348 Thought, Language, andCommunication inSchizophrenia: Diagnosisand Prognosis
by Nancy C. Andreasen andWilliam M. Grove
Abstract
Using the Scale for the Assessment ofThought, Language, and Communi-cation (TLC), we examined thefrequency of "thought disorder" in 94normal volunteers and 100psychiatric patients (25 each sufferingfrom manic disorder, schizoaffectivedisorder, schizophrenic disorder,disorganized type, and schizophrenicdisorder, paranoid type). Weobserved the manics to have asubstantial amount of thoughtdisorder and the normals to have amodest amount, suggesting thatthought disorder is probably notpathognomonic of schizophrenia. Thepatients with affective illness did,however, show a somewhat differentpattern of abnormality. In particular,patients with affective psychosis havemore prominent positive thoughtdisorder, while the schizophrenicpatients tend to have more negativethought disorder. Evaluation of thepatients 6 months later indicated thatmost types of thought disorder remitin the manics, while they persist inthe schizophrenics; patients withschizoaffective disorder also tend toimprove substantially. The strongestpredictor of outcome was thepresence of negative thoughtdisorder.
Ever since the concept of thoughtdisorder was given preeminence inBleuler's (1950) conceptualization ofschizophrenia, the study of thisimportant symptom (or sign) hasbeen plagued by the absence of acommon ground of agreementconcerning its definition or the bestmethods for assessing it. Bleuler'sdescription of associative loosening isrelatively vague, and until the recentneo-Kraepelinian revival, Kraepehns(1919) early descriptions of thoughtand language abnormalities wererarely consulted.
The Scale for the Assessment ofThought, Language, and Communi-cation (TLC) was originallydeveloped in order to remedy thisproblem and to provide a consistentset of definitions that could be usedclinically and that would have highreliability (Andreasen 1978, 1979a,1979b). Because these definitionswere designed for use in a clinicalsetting, they rely heavily on thenaturalistic observation of languagebehavior. While it is also possible toobtain indices of cognitiveperformance or "thought" usingformal tests to elicit disorderedthinking, such as proverb inter-pretation or projective tests (Gorham1956; Shimkunas, Gynther, andSmith 1967; Andreasen, Tsuang, andCanter 1974; Johnston and Holzman1979), in a clinical setting we usuallyinfer a person's thoughts directlyfrom his speech. Consequently, theset of definitions in the TLC empha-sizes the observation of languagebehavior as a way of evaluatingthought disorder.
In our early work with thesedefinitions, we studied a sample of113 patients suffering from mania,depression, and schizophrenia(Andreasen 1979a, 1979b). Weobserved that the traditional conceptof thought disorder, which empha-
1 sizes the importance of associativeloosening, was not useful diagnos-tically, since many patients sufferingfrom mania also exhibited associativeloosening. On the other hand, weobserved some differences betweenmanic and schizophrenic patients intheir overall pattern of subtypes ofthought disorder. In particular, adistinction between positive andnegative formal thought disorderseemed to be useful, since patients
Repnnt requests should be sent toDr N C Andreasen, Dept of Psychiatry,University of Iowa, Iowa City, IA 52242
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO. 3, 1986 349
suffering from schizophrenia tendedto have prominent poverty of speechand poverty of content while manicpatients did not. We concluded that"thought disorder" is probably notpathognomonic of schizophrenia, asBleuler originally suggested, althoughsome patients with schizophreniamay show a type of thought disorderthat is characterized by intellectualimpoverishment and that issomewhat more characteristic ofschizophrenia.
Since our original development ofthe TLC, a number of other investi-gators have explored its usefulness(Harvey 1983; Harvey, Earle-Boyer,and Wielgus 1984; Berenbaum,Oltmanns, and Gottesman 1985;Simpson and Davis 1985; Davis etal., in press). They have found it tobe highly reliable in a variety ofdifferent settings and have supportedthe distinction between positive andnegative formal thought disorder.
In our own work on thought andlanguage in schizophrenia, we felt itwas important to amplify ouroriginal studies of thought, language,and communication in severaldifferent ways. In particular, wewere interested in observing theextent to which normal individualsexhibit the types of TLC abnormalityobserved in schizophrenic patients.Further, we were interested inexamining the frequency of thought,language, and communication abnor-malities in a transitional diagnosticgroup, patients with schizoaffectivedisorder, and in subtypes of schizo-phrenia. Finally, we were interestedin exploring whether either severityor type of thought disorder ispredictive of outcome when patientsare followed longitudinally.
Methods
The patients in this study werestratified into four diagnostic
groups: 25 each suffering from manicdisorder; schizoaffective disorder;schizophrenic disorder, disorganizedor hebephrenic subtype; and schizo-phrenic disorder, paranoid subtype.Patients for each of these groupswere recruited from consecutiveadmissions to the University of IowaPsychiatric Hospital. All diagnoseswere made using the ResearchDiagnostic Criteria (RDC) (Spitzer,Endicott, and Robins 1978). RDCdiagnoses were made based on datacollected using the Schedule forAffective Disorders and Schizo-phrenia (SADS) (Endicott and Spitzer1978).
