Language impairment was once thought to be uncommon in PSPs/PSPd and CBS/CBD [
8,
54], but some early studies may have overlooked or downplayed language disturbances to emphasise the importance of motor features [
55]. Additionally, formal language testing was often not included, or described, in early studies. More recent studies suggest that language impairment is common in CBS [
16,
17,
51,
56] and PSPs [
57‐
59], which we summarise here.
Fluency
PSPs and CBS typically result in severely impaired initial letter fluency [
9,
60‐
77] and moderate deficits in category fluency [
9,
41,
60,
64,
67‐
69,
74,
76‐
78]. The fact that letter fluency is typically more affected than category fluency [
9,
64] is consistent with the finding, in functional imaging with healthy adults, that letter fluency principally activates frontal regions whereas category fluency produces mainly temporal-lobe activation [
79,
80].
Additionally, the types of words generated are revealing: PSPs and CBS patients sometimes, puzzlingly, generate low-frequency words [
81]. For example,
ptarmigan, pterodactyl, and
farinaceous were the only p- and f-words elicited from a PSP-RS patient in our clinic. This is distinct from the strong bias towards high-frequency words in svPPA and AD [
82].
Executive vs. language accounts of fluency impairment
Fluency scores may relate to other measures of language ability [
83], but it is commonly accepted that fluency performance depends upon additional cognitive processes including executive function, initiation, working memory and attention [
84]. Therefore, without more detailed language assessment, it would be unclear whether a fluency deficit reflects a breakdown of language per se, other cognitive processes, or both.
Schofield et al. reported impaired letter fluency in 100% of 11 autopsy-confirmed PSP-RS patients (PSPd), while 55% were impaired on a composite score derived from various language tests including category fluency [
75]. Similarly, confrontation naming was dramatically better than letter fluency in a PSPs study by Gerstenecker et al. [
85]. This suggests that, in at least a subset of patients, fluency impairment in PSP is independent from problems in the principal aspects of language, although of course language deficits, when present, may exacerbate the fluency impairment. Indeed, Sitek et al. [
86] reported language impairment (based on assessment of spontaneous speech, naming, comprehension, and repetition) in 80% of 20 patients with PSP-RS and even hypothesised that this might underlie the low Frontal Assessment Battery scores observed in PSPs.
Patients with PSPs and CBS have reduced initiation and speed of processing which might in part explain their poor fluency, though this would likely have a similar impact on letter and category fluency. In a study of over 300 PSPs cases [
62], the most salient cognitive impairment was on the ‘initiation and perseveration’ subscale of the Dementia Rating Scale, although this is a composite score derived from tasks which include verbal fluency. While the slowed speed of processing in PSPs and CBS patients [
57,
73,
87] could contribute to their fluency impairment (since these tasks are timed), extending the time limit beyond one minute rarely improves the number of words elicited in letter fluency from people with PSPs: they appear to dry-up after a few words regardless of the time available. Consistent with this are findings from Robinson and colleagues [
88] showing that the majority of words generated over a range of fluency tasks by a PSPs patient with frontal dynamic aphasia were produced within the first 20–50% of the allotted time.
Neural correlates of fluency impairment
Investigations of neuroanatomical correlates of fluency impairment in PSPs seem to implicate the dorsolateral prefrontal cortex (DLPFC), concordant with both lesion [
89] and functional imaging studies [
90] associating letter fluency with DLPFC, especially on the left side. Verbal fluency (combining letter and category fluency) in PSPs has been shown to correlate with a component of the informant-rated Katz Adjustment Scale which is purported to correspond to DLPFC functioning [
91]. Further, impaired letter fluency apparently relates to increased neuronal tau deposition in the superior frontal gyrus [
75], an area associated with the executive aspects of the task [
92], while a composite language score (including category fluency) was related to pathology in inferior frontal and perirhinal cortices, areas associated with word retrieval [
92,
93]. These findings suggest that executive dysfunction underlies the letter fluency impairment in PSPs, though they do not definitively establish whether executive function is vital in fluency tasks generally or is differentially critical to the initial-letter version.
