�WPC�K �����dCxc�8e�1���Ne�Z홅�ITp�de��^nn����Q�}~�7 �=QN�b$�c���V�t^,�����H�v��n����[B�a$�sH{U}���D=#���W��k� ����}�����u��(�Id�I���_����uQ�h�ov�P �ƪ�kj%�'9�d8�� ��:Q/R�`S��u���AJ���::)�����ӪH5Ѧ�E�y��B�QsV}/���T�����8w�dµ�.v�$Xa���6��N.S�)����T&����s�b���� �%�v��68��O�b�ry76��'���Ju�xg��cP�&����N��[q�jnR� ��y��b�0Oj{!��&~���5/NV4>�_��y�� -"��m��F�#�4 ��6��&��ٴ�+/�{�L�@�&�o�AN�B�шFF��ۖ5��?�������=�,�,K��4���-`�Dz���:i������5�>�rz�|AU:� %�U@�4 0�"8�UN�5 nE 0rH U >�  � �W�, 3-J6�M7M7 -8�6AB4H4H4H4H4H4H4H4H4H4H4H4H4H4H4H4H4H B5I5I5I5I5I5I5I 0DRIRIRI AO�I��I 0��J 0D9K D/}K}K}K}K}K}K}K}K}K D3�K B�K<�6X9`("Courier 10cpi X*��� `(CG TimesScalable]������/����4(e2�_$��� � �!��� X� XX X���X|XX X� ��  ��  �ZUMA Talk (Revised Version)� ' '  (@�� Z �6Times New Roman Regular 3|X%�<�6X9`("Courier 10cpiX�x�6X@��8�;X@*��� `(CG TimesScalableXXw��� P���7�XP<�6X9`("Courier 12cpile��d�6X@��8�;�@(J�$����X�"XXX���X)�XXX�"��  �3ZW� Z &Tms Rmn Regular  � �40�"��� XɱXX X���XɱXX Xɱ��  ��  ��@77*<�TEXT��  ��0 X �� �MINISTER:��"The�church�is�God's�house�where�the�children�of�God�come�together�to�praise�and�worship�Him."� XxXx ��0 X �� �ARCHITECT:��"We�would�like�to�build�that�church.��Its�maintenance�costs�and�energy�loss�is�low.��Spaces�have�multiple�uses.��It's�cost�effective�and�impressive."� XxXx ��0 X �� �PARISHIONER:��"My�church�is�a�friendly�place��but�it's�sometimes�hard�to�hear�the�sermons."� XxXx ��0 X �� �RESEARCHER:��"The�church�is�a�good�example�of�a�social�institution�being�a�source�of�cultural�mores�and�normative�behavior."� XxXx ���@K K (<�EMPHASIS��� � ����X�Idea�Category��9 9 � �Minister�����*��Architect�����5%�Parishioner���::C��Researcher��    ��Have��9 9 � ��5.37�����.��17.00���ss;%��5.37���H���5.37�� � �  �Being��| | !� �3.18���/��0.28�����<%�5.49�����H��13.55�� � �  �Object��� � � ��13.84�����.��41.54���RR;%�15.70�����G���13.84�� � �  �Sense��9 9 � ��2.65�����.���2.65���RR;%�32.91���H���2.65�� � �  �Expression�Arena��  � �70.78�����.���4.21���ss;%��4.21���H���4.21�� �� �Activity��9 9 � ��4.01�����.��32.14���ss;%��4.01�����H��41.65�� yy �Science��9 9 � ��1.77�����.���1.77���ss;%��1.77���SSI��7.85�� bb �Structure��| | !� �7.88���/��5.39�����<%�9.87�����H��16.79�� KK �Processing�Things��9 9 � ��6.05���/��9.22���RR;%�18.39���H���6.05�� 44 ���@a a )<�CONTEXT��Role��e e � �Traditional�����*��Practical�����7%�Emotional���E��Analytic�� �� ��Minister��  � �25.00���rr-���13.17���ss;%��2.49���H���9.33�� �� �Architect��9 9 � ��8.77�����.��25.00���ss;%��7.25���H���8.98�� �� �Parishioner��| | !� �4.65���rr-���20.72���RR;%�20.35���H���4.28�� �� �Researcher��| | !� �3.84�����.���1.45���:%��23.55�����H��21.16�� || � �� �40�"��� XɱXX X���XɱXX Xɱ��  ��  �� X �� � ��  �� ` �� � ��  �� h �Manager�  �Manager� � �Staff�  �� X �� � ��  �� ` �� � �Mean�  �� h �to�Staff�  �to�Resident� � �to�Resident� �� �Idea�Category�  �� ` �� � �E�Score� h �Distance�  �Distance� � �Distance�� �� �� X �� � ��  �� ` �� � ��  �� h �� � �C�o�r�r�e�l�a�t�i�o�n�s��� X �Good�� X �� � �Manager� ` �� � ��9.5�  �� h �0.13� � ��  �0.36� p �� � ��0.22� �� �� X �� � �Staff�  �� ` �� � �14.4�  �� h �0.64� � ��  �0.63� p �� � ��0.29� vv �� X �� � �Resident� ` �� � �16.6�  �� h �0.08� � ��  ��0.13� p �� � �0.09� __ �� X �Happy�� X �� � �Manager� ` �� � ��0.1�  �� h �0.12� � ��  �0.26� p �� � ��0.55� 11  �� X �� � �Staff�  �� ` �� � �0.8�  �� h �0.26� � ��  �0.19� p �� � ��0.27�    �� X �� � �Resident� ` �� � �3.5�  �� h ��0.50� � ��  ��0.35� p �� � �0.10�    �� X �Depressed��������� X �� � �Manager� ` �� � ��1.1�  �� h �0.01� � ��  ��0.20� p �� � �0.16� � �  �� X �� � �Staff�  �� ` �� � ��0.6�  �� h ��0.24� � ��  ��0.28� p �� � �0.02� � �  �� X �� � �Resident� ` �� � �0.2�  �� h �0.30� � ��  �0.16� p �� � �0.29� � �  �� X �Community���������� X �� � �Manager� ` �� � �3.8�  �� h �0.44� � ��  �0.73� p �� � �0.05� yy �� X �� � �Staff�  �� ` �� � �3.2�  �� h �0.09� � ��  �0.06� p �� � ��0.22� bb �� X �� � �Resident� ` �� � �1.6�  �� h ��0.17� � ��  �0.33� p �� � �0.08� KK �� X �We������������ X �� � �Manager� ` �� � �1.8�  �� h �0.07� � ��  �0.15� p �� � ��0.35�  �� X �� � �Staff�  �� ` �� � �3.0�  �� h �0.24� � ��  ��0.04� p �� � �0.25�  �� X �� � �Resident� ` �� � �1.3�  �� h ��0.33� � ��  �0.28� p �� � �0.28� �� �� X �They����������������� X �� � �Manager� ` �� � �4.6�  �� h �0.20� � ��  �0.25� p �� � �0.00� �� �� X �� � �Staff�  �� ` �� � �6.8�  �� h ��0.40� � ��  ��0.38� p �� � ��0.14� �� �� X �� � �Resident� ` �� � �7.3�  �� h ��0.09� � ��  ��0.33� p �� � �0.08� �� �� X �Fellow�Feeling�������� X �� � �Manager� ` �� � �3.6�  �� h ��0.04� � ��  ��0.28� p �� � �0.30� ee �� X �� � �Staff�  �� ` �� � �3.0�  �� h ��0.03� � ��  ��0.04� p �� � �0.38� NN �� X �� � �Resident� ` �� � �1.6�  �� h ��0.30� � ��  ��0.27� p �� � ��0.08� 77 �� X �Implication��������� X �� � �Manager� ` �� � �4.0�  �� h �0.37� � ��  �0.47� p �� � �0.14�   ! �� X �� � �Staff�  �� ` �� � �6.2�  �� h ��0.43� � ��  ��0.25� p �� � �0.42� ��" �� X �� � �Resident� ` �� � �6.0�  �� h �0.67� � ��  �0.74� p �� � ��0.13� ��# �� X �Cognition������ X �� � �Manager� ` �� � �10.1�  �� h ��0.62� � ��  ��0.36� p �� � �0.63� �!�!% �� X �� � �Staff�  �� ` �� � �14.0�  �� h �0.02� � ��  �0.00� p �� � ��0.32� �"�"& �� X �� � �Resident� ` �� � �10.4�  �� h ��0.13� � ��  ��0.23� p �� � �0.34� ##' ���򀀀��������������������������������� Q%Q%) ���*��Source:�Metropolitan�Minnesota�data�on�11�homes,�1984.�� M� �/�#��  �Table�1.