A comparison or control groupconsisted of 94 normal individualsrecruited from Iowa City and nearbycommunities. Recruits were obtainedthrough advertising in a localnewspaper. In order to match thecontrols to the patients as closely aspossible, normal individuals wereincluded only if they were under age40 and had less than 2 years' collegeeducation. The normals werescreened for past history of majorpsychiatric illness (limited to thosediagnoses defined by the RDC), andvolunteers who had a current or pasthistory of these illnesses wereexcluded.
All subjects in the study wereevaluated using the Scale for theAssessment of Thought, Language,and Communication (TLC)(Andreasen 1978). This instrumentcontains definitions of 18 subtypes of"thought disorder" and has beenfound to have good interraterreliability. The scale contains boththe definitions and instructions forrating severity on a 0-3 or 0-4 scale(depending on the item). To ensuregood reliability, definitions arerelatively specific, and judgment asto severity for each of the individualitems usually depends on thefrequency with which a particular
phenomenon has been observed.Since such frequencies may vary withthe length of the interview, we haveused a standard 45-minute interview.
All subjects were evaluated with astandardized interview designed toavoid a discussion of the patient'spsychopathology (available fromN.C.A. on request). This interviewbegins by having each subject talkabout himself for as long as possible(for at least 5 minutes) without inter-ruption, in response to the prompt,"Could you tell me a little bit aboutwhat you're like, such as whereyou're from, what you're interestedin, and things like that?" If thepatient had difficulty in elaboratinghis response, nonspecific promptssuch as "Could you tell me a littlemore about that7" were used. Afterthe initial monologue, additionalprespecified questions in theinterview covered a variety of topics,such as politics, religious beliefs, andfamily life. Ratings of thought,language, and communication weredone "live," and score sheets werefilled out immediately after theinterview, and after the patient hadleft the room. Because it defines 18types of thought disorder, the TLC isa complicated scale; consequently,raters usually make notes during theinterview about the frequency ofindividual phenomena and completethe score sheet immediately after theinterview. Interviews were conductedeither by one of the authors(W.M.G.) or by a research assistantwith a bachelor's degree inpsychology who had been trained inthe use of the scale by one of theauthors. All interviews were alsotape-recorded and transcribed forlater analysis, but all the data in thisarticle are based on the live inter-views.
The standardized interviewcovering neutral topics was used inorder to keep the interviewer as blind
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
350 SCHIZOPHRENIA BULLETIN
to the patient's diagnosis as possible.Discussion of the patient's psycho-pathology was avoided. Never-theless, in a "live" situation, someclues are inevitably provided. Quitesimply, manics tend to talk more andmore rapidly, and normals tend toappear relatively normal andunmedicated. Thus, the use of liveratings does introduce some potentialbias which could be avoided by usingeither tapes or transcribed samples.For the present investigation,however, our primary goal was toobtain the most accurate ratings ofthought disorder that would bepossible, and we felt this goal wasbest achieved through the use of liveinterviews. We have not as yetcompleted ratings of thought disorderusing live versus taped versustranscribed samples, but weanticipate that estimated severity islikely to increase as one becomesmore removed from the livesituation, since intonations, gestures,and interpersonal interaction tend tomake speech appear more compre-hensible. Thus, in this investigation,we used all possible safeguards tomaximize blindness, but we recognizethat blindness is probably notcomplete. Evaluation of transcriptsmight provide the most rigorous test
of the frequency of abnormalities innormals versus various groups ofpatients, although with the possiblehazard of overestimating the degreeof abnormality in all groups.
To give some indication of generalintelligence, all normals and patientswere also evaluated with the ShipleyInstitute of Living Scale.
Patients were evaluated during thefirst week after admission, usually inthe first 3 days. Nearly all werereceiving medication at the time ofevaluation, since medications areusually prescribed at IowaPsychiatric Hospital within the firstday or two after admission to reducelength of stay. Despite this, all werestill severely symptomatic andmanifested a full and typical manic,schizophrenic, or schizoaffectivesyndrome. In a few instances,patients who were admitted consecu-tively and met inclusion diagnosticcriteria had responded to medicationso quickly that they no longermanifested the full manic syndromeduring the first week; these patientswere therefore not interviewed withthe TLC and not included in thestudy.
We conducted a followupevaluation with the patients 6months after the index evaluation. At
followup, all available patients wereexamined with an abridged versionof the SADS (SADS-C) and with theGlobal Assessment Scale (GAS)(Endicott et al. 1976) and the TLC.We were able to obtain followupevaluations on approximately 70percent of the original sample. Thesewere divided evenly across thevarious diagnostic cells. Nosignificant differences in diagnosis,sex ratio, age, or severity of illnesswere observed between those patientswhom we were successful infollowing up and those who couldnot be interviewed at the followupevaluation.
