Exploring Harmful Illocutionary Forces Expressed by Older Adults with and Without Alzheimer’s Disease: A Multimodal Perspective

2.1 Illocutionary Forces and Multimodal Corpus Linguistics

Against the background of the ‘linguistic turn’ in Western philosophy, Austin (1962) proposed the Speech Act Theory in the 1950s, and further elaborated by Searle (1969), thriving as a classic topic in pragmatics. Austin (1962) holds that as far as the speakers are sincere when they make the utterance, they are ‘doing things with words’ and producing the corresponding illocutionary force. As the first scholar to put forward a set of speech act theories and the concept of an ‘indirect speech act,’ Searle made a unique contribution to speech act theory. Following Searle (1969, 21), speech act is the most basic unit of human communication. Based on the discussion of (1) the illocutionary point of a speech act, (2) the direction of fit between discourse and the objective world, (3) the psychological states of discourse expression, (4) force or strength of the point, (5) status or position of the hearer or speaker, (6) way the utterance relates to the interests of the hearer or speaker, (7) relations to the rest of the discourse, (8) propositional content, (9) requirement that some acts must be speech acts while others need not be, (10) lack or need for extra-linguistic institutions, (11) lack of or need for illocutionary verbs, and (12) style of performance, Searle (1979) divided speech acts into five categories: assertives, directives, commissives, expressives, and declarations. However, the criteria of his taxonomy was based on the speaker’s ‘intentionality’, overlooking the fact that speech acts are social behaviours in nature (Huang 2021b) and serve as pivotal benchmarks for assessing the pragmatic competence in older adults (Huang and Yang 2023). Correspondingly, the investigation of illocutionary force indicating devices (IFID) was narrow and inadequate, mainly focusing on the linguistic means (Huang 2018).

The development of modern multimedia technology reveals the essence of language activities, and the data that linguists concerned changed from monomodal to multimodal. Linguists began to pay increasing attention to the fact that verbal and nonverbal cues fit together into integrated messages in daily interaction. With such a background, multimodal corpus comes into being as the ‘Corpus 4.0’ (Knight 2009), which provides a richer representation of different aspects of discourse, including participants’ utterance content, prosody, gestures, and also the context or co-text in which interaction takes place (Huang 2021a). The adoption of the multimodal corpus in pragmatic research provides the classic theories with fresh perspectives and new practices (Knight and Adolphs 2008).

So far, few studies utilized multimodal corpus-based approaches. Nevertheless, Gu (2013a) made a breakthrough in this regard, presenting a conceptual model for empirically studying the tripartite interaction between illocution, emotion and prosody based on cases taken from the Spoken Chinese Corpus of Situated Discourse (SCCSD). Gu unprecedentedly investigated speech acts as human behavior rather than treating language use as sheer information-exchanging activity, while such non-verbal cues as facial expressions, gestures, and bodily movements were absent from the analysis (Huang 2018).

In this study, Gu’s analytic framework is refined to some extent, and other cues including gestures and bodily movement, indispensable components of the felicitousness of illocutionary acts performed by ‘the whole person,’ have been taken into account. To bring to the fore the close relations between illocutionary force tokens, the speaker’s individuality, and the discursive environment, illocutionary forces are examined in the context of Chinese situated discourse.

Filiou et al. (2024) suggest that speech acts serve as a valuable tool for gaining insights into how AD older adults perceive and tackle daily activities, as well as the consequences of their choices of actions. Our study examines the harmful illocutionary forces in speech by elderly individuals with Alzheimer’s Disease (AD), noting their increased expression of negative emotions and unpredictable behaviours, such as irritability and pessimism (Huang and Che 2023). By identifying how these individuals’ speech acts are marked by sudden emotional shifts, it may inform tailored support and communication strategies for those with AD.