��An�Illustration�of�E�Scores�and�C�Scores � �40�"��� XɱXX X���XɱXX Xɱ��  ��  ��� X �� � ��  �� ` �� � ��  �� h �� � �Urban�  �� p �Rural� �� �� X �� � ��  �� ` �� � ��  �� h �� � ���Sample��� p ���Sample��� �� �� X �Manager�to�Resident�Social�Distance�� X �� � �Overall�Distance� � ��  �mean� h �� � ���30.1�  �� p ���22.7� �� �� X �� � �Between�Homes� � ��  �mean� h �� � ���17.7�  �� p ���13.7� �� �� X �� � ��  �� ` �� � ��  �range� h �� � �(50.2)�  �� p �(31.7)� vv �� X �� � �Within�Homes� � ��  �mean� h �� � �����2.1� p ����1.8� __ �� X �� � ��  �� ` �� � ��  �range� h �� � ��(�6.0)� p �(�5.9)� HH �� X �Staff�to�Resident�Social�Distance�� X �� � �Overall�Distance� � ��  �mean� h �� � ���15.9�  �� p ����7.4�    �� X �� � �Between�Homes� � ��  �mean� h �� � ���10.2�  �� p ����4.2�    �� X �� � ��  �� ` �� � ��  �range� h �� � �(29.3)�  �� p �(14.6)� � �  �� X �� � �Within�Homes� � ��  �mean� h �� � ����3.6�  �� p ����2.4� � �  �� X �� � ��  �� ` �� � ��  �range� h �� � �(�7.2)�  �� p �(�8.0)� � �  ��򀀀���������������������������������� � �  ���*��Source:�Metropolitan�data�on�11�nursing�homes�include�11�managers,�11�staff,�and�28�residents�interviewed�in�1984.�� �� �Rural�data�include�4�managers,�22�staff,�and�20�residents�interviewed�in�1990.��Both�studies�were�conducted�in�Minnesota.�� �� �/�#��  �Table�3.����XɱXX X�Average�Social�Distance�of�Residents�From�Each�Other�And�From�Staff�and�  �Mangers�Within�And�Between�Homes*�#� XɱXXXɱ%#�� �� � �� �/�#��  �Table�5.����XɱXX X�Correlates�of�Contextual�Distance�Between�Roles�And�Selected�Idea�Category�  �Emphasis�Scores*�#� XɱXXXɱ%#� �� �40�"��� XɱXX X���XɱXX Xɱ��  ��  �� X �� � ��  �� ` �� � ��  �Manager� � �Manager� p �Staff�  �� X �� � ��  �� ` �� � ��  �to�Staff� � �to�Resident� p �to�Resident� �� �� X �� � ��  �� ` �� � ��  �Distance� � �Distance� p �Distance�� �� �� X �Overall�Social�Distance� � ��  �25.1� h �� � �30.1�  �� p �15.9� �� �� X �Ownership�� X ����Proprietary�(N=2)� � ��  �14.2� h �� � �19.9�  �� p ��9.9� �� �� X ����Church�Sponsored�(N=4)�  �21.6� h �� � �29.7�  �� p �10.9� vv �� X ����Foundation�(N=3)� � ��  �24.6� h �� � �42.0�  �� p �20.9� __ �� X ����Public�(N=4)� ` �� � ��  �31.4� h �� � �23.0�  �� p �24.3� HH �� X �Decision�making�Autonomy�� X ����Local�Decisions�(N=2)� � ��  ��4.7� h �� � �11.0�  �� p �10.0�    �� X ����Centralized�Decisions�(N=8)�  �32.5� h �� � �35.2�  �� p �15.6�    ��� X �Correlations�With�Social�Distance��� X �Number�of�Beds� ` �� � ��  �0.28� h �� � ��0.02�  �� p ��0.33� � �  �� X �Nursing�Home�Reputation�(rank�r)�  �0.32� h �� � �0.41�  �� p �0.17� �� �� X �Manager�to�Staff�Distance� � ��  ������ h �� � �0.86�  �� p ��0.28� yy �� X �Manager�to�Resident�Distance�  ������ h �� � ������  �� p ��0.20� bb ��򀀀������������������������������������� KK ���*��Source:�Metropolitan�Minnesota�data�on�11�homes,�1984.��� 44 � �� �/�#��  �Table�4.����XɱXX X�Relationship�Between�Social�Distance�and�Selected�Characteristics,�Urban�  �Minnesota�Sample�*�#� XɱXXXɱ%#� � �40�"��� XɱXX X���XɱXX Xɱ��  ��  ��@" " $<�Context�Dimension��  ��$�� �����X$�� � ���e e � ���Traditional������*����Practical������7%���Emotional����$$C����Analytic��� �� �Managers�� � �Urban��| | !� �2.76���//,����11.31�����<%�5.13�����E���19.20�� �� �� � �Rural��| | !� �3.80�����-����3.10�����9%���19.58�����E���20.28�� �� �Staff�� � �Urban��| | !� �1.79�����-�����.23�����9%���18.64�����E���20.20�� vv �� � �Rural��| | !� �3.52�����-����7.75�����9%���21.48�����E���17.24�� __ �Residents� � �� HH �� � �Urban��� � !� ��.03���rr-���10.63�����9%���24.31�����E���13.64�� 11  �� � �Rural��� � "� �.98�����-����9.93�����9%���23.97�����E���15.01��    ��򀀀������������������������������������������    ���*��Source:�Metropolitan�data�include�11�managers,�11�staff,�and�28�residents�interviewed�in�1984.��Rural�data�include� � �  �4�managers,�22�staff,�and�20�residents�interviewed�in�1990.��Both�studies�were�conducted�in�Minnesota.��� �� �/�#��  �Table�2.����XɱXX X�Social�Context�Score�Profiles�for�Administrators,�Staff�and�Residents�of�Urban�and�  �Rural�Nursing�Homes*�#� XɱXXXɱ%#�� �� ���P Pd(3�$��� � !��  �3/56C��<< C���Level 1Level 2Level 3Level 4Level 5(f30�$��� � �!��� XoxXX X���X��XX Xox��  ��  �($$��� ��  �1�  �� �(('�� dxd@@@@'��dxd�� d� � �!��� XɱXX X���XɱXX Xɱ��  ��  ��0���z�z�&�z�<C>��X0��@d d &��A�COMPUTER�CONTENT�ANALYSIS�APPROACH�TO�� � ��@� � %��MEASURING�SOCIAL�DISTANCE�IN�RESIDENTIAL���@� � *��ORGANIZATIONS�FOR�OLDER�PEOPLE����  �Donald�G.�McTavish�Department�of�Sociology�University�of�Minnesota�Minneapolis,�MN�55455��  ���@88/���Kenneth�C.�Litkowski���@��4��CL�Research���@KK/��20239�Lea�Pond�Place���@��.��Gaithersburg,�MD�20879����@uu2��Susan�Schrader���@��.��Department�of�Sociology���@��1��Augustana�College���@.��Sioux�Falls,�SD��57105����8 * �Computer�content�analysis�provides�another�approach�to�measuring�aspects�of�social�structure.��Different�social�positions�imply�different�social�perspectives�that�are�evident�in�language.��A�language-based�measurement�of�distance�between�positions�in�an�organization�is�described,�using�as�data�verbatim�transcripts�of�interviews�with�occupants�of�positions�in�nursing�homes�talking�about�their�organizational�situation.��Minnesota�Contextual�Content�Analysis�(MCCA),�a�computer�content�analysis�approach,�scores�social�perspectives�in�these�texts�and�computes�social�distance�as�a�function�of�differences�between�perspectives,�facilitating�an�examination�of�social�distance�with�other�organizational�and�personal�outcomes.��Correlates�of�distance�between�roles�across�nursing�homes�suggest�consequences�for�organizational�structure�and�the�meaning�residents�express�about�their�experience.