg, 1986).
Management literature does not assume that the linguistic skills are innate. Prior studies suggest t
hat leaders can be trained to improve their language with significantly and positively related changes in such subordinate outcomes as productivity, overall job satisfac-tion, loyalty to one’s superior, and reduction in j ob stress variables (Graen, Novak, & Sommerkamp, 1982; Graen, Scandura, & Graen, 1986; Graen & Scandura, 1987). Spe-cific interventions included conversational training objectives in (1) performance and goal clarification and, (2) empathy, in the forms of active listening and attempts to respond to the subordinate’s experience while sharing one’s own (the leader’s, in this ca). Clearly, the objectives reflected the consideration and ini-tiating structure constructs from the Ohio State studies and related theoretical perspec-tives (Daft, 1988; Yukl, 1989). Moreover, spo-ken language was the critical transmission of behavioral intent despite the important role that relational context (the leader-member exchange model [Graen & Cashman, 1975]) played in the studies.
Building from similar insights, Sullivan (1988) conceptualized motivating language, a model of effective leadership speech. This “strategic talk” has the goal of bridging the distance between leader intent and employee understanding to favorably influence employee outcomes. Motivating language is relatively simple yet firmly bad in widely accepted leadership and communication theories. In brief, motivating language theory (ML) hy-pothesizes that deliberate variance in leader speech can b
e ud as a motivational tool to help employees meet desired organizational and personal objectives. The strategic variance in leader language is rooted with three uni-versal types of speech acts or “the basic or minimal units of linguistic communication . . . where language takes the form of `rules gov-erned, intentional behavior’” (Searle, 1969, p.
16). It is important to note that ML only ex-plains subordinate respons to superior- ini-tiated language and not the counterpart (i.e., comparable superior respons to subordinate-initiated language).
In defining the ML model, the three types of speech acts were conceptualized by Sullivan (1988) as the following:
1.Perlocutionary language is direction-
giving and uncertainty reducing.
Sullivan (1988) hypothesized that
when language minimized worker role
and task ambiguity, performance and
job satisfaction would increa. This
type of speech act is similar to the
structure dimension of the Ohio State
and path-goal theories (Yukl, 1989).
A form of direction-giving speech oc-
curs when a boss clarifies tasks, goals,
and rewards to an employee.
2.Illocutionary language occurs when a
leader is willing to share his or her
affect with a subordinate. Unlike as-
signment clarification, illocutionary
language is an expression of human-
ity. This form of speech act occurs
when a manager compliments a
“Strategic talk”has the goal of bridging the distance between leader intent and employee understanding to favorably influence employee outcomes.
worker for a job well done. In some
respects, this form of speech act par-
allels the consideration dimension of
a number of maj or leadership theo-
ries including the previously cited
path-goal and Ohio State studies
(Daft, 1988; Yukl, 1989).
3.Locutionary or meaning-making lan-
guage happens when a leader explains
the organization’s cultural environ-
ment to a worker, including its struc-
layering
ture, rules, and values. Locutionary
language also alerts worker n-
making to incorporate cultural norms
(Sullivan, 1988). Frequently, mean-
ing-making language is indirectly
transmitted with metaphorical stories
and rumors (Cooke & Rousau,
1988). For instance, the advice that
“Even the President started in the
mailroom” could be interpreted as
“Everyone is expected to pay his or her
dues here.” Management rearchers
have tied the importance of cultural
transmission with key process and per-
formance outcomes (Cooke &
Rousau, 1988; Deal & Kennedy,
1982).
In sum, cultural meaning-making in combination with direction-giving and empathetic speech are the
principal com-ponents of ML. Furthermore, the prence of meaning-making language within this combination distinguishes ML from most leader speech theories.
Motivating language theory draws from a few more primary assumptions. First, the three basic speech acts reprent most verbal ex-pressions that can occur in leader-worker talk. Second, leader behavior strongly influences the effect of motivating language on subordi-nate outcomes (Sullivan, 1988; J. Sullivan, personal communication, March 3, 1992). Subordinates rely more on behavioral mes-sages than speech when the two are incon-gruent (Dulek & Fielden, 1990; Goffman, 1959; Ober, 1992). Talk is viewed as cheap when it conflicts with actions. Subordinates view the leader’s speech as part of a behav-ioral framework, and motivating language is only a part of this framework.
Motivating language’s third basic assump-tion is linked to the expectation that workers
long time no egive actions greater credence than they do ver-
bal communications. Leader communication
is a dyadic process. Consistent with the inter-
pretive perspective (Putnam, 1983), worker
perceptions determine whether the leader’s
language is in fact motivating. A worker must
understand the leader’s intended message to
achieve ML’s inferred goal of mutual n-
making between boss and employee.
The theory’s fourth assumption is that all
three types of speech form an integral whole.