Neural correlates of fluency impairment in CBS have not been extensively investigated. Positron emission tomography indicates left frontoparietal hypoperfusion in CBS patients with impaired verbal fluency [
94], suggesting the presence of a broader dysexecutive syndrome. In pathologically-confirmed CBD [
19], histopathologic abnormalities were most prominent in frontal and parietal regions in patients who also showed neuropsychological impairment, again consistent with an account involving executive dysfunction.
Naming
Confrontational naming appears to be relatively well preserved in PSPs. A mild naming impairment has been reported when an extended naming task is used such as the full 60-item version of the Boston Naming Test (BNT) [
45,
67,
78,
85,
95,
96], while some studies using shortened naming tasks have found performance to be normal [
69,
97] or only mildly impaired [
57,
59]. This may suggest that the problem relates more to sustained attention than to the processes required for object naming per se.
Naming difficulty should not be confused with the more common description “word finding difficulty” during conversational speech, which is likely to reflect the fluency difficulties outlined above. Maher, Smith and Lees [
71] reported mild word-finding difficulty in seven out of 25 PSPs patients without severe dysphasia or comprehension deficits. Nine of the 25 patients completed the Graded Naming Test [
98], with all but one in the average or superior ranges. Impaired word-finding in connected speech was also reported by Kobylecki et al. [
99] in 33% of 60 PSPs patients.
Where frank naming errors do occur in PSPs, they are often visually related to the target objects [
36] and thus may be indirectly due to gaze palsy or other visual problems that are common in PSPs [
36,
58,
100]. There are also a few reports of errors at the semantic or lexical retrieval stage [
59,
95], though these could reflect attentional deficits rather than a genuine degradation of semantic knowledge.
Confrontational naming is impaired in some patients with CBS [
17,
19,
56,
101,
102] but often only mildly so [
101]. It has been suggested that this reflects impaired retrieval rather than a semantic deficit, since performance is usually aided by phonemic cueing [
102,
103], but semantic errors have been reported [
101]. Naming impairment in CBS has been shown to correlate with impaired lexical retrieval (assessed using a category fluency task), as well as with impaired visual-spatial functioning [
102]. Further, in this study, naming impairment correlated with volume of the left lateral temporal cortex and the left frontal cortex, which, according to the authors’ hypotheses, contribute to lexical retrieval and verbal working memory, respectively.
The types of errors made during a naming task are rarely reported, leaving the basis for these errors in PSPs and/or CBS unclear. We suggest that future studies should analyse and report the types of naming errors in an attempt to specify the contribution of lexical, semantic, attentional, and visuospatial factors.
Some studies report an intriguingly disproportionate impairment for action naming in PSPs and CBS [
104‐
106]. For example, in a study of frontotemporal dementia, PSPs, and CBS [
104], the latter two groups had more difficulty with action than object naming which was attributed to disruption of frontoparietal-subcortical circuits involved in action knowledge and representation. Similarly, Chow et al. [
105] found that PSPs and CBS patients were impaired on sound naming (naming the objects which produce various sounds), particularly for manipulable objects (defined as objects for which a goal-directed hand movement is required to produce the sound; e.g., guitar). Further, performance on this task was associated with atrophy of the left pre-motor region. This group was not significantly impaired on naming using the short form of the BNT (note that all but one of the items on this test are non-manipulable objects). These results might reflect the role of classical motor networks in speech production and language comprehension for actions [
107].
Comprehension and semantic association
Systematic investigation suggests that single-word comprehension is relatively spared in PSPs [
100,
104,
108], while sentence comprehension may be mildly impaired [
47,
57,
100,
109], reinforcing the impression in early studies [
71,
110]. However, five out of six patients investigated by Podoll, Schwarz and Noth [
36] and more than 50% of patients in a study by Catricalà et al. [
59] showed impaired word as well as sentence comprehension. Comprehension of action-verbs may be particularly affected in PSPs, in keeping with the deficits in action naming above [
106,
111,
112].