Results
Demographic and other charac-teristics of the subjects aresummarized in table 1. In general, allsubjects are relatively similar in age,educational status, sex ratio, andintelligence (as assessed by theShipley). As might be expected, thepatients do differ substantially fromone another in duration of thecurrent episode and total amount oftime spent in the hospital duringprevious episodes. In particular, thetwo schizophrenic groups had consid-
Table 1. Demographic and other characteristics of subjects
Agen(%) femaleEducation, yearsShipley-HartfordDuration of episode,
weeksTime in hospital,
total weeks
Normals(n =
Mean
28.925614.10
115.48
: 94)
SD
9.13(59)
3.427.71
Manics(n =
Mean
35.041313.38
102.88
9.88
28.46
25)
SD
14.01(52)
2.2411.98
9.63
30.84
Schizo-affectives(n =
Mean
28.161413.50
108.00
64.48
41.72
25)
SD
9.50(56)
1.8910.09
164.44
76.27
Hebephrenics(n =
Mean
26.681212.5695.26
246.96
66.71
25)
SD
7.41(48)
1.6612.88
239.55
160.85
Paranoids(n =
Mean
27.758
13.74104.00
256.58
79.71
25)
SD
6.21(32)
2.779.80
262.24
139.76
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO 3, 1986 351
erably greater chronicity, while themanics had the least chronicity.
Table 2 indicates the frequencieswith which the various types ofthought, language, and communi-cation abnormalities were observedin the normal individuals and thefour patient groups. It is interestingto note that some TLC "abnor-malities" occur in the speech ofnormal individuals; in particular, thenormals showed a relatively largeamount of derailment and loss ofgoal. The patient groups differsignificantly from one another onnearly all types of language abnor-mality that occur with anyfrequency. Manic speech tends to befluent and disorganized, asmanifested by a high rate of pressure
of speech, derailment, loss of goal,circumstantiality, incoherence, andillogicality. The schizoaffectives showa less severe but similar type ofabnormality, with a somewhatgreater frequency of poverty ofspeech and a lesser frequency ofpressure of speech. Hebephrenicspeech tends to be empty (highratings on poverty of speech andpoverty of content), but also verydisorganized (high frequency ofderailment, incoherence, andillogicality). The paranoid patientsshow a similar but less severepattern, particularly with lesspoverty of content and incoherence.
Following are illustrative examplesof some of the abnormalitiesobserved in these various groups:
Normal
I: Tell me about your work. What doyou like about it the most, and why7
P: Oh, it's easy work for the pay. I guessthat's what I like about it the best. Ithas good benefits. Uh, my hours arepretty good, except when I have towork 8-hour days. Sometimes I getstuck with the 10 to 7 shift. That'skind of a drag. Urn, it's mostly olderpeople that come into the grocerystore, since it's up in the Polish neigh-borhood, I guess, of Iowa City. I thinkthey come there mostly for somebodyto talk to. They don't have anythingelse to do, most of them are retired.Uh, I like to sew. I went and got anew sewing machine this week, and Iplan on making a dress for Noelle forChristmas. 1 like to swim, I haven'thad the chance since I was pregnant
Table 2. Frequencies of thought, language, and communication abnormalities ' among groups
Poverty of speechPoverty of contentPressure of speechDistractible speechTangentialltyDerailmentIncoherenceIllogicalityClangingNeologismsWord approximationsCircumstantialityLoss of goalPerseverationEcholaliaBlockingStilted speechSelf-reference
Global rating
Normals
n
51632
30000026
1770111
6
%
51632
32000026
1880111
6
Manics
n
12
2251
2069104
1113110041
19
%
48
8820
4802436
40
1644524400
164
76
Schizo-affectives
n
63832
14230004550020
15
%
241232128
568
12000
1620200080
60
Hebe-phrenlcs
n
612636
171012
1014880010
20
%
2448241224684048
404
163232
0040
80
Paranoids
n
92404
14531024740000
14
%
368
160
16562012408
1628160000
56
Chi-square*
22.0447.1373.9411.9116.1424.6336.7649.394.80—
10.4221.8512.6821.69
—1.08
13.072.93
81.61
P
.01
.01
.01
.02
.01
.01
.01
.01
.31—.03.01.01.01—.90.01.57
.01
'Criterion: Rating > 1.'Between patient groups, normals excluded.
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
352 SCHIZOPHRENIA BULLETIN
and I haven't gone back yet I likelittle kids. I hope I'll have some more,but I don't know whether I'll be ableto afford it Um, let me think. I votedtoday.