2.2 Pragmatic Impairment in Dementia and Multimodal Compensation

Recent decades have witnessed an escalating scholarly interest in the pragmatic deficits exhibited in dementia patients, with the contrast between healthy and cognitively impaired older adults occupying the spotlight of academic inquiry. Drawing upon empirical evidence sourced from the Chinese context, seminal investigations conducted by Huang, Yang, and Liu (2021), alongside Zhou and Huang (2023), have begun to scrutinize the speech act execution among cognitively healthy and impaired older adults. Earlier research has encapsulated an array of pragmatic deficits associated with elderly individuals afflicted by Alzheimer’s Disease (AD). These encompass challenges with forward-and-backward interdependency, speech act repetition, vague deixis, compromised emotional expression, vague self-awareness, and the need for multimodal communicative strategies (Huang and Che 2023).

However, the pragmatic compensation mechanisms advance the communication under some circumstances. Research has found that cognitively impaired individuals often rely on physical contacts to initiate, maintain or strengthen interactions (Magai, Cohen, and Gomberg 1996); and those with aphasia use non-verbal resources to compensate for their compromised verbal abilities (Perkins 2007; Simmons-Mackie and Stoehr 2016). Both interpersonal and intrapersonal compensatory adaptation operates in the interaction of the elders with AD (Huang, Yang, and Liu 2021; Zhou and Huang 2023). Among these studies, some focus on the production of co-speech gestures. Older adults reportedly deploy more referential nonverbal behaviors to communicate complex content (Asplund, Jansson, and Norberg 1995), with an increase in iconic gestures, which have a high semantic load, to support their verbal discourse that might probably otherwise be perceived as more affected by the memory deficits (Geladó, Gómez-Ruiz, and Diéguez-Vide 2022). Alzheimer’s patients usually favor congruent co-speech gestures over incongruent ones, which facilitate their comprehension (Eggenberger et al. 2016). In addition, AD patients often resort to non-verbal resources to communicate and express their emotions, though they cannot directly deliver the propositional content in performing speech act (Cummings 2009). It has been observed that cognitively impaired seniors employ laughter as a compensatory act to overcome communication disorders, and to acknowledge their troubles in comprehension and expression of utterances (Kasper 2006).

Preliminary investigations into the pragmatic deficits of older individuals with cognitive disabilities have commenced from a multimodal standpoint. However, studies detailing the specific attributes and mechanisms of pragmatic disorder and multimodal compensation in communication, remain scant.

2.3 The Multimodal Perspective of Investigating Illocutionary Forces and Emotions

The enactment of speech acts involves a multimodal interplay – not merely verbal, but also dynamic and expressive – through which illocutionary force is generated, transmitted, and received, as postulated by Huang (2017). Among those non-verbal cues, emotions are fundamental to the performance and interpretation of speech acts, linking intentionality with communicative action (Alba-Juez and Larina 2018; Goldie 2002; Gu 2017). Additionally, emotions are usually related to the interest of interlocutors and emotions further influence and are conveyed by facial expressions, prosodic features and gestures (Huang 2021a; Johar 2016; Martin and Devillers 2009). Many studies pointed out that prosodic features such as intonation, pitch, speech rate and stress are not only alternatives to realizing illocutionary acts but also IFIDs (Austin 1962; Searle 1969), which play an essential role in the older adults’ expression of IF.

Given this interrelation, it is quite necessary to investigate how multimodal information has been impacted by speakers’ occurrent emotions, and how the triadic interaction of emotion, prosody, and non-verbal acts demonstrates harmful illocutionary forces.

This study follows the hierarchical method of Gu (2013a) and divides emotional states into three tiers: background emotion, primary emotion, and social emotion. Background emotions hinge on the speaker’s physical conditions, such as well-being, fatigue, anxiety, etc.; primary emotions refer to universal human emotions, including happiness, sadness, fear, and even no emotions; and social emotions are emotions learned socially, which can be divided into two main sources (other-directed and self-directed) and three types (positive, neutral, and negative). This study will annotate the emotions in speech acts based on the classification.