While semantic knowledge is relatively preserved in PSPs there can be a mild impairment on formal testing. For example, on the Sydney Language Battery (SYDBAT) PSPs patients were mildly impaired relative to controls, and not significantly different from patients with nfvPPA [
57]. There are also reports of impaired performance on tests of semantic association [
57,
59,
95], although there is a significant executive component to this test, and furthermore, scanning the test items requires vertical eye movements which may disadvantage patients with this disorder. Indeed, many assessments of semantic knowledge use word-picture or sentence-picture matching tasks on which deficits in visual attention or visual scanning may impact performance [
100,
109].
In CBS, single-word comprehension was impaired in 52% of patients in Di Stefano et al. [
17], but in only two of fifteen patients in Huang et al. [
51], and no significant difference from controls was reported by Cotelli et al. [
104]. However, differences between these studies in (a) the criteria for diagnosing CBS (b) the tests used to evaluate single-word comprehension (e.g., matching of names to real objects vs. pictures), and (c) the methods for classifying impairment (e.g., direct comparison to a group of control participants vs. normative values) make it difficult to draw generalised conclusions.
Patients with a clinical pattern of nfvPPA and pathologically proven CBD show significant impairment in sentence comprehension [
113], particularly in later stages of the disease course [
19]. The researchers argued that this deficit reflected genuinely impaired understanding of grammatically complex sentences rather than problems of single-word knowledge. Further detail was provided by Cotelli et al. [
114] who reported unimpaired sentence comprehension in a CBS group as measured using a sentence-picture matching task, but impaired ability to detect violations of grammar during a sentence judgment task. The interpretation of a specific syntactic deficit may, however, be complicated by the fact that patients with CBS [
101,
115] and pathologically proven CBD [
19] can show impairments on tasks of semantic association. The difficulty in judging a sentence as ungrammatical might therefore be, in part, semantic.
Sentence production
Agrammatism is uncommon in early PSP-RS, but features of nfvPPA or PSP-SL, including agrammatism, have been reported in cases of PSPd [
21,
116]. Tasks requiring sentence production, such as picture description, can reveal other abnormalities, including: perseverations; fewer morphemes, words, and sentences; and fewer novel words and sentences in the context of normal syntactic structure [
36,
58,
59,
100]. CBS, on the other hand, can present with syntactic errors, phonological errors, paraphasias, and speech apraxia [
17,
51,
101] as well as impaired syntactic knowledge [
114] and phonological processing [
101]. Unfortunately, differences between these studies hinder clear and generalisable conclusions, including differences in diagnostic criteria, sample sizes, test materials and threshold values for impaired performance.
Repetition
While repetition has not been extensively investigated in PSPs, one study reported that single-word repetition in this group was significantly worse than controls, but significantly better than a group with nfvPPA [
57]. Four patients presenting with nfvPPA who later developed clinical features of PSP [
117] were not significantly impaired on either single-word or sentence repetition relative to controls. Further, they performed significantly better on both repetition tasks than a group of patients with nfvPPA without PSP syndrome.
Deficits of word and sentence repetition have been noted in some CBS patients [
17,
51,
53]. Impaired sentence repetition, which is characteristic of lvPPA, seems to be more common in CBS with underlying amyloid pathology [
53].
Reading
In PSPs, where tested, reading is described as slow with poor pronunciation due to dysarthria [
36,
41,
117]. There are also reports of difficulties in deciphering words and/or in finding the beginning of each new line in a text, suggesting that visual scanning deficits are a significant factor in the reading difficulties in PSPs [
36].
Patients with CBS may be unimpaired on single word reading [
101], although a minority of patients make regularisation errors [
51,
101] (e.g., pronouncing a written word like
sew as “sue”, in line with more typical spelling-to-sound correspondences). One patient in a study by Graham et al. [
101] who was severely aphasic made errors which were characterised as non-word approximations of the target (e.g., girl → “dirl”) or visual errors (e.g. loss → “lost”). The majority of CBS patients in this study were impaired on a task of non-word reading, consistent with their impaired performance on other tasks of phonological skills such as phoneme blending and segmentation. Larger studies with systematic analysis of reading in PSPs and CBS are so far lacking.