Manic
I: So, to begin with, could you tell me alittle bit about yourself, things likewhere you're from—
P: I'm from Cedar Rapids, and I live withmy parents. My parents are the holyfamily, part of it. My brother lives inOmaha, and he's part of the holyfamily He's Jesus I'm the daughter ofGod, and my sister's pregnant and shehas a part in the holy family and Godknows what I'm gonna do next. Godhas preplanned, has planned my life,and I'm not afraid to speak for the, forthe unfortunate as for the fortunate,and for the stars and for Stella andfor, and for, and for myself. I'm a bigstar I just haven't grown enough, um,God and the Holy Spirit help me. Andwithout, without my friends I wouldbe nothing. Without my music I'm avegetable. Without my, my nurseswhich have, heroes, the women arethe heroes, the men are the heroes, butthere's a champion that, that's beensuffering. That all now?
Schizophrenic
I: Why don't you tell me a little bitabout what you think about currentpolitical issues, like the energy crisis,for example.
P: They're destroying too many cattleand oil ]ust to make soap. If we needsoap when you can |ump into a poolof water and then when you go to buyyour gasoline, m-my folks alwaysthought they should, get pop but thebest thing to get, is motor oil, andmoney. May as well go there andtrade in some, pop caps and, uh, tires,and tractors to grup, car garages, sothey can pull cars away from wrecks,is what I believed in
To obtain a perspective on howpatients' speech differs from normal
speech in its descriptive character-istics, we compared the mean scoresfor each of the patient groups to thatof the normals. Because of the largenumber of comparisons, we used asomewhat conservative .01 signif-icance level, rather than the morecustomary .05 level. At this level, themanics showed significant differenceson pressure of speech, distractibility,derailment, incoherence, illogicality,circumstantiality, loss of goal, andperseveration. The schizoaffectivesshowed fewer differences, havingonly more pressured speech,derailment, and illogicality. Thehebephrenic patients had significantlyhigher ratings on poverty of speech,poverty of content, tangentiality,derailment, incoherence, illogicality,word approximations, and persev-eration. The paranoids had signif-icantly higher ratings on poverty ofspeech, poverty of content,derailment, incoherence, and illogi-cality.
Tables 3-6 explore our ability toreplicate our previous work. Onemajor aspect was the division of TLCabnormalities into positive versusnegative thought disorder. In ourprevious study, this concept wasmore useful in distinguishing manicsfrom schizophrenics than was adescription involving looseassociations. Table 3 shows the meanscale scores of the four diagnosticgroups for positive versus negativeformal thought disorder and looseassociations. The score for positiveversus negative formal thoughtdisorder was developed bysubtracting two indices of negativeformal thought disorder (poverty ofspeech and poverty of content) fromthe sum of pressure of speech,tangentiality, derailment,incoherence, and illogicality. Thescore for loose associations consistsof a sum of five types of thoughtdisorder often considered to be
manifestations of loose associations:tangentiality, derailment,incoherence, illogicality, andclanging.
Table 3 also shows an analysis ofvariance (ANOVA) for positiveversus negative formal thoughtdisorder. The four groups differsignificantly on this variable at the.01 level. Followup tests indicate thatthe significance is due to differencesbetween manics and other patients,manics and schizoaffectives, andmanics and schizophrenics. Thus, theresults of this second study confirmthe usefulness of the positive versusnegative distinction in discriminatingbetween manic formal thoughtdisorder and various types of schizo-phrenic formal thought disorder.
Table 3 also shows the ANOVAfor loose associations. This ANOVAis also significant, although to asomewhat lesser degree. Thefollowup tests indicate that thisconstruct may be more useful indiscriminating between various tradi-tional subtypes of schizophrenia. Thefollowup tests indicate significantcontrasts between the schizoaffectivesand the schizophrenics, and betweenthe hebephrenics and the paranoids.In our 1979 article (Andreasen 1979a,1979b), we did not find significantdifferences between manics andschizophrenics when we used asimilar measure of "loose associ-ations. We did not, however,stratify by schizophrenic subtype orinclude schizoaffectives in thesample; therefore, we could not makesubtype comparisons.
Table 4 compares the discriminantfunction weights generated from ourearlier work and from the presentinvestigation. The weights developedfrom the two samples are somewhatdifferent. In our first sample, povertyof speech and derailment weresomewhat more important asdiscriminators than they are in the
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO. 3, 1986 353
Table 3. Comparison of diagnostic groups on formal thought disorder ratings1
Manics(n = 25)
Schlzoaffectives(n = 25)
Hebephrenlcs(n = 24)
Paranoids(n = 25)
Variable
Positive-negative formal thoughtdisorder'
Loose associations1
Source
Groups
Error
Total
Contrasts
Manic vs. other patientsSchlzoaffective vs.
schizophrenicHebephrenic vs. paranoidAffective symptoms vs.