��This�content�analysis�permitted�a�relatively�accurate�identification�of�each�respondent�with�a�particular�nursing�home,�a�measurable�aspect�of�organizational�culture.��The�structure�of�these�differences�reveals�important�facets�of�organizational�structure.��The�methodology�used�here�is�compared�to�techniques�in�information�retrieval�for�characterizing�documents�by�semantic�vectors.��This�comparison�suggests�that�MCCA�captures�finer�grained�concepts.��Semantic�analysis�of�the�MCCA�categories�using�WordNet�shows�that�they�constitute�semantic�domains,�whose�further�refinement�may�lead�to�better�characterization�of�the�identified�differences.� *�!*�! ��1� * �Introduction� /(#' �� * �This�paper�describes�a�language�based�approach�to�the�measurement�of�the�social�distance�between�positions�in�an�organization,�as�an�example�of�using�a�content�analysis�technique�that�may�usefully�be�extended�to�numerous�other�applications.��In�everyday�interaction,�differences�in�stylistic�register�and�pragmatics�are�evident�between�managers�and�employees,�doctors�and�patients,�accountants�and�sales�staff�in���what��they�commonly�discuss�as�well�as���how��these�topics� �, (, �are�presented.��The�approach�discussed�in�this�paper�uses�conversational�interviews�with� �-�(- �occupants�of�positions�in�organizations�in�which�they�talk�about�their�organizational�situation.��Computer�content�analysis�provides�a�practical�tool�for�reliably�coding�social�perspectives�using�these�verbatim�transcripts.��Differences�between�perspectives�is�a�function�of�social�distance�between�statuses.��This�approach�to�the�measurement�of�social�distance�in�organizations�facilitates�an�examination�of�the�impact�this�feature�of�social�structure�upon�other�organizational�and�personal�outcomes.�� * �In�this�paper�we�describe�a�contextual�content�analysis�approach�to�the�measurement�of��cultural�aspects�of�social�distance�between�statuses�in�organizations.��These�distances�are�shown�to�be�related�to�selected�other�features�of�organizations�and�individuals�in�the�organizations.��Conversational�interview�data�from�a�study�of�administrators,�staff�and�residents�of�nursing�homes�in�Minnesota,�USA�are�used�to�illustrate�the�approach.��2� * �Differences�in�Perspective�Between�Statuses� ��  �� * �Several�measures�of�social�distance�have�been�developed�and�various�approaches�have�been�used�in�analyzing�them.��(See,�for�example,�Kadushin�(1962),�Kidwell�&�Booth�(1977),�McPhersion�et�al.�(1987),�Miller�(1983),�and�Reiss�(1961)�for�further�details,�as�well�as�McTavish�&�Felt�(1987)�and�McTavish�&�Schrader�(1992)�for�consideration�of�these�measures�and�approaches�in�the�genesis�of�this�study.)��Here�we�focus�on�relevant�concepts�of�social�perspectives�necessary�to�understand�our�results�and�to�explicate�our�content�analysis�approach.�� * �Qualified�incumbents�of�different�social�positions�learn�expected�role�behavior�and�interact�in�somewhat�different�social�circles�by�virtue�of�their�positions;�understand�differing�priorities,�objectives,�and�obligations;�and�because�of�these�differences,�typically�see�their�organizational�situation�in�somewhat�different�ways.��By�virtue�of�their�social�position�in�a�structure,�they�have�somewhat�different�social�perspectives.�� * �The�administrator�of�a�nursing�home,�for�example,�is�involved�with�residents�and�their�kin,�state�regulators,�volunteers,�owners,�staff�and�media,�to�name�a�few.��A�typical�responsibility�is�to�represent�the�organization�to�these�various�stakeholders.��Residents,�on�the�other�hand,�relate�to�other�residents�and�to�their�own�kin,�to�staff�and�administrators,�and�to�various�volunteers�who�may�be�around�from�time�to�time,�primarily�as�the�recipient�of�services.��Different�statuses�promote�different�perspectives�on�the�nursing�home�situation.�� * �Particular�statuses�can�be�located�in�more�general�institutional�structures.��Thus�we�can�identify�a�particular�status�such�as�"father"�with�a�specific�family�organization�as�well�as�with�a�type�of�social�institution�(e.g.�"the�family").��More�broadly,�a�status�can�be�characterized�in�terms�of�the�more�general�themes�of�institutional�sectors.�"Traditional"�institutions�(e.g.�religion,�family,�legal�institutions)�emphasize�normative�standards�for�appropriate�behavior�and�sanctions�for�deviance.��Practical�accomplishment�of�goals�and�success�are�themes�emphasized�in�economic�and�production�institutions.��A�given�social�position�can�be�identified�to�a�greater�or�lesser�degree�with�the�broader�institutional�sectors�whose�themes�it�most�emphasizes.��These�institutional�themes�characterize�a�social�position�and�are�evident�in�the�language�used�by�occupants�of�a�status.��For�example,�while�a�minister�might�be�expected�to�approach�topics�in�a�more�traditional�than�pragmatic�way,�a�church�treasurer�might�typically�emphasize�pragmatic�themes�a�bit�more,�stemming�from�the�somewhat�different�perspectives�involved�in�their�different�statuses�in�a�church�organization.��The�"same"�types�of�status�may�have�a�somewhat�different�perspective�in�different�social�structures�(societies,�times,�places).��A�salesperson�for�IBM�has�a�different�perspective�than�a�salesperson�for�a�car�dealership�and�these�have�changed�through�time.� �-�(- �� * �In�this�paper�we�measure�social�distance�between�statuses�in�terms�of�institutional�perspectives,�using�the�role�behavior�of�participants'�talking�about�their�social�position�as�an�indicator�of�the�institutional�perspective�of�their�social�position.