Writing
Dysgraphia is common in PSPs and CBS [
36,
51], and may be due to linguistic, cognitive, visual, or motor impairment. Writing features noted in PSPs include micrographia, abnormal slanting, omission of letters and words, addition of letters, and omission of diacritic marks in a Polish-language group [
36,
59,
96,
118,
119]. The micrographia of PSPs differs from Parkinson’s disease by the absence of progressive fatigue (reducing letter size). The dysgraphia in PSPs may result from visuo-constructional or oculomotor deficits, whereby abnormal vertical and horizontal saccades disrupt the visual monitoring of writing [
36,
118], and may be exacerbated by severe limb akinesia.
CBS and autopsy-proven CBD are associated with constructional apraxia and impaired handwriting, in keeping with the limb apraxia which affects most patients [
19,
34]. Dysgraphia was claimed to be the most common language-related abnormality in CBS by Huang et al. [
51], although its nature was not described and limb apraxia was noted in all patients. Nevertheless, errors both in writing and oral spelling (the latter used by Graham et al. [
101] to avoid the impact of motor deficits on handwriting) in CBS have included article omissions, letter substitutions and omissions, and a mixture of phonologically plausible and non-phonologically plausible spelling errors [
101,
120‐
122].
Summary of language deficits
The language impairment in PSP seems largely consistent with a “frontal” deficit, possibly linked to executive dysfunction and impaired initiation. It has certainly been argued that executive deficits underlie the verbal fluency impairment in PSP [
64,
123]. In contrast to the typically more severe deficit of category than letter fluency impairment in AD that suggests a breakdown of semantic memory [
9,
64], the reverse is typical in PSP and could be explained by problems of initiation or general executive function. Indeed, patients with PSPs show impairments in other tests of ‘frontal functioning’ such as planning, orienting attention, and set-shifting [
59,
67,
124‐
126]. Consistent with this are reports of an “adynamic aphasia” in PSP, characterised by reduced verbal output and cognitive processing speed [
13,
57,
108]. In a recent study [
57], the finding of only mild or borderline correlations between tests of language and those of executive function led the authors to conclude that language impairment in PSPs is not solely attributable to executive dysfunction. It is worth noting, however, that the PSPs cohort in this study were recruited from a predominantly cognitive disorders clinic and may therefore have presented with more cognitive and behavioural features than motor features. Future research is clearly needed to delineate the nature of the language impairment in PSP, including an analysis of the types of errors made by patients on language tasks to determine whether these can be “explained away” by visual or motor difficulties.
While visual and motor problems exacerbate some of the language impairments documented in CBS, these patients appear to experience a true breakdown of language ability [
127]. Aphasic syndromes reported in CBS/CBD include a surprisingly wide variety: non-fluent aphasia, anomic aphasia, fluent aphasia, Broca’s aphasia, mixed aphasia, and AOS [
17,
19,
28,
34,
55,
128]. These different patterns may be a function of the stage at which assessment occurs, or perhaps they will be found to relate to the different pathologies which may underlie CBS when these are better delineated.
Most of the evidence to date comes from studies of classic phenotypes. It is becoming clear from the few available larger clinicopathological series that phenotypic variation is common, including clinical overlap (e.g., PSP-CBS, and CBS-PSP) and non-classical phenotypes (e.g., PSP-SL, CBS-NAV) [
11,
18,
20,
21,
28,
29], and more such studies are clearly needed.
In addition, most studies have been cross-sectional and therefore may not adequately characterise the onset and evolution of language deficits. Early vs. late language loss in PSPs and CBS may turn out to have important clinical and pathological implications. From the few longitudinal clinicopathologic studies, it seems that patients with PSPd who present initially with language impairment (PSP-SL) later develop typical motor features (such as falls, abnormal saccades or pursuit, and supranuclear gaze palsy); this distinction may help to distinguish PSP-SL from nfvPPA [
11]. Similarly CBD patients presenting with nfvPPA subsequently often show extrapyramidal involvement before death [
15]. Conversely, in CBS at least, patients presenting with typical movement disorders also often develop aphasia or behavioural change [
129].
Given that language symptoms can vary by stage, different clinical tools may be required to elicit and characterise language disorders at different stages. For example, patients may write answers in early but not late stages; significant gaze palsy and visual disturbance in later stages of PSPs may limit the usefulness of language tests with visual stimuli; and the assessment of aphasia can be complicated by the development of dysarthrophonia.