**3. Results**

#### **3.1. Main outcomes**

A total of 15 cases with a value different from 3 in the TS scale were observed. Only one case was observed with a score of 0 and four cases with a score of 2. Due to such small number of cases, and in order to get better inferences, TS was recoded using 0 when the symptom was absent (non-TSP) and 1 (TSP) when the symptom was present (prevalence 16%).

**Table 1** shows that TSP, non-TSP and HP did not differ in their demographic data.

**Table 2** shows that there were no significant differences on type of lesion between non-TSP and TSP. Malignant tumors represented the most frequent type of lesion. By grouping the cells with fewer cases (i.e., the cells with the rest of the lesions), a non-significant difference


Both groups of patients did not differ on disease duration [(mean ± SD): non-TSP: 29.04 ± 65.25, TSP: 35.66 ± 79.31 [F (1, 93) = 0.12, p = 0.73], in the number of any additional risk for cognitive impairment [non-TSP: 1.15 ± 1.03, TSP: 1.4 ± 1.06 [F (1, 93) = 0.74, p = 0.39], or in the presence of hemiparesis [brachial: non-TSP: 31% (partial/total count) (25/80), TSP: 27% (4/15): (χ2 = 0.12, df = 1, p = 0.72); crural: non-TSP: 31% (25/80), TSP: 20% (3/15) (χ2 = 0.77; df = 1, p = 0.38)].

Neurocognitive Implications of Tangential Speech in Patients with Focal Brain Damage

http://dx.doi.org/10.5772/intechopen.71904

203

The Cronbach alpha coefficient for all the indicators of the comprehensive battery considered as a whole (representing GNP) was 0.94. The Cronbach alpha coefficients for CR and E&T were 0.92 and 0.77, respectively. Therefore, GNP, on the one hand, in addition to CR and E&T, on the other hand, were analyzed. The Cronbach alpha coefficient for all the indicators of DP was 0.70; therefore, a representative measure of DP was also analyzed. (Note: In all the significant differences reported from now on, TSP was always more impaired than non-TSP,

The ANOVA with TS as grouping variable and GNP as dependent variable (see **Figure 1**) indicated a main effect of TS (F (2, 172) = 25.55, p < 0.0001) with significant pairwise comparisons (Bonferroni post-hoc tests: non-TSP vs. TSP: p = 0.0017; HP vs. either non-TSP or TSP:

The bivariate MANOVA with TS as grouping variable and CR and E&T as univariate dependent variables (see **Figure 2**) indicated that CR and E&T produced significant effect on GNP (Wilks lambda = 0.73, F (4, 342) = 14.77, p < 0.0001) and that a main effect of TS was produced on the two components of performance (univariate effect of CR: F (2, 172) = 12.32, p < 0.0001, univariate effect of E&T: F (2, 172) = 28.35, p < 0.0001). However, non-TSP and TSP did not differ in CR but they did differ in E&T when pairwise comparisons were analyzed (Bonferroni post-hoc tests: non-TSP vs. TSP in CR: p = 0.2439, non-TSP vs. TSP in E&T: p < 0.0001; HP vs.

**Figure 1.** GNP (total score) as a function of TS (HP, non-TSP and TSP). LS means effective hypothesis decomposition.

and both groups of patients were more impaired than HP).

p < 0.0001).

Vertical bars denote 0.95 confidence intervals.

**Table 1.** Demographic data.

between non-TSP and TSP was also observed when malignant tumors were compared with the rest of the lesions (χ<sup>2</sup> = 0.02; df: 1; p = 0.87).

Non-significant differences were observed between non-TSP and TSP when A versus P lesions, excluding other lesions, were compared [non-TSP: A = 53% (A lesions/A + P lesions): (31/59), P = 47% (P lesions/A + P lesions): (28/59); TSP: A = 45% (5/11), P = 55% (6/11) (χ<sup>2</sup> = 0.19; df: 1; p = 0.67)], or when R versus L lesions, excluding other lesions, were compared [non-TSP: R = 59% (R lesions/R + L lesions): (34/58), L = 41% (L lesions/R + L lesions): (24/58); TSP: R = 56% (5/9), L = 44% (4/9) (χ<sup>2</sup> = 0.03; df: 1; p = 0.86)]. Non-significant differences were observed between non-TSP and TSP when specific site lesions [i.e., A, P, AP, SC (χ<sup>2</sup> = 2.89; df: 3; p = 0.41); L, R, B (χ<sup>2</sup> = 0.97; df: 2; p = 0.61)] were compared, or even when specific lobe lesions were compared (χ<sup>2</sup> = 8.48; df: 9; p = 0.49) (percentages not shown, but available upon request).