��This�is�scored�in�a�way�that�indicates�the�relative�emphasis�on�different�institutional�themes.��Differences�between�scores�in�terms�of�these�institutional�perspectives�is�proposed�as�a�measure�of�social�distance.��3� * �A�Language�based�Measure�of�Social�Distance.� & v �� * �Occupants�of�a�social�position�are�asked�to�talk�in�an�open�ended�fashion�about�what�their�context�is�like�from�their�positional�point�of�view.��Open�ended�conversations�are�a�more�relevant�and�direct�solution�to�the�measurement�of�differences�in�social�perspective�because�they�readily�capture�meanings�and�emphasis�that�the�respondent�wishes�to�express�than�many�structured�approaches.��Ideas�which�the�respondent�introduces�are�conditioned�by�their�experience�in�a�particular�status.��This�contributes�to�face�validity�by�utilizing�meanings�a�position�holder�chooses�to�use.��A�verbatim�transcript�of�these�conversations�is�scored�using�a�computer�content�analysis�procedure�to�indicate�similarity�to�themes�characteristic�of�broader�institutional�perspectives.��A�social�distance�score�is�then�computed.��Computer�scoring�permits�a�reliable�contrast�of�status�centered�dialogue�by�avoiding�coder�reliability�problems�affected�by�a�coder's�own�position,�experience�and�fatigue.��4� * �The�MCCA�Approach.� �K �� * �The�computer�content�analysis�approach�used�here�is�called�Minnesota�Contextual�Content�Analysis�(MCCA).��MCCA�is�described�in�McTavish�and�Pirro�(1990)�and�early�work�on�a�similar�approach�was�reported�by�Cleveland,�McTavish�and�Pirro�(1974).��Two�kinds�of�normed�scores�are�generated�for�each�analyzed�text.��One�kind�of�score�shows�the�emphasis�(called�E-scores)�placed�on�each�of�many�idea�categories.��An�idea�category�consists�of�a�group�of�words�which�reflect�a�given�idea�or�meaning.��For�example,�the�idea�of�"control"�occurs�when�words�such�as�"allow",�"authorize",�"insist",�and�"command"�are�used.��The�MCCA�dictionary�distinguishes�116�idea�categories;�words�may�be�assigned�to�more�than�one�category�since�they�may�have�more�than�one�sense;�the�assignments�to�categories�were�made�judgmentally.��Scores�are�"normed"�against�expected�usage�of�the�words�in�an�idea�category�so�that�positive�E�scores�indicate�an�over�emphasis�and�negative�E�scores�indicate�a�relative�omission�of�a�given�idea�in�the�text.��Normed�scores�are�computed�in�a�z�score�like�fashion,�contrasting�category�proportions�with�the�expected�probability�of�use�of�a�given�idea�category,�divided�by�a�standard�deviation�of�expected�category�usage�across�various�social�contexts.��Expectations�are�based�on�the�Kucera�and�Francis�(1967)�word�counts�and�percentages.�� * �The�E�scores�are�computed�for�each�of�the�categories�and�are�the�basis�for�the�conceptual�analysis.��The�pattern�of�connectedness�of�various�ideas�in�a�text�is�examined�using�a�clustering�routine.��Similarity�and�distinction�between�texts�in�terms�of�emphasized�patterns�of�ideas�can�be�quantified�as�well.��A�distance�between�texts�can�be�measured�as�a�discrepancy�between�texts�on�their�profile�of�relative�use�of�the�117�categories�(the�117th�category�is�the�"leftover"�list�of�uncategorized�words).��The�structure�of�conceptual�differences�shown�in�this�proximity�matrix�can�also�be�examined�by�clustering�and�other�statistical�techniques.�� * �The�second�kind�of�score�consists�of�a�profile�of�four�"social�context"�scores�(C-scores).��Organizational�settings�can�be�distinguished�by�the�pattern�of�language�used�in�these�contexts,�in�part,�because�of�different�problems�or�tasks�an�organization�typically�encounters.��Contextual�content�analysis�utilizes�these�differential�patterns�of�emphasis.��Four�vectors�or�contextual� �-�(- �themes�have�been�identified�by�factoring�and�experience�with�texts�from�different�institutional�sectors.��Each�context�dimension�is�a�function�of�the�emphasis�in�the�text�across�a�large�number�of�idea�categories,�and�is�represented�in�MCCA�by�a�vector�of�weights;�these�marker�contexts�are�experimental,�empirically�derived�profiles�of�relative�emphasis�on�each�idea�category.��For�example,�although�the�idea�of�"deviance"�can�be�used�in�any�organization,�it�receives�relatively�greater�emphasis�in�more�traditional�institutions.��Together�the�profile�of�four�social�C�scores�for�a�text�measures�how�closely�the�text's�pattern�of�language�matches�each�of�these�four�broad�institutional�themes.��The�reference�themes�are�labeled�as�follows:��0 * �a)�Traditional�-�a�focus�on�norms�and�expectations�for�appropriate�behavior.��This�is�especially�emphasized�in�text�from�judicial�and�religious�organizations.� *(#*(# ��0 * �b)�Practical�-�a�focus�on�successful�(efficient)�goal�accomplishment.��This�theme�is�most�evident�in�business�and�work�organizations.� *(#*(# ��0 * �c)�Emotional�-�a�focus�on�personal�involvement,�comfort,�enjoyment,�or�leisure.��This�theme�is�typical�of�leisure�or�recreational�organizations.� *(#*(# ��0 * �d)�Analytic�-�a�focus�on�objectivity,�curiosity�or�interest.��This�emphasis�is�more�pronounced�in�research�and�educational�settings.� *(#*(# �Usually�there�are�no�"pure�types"�and�people�describe�their�orientation�to�their�setting�in�a�"blend"�or�"profile"�of�emphases�across�the�four�contextual�dimensions�which,�thus,�locate�the�social�context�of�the�text.��The�social�context�of�the�text�is�used�to�disambiguate�ambiguous�words�by�accumulating�contextual�scores�using�weights�reflecting�the�relative�use�of�each�conceptual�category�in�the�four�social�contexts.