nonaffectlveManic vs. schizoaffectiveManic vs. schizophrenic
df
3
96
99
df
1
11
111
Mean
169.55252.49
SD
30.3229.85
Mean
141.13236.99
ANOVA for positive-negativeformal
Sums ofsquares
14984.9-7
114945.6-0
129930.5-7
thought disorderMean
square
4994.99
1197.35
F
4.17
t
3.26
- . 4 21.31
1.542.902.94
P
.01
P
.01
.68
.19
.13
.01
.01
SO
28.3618.66
df
3
96
99
df
1
11
111
Mean
151.05264.81
SD Mean SD
40.11 138.27 38.1738.75 245.71 33.32
ANOVA for loose
Sum ofsquares
10329.46
92284.18
102613.64
associationsMean
square F
3443.15 3.58
961.29
t
.46
-2 .412.18
-1 .701.77
.37
P
.02
P
.65
.02
.04
.10
.08
.72
'Individual ratings were transformed to have a mean of 50 and a standard deviation of 10.'Sum of pressure of speech, tangentlality, derailment, Incoherence, and Illogicality, less sum of poverty of speech and poverty of content.'Sum of tangentlality, derailment, incoherence, Illogicality, and clanging.
present sample, while in the currentsample, illogicality has been includedwith a negative weighting. However,the 1979 discriminant functionperformed very successfully in classi-fying the second replication sample.Using the 1979 function on the 1979sample, we correctly classified 84percent of manics and schizo-phrenics. When applied to thepresent sample for class cross-validation, this function correctlyclassifies 80 percent of the cases. Wealso used jackknife procedures inorder to apply the new functionderived from the current sample andfound that it correctly classified 83
percent of that sample. These resultsprovide a relatively strongconfirmation of our earlier work,suggesting that examining varioussubtypes of TLC abnormality may bequite useful in distinguishing manicsfrom schizophrenics on the basis oftheir language behavior.
Tables 5 and 6 compare thefrequency of TLC abnormalities inthe 1979 and the current sample inorder to explore what might be calledreplicability. The ratings on thecurrent sample were completed bydifferent interviewers from those whocollected the 1979 sample. Thepatients were, of course, completely
different, although the size of thediagnostic cells for mania and schizo-phrenia is quite similar. Thefrequency of ratings shows surprisingstability from one study to another.The only significant differencesbetween the two studies are theglobal rating for schizophrenics andthe rating of tangentiality for manics.
Table 7 contains an orthogonalfactor analysis of the TLC variables.The results of this analysis yieldedthree factors from Cattell's scree test.After rotation to simple structure bynormalized Varimax rotation, thefirst factor might be termed a "fluentdisorganization" factor, with high
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
354 SCHIZOPHRENIA BULLETIN
Table 4. Comparison of discriminant functions between manicsand schizophrenics
Standardized discriminant weightFunction from
Function from replication1979 sample sample
Variable (n = 77) (n = 75)
Poverty of speechPoverty of contentPressure of speechDistractible speechDerailmentIllogicalityLoss of goal
% Correctly classified
-.261-.351
.634
.422-.406—.223
80'
—-.457
.898
.685—
-.390—
82.72
'Cross-validation performance•Estimated performance by Jackknife
positive loadings on pressure ofspeech, derailment, incoherence,illogicality, loss of goal, and persev-eration, and a high negative loadingon poverty of speech. Factor 2appears to be an "emptiness" factor,with high positive loadings onpoverty of speech, poverty ofcontent, and tangentiality and anegative loading on pressure ofspeech and circumstantiality. Factor3 might be considered to be a"linguistic control" factor, with highpositive loadings on stilted speechand persistent self-reference andnegative loadings on clanging andneologisms. The components in thisthird factor occur so infrequently,
Table 5. Comparison of thought, language, and communication abnormalities1 forschizophrenics: 1979 vs. replication sample
Variable
Poverty of speechPoverty of contentPressure of speechDistractible speechTangentialityDerailmentIncoherenceIllogicalityClangingNeologismsWord approximationsCircumstantialityLoss of goalPerseverationEcholaliaBlockingStilted speechSelf-reference
Global rating
1979(n
n
1318121
16257
120102
20112216
41
sample= 45)
%
294027
236561627
0204
4424
442
13
91
n
1514103
103115152038
15120010
34
Replicationsample(n = 50)
%
3028206
20623030
406
1630240020
68
Chi-square3
.041.53.59—
2.88.41
2.78.13———
3.362.122.41————
7.61
P'
.84
.22
.56
.35
.09
.52
.10
.72
.27
.47
.14
.07
.15
.12
.22
.22
.77
.01
.01
'Criterion rating > 1'Where no statistic is reported, Fisher's exact test was used
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO. 3, 1986 355
Table 6. Comparison of thought, language, and communication abnormalities1 for manics: 1979 vs.replication sample
Variable
Poverty of speechPoverty of contentPressure of speechDistractible speechTangentialityDerailmentIncoherenceIllogicalityClangingNeologismsWord approximationsCircumstantialityLoss of goalPerseverationEcholaliaBlockingStilted speechSelf-reference
Global rating
1979(n
n
26
23101118583118
1411
1127
28
sample= 32)
%
619723131561625
933
254434
336
22
88
n
12
2251
2069104
1113110041
19
Replicationsample
(n = 25)
%
48
882-
4802436
40
16445244
00
164
76
Chl-square*
——
2.20.92
7.793.56
.63
.81———
2.28.38.55————
1.28
P1
.59
.22
.14
.33
.02
.06
.43
.37
.40
.56
.11
.13
.64
.46
.56
.56
.22
.06
.74
'Criterion: rating > 1.•Where no statistic is reported, Fisher's exact test was used.
however, that it cannot be consideredto be a very important one.