AVM: arteriovenous malformation, BEN TU: benign tumor, MAL TU: malignant tumor, ISQ STR: ischemic stroke, HEM STR: hemorrhagic stroke, TBI: traumatic brain injury, OTHER [name (N)]: subdural hematoma (1), aneurysm (2), mesial temporal sclerosis (4), abscess (2) and cyst (4).

**Table 2.** Classification of the focal cerebral lesions based on their type.

Both groups of patients did not differ on disease duration [(mean ± SD): non-TSP: 29.04 ± 65.25, TSP: 35.66 ± 79.31 [F (1, 93) = 0.12, p = 0.73], in the number of any additional risk for cognitive impairment [non-TSP: 1.15 ± 1.03, TSP: 1.4 ± 1.06 [F (1, 93) = 0.74, p = 0.39], or in the presence of hemiparesis [brachial: non-TSP: 31% (partial/total count) (25/80), TSP: 27% (4/15): (χ2 = 0.12, df = 1, p = 0.72); crural: non-TSP: 31% (25/80), TSP: 20% (3/15) (χ2 = 0.77; df = 1, p = 0.38)].

The Cronbach alpha coefficient for all the indicators of the comprehensive battery considered as a whole (representing GNP) was 0.94. The Cronbach alpha coefficients for CR and E&T were 0.92 and 0.77, respectively. Therefore, GNP, on the one hand, in addition to CR and E&T, on the other hand, were analyzed. The Cronbach alpha coefficient for all the indicators of DP was 0.70; therefore, a representative measure of DP was also analyzed. (Note: In all the significant differences reported from now on, TSP was always more impaired than non-TSP, and both groups of patients were more impaired than HP).

The ANOVA with TS as grouping variable and GNP as dependent variable (see **Figure 1**) indicated a main effect of TS (F (2, 172) = 25.55, p < 0.0001) with significant pairwise comparisons (Bonferroni post-hoc tests: non-TSP vs. TSP: p = 0.0017; HP vs. either non-TSP or TSP: p < 0.0001).

between non-TSP and TSP was also observed when malignant tumors were compared with

**(three-level frequency)**

non-TSP 41.90 ± 14.50 42 32 6 49 80 TSP 42.26 ± 14.23 6 8 1 8 15 HP 44.40 ± 16.85 34 40 6 40 80 Total 43.07 ± 15.56 82 80 13 97 175

> χ<sup>2</sup> = 2.11; df: 4 p = 0.71

**Gender (men frequency)**

χ<sup>2</sup> = 2.08; df: 2 p = 0.35

**N**

Non-significant differences were observed between non-TSP and TSP when A versus P lesions, excluding other lesions, were compared [non-TSP: A = 53% (A lesions/A + P lesions): (31/59), P = 47% (P lesions/A + P lesions): (28/59); TSP: A = 45% (5/11), P = 55% (6/11) (χ<sup>2</sup> = 0.19; df: 1; p = 0.67)], or when R versus L lesions, excluding other lesions, were compared [non-TSP: R = 59% (R lesions/R + L lesions): (34/58), L = 41% (L lesions/R + L lesions): (24/58); TSP: R = 56% (5/9), L = 44% (4/9) (χ<sup>2</sup> = 0.03; df: 1; p = 0.86)]. Non-significant differences were observed between non-TSP and TSP when specific site lesions [i.e., A, P, AP, SC (χ<sup>2</sup> = 2.89; df: 3; p = 0.41); L, R, B (χ<sup>2</sup> = 0.97; df: 2; p = 0.61)] were compared, or even when specific lobe lesions were compared (χ<sup>2</sup> = 8.48; df: 9; p = 0.49) (percentages not shown, but available upon request).