��Table�1�illustrates�E�scores�and�C�scores�assigned�to�different�texts,�showing�how�different�vectors�are�generated.�� * �Where�texts�are�generated�by�people�occupying�different�status�locations�in�an�organization,�the�vector�of�four�C�scores�identifies�the�social�location�of�their�statuses.��Distance�between�two�C�score�profiles�can�be�represented�by�a�standard�euclidean�distance�computation.��If�the�distance�measure�is�zero,�the�two�positions�have�the�same�profile�of�C�scores.��The�larger�the�distance�measure,�the�greater�the�social�context�distance�between�the�two�positions.��It�becomes�larger�as�they�take�on�quite�different�contextual�perspectives.��If�the�measure�between�two�organizational�positions�is�very�large,�it�is�hypothesized�that�communication�difficulties�are�likely�to�be�encountered�because�there�is�little�shared�perspective.��These�contextual�distance�measures�have�been�used�in�a�wide�variety�of�research.�� * �It�is�this�euclidean�distance�measure�between�pairs�of�C�score�profiles�which�we�propose�to�measure�social�distance�between�statuses.��It�is�based�on�perspectives�that�incumbents�of�a�status�express.��This�measure�facilitates�studies�of�the�relative�similarity�of�social�distance�across�similar�statuses,�changes�through�time,�as�well�as�studies�of�the�consequences�of�differing�social�distances�for�the�same�positions�in�different�organizations.��5� * �Social�Distance�In�Nursing�Homes:�An�Illustration.� � �� ]&�!% �� * �Conversational�interviews�were�conducted�with�98�administrators,�staff�and�residents�from�15�selected�residential�settings�for�older�people�in�two�studies:�one�from�a�metropolitan�region�(McTavish�and�Felt,�1987)�and�a�1990�replication�in�a�rural�county�(McTavish�and�Schrader,�1992).��The�two�studies�include�very�large�and�small�nursing�homes�and�retirement�apartments�as�well�as�in�home�care�in�a�rural�county�and�a�variety�of�ownership�arrangements.��Given�the�� �small�sample�size,�the�findings�suggested�below�should�be�viewed�as�suggestive�rather�than� �+#'+ �� �����(#�(#�� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� ���������� �� �(#�(#�����definitive.��However,�systematic�patterns�are�found,�generally�in�hypothesized�directions.�Q�2 >.*x p  `���Q��� �� * �Subjects�were�asked,�"What�is�it�like�for�you�around�here?"�and�their�responses�(some�264,000�words)�were�transcribed�verbatim�onto�a�computer�text�file�so�that�overall�similarities�and�differences�in�perspective�from�each�of�the�three�positions�could�be�examined.��5.1� * �Social�Structure� ]&�!% �� * �Table�2�shows�the�profile�of�context�scores�for�administrator,�staff�and�resident�groups.��Overall,�the�people�we�interviewed�describe�their�nursing�home�in�emotional�terms�(the�highest�C-score�in�most�cases),�that�is,�in�terms�of�personal�involvement,�reactions,�preferences�and�concerns.��This�is�much�more�the�case�for�residents�who�are�involved�continuously,�than�for�either�staff�or�administrators�for�whom�it�is�a�job.��As�expected,�administrators�talk�in�more�pragmatic�terms�about�goal-accomplishment�and�achievement�than�is�true�of�the�other�two�groups.��Administrators�also�have�a�more�traditional�perspective�than�the�other�two�groups.��� * �It�is�interesting�to�note�the�increased�traditional�and�pragmatic�scores,�and�decreased� �-�(- �� �����(#�(#�� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� �(#�(#�����emotional�and�analytic�scores�as�one�moves�from�residents�to�staff�to�administrators.��This�ordering�fits�with�expectations�about�the�intermediate�position�of�staff�in�the�nursing�home�context.��The�difference�in�context�scores�between�rural�and�urban�nursing�homes�indicates�that�rural�managers,�staff�and�residents�are�more�traditional�and�managers�and�staff�are�less�practical�and�more�involved�personally�(emotional�C�score),�for�example.�Q�2>.*x p  `���Q��� * �Table�3�presents�the�social�distances�of�managers�and�staff���from��residents�in�both�urban�and� �� �rural�samples.��While�social�distances�are�somewhat�smaller�for�the�rural�sample,�the�striking�contrast�is�not�the���difference��in�social�distance�between�nursing�homes�but�the���uniformity��of� s� �social�distance�within�a�given�organization.��We�interpret�this�as�a�result�of�the�social�structure�of�a�given�home.��For�example,�in�the�urban�sample,�social�distances�between�residents�and�managers�in�different�nursing�homes�averaged�17.7�(range�50.2�points),�but�within�the�same�home�was�only�2.1�(range�6.0�points).��While�different�homes�are�quite�different�in�the�social�distances,�within�a�given�home�there�is�great�uniformity,�suggesting�an�organizational�constraint:�a�typical�distance�within�any�given�home.��Different�organizations�seem�to�have�different�typical�social�distances,�a�phenomenon�that�is�not�adequately�dealt�with�in�organizational�literature.�� * �Finally,�Figure�1�shows�the�pattern�of�social�distance�between�these�three�statuses.��The�distances�suggest�a�"triangular"�structure:�the�distance�between�managers�and�residents�is�not�simply�the�sum�of�distance�between�manager�and�staff�plus�staff�to�resident�distances.��For�only�one�of�our�homes,�the�most�bureaucratically�structured,�was�the�more�"linear"�pattern�evident.��Interestingly,�proprietary�homes�have�the�smallest�social�distances�between�all�statuses,�perhaps�because�these�structures�underlie�some�of�the�typical�strains�felt�by�staff,�residents�and�managers.