Sullivan makes this aspect of motivating lan-
guage clear when he states that leaders must
u a combination of all three speech acts in
order to gain the full benefit of motivating
language. Under this assumption, motivating
language u cannot be piecemeal. The full
power of motivating language will only be re-
alized by managers adept in all three speech
acts. As such, the three types of speech can
be en as reflecting an underlying construct
of leader motivating language ability (Sullivan,
1988; J. Sullivan, personal communication,
March 3, 1992). Sullivan’s theory is also sup-
ported by such work as Pincus’ (1986) study,
which shows strong empirical relationships
between different aspects of leader commu-
nication.
Hypothes
As stated earlier, ML is important becau it
links strategic leader communication with the
key employee outcomes of performance and
job satisfaction. The predictions lead to the
following hypothes:
Hypothesis 1: There is a significant
and positive relationship between a
leader’s u of motivating language
and a subordinate’s performance.
Hypothesis 2: There is a significant
and positive relationship between a
leader’s u of motivating language
and a subordinate’s job satisfaction.
Hypothesis 3: The latent motivating
language variable is significantly
reflected through the measures of
direction giving, empathetic, and
meaning-making language.
Talk is viewed asproxyfox
qbasiccheap when it
conflicts with
actions.
Methods
Procedures and Sample
The sample ud for this initial test of ML outcomes consisted of the nursing staff in a large government health care facility located in the southeastern United States. This orga-nization has a stated obj ective to improve worker satisfaction and efficiency through an ongoing commitment to continuous quality improvement. The organization also has de-cided that improving communication through-out the facility and especially within the nurs-ing staff is a major means of achieving this goal, which is in line with Barbour’s (1996) work with improving health care facilities.
We surveyed workers through a written questionnaire distributed at the worksite. Each subordinate rated his or her supervisor’s u of motivating language and his or her own level of job satisfaction. In turn, the superiors rated subordinates’ j ob performance. We then matched superior and subordin
ate respons through identifying codes given by the work-ers. We provided all necessary information through both written and oral instructions.
First, we met personally with all supervi-sors and available subordinates to give verbal instructions and answer any subquent ques-tions about the survey. In addition, we provided all respondents with written directions included with their surveys. We took special care to as-sure all subjects of strict respon confidential-ity. Respondents returned their surveys to a designated cure central location.
Our respon rate compared favorably to the norm for lf-administered questionnaires, approximately 44 percent (Cooper & Emory, 1995; Edwards, Thomas, Ronfeld, & Booth-Kewley, 1997). The total sample group con-sisted of approximately 450 nurs. The workers returned 198 surveys. From the surveys, we were forced to discard 34 due to insufficient information, leaving 164 usable surveys. Within this group, 151 are worker surveys and 13 (out of 25) are supervisor sur-veys. Our final respon rate was 34 percent of the superior-subordinate dyads.
We next met with top organizational offi-cials (the head of nurs and the human re-sources vice president) to e if our sample appeared to be reprentative of the institu-tion as a whole. After we
discusd the sample’s demographic and performance characteristics, the officials determined that there were no apparent differences between our sample and the hospital’s population. We were not able to make direct comparisons between motivating language u and job satisfaction since the characteristics are not generally surveyed at the institution.
Our sample’s demographic breakdown ems to be consistent with most nursing or-ganizations. The majority of respondents were women, with 68.9 percent of the subordinates and 84.6 percent of the superiors classifying themlves as female. Most respondents clas-sified themlves as caucasian, with 64.4% and 69.2 percent of the subordinates and superi-ors choosing this category. Also, the majority of both superiors (92.3 percent) and subordi-nates (71.9 percent) had at least a college edu-cation.
Model
We tested our hypothes using a structural equation model analysis (Hair, Anderson, & Tatham, 1987). W e bad model specifications on our preceding hypothes and motivating language theory, as conceptualized by Sullivan (1988), and further developed by Mayfield (1993) and Mayfield, Mayfield, and Kopf (1994; 1995). Figure 1 gives a graphic repre-ntation of this model.
Sullivan hypothesized a single latent fac-tor reprenting a superior’s individualized u of motivating language. Sullivan also theorized that the latent motivating language factor could be wholly captured through the mea-surement of three obrvable factors; namely, the indicants of direction-giving, empathetic, and meaning-making language. The body of literature associated with motivating language clearly states that a leader’s u of motivating language should affect worker performance and satisfaction outcomes. If the supposi-tions are true, the latent motivating language factor should be significantly and positively linked with measures of a worker’s perfor-mance and job satisfaction.
withyouallthetimeMeasures
All measures of motivating language showed
The organization also has decided that improving communication throughout the facility and especially within the nursing staff is a major means of achieving this goal.
high levels of reliability (Churchill, 1979). Di-rection-giving language had a reliability of .95; empathetic language had a reliability of .97; and meaning-making language had a reliabil-ity of .93. We ud the Employee Rating Scale to measure worker performance (Cashman, Danreau, Graen,
& Haga, 1976), and the Hoppock scale (Hoppock, 1935) to measure job satisfaction. Both of the scales are widely adopted to measure worker performance and satisfaction in leader/subordinate rearch (Cashman, et al., 1976; Robbins, 1993). The scales had reliabilities of .96 and .71, respec-tively, in our study. Further information on the study measures is prented in Table I.