AVM: arteriovenous malformation, BEN TU: benign tumor, MAL TU: malignant tumor, ISQ STR: ischemic stroke, HEM STR: hemorrhagic stroke, TBI: traumatic brain injury, OTHER [name (N)]: subdural hematoma (1), aneurysm (2), mesial

the rest of the lesions (χ<sup>2</sup> = 0.02; df: 1; p = 0.87).

F(2, 172) = 0.53 p = 0.57

**Table 1.** Demographic data.

202 Gerontology

**Group Age (mean ± SD) Education** 

**Lesion Group**

**χ<sup>2</sup> = 3.27; df: 6; p = 0.77**

temporal sclerosis (4), abscess (2) and cyst (4).

**Table 2.** Classification of the focal cerebral lesions based on their type.

**Type Non-TSP TSP AVM 8 2 BEN TU 18 4 MAL TU 25 5 ISQ STR 6 1 HEM STR 4 1 TBI 6 2 OTHER 13 0 Total 80 15**

The bivariate MANOVA with TS as grouping variable and CR and E&T as univariate dependent variables (see **Figure 2**) indicated that CR and E&T produced significant effect on GNP (Wilks lambda = 0.73, F (4, 342) = 14.77, p < 0.0001) and that a main effect of TS was produced on the two components of performance (univariate effect of CR: F (2, 172) = 12.32, p < 0.0001, univariate effect of E&T: F (2, 172) = 28.35, p < 0.0001). However, non-TSP and TSP did not differ in CR but they did differ in E&T when pairwise comparisons were analyzed (Bonferroni post-hoc tests: non-TSP vs. TSP in CR: p = 0.2439, non-TSP vs. TSP in E&T: p < 0.0001; HP vs.

**Figure 1.** GNP (total score) as a function of TS (HP, non-TSP and TSP). LS means effective hypothesis decomposition. Vertical bars denote 0.95 confidence intervals.

with non-significant pairwise comparisons between non-TSP and TSP (Bonferroni post-hoc test: p = 1); significant pairwise comparisons between HP and either non-TSP or TSP were observed (Bonferroni post-hoc tests: p < 0.003). The ANOVA with TS as grouping variable and the number of words in oral verbal fluency as dependent variable indicated a main effect of TS (F (2, 172) = 25.02, p < 0.0001), with non-significant pairwise comparisons between non-TSP and TSP (Bonferroni post-hoc test: p = 0.621); significant pairwise comparisons between

Neurocognitive Implications of Tangential Speech in Patients with Focal Brain Damage

http://dx.doi.org/10.5772/intechopen.71904

205

Regarding the report of the caregiver during the initial interview, the r (rank-biserial) between TS and: (a) sensory deficits, (b) motor deficits, (c) sleeping disorders, (d) perceptual disorders, (e) language disorders, (f) behavioral disorders and (g) thought disturbances were: 0.06, −0.14, 0.07, 0.08, 0.00, 0.29 and 0.43, respectively, the last two coefficients being statistically significant. The association between TS and the presence of seizures was non-significant: non-

Regarding the complementary behavioral observations registered during the administration of the neuropsychological battery, the r (rank-biserial) between TS and: (a) degree of cooperation, (b) emotional state, (c) language speed, (d) disability awareness, (e) voice volume and (f) prosody were: −0.36, 0.23, 0.32, −0.19, 0.14 and 0.08, respectively. Of these, the first three coefficients were statistically significant. A non-significant association with the presence of emotional lability was observed: non-TSP: 6% (5/80), TSP: 20% (3/15) (χ2 = 3.09, df = 1, p = 0.08). A non-significant association with the presence of verbal perseverations was also observed: non-TSP: 8% (6/80), TSP: 13% (2/15) (χ2 = 0.56, df = 1, p = 0.46). There were no patients who showed aggression, hallucinations or delusions, during the administration of