��5.2� * �Correlates�of�distance�between�roles� )l$( �� * �Table�4�shows�some�distances�between�a)�managers�and�staff,�b)�managers�and�residents,�and�c)�staff�and�residents,�within�urban�nursing�homes�for�which�we�have�complete�data.��� �Overall,�managers�are�more�distinct�from�residents�(distance�=�30.1)�than�they�are�from�staff� �+''+ �(distance�=�25.1),�and�than�staff�are�from�residents�(distance�=�15.9),�as�expected.�Q�2>.*x p L `���Q��� �� * �Social�distance�measured�in�this�way�is�systematically�related�to�several�organizational�features,�as�shown�in�table�4.��As�to�type�of�ownership,�large�public�institutions�show�the�largest�manager�to�staff�and�staff�to�resident�distances,�while�non�profit�foundation�owned�homes�have�the�largest�manager�to�resident�distance.��To�some�extent,�this�difference�is�reflected�in�the�structure�of�the�ownership.��As�expected,�homes�which�are�a�part�of�a�larger�corporate�structure�with�centralized�decision�making,�have�a�considerably�larger�social�distance�internally�between�the�manager�and�both�staff�and�residents�than�do�homes�which�are�not�part�of�a�corporate�chain.�� * �Correlations�with�the�number�of�beds�in�a�home�show�that�managers�in�larger�homes�are�more�remote�structurally�and�in�the�perspectives�they�share�with�those�in�other�statuses.��This�would�create�a�sense�of�remoteness�and�dissimilarity�among�statuses,�and�impede�open�discussion�(what�is�clear�expression�for�one�becomes�odd�and�encoded�response�from�other�perspectives).��The�opposite�pattern�is�evident�for�staff�resident�distances;�the�larger�the�home�the�smaller�the�distance.��This�is�perhaps�due�to�hiring�practices�if�larger�metropolitan�Minnesota�homes�seek�staff�who�are�more�similar�to�residents.�� * �Nursing�home�reputation,�as�judged�by�outside�professionals,�is�positively�correlated�with�social�distance�and�strongest�for�the�social�distance�of�managers�from�both�residents�and�from�staff.��Perhaps�the�greater�distance�for�higher�ranked�homes�represents�a�distinctive�managerial�perspective�or�special�professional�training�or�the�function�of�the�manager�in�representing�the�� �����(#�(#�� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� �(#�(#�����home�to�the�community.��Finally,�manager's�distance�from�staff�is�positively�related�to�their�� �distance�from�residents,�with�greater�distance�of�managers�from�staff�and�residents�negatively� �*:&* �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� �� ��$�� �-�(-���� ��$�� �����(#�(#�� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� ���������� �(#�(#�����correlated�with�closer�distance�between�staff�and�residents.�Q�2>.*x p  `���Q��� �� * �Social�distance�also�appears�to�be�systematically�related�to�meanings�that�a�nursing�home�has�for�those�who�work�and�live�in�it,�suggesting�structural�effects�that�need�further�investigation.��Table�5�presents�nine�idea�(E�score)�categories�that�illustrate�some�correlates�of�social�distance�in�urban�nursing�homes.��Themes�like�"good",�"happy"�and�"depressed"�receive�more�emphasis�by�residents�than�managers�(E�scores�for�each�of�the�nine�conceptual�areas�are�given�for�each�status�in�column�one�of�Table�5).��Emphasis�on�the�idea�of�"good"�in�staff�interviews�becomes�more�pronounced�where�managers�are�at�a�greater�social�distance�from�staff�and���residents��(correlations�of�+.64�and�+.63),�but�greater�staff�resident�social�distance�is� �" ! �negatively�related�to�emphasis�on�"good"�by�both�staff�and�managers�(but�essentially�uncorrelated�with�this�theme�for�residents).��The�greater�the�manager�staff�distance,�the�less�residents�express�themes�of�"happy",�"we",�and�"fellow�feelings"�and�the�more�they�express�"depressed"�themes.��Also,�larger�distance�between�staff�and�residents�is�related�to�greater�resident�emphasis�on�"depression".��Greater�managerial�distance�seems�to�be�negatively�related�to�emphasis�on�"cognition",�that�is,�talk�of�knowing�or�thinking,�suggesting�an�"out�of�sight,�out�of�mind"�aspect�of�social�distance.��It�is�interesting�to�note�the�positive�relationship�of�staff�distance�from�residents�on�the�cognitive�concerns�of�both�managers�and�residents.�� * �The�greater�managerial�distance�from�staff�or�residents,�the�more�managers�emphasize�"community".��On�the�other�hand,�the�greater�staff�resident�distances�the�less�emphasis�on�� �"community"�by�staff,�suggesting�that�distance�from�residents�has�a�different�effect�on�manager� �,(, Ѐ�Q�2 >.*x p  `���Q�� � �� ��%��  � ��%�and�staff.��Greater�staff�distance�is�associated�with�greater�resident�emphasis�on�"we",�and,�likewise,�the�greater�manager�staff�distance,�the�more�likely�staff�are�to�emphasize�"we"�and�more�likely�they�are�to�refer�to�"they".��These�indicate�trends�toward�separation�and�recognition�of�their�separate�status�in�the�organization.��Managerial�social�distances�(from�staff�and/or�residents)�seems�to�correlate�more�strongly�with�meanings�expressed�by�residents�than�do�staff���%��%�resident�distances,�suggesting�a�greater�impact�of�the�manager�on�meanings�residents�express�about�their�setting.��These�relationships�point�up�the�need�for�further�investigation�of�implications�of�social�distance�for�meanings�others�express�about�their�organization.�� * �In�sum,�the�manager�(or�staff)�to�resident�social�distance�appears�to�be�a�characacteristic�which�distinguishes�these�orgranizations�but�is�relatively�uniform�within�a�given�home.��The�correlations�lead�us�to�expect�consequences�of�the�distances�for�communication�and�differentiation�within�nursing�homes,�for�staff�training�and�effectiveness,�and�for�the�ability�of�staff�and�managers�to�respond�to�changes�in�their�client�population.