Results
四级报名We tested our model using covariance analy-sis techniques, also known as structural equa-tion modeling. This strategy asss both how well our model fits our data and the strength of the variable’s relations. We ud the statis-tical software package PC LISREL 7.16 to perform the actual analysis. This software package provides three widely accepted tests of a model’s overall fit. The package also cal-culates estimates of the variable relationships.
We tested the overall fit of the model through a general goodness-of-fit index, a chi-square analysis, and the root mean residual analysis. The goodness-of-fit index ranges be-tween a low of 0 and a high of 1. Joreskog and Sorbom (1989) suggest that a good fit is indi-cated by an index above .90. In LISREL analy-sis, there are two ways of interpreting the chi-square analysis. The traditional way is that a significant result indicates a model that dif-fers from the obrved data (Joreskog & Sorbom, 1
989). Alternately, Wheaton and his colleagues suggest that the chi-square test indicates a good model-to-data fit when the ratio of the chi-square statistic to its degrees of freedom is five or less (Wheaton, Muthen, Alwin, & Summers, 1977). Finally, the root mean residual should be less than .05 to indi-cate a good fit between model and data (Fulk, 1993; Hughes, Price, & Marrs, 1986; Joreskog & Sorbom, 1989).
Results show a good fit between the hy-pothesized model and the data. There is an
有理数的混合运算练习题overall adjusted goodness-of-fit index of .975 FIGURE 1. Hypothesized motivating language model with outcomes.
and an unadj usted goodness-of-fit index of .993. The chi-square test shows no significant difference in the predicted model and the model derived from the data. The chi-square result is 2.55 with 4 degrees of freedom for a p-value of .636. The ratio of chi-square value to degrees of freedom also falls within the ac-ceptable range for a good model of the data. Finally, the root mean square residual of .013 indicates a good fit of data to model.
W e also tested hypothes about individual model parameters using LISREL. Specifically, we ud t-tests of the links between the model’s parameters and examined the standardized path coefficients of the links. We ud the t-tests to test for significant relationships be-tween the latent and obrvable measures. Examining path coefficients helps us deter-mine the relative effect of each variable.
Our analysis supports the hypothes about the relationships between the individual variables and the predictive power of motivat-ing language theory. T-tests indicate significant relationships between the latent ML factor and subordinates’ j ob performance and satisfac-tion. The results prented in T
able II sup-port predictions that a leader’s motivating language u significantly improves a worker’s performance and job satisfaction.
Further insights ari from examination of the standardized path coefficients (pre-nted in Figure 2). A leader’s u of motivat-ing language appears to very strongly influence a worker’s job satisfaction (with a path coeffi-cient of .67) and less strongly, though signifi-cantly, a worker’s performance (with a path coefficient of .22). In practical terms, for ev-ery 10% increa in a leader’s motivating lan-guage u we can expect an approximate 7% increa in worker job satisfaction and a 2% increa in worker performance.
W e should, however, view the relationship between ML and j ob satisfaction with some caution. This relationship may be somewhat inflated due to common methods variance. Al-though we do not expect such inflation to be great, we must wait until further studies are per-formed before fixing the exact relationship.
Motivating language itlf appears to be most strongly affected by a leader’s direction-giving language (with a path coefficient of .93) and empathetic factors (with a path coefficient of .92), and slightly less strongly reprented
Descriptive Measures and Correlations of Analysis Scales.*
The results prented in Table II support predictions that a leader’s motivating language u significantly improves a worker’s performance and job satisfaction.
Results from Lisrel Analysis Lisrel
Estimates of Path Coefficients and
t-tests.
Direction-Giving.9314.16
Empathetic.9214.05
Meaning-Making.689.25
Performance.22 2.48重庆培训班
Job Satisfaction.67 5.92
wenger
*All t-tests are significant at the .05 level.
Exogenous Path Coefficient to
V ariable Latent Variable t-test*
I
n
d
e
p
e
n
d
e
n
t
D
e
p
e
n
d
e
n
t
Job
Sat Perf Dir Emp Mean Job Satisfaction.71
Worker Performance.13.96高考短文改错
Direction-Giving Language.43.16.95
Empathetic Language.44.22.85.97
Meaning-Making Language.34.13.64.62.93 Scale Mean 4.19 3.81 3.54 3.61 2.19 Standard Deviation.84 1.02 1.00 1.05 1.11 Questions in Scale46554 *Reliabilities on the diagonals.