Since the GNP component that produced a significant difference between TSP and both non-TSP and HP in pairwise comparisons was E&T, its individual indicators were analyzed. The MANOVA with TS as grouping variable and the E&T indicators as dependent variables indicated that all the dependent variables produced a significant effect on the multivariate measure of performance (Wilks lambda = 0.38, F (34, 312) = 5.75, p < 0.0001). A main effect of TS was observed in all the components of the model except for E in time orientation to year as well as two types of E in hidden objects (all significant univariate effects: F (2, 172) ≥ 3.11, p < 0.05; all non-significant univariate effects: F (2, 172) ≤ 1.41, p ≥ 0.2466). Significant differences between non-TSP and TSP according to Bonferroni post-hoc test involved E in verbal auditory sustained attention [omission E (p = 0.004) and commission E (p = 0.0002)], T in nonverbal visual sustained attention (p < 0.0001), as well as E in the time of day (p = 0.0003). The two groups of patients did not differ in the rest of the indicators (Bonferroni post-hoc tests: non-TSP vs. TSP all p ≥ 0.0627). [Note: Considering pairwise comparisons, when a significant difference between non-TSP and TSP was observed, a significant difference between TSP and HP was also observed with p < 0.0001; besides, by taking the P25 of the whole sample as cutoff point for these significant indicators, 93% of the HP (74/80) and 68% of the non-TSP (54/80) had a score greater

HP and either non-TSP or TSP were observed (Bonferroni post-hoc tests: p < 0.0001).

TSP: 51% (41/80), TSP: 33% (5/15) (χ2 = 1.62, df = 1, p = 0.20).

**3.2. Complementary statistical information**

the battery.

**Figure 2.** CR and E&T (total score) as a function of TS (HP, non-TSP and TSP). LS means effective hypothesis decomposition. Vertical bars denote 0.95 confidence intervals. E&T were multiplied by (−1).

either non-TSP or TSP: p < 0.0003 in any of the univariate measures). (Note: Since CR and E&T have different units of measurement, z-scores were used to show GNP results, which were identical to the results obtained with the raw scores).

The ANOVA with TS as grouping variable and the representative measure of DP as dependent variable indicated a main effect of TS (F (2, 172) = 34.61, p < 0.0001), with significant pairwise comparisons among the three groups using Bonferroni post-hoc test: that is, difference between non-TSP and TSP: p = 0.002, difference between HP and either non-TSP or TSP: p < 0.0001. A significant association between TS and DP was demonstrated (see **Table 3**): By taking the 25th percentile (P25) of the whole sample as cutoff point, 95% of the HP and 63% of the non-TSP had a score greater than the P25, while 80% of the TSP had a score equal or less than the P25.

The ANOVA with TS as grouping variable and the number of words in written verbal fluency as dependent variable indicated a main effect of TS (F (2, 172) = 12.63, p < 0.0001),


**Table 3.** Distribution of frequencies according to 25th percentile on the representative measure of DP.

with non-significant pairwise comparisons between non-TSP and TSP (Bonferroni post-hoc test: p = 1); significant pairwise comparisons between HP and either non-TSP or TSP were observed (Bonferroni post-hoc tests: p < 0.003). The ANOVA with TS as grouping variable and the number of words in oral verbal fluency as dependent variable indicated a main effect of TS (F (2, 172) = 25.02, p < 0.0001), with non-significant pairwise comparisons between non-TSP and TSP (Bonferroni post-hoc test: p = 0.621); significant pairwise comparisons between HP and either non-TSP or TSP were observed (Bonferroni post-hoc tests: p < 0.0001).

Regarding the report of the caregiver during the initial interview, the r (rank-biserial) between TS and: (a) sensory deficits, (b) motor deficits, (c) sleeping disorders, (d) perceptual disorders, (e) language disorders, (f) behavioral disorders and (g) thought disturbances were: 0.06, −0.14, 0.07, 0.08, 0.00, 0.29 and 0.43, respectively, the last two coefficients being statistically significant. The association between TS and the presence of seizures was non-significant: non-TSP: 51% (41/80), TSP: 33% (5/15) (χ2 = 1.62, df = 1, p = 0.20).