��There�are�also�implications�for�factors�which�may�affect�organizational�"climate".̀�5.3� * �Evidence�of�Organizational�Climate� W�  �� * �Figure�2�presents�the�result�of�a�discriminant�function�analysis�in�which�selected�E�scores�from�the�conversation�of�managers,�staff�and�residents�are�used�to�predict�which�organization�they�come�from.��The�structure�of�the�discriminant�function�analysis�reflects�differences�between�organizations.��Three�dimensions�were�identified�(the�two�main�ones�are�shown�in�Figure�2).�� * �The�first�dimension�(top�to�bottom)�emphasizes�choices�or�reasoning�about�the�setting.��At�the�top,�reasoning�about�organizational�limitations�is�emphasized�("Everybody�had�a�place�to�sit�and�they�didn't�welcome�anyone�else"),�and�toward�the�bottom�personal�reasoning�related�to�self�and�the�outside�("Because�we�exercised�together").��The�second�dimension�from�left�to�right�suggests�the�basis�of�integration�into�the�setting.��At�the�left�are�themes�emphasizing�a�normative�integration,�with�a�sense�of�guest�like�attachment�("They�do�keep�it�nice�and�clean"�and�"I�like�the�friends�here").��At�the�right�contractual�integration�appears�to�be�emphasized�with�talk�about�changing�arrangements�and�what�"they"�want�("So�I've�been�able�to�act�quickly�and�responsibly"��to�arrange�for�the�home).��Finally,�a�third�dimension�extends�toward�the�reader�from�the�graph,�running�from�an�emphasis�on�supervised�care�and�assistance�toward�relative�freedom�of�choice�at�the�other�end�of�the�continuum.�� * �The�fact�that�meaning�scores�permit�an�accurate�identification�of�the�speaker's�organization�suggests�the�existence�of�an�organizational�"climate"�which�pervades�the�speech�patterns�of�those�in�each�residential�organization.��6� * �Discussion� �$�# �� * �Using�computer�content�analysis�tools,�open�ended�conversations�with�occupants�of�structural�positions�can�be�coded�in�terms�of�their�contextual�perspective.��Distances�between�these�perspectives�are�proposed�as�a�measure�of�social�distance.��Such�measures�build�upon�the�meanings�expressed�about�a�social�context�in�ways�which�permit�comparisons�in�a�general�framework�of�institutional�themes,�across�social�structures�and�through�time.��Thus,�one�could�empirically�raise�questions�about�the�effect�of�structural�versus�personal�characteristics�in�altering�the�social�distance�between�positions�in�an�organization.��Social�distance�appears�to�be�related�to�structural�features�of�nursing�homes�and�to�meanings�those�in�different�positions�in�the�home�have�about�their�organizational�situation.�We�suggest�that�contextual�content�analysis�and�other�language�based�measures�of�social�distance�can�be�explored�for�use�in�organizational�research.� �-�(- Ї7� * �Future�Prospects� � �� * �The�E�scores�used�as�the�basis�for�the�foregoing�analysis�are�similar�to�the�use�of�semantic��XI�XXX6�Ԁ�#�X6�XXXI�z#�� �� �codes�from�a�machine�readable�dictionary�for�filtering�documents�for�their�broad�subject�appropriateness�to�a�topic�of�interest�(Liddy�et�al.,�1993).��In�that�study,�semantic�codes�(such�as�"business,"�"occupations,"�and�"political�science")�categorizing�the�use�of�many�(but�only�specialized)�words�into�124�major�fields�are�used.��After�processing�a�text�to�disambiguate�words�(including�stemming,�part�of�speech�tagging,�using�sentence�level�context�heuristics,�analyzing�co�occurrence�probabilities�with�a�correlation�matrix),�a�vector�of�frequencies�is�generated�and�normalized�(to�control�for�document�length).��A�query�is�coded�using�the�same�technique�and�is�matched�with�semantic�vectors�characterizing�a�document�collection�based�on�a�computed�predicted�similarity�value.��(The�document�collection�has�also�been�clustered�using�agglomerative�clustering�algorithms�so�that�cluster�centroids�can�be�compared�to�the�query�to�provide�a�searcher�with�a�"semantically�cohesive�cluster"�for�browsing.)��This�approach�has�been�found�to�provide�enhanced�information�retrieval�results.�� * �As�noted�above,�E�scores�provide�a�mechanism�for�identifying�a�speaker's�organization.��To�understand�why�this�mechanism�is�similar�to�the�use�of�semantic�codes,�we�have�begun�a�semantic�analysis�of�the�words�in�MCCA's�idea�categories�(as�part�of�a�longer�term�effort�to�extend�the�words�and�categories�that�can�be�used�and�to�port�the�system�to�a�different�computing�environment).��MCCA's�dictionary�consists�of�11,000�distinct�words,�with�an�average�of�about�95�words�per�category.��Using�DIMAP�2�(CL�Research,�1992)�and�DIMAP�3�(CL�Research,�1995),�the�MCCA�dictionary�was�uploaded�into�machine�tractable�form�for�study�with�other�lexical�resources.��In�particular,�sublexicon�dictionaries�were�created�for�individual�idea�categories�by�converting�WordNet�)��1.4�(Miller�et�al.,�1993)�entries�for�the�words�in�the�category�into�DIMAP�format.��The�hypernymic,�hyponymic,�and�other�relations�in�WordNet�could�then�be�examined�to�assess�the�semantic�characteristics�of�these�sublexicons.�� * �WordNet�is�a�semantic�network�of�about�100K�words�grouped�into�"synonym�sets"�(synsets)�of�about�5�words�or�phrases.��These�synsets�are�connected�by�various�semantic�relations�with�one�another,�identifying�more�general�and�more�specific�concepts,�part-of�relations,�synonymic�relations,�antonymic�relations,�etc.��These�are�among�the�more�general�relations�that�exist�among�words�as�identified�in�lexical�semantics�(Cruse,�1986�and�Nida,�1975).