Table 8 shows the frequency ofTLC abnormalities in the variousdiagnostic groups at the time offollowup. We did not follow up thenormal individuals, but their rates atthe index evaluation are included intable 8 for comparison. As table 9indicates, the manic patients haveremitted significantly at followup, ashave the schizoaffectives. The speechof the manics has essentiallynormalized. So, too, has that of theschizoaffectives, although theycontinue to have a somewhat highrate of poverty of speech. On theother hand, the schizophrenicpatients continue to have relatively
persistent disorganization. Whenmean ratings between index evalu-ation and followup evaluation arecompared, the hebephrenics show asignificant change only in pressure ofspeech and incoherence (in thedirection of improvement), while theparanoids show no significant differ-ences. Thus, the language abnormal-ities in the schizophrenics tend topersist at followup.
We were interested in determiningwhether the type and amount ofthought disorder shown by thesepatients was in any sense predictiveof the severity of their syndrome atfollowup. Therefore, we examinedthe correlation between outcome, asmeasured by the Global Assessment
Scale, and positive versus negativethought disorder (as previouslydefined). Table 9 shows that negativeformal thought disorder is moder-ately correlated with outcome, asindicated by absolute GAS raringand by the degree of change in theGAS from index evaluation tofollowup. Thus, the presence ofnegative formal thought at indexevaluation predicts a poor outcomeat followup. On the other hand,positive formal thought disorder haslittle prognostic significance whenused to predict outcome amongpsychotic patients in general. Thelack of correlation appears to occurbecause manic patients, who have arelatively high rate of positive formal
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
356 SCHIZOPHRENIA BULLETIN
Table 7. Factor analysis of
Variable
Poverty of speechPoverty of contentPressure of speechDistractible speechTangentialityDerailmentIncoherenceIllogicalityClangingNeologismsWord approximationsCircumstantialityLoss of goalPerseverationEcholaliaBlockingStilted speechSelf-reference
% of variance
thought, language,
1Fluent
disorganization
- .592.236.648.428.251.775.632.547.443.227.501.501.740.746.099.036.531.269
25.5Cumulative % of variance 25.5
and communication
Loading on factor
2
Emptiness
.416
.609- .462- .158
.464
.021
.365
.128
.323
.238
.256-.493
.019-.052-.277
.102
.087-.149
9.735.2
variables among
3Linguistic
control
.167
.055- .078- .208- .135- .112- .089
.265- .466- .353
.358- .017- .244
.137- .223
.288
.531
.673
8.944.1
patients
Commonality
.551
.430
.639
.251
.297
.614
.541
.386
.518
.233
.445
.494
.607
.578
.136
.095
.571
.547
thought disorder, have a goodoutcome, but its predictive powerwashes out when manic patients arepooled with schizophrenic patients(who may have a high rate ofpositive formal thought disorder anda relatively poor outcome).
DiscussionThis larger and more extensive studyof thought, language, and communi-cation in schizophrenia, schizoaffec-tive disorder, and mania has repli-cated a number of our previousfindings and has elaborated themfurther.
Taken together with our previouswork, the current study suggests thatthe global concept of "thoughtdisorder" should not be considered tobe pathognomonic of schizophrenia
or diagnostic of it. The present studyindicates that mild abnormalities inlanguage behavior even occur innormal individuals. Derailment, theclosest single equivalent to Bleuler's"associative loosening," is the mostcommon type of "abnormality"observed. Further, patients withmania and schizoaffective disorderalso display prominent abnormalitiesin thought, language, and communi-cation. If anything, the abnormalitiesare more pronounced in patients withmania than in those with schizo-phrenia.
The present study providesadditional support for the utility ofsubdividing the global concept ofthought disorder into a variety ofsubtypes. The TLC distinguishesbetween 18 different subtypes. Someof these, however, are relatively
uncommon; some types of thoughtdisorder that have frequently beenwidely described and discussed in thepast are among those that arerelatively uncommon, such asblocking or neologisms. When therate of the various subtypes ofthought disorder is observed acrossfour diagnostic groups, interestingdifferences in quality and severity offormal thought disorder areobserved. The abnormalities seen inmania appear to be qualitativelydifferent from those observed inschizophrenia. In particular, patientswith mania have a more prominentpositive thought disorder, whilepatients with schizophrenia have amore prominent negative formalthought disorder.