Regarding the complementary behavioral observations registered during the administration of the neuropsychological battery, the r (rank-biserial) between TS and: (a) degree of cooperation, (b) emotional state, (c) language speed, (d) disability awareness, (e) voice volume and (f) prosody were: −0.36, 0.23, 0.32, −0.19, 0.14 and 0.08, respectively. Of these, the first three coefficients were statistically significant. A non-significant association with the presence of emotional lability was observed: non-TSP: 6% (5/80), TSP: 20% (3/15) (χ2 = 3.09, df = 1, p = 0.08). A non-significant association with the presence of verbal perseverations was also observed: non-TSP: 8% (6/80), TSP: 13% (2/15) (χ2 = 0.56, df = 1, p = 0.46). There were no patients who showed aggression, hallucinations or delusions, during the administration of the battery.

#### **3.2. Complementary statistical information**

either non-TSP or TSP: p < 0.0003 in any of the univariate measures). (Note: Since CR and E&T have different units of measurement, z-scores were used to show GNP results, which were

**Figure 2.** CR and E&T (total score) as a function of TS (HP, non-TSP and TSP). LS means effective hypothesis

The ANOVA with TS as grouping variable and the representative measure of DP as dependent variable indicated a main effect of TS (F (2, 172) = 34.61, p < 0.0001), with significant pairwise comparisons among the three groups using Bonferroni post-hoc test: that is, difference between non-TSP and TSP: p = 0.002, difference between HP and either non-TSP or TSP: p < 0.0001. A significant association between TS and DP was demonstrated (see **Table 3**): By taking the 25th percentile (P25) of the whole sample as cutoff point, 95% of the HP and 63% of the non-TSP had a score greater than the P25, while 80% of the TSP had a score equal or

The ANOVA with TS as grouping variable and the number of words in written verbal fluency as dependent variable indicated a main effect of TS (F (2, 172) = 12.63, p < 0.0001),

**Group ≤ Percentile 25 > Percentile 25 Total** HP 4 76 80 Non-TSP 30 50 80 TSP 12 3 15

**Table 3.** Distribution of frequencies according to 25th percentile on the representative measure of DP.

identical to the results obtained with the raw scores).

decomposition. Vertical bars denote 0.95 confidence intervals. E&T were multiplied by (−1).

less than the P25.

204 Gerontology

χ<sup>2</sup> = 46.24; df: 2; p < 0.0001 (25th percentile = 36.00)

Since the GNP component that produced a significant difference between TSP and both non-TSP and HP in pairwise comparisons was E&T, its individual indicators were analyzed. The MANOVA with TS as grouping variable and the E&T indicators as dependent variables indicated that all the dependent variables produced a significant effect on the multivariate measure of performance (Wilks lambda = 0.38, F (34, 312) = 5.75, p < 0.0001). A main effect of TS was observed in all the components of the model except for E in time orientation to year as well as two types of E in hidden objects (all significant univariate effects: F (2, 172) ≥ 3.11, p < 0.05; all non-significant univariate effects: F (2, 172) ≤ 1.41, p ≥ 0.2466). Significant differences between non-TSP and TSP according to Bonferroni post-hoc test involved E in verbal auditory sustained attention [omission E (p = 0.004) and commission E (p = 0.0002)], T in nonverbal visual sustained attention (p < 0.0001), as well as E in the time of day (p = 0.0003). The two groups of patients did not differ in the rest of the indicators (Bonferroni post-hoc tests: non-TSP vs. TSP all p ≥ 0.0627). [Note: Considering pairwise comparisons, when a significant difference between non-TSP and TSP was observed, a significant difference between TSP and HP was also observed with p < 0.0001; besides, by taking the P25 of the whole sample as cutoff point for these significant indicators, 93% of the HP (74/80) and 68% of the non-TSP (54/80) had a score greater than the P25, while 80% of the TSP (12/15) has a score equal or less than the P25 (results available upon request).] As shown, the presence of TS for the E&T indicators (individually considered) appeared as a rather nonspecific factor in terms of the stimuli involved, that is, TS was not related to a certain modality of cognitive impairment such as the verbal or the nonverbal one.