��Moreover,�the�hierarchical�structure�of�WordNet�is�such�that�all�nouns�and�verbs�are�grouped�into�about�150�semantically�coherent�categories.��In�examining�the�MCCA�sublexicons,�after�removing�some�senses�from�the�WordNet�entries�that�were�clearly�not�members�of�a�particular�idea�category,�an�(intuitively)�high�degree�of�coherence�was�found.��In�particular,�most�of�the�words�were�connected�to�other�words�in�the�sublexicon�by�morphological�and�derivational�relations,�taxonomic�relations,�and�partitive�relations,�thus�containing�common�semantic�components�and�making�them�semantic�domains.��In�addition,�using�the�WordNet�data,�it�was�possible�to�extend�the�words�that�might�be�associated�with�each�idea�category,�primarily�by�identifying�hyponyms�(narrower�terms).�� * �These�relations�provide�a�stronger�unity�than�do�the�subject�categories�of�a�dictionary,�which�only�indicate�specialized�usages�of�particular�words.��These�subject�categories�do�not�capture�semantic�components�which�reflect�fine�grained�meanings�inherent�in�particular�words.��It�may�be�suggested�that�representing�texts�with�such�meanings�will�provide�better�characterizations�of�texts�and�result�in�even�better�information�retrieval.��Krovetz�and�Croft�(1992)�have�already�shown�that�using�such�derivational�and�other�lexical�relations�results�in�improved�identification�of�word�senses�in�disambiguation�with�an�accompanying�improved�retrieval,�as�measured�by� �-)- �improved�recall�and�precision.�� * �The�use�of�WordNet�and�other�lexical�resources�allows�the�further�refinement�of�the�idea�categories�into�semantic�components.��Considerable�research�is�now�identifying�such�components�(as�well�as�features�characterizing�stylistic�register,�other�pragmatic�information,�and�words�that�are�associated�with�particular�functions��such�as�expressive,�interpersonal,�and�emotive��of�language)�and�incorporating�them�into�machine�readable�dictionaries.��In�fact,�it�is�possible�that�the�functions�of�language�may�correspond�to�the�contexts�set�up�in�MCCA.��The�effect�of�this�work�on�the�content�analysis�of�the�present�study�of�nursing�homes�will�be�examined�next;�it�is�expected�that�this�will�lead�to�sharper�and�better�defined�characterizations�of�the�results�reported�herein.��These�results�are�likely�to�support�other�similar�analyses�using�MCCA�contextual�content�analysis�on�numerous�other�kinds�of�texts.�� * ��8� * �References� ��  Ѐ�Cleveland,�C.E.,�McTavish,�D.�&�Pirro,�E.B.�(1974):�Contextual�Content�Analysis.�In:�Proceedings�of�the�ISSC/CISS�Workshop�on�Content�Analysis�in�the�Social�Sciences,�a�conference�sponsored�by�the�Standing�Committee�on�Social�Science�Data�of�the�International�Social�Science�Council,�UNESCO,�Centro�Nazionale�Universitario�del�Calcolo�Eletronico�(CUNCE),�Pisa,�Italy�September�5�13.��CL�Research�(1992):�DIMAP�2�(Dictionary�Maintenance�Programs)�User's�Manual.��Gaithersburg,�MD:�CL�Research.��CL�Research�(1995):�DIMAP�3�(Dictionary�Maintenance�Programs)�User's�Manual.��Gaithersburg,�MD:�CL�Research.�(In�preparation.)��Cruse,�D.�A.�(1986):�Lexical�Semantics.�Cambridge:�Cambridge�University�Press.��Kadushin,�C.�(1962):�Social�Distance�Between�Client�and�Professional.�American�Journal�of�Sociology�67(5):�517�531.��Kidwell,�I.�J.�&�Booth,�A.�(1977):�Social�Distance�and�Intergenerational�Relations.�Gerontologist�17(5):�412�420.��Krovetz,�R.�&�Croft,�W.�B.�(1992):�Lexical�Ambiguity�and�Information�Retrieval.�ACM�Transactions�on�Information�Systems�10(2):�115�41.��Liddy,�E.�D.,�Paik,�W.,�&�Yu,�E.�S.�(1993):�Document�Filtering�Using�Semantic�Information�from�a�Machine�Readable�Dictionary.��In:�Proceedings�of�the�Workshop�on�Very�Large�Corpora:�Academic�and�Industrial�Perspectives,�sponsored�by�the�Association�of�Computational�Linguistics,�Ohio�State�University,�Columbus,�Ohio,�USA,�June�22,�1993:�20�29.̀�McPherson,�J.�M.�&�Smith�Lovin,�L.�(1987):�Homophily�in�Voluntary�Organizations:�Status�Distance�and�the�Composition�of�Face�to�Face�Groups.�American�Sociological�Review�52:��370���!��!�379.�� �-�(- �McTavish,�D.�G.�&�Felt,�D.�(1987):�Role�Distances�and�Atmosphere�in�Nursing�Homes.�Sociology�of�Rural�Life�9(1):�3�4,8.��McTavish,�D.�G.�&�Pirro,�E.�B.�(1990):�Contextual�Content�Analysis.�Quality�and�Quantity�24:�245�265.��McTavish,�D.�G.�&�Schrader,�S.�L.�(1992):�Residential�Options�for�Rural�Minnesota�Elderly.�Sociology�of�Rural�Life�12:�3�4,7�8.��Miller,�D.�C.�(1983):�Handbook�of�Research�Design�and�Social�Measurement,�4th�edition,�New�York:�Longman�329�333.��Miller,�G.�A.,�Beckwith,�R.,�Fellbaum,�C.,�Gross,�D.,�Miller,�K.�&�Tengi,�R.�(1993):�Five�Papers�on�WordNet�)�.��CSL�Report�43.�Princeton,�NJ:�Princeton�University�Cognitive�Science�Laboratory.��Nida,�E.�A.�(1975):�Componential�Analysis�of�Meaning.�The�Hague:�Mouton.��Reiss,�A.�J.,�Jr.�(1961):�Occupations�and�Social�Status,�New�York:�Free�Press�of�Glencoe.��9� * �Biographical�Sketches� �4 ��Donald�McTavish,�PhD,�is�a�Professor�of�Sociology�at�the�University�of�Minnesota.��Areas�of�interest�include�computer�content�analysis,�research�methods�and�statistics,�and�the�sociology�of�age.��Kenneth�Litkowski�is�the�owner�of�CL�Research,�which�markets�software�for�use�in�developing�and�maintaining�lexicons�for�natural�language�processing.��He�uses�this�software�for�research�in�the�semantic�structure�of�language.��Susan�Schrader,�PhD,�is�Assistant�Professor�of�Sociology�at�Augustana�College.��Areas�of�interest�include�institutional�settings�for�older�people�and�the�sociology�of�age.�� �