The distinction between positiveand negative formal thought disorder
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO. 3, 1986 357
Table 8. Frequencies of thought, language, and communication abnormalities1 among diagnosticgroups at followup
Variable
Poverty of speechPoverty of contentPressure of speechDistractible speechTangentialityDerailmentIncoherenceIllogicalityClangingNeologismsWord approximationsCircumstantialityLoss of goalPerseverationEcholallaBlockingStilted speechSelf-reference
Global rating
Normal(n =
n
51632
30000026
1770111
6
: 94)
%
51632
32000026
1880011
6
Manic(n =
n
214126110012530011
6
: 18)
%
115
226
1133
66006
1128170066
33
Schlzo-affective(n =
n
612034210016130030
5
•• 1 9 )
%
325
110
1621115005
325
1600
160
26
Hebe-phrenlc(n *
n
511209
10270116470022
14
= 17)
%
2955120
53591141
066
35244100
1212
82
n
560005120010010010
5
%
3340000
337
130070070070
33
Paranoid(n = 15)
Chl-square1
18.5366.23
6.482.22
44.086.229.70
39.66—8.581.53
19.897.13
15.31—.75
9.528.42
51.49
P
.01
.01
.17
.69
.01
.18
.05
.01—.07.82.01.13.01—.95.05.08
.01
'Criterion: Rating > 1.•Between patient groups, normals excluded.
may have more clinical utility thanthe traditional concept of associativeloosening. Both approaches toconceptualizing thought disorderwere examined in the four diagnosticgroups. The distinction betweenpositive versus negative thoughtdisorder was useful in separatingmanics from all other patients, from
schizoaffectives, and from schizo-phrenics. On the other hand, associ-ative loosening is significantly moresevere in schizophrenics than inschizoaffectives, and within theschizophrenic group more severe inthe hebephrenics than in theparanoids. These results suggest that,within the schizophrenia spectrum,
Table 9. Correlation between index formal thought disorder andoutcome as measured by Global Assessment Scale
GAS Change in GAS' P
Positive FTDNegative FTD
-.13-.38
.279
.002- . 0 3- . 2 8
.796
.02
Note.—GAS = Global Assessment Scale; FTD = formal thought disorder.
the degree of disorganization (asmanifested by associative loosening)may be useful in identifying moresevere illnesses (i.e., pure schizo-phrenia versus schizoaffectivedisorder or hebephrenic versusparanoid schizophrenia). On theother hand, within the broadspectrum of psychosis, the fluencyand productivity of languagebehavior may be more important fordistinguishing between affectivepsychosis and "core" schizophrenia.
When language behavior isobserved longitudinally in psychoticpatients, the patients appear to differsubstantially in course and outcome.Patients suffering from mania tend toimprove markedly and to have essen-
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
358 SCHIZOPHRENIA BULLETIN
tially reversible abnormalities. Onthe other hand, the schizophrenicpatients tend to have relativelypersistent abnormalities. Patientswith schizoaffective disorderresemble the manics and improve,but to a somewhat lesser degree. Thisfinding is interesting since, at indexevaluation, the schizoaffectives aremore similar to the schizophrenics; atoutcome, however, they more closelyresemble the manics. This finding isprobably indicative of some hetero-geneity within the schizoaffectivegroup, but it may also indicate thatschizoaffective disorder is a transi-tional diagnosis and (as defined bythe RDC) represents an illness thatresembles schizophrenia phenomeno-logically but which has a relativelybetter outcome. The overall goodoutcome in patients with pure manicphenomenology or some tincture ofmanic phenomenology suggests thatthe language disorganization inmania is probably due to a differentmechanism than that occurring inschizophrenia.
The type of thought disorderobserved initially may provide someprognostic indication. Patients whomanifest a prominent negativethought disorder are less likely tohave global improvement orremission, as indicated by the GlobalAssessment Scale. Thus, negativethought disorder, when present, maybe predictive of a poor outcome. Onthe other hand, positive thoughtdisorder is not a useful prognosticindicator. Some patients withpositive thought disorder will remitor improve, while others will not
Based on these observations, theScale for the Assessment of Thought,Language, and Communication(TLC) appears to be a useful tool forthe clinical assessment of languagebehavior in the major psychoses. Theresults show good consistency fromone study to another, with relatively
stable rates of the various TLCabnormalities across two differentsamples and good cross-validation ofa discriminant function. Thus, thedefinitions that it contains appear tohave utility both in the real world ofthe clinic and as a research tool.
References
Andreasen, N.C. The Scale for theAssessment of Thought, Language,and Communication (TLC). IowaCity: The University of Iowa, 1978.Andreasen, N.C. Thought, language,and communication disorders:I. Clinical assessment, definition ofterms, and evaluation of their relia-bility. Archives of GeneralPsychiatry, 36:1315-1321, 1979a.
Andreasen, N.C. Thought, language,and communication disorders:II Diagnostic significance. Archivesof General Psychiatry, 36:1325-1330,1979b.Andreasen, N.C, Tsuang, M.T.; andCanter, A. The significance ofthought disorder in diagnostic evalu-ation. Comprehensive Psychiatry,15:27-34, 1974.
Berenbaum, H.; Oltmanns, T.F., andGottesman, I.I. Formal thoughtdisorder in schizophrenics and theirtwins. Journal of AbnormalPsychology, 94:3-16, 1985.