to the time of day. On the contrary, the effect of TS on the tasks related to DP involved, specifically, complex verbal comprehension and the spontaneous, implicit and delayed recall of a story, excluding cued recall. The association between TS and recognized measures of discourse processing (such as narrative comprehension, memory and production [8–10]) provided support for the viability and validity of the present screening scale to assess TS. A satisfactory inter-rater reliability for the TS scale was also observed. Additionally, TS was associated with emotional and behavioral alterations in the clinical sample: significant correlations were observed between TS and the emergence of behavioral and thought disturbances, as reported by the caregiver during the initial interview, as well as between TS and the complementary behavioral observations of emotional excitement, rapid speech and diminished cooperation, as reported by the neuropsychologist during the battery administration. Regarding intervening clinical variables, and aside from the type and site of injury (see above), TS was independent of demographic variables, presence of neuro-

Neurocognitive Implications of Tangential Speech in Patients with Focal Brain Damage

http://dx.doi.org/10.5772/intechopen.71904

207

In the most recent research on verbal communication, AbdulSabur et al. [8] and Awad et al. [40] analyzed the processes involved in sharing knowledge through narrative processing, and described extended brain networks and bilateral involvement in their neurofunctional studies with healthy participants. Consistent with this view, Jouen et al. [5] observed in a combined fMRI and DTI study with healthy participants that understanding sentences and pictures revealed bilateral involvement and a common fronto-temporo-parietal network for both modalities. This semantic network was not limited to sensorimotor systems but extended to the highest levels of cognition, including autobiographical memory, scene analysis, mental model formation, reasoning and theory of mind. Present results agree with those studies since no focal brain lesions were identified for TS. To be precise, present results agree with the hypothesis that this neurogenic inability to spontaneously find, organize and communicate verbal information for a specific topic, and beyond single words, may be caused by several sites of brain damage or, most probably, by aberrant interactions within extended

In trying to understand how TS is integrated with the rest of the cognitive functions, present results indicated that TSP showed an overall impaired cognitive performance relative to non-TSP; however, the contribution of the GNP components of E&T and CR showed distinctive patterns: the relative weight of E&T was superior to that observed in CR, thus finally producing a significant effect only on E&T. More thorough analyses carried out to discover the nature of the cognitive impairments associated with TS provided illustrative results: by considering just those individual indicators of E&T related to sustained attention, which were significantly associated with TS, it can be noticed that one task involved verbal auditory stimuli and the other task involved nonverbal visual stimuli; besides, one

logical risks and disease duration.

**5. Discussion**

brain networks.

Additionally, and given that the representative measure of DP also produced a significant difference between TSP and both non-TSP and HP, its individual indicators were analyzed. The MANOVA with TS as grouping variable and the indicators of DP as dependent variables indicated that all the dependent variables produced a significant effect on the multivariate measure of performance (Wilks Lambda = 0.52, F (12, 334) = 10.71, p < 0.0001). A main effect of TS was produced on all the components of the model except for the interviewer's global impression during administration of the cued story recall (all significant univariate effects: F (2, 172) ≥ 8.78, p < 0.0002; the non-significant univariate effect: F (2, 172) = 1.48, p = 0.2305). When pairwise comparisons were analyzed, non-TSP and TSP did not differ in the standardized and detailed scoring of the 25 passages after administration of the cued story recall (Bonferroni post-hoc test: p = 0.3179) but they did differ in the rest of the components (Bonferroni post-hoc tests: non-TSP vs. TSP all p ≤ 0.0220). Considering HP, significant differences between HP and TSP were observed in any of the univariate measures, excluding cued recall (Bonferroni post-hoc tests: HP vs. TSP all p < 0.0001). Therefore, not all the indicators of DP, individually considered, produced significant differences between non-TSP and TSP as it can be inferred from the cued recall in its both indicators. [Note: all univariate results described in the present study were confirmed with nonparametric tests (results available upon request)].

Regarding inter-rater reliability, the ICC was 0.86 without difference between both raters according to the Wilcoxon paired-sample test (z = 0.00, N = 95, p = 1). [Note: the ICC for the dichotomized TS scale was 0.79 [Wilcoxon paired-sample test (z = 0.40, N = 95, p = 0.685)].