3.1 Fundamental Principle

Primarily, the Discovery Procedure is developed according to the ‘live, whole person’ principle proposed by Gu (2013b), which advocates that cues like prosody, bodily movement, and postures are all indispensable components of the felicitous performance of illocutionary acts in situated discourse. Therefore, the eloquent ‘live, whole person’ in discourse activities constitute the object of Simulative Modelling (Huang 2021b). Accordingly, such cues as prosody, facial expressions, gestures, postures, and bodily movements should be considered.

Under the methodology of Simulative Modelling, illocutionary forces in situated discourse are modelled and investigated through the four perspectives, i.e. what is said, what is thought, what is felt, and what is embodied (STFE Principle).

3.2 Analytical Framework

To analyse the interactive relationships between illocutionary force, emotions, and prosody, Gu (2013a) proposed a ‘Discovery Procedure for the tripartite interaction’. Aiming to improve such procedure of analysis, Huang (2021b) established his refined procedure with the distinctive addition of gestures (including facial expressions, physical attributes, and postures) and situational factors. The Discovery Procedure contains five programme segments, including the conceptual model of illocutionary force, situational factors, emotional states, prosodic features, and gestures. In the following sections, elaboration is given on the basic framework, connotations, and analytic methods of the Discovery Procedure.

1. The conceptual model

   Illocutionary act-tokens in situated discourse are abstracted and generalized from different angles through concept modelling. It should be noted that here we adopt Gu’s (2013a) ‘octet scheme’ to shape our Conceptual Model. Having considered the basic forms of illocutionary force and the relevant characteristics and influential factors of the discourse it situated, the ‘octet scheme’ is represented. The Conceptual Model established by this formalized method describes and defines the characteristics and properties of a certain type of illocutionary force and provides perspectives for investigating the felicity/infelicity of certain illocutionary forces type in situated discourse.

   The Conceptual Model of illocutionary force adopts the ‘octet scheme,’ as shown in Figure 1.

2. Data modelling

   In this study, data modelling aims to prepare the required data types for the study of live illocutionary forces (IFs) from the perspective of the STFE-match principle, i.e., what is said, what is thought of, what is felt, and what is embodied.

   Specifically, the data sets of text layer, audio layer, and video layer were constructed based on AntConc, Praat, and Elan. The implementation and evaluation stage are reflected in the annotator using the above-mentioned software to annotate and present the data of multiple layers according to the dimensions of concept modelling, to check whether the resultant data conform to concept modelling, and thus to lay the data foundation for the subsequent descriptive analysis of speech acts.

   In this corresponding and multiple-tier scheme, eight tiers have been established to annotate speech acts (see Table 1) with the information from different perspectives conforming to our research target. The fundamental purpose of concept modelling and data modelling is to depict and reveal the characteristics and compensatory mechanisms of the language impairment of older adults with AD.

Figure 1:

Conceptual model of illocutionary force.

4.1 Data Sources and Collecting Methods

The corpus was collected in community senior activity centres, senior service centres, community activity centres for senior Communists, and Grade A tertiary hospitals. All participants had signed informed consent for data collection. The life course data were collected using a wireless digital voice recorder (SONY SX2000-A10) with sampling rates of 16 kHz and 44.1 kHz, and a high-sampling rate CNC camera (SONY FDR-AX60) to capture video streams.

This study’s data collection and research scheme were reviewed by the research ethics committee of the author’s institution under the review number tjsflrec202101.

4.2 Data Processing and Annotation

The annotation scheme follows the STFE-match principle (Gu 2013b) and adopts the primary thought of ‘Simulative Modelling’ to compile a multimodal corpus that approximates, as faithfully, as accurately, and as fully as possible, the whole person’s total saturated experience with total saturated signification in situated discourse.

Faced with miscellaneous illocutionary force tokens in situated discourse, Gu (2013a) suggested that they should be classified from the perspective of the relations between ‘what is talked about’ and the speaker/hearer’s interests. In this regard, speech acts are classified according to discourse activities which carry illocutionary forces/acts, following the method of examining speech acts at the level of behaviour activities (discourse activities). Classifying Speech act from a discourse perspective, oriented toward the speech acts in natural discourse, has built a bridge between speech acts and the discourse in which they are located (Adolphs 2008, 47). This study will focus on the indication of harmful illocutionary forces.