Bleuler, E. Dementia Praecox, or theGroup of Schizophrenias. (1911)Translated by J. Zinkin. New York:International Universities Press,1950.
Davis, G.C.; Simpson, D.M.;Foster, D.; Arison, Z.; and Post, M.Reliability of Andreasen's Thought,Language, and CommunicationScale. Journal of Clinical Psychiatry,in press.
Endicott, J., and Spitzer, R.L. Adiagnostic interview: The Schedule
for Affective Disorders and Schizo-phrenia (SADS). Archives of GeneralPsychiatry, 35:837-844, 1978.
Endicott, J.; Spitzer, R.L.; Fleiss, J.;and Cohen, J. The GlobalAssessment Scale: A procedure formeasuring overall severity ofpsychiatric illness. Archives ofGeneral Psychiatry, 33.766-771,1976.
Gorham, D.R. Use of the proverbstest for differentiating schizophrenicsfrom normals. Journal of ConsultingPsychology, 20:435-440, 1956.
Johnston, M.H., and Holzman, P.S.Assessing Schizophrenic Thinking.San Francisco: Jossey-Bass,Publishers, 1979.
Kraepelin, E. Dementia Praecox andParaphrenia. Translated by R.M.Barclay. Edinburgh: E. & S. Living-stone, 1919.
Harvey, P.D. Speech competence inmanic and schizophrenic psychoses:The association between clinicallyrated thought disorder and cohesionand reference performance. Journalof Abnormal Psychology,92:368-377, 1983.
Harvey, P.D.; Earle-Boyer, E.A.; andWielgus, M.S. The consistency ofthought disorder in mania andschizophrenia: An assessment ofacute psychotics. Journal of Nervousand Mental Disease, 172:458-463,1984.
Shimkunas, A.M.; Gynther, M.D.;and Smith, K Schizophrenicresponses to the proverbs test:Abstract, concrete, or autistic?Journal of Abnormal Psychology,72:128-133, 1967.
Simpson, D.M., and Davis, G.C.Measuring thought disorder withclinical rating scales in schizophrenicand nonschizophrenic patients.Psychiatry Research, 15:313-318,1985.
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from
VOL. 12, NO. 3, 1986 359
Spitzer, R.L.; Endicott, J.; andRobins, E. Research DiagnosticCriteria: Rationale and reliability.Archives of General Psychiatry,35:773-782, 1978.
Acknowledgment
The research reported was supportedin part by National Institute of
Mental Health, Grant MH-31593; aScottish Rite Schizophrenia ResearchGrant; the Nelle Ball Foundation;and Grant RR59 from the GeneralClinical Research Centers Program,Division of Research Resources,National Institutes of Health.
The Authors
Nancy C. Andreasen, M.D., Ph.D.,is Professor of Psychiatry, andWilliam M. Grove, Ph.D., isResearch Associate, Depart-ment of Psychiatry, University ofIowa, Iowa City, LA.
Audio Tapes onSchizophreniaAvailable
Guilford Publications announcesSchizophrenia: Interview Strategiesfor Detecting CharacteristicSymptoms, a training model byBMA Audio Cassettes. Developed atthe UCLA Neuropsychiatric Instituteby Dr. Ian Falloon and Dr. DavidLukoff, the package includes twocassette tapes and a useful manualwith instructive information designedfor researchers, psychiatric residents,and psychology interns.
The program focuses onperceptual, thought, and communi-cation dysfunctions specific to schizo-phrenia. Drs. Falloon and Lukoff usean experiential, interactive approachto demonstrate how to elicit andevaluate distinctive qualities of apatient's experience and to relate thisinformation to today's complexdiagnostic standards. The programaddresses the assessment of thoughtinterference, hallucinations,delusions, and speech and behavioraldisturbances. Clear definitions ofsymptoms, dramatized interviews,and critical commentary encouragelisteners to develop their ability to
detect characteristic indicators ofschizophrenia, as they compare theirown conclusions with those of theauthors.
BMA also offers The InitialInterview: Evaluation Strategies forDSM-III Diagnosis, by RobertSpitzer, M.D., and Janet Williams,D.S.W., A.C.S.W. This set of tapespresents interviewing strategies whichcan be applied across the full rangeof DSM-III, while focusing on fourrepresentative symptom types(anxiety, depression, crisisadjustment, and psychosis). Theprogram demonstrates the StructuredClinical Interviews for DSM-III(SCID) method which Drs. Spitzerand Williams pioneered to gathermore reliable information fromconfused or relatively inarticulatepatients and to establish proceduralguidelines for interviewing.
For more information or to order,write to BMA Audio Cassettes, 200Park Avenue South, New York, NY10003 or call toll-free 800-221-3966(in New York State and Canada, call212-674-1900).
by guest on Novem
ber 8, 2016http://schizophreniabulletin.oxfordjournals.org/
Dow
nloaded from