In order to ensure the accuracy, objectivity, and consistency of corpus annotation, this study used internal testing (intra-annotator consistency) and external validation (expert reliability and validity tests) to evaluate the quality of annotation. It shows that the intra-annotator consistency is 0.813, the professional verification of validity is 0.820, and the reliability is 0.631, all statistically significant. Overall, the annotation was reliable and valid, which effectively reflected the illocutionary force in Chinese situated discourse.

5.1 The Multimodal Features of the Older Adults with Normal Aging in the Expression of Harmful Illocutionary Force

In 113 tokens of the harmful illocutionary force, the five most frequent tokens are 44 instances of the IF of bao’yuan (complaining), 31 of su’ku (grumbling), 18 of ju’jue (refusing), 8 of fan’si (self-reflection), and 4 of hai’pa (feeling scared).

This section draws on the multimodal nature of IF and the multidimensionality of IFIDs. Such IF expressive devices as emotions, prosodic features and gestures bear different significance and collaborate in the performance of illocutionary acts and then the production of IFs.

5.1.4 Correlation of Emotions, Prosodic Features and Gestures in the Harmful Class

In this section, 113 dummy variables of multimodal features in the harmful classification were input into SPSS for multidimensional scaling analysis to investigate the clustering of multimodal features. The analysis was terminated normally at iteration step 7 with stress = 0.148 and RSQ = 0.927, indicating that the results are ideal.

As we can see from Figure 2, among the tokens of harmful IF, neutral facial expression, low pitch, sitting upright, fast speed pace, stress, even tone, neutral expression and neutral background emotion belong to the same classification (in the first quadrant), reflecting the multimodal features of harmful class. Under the background emotion of “relaxed” or “well-being”, cognitively healthy older adults tend to speak faster and in an even tone. Since the discourse content is against the individual’s interest, the speech is more likely to be presented at a low pitch. In the fourth quadrant, self-directed negative social emotion, long pause, middle pace, falling tone, high pitch, and negative primary emotion are basically grouped into one class. It indicates that due to the influence of negative primary emotion, the speaker’s speech pace changes from fast to middle, the tone changes to a descending manner, and there is a high frequency of pause in the discourse. Low pitch, fast speech pace, and middle pitch, fast speed rate are the most frequent modal types.

Figure 2:

Multidimensional scaling analysis of cognitively healthy older adults’ IF expression in the harmful class.

5.2 The Multimodal Features of the Older Adults with Alzheimer’s Disease in the Expression of Harmful Illocutionary Force

In 103 instances of the harmful IF, the top five most frequent instances are 31 instances of the IF of bao’yuan (complaining), 30 of su’ku (grumbling), 19 of ju’jue (refusing), 7 of hai’pa (feeling scared), and 5 of fan’si (self-reflection).

5.2.4 Correlation of Emotions, Prosodic Features and Gestures in the Harmful Class

As previously discussed, speech acts encompass multimodal behaviors. With the decline of linguistic competence, AD seniors are inclined to ‘compensate’ their pragmatic communication from other resources, including gestures, gaze, body pose, etc. Therefore, examining IFs in older adults with AD from a multimodal perspective and exploring correlations among the multimodal features is crucial.

Figure 3 shows the clustering of the multimodal features of the 103 instances of harmful IFs. This analysis terminated naturally at the 4th iteration with stress = 0.178 and RSQ = 0.882, and the fitting results are rather ideal.

Figure 3:

Multidimensional scaling analysis of AD older adults’ IF expression in the harmful class.

As can be seen in Figure 3, the emotional state of older adults with AD is predominantly negative when expressing harmful IFs, with features such as negative primary emotions and self-directed negative social emotions appearing on the left side of the first modal. Statistically (see Table 4 in Appendix 3), 68.0 % of these 103 utterances convey negative primary emotions and 24.3 % no primary emotions. The background emotions in these utterances are also predominantly neutral (57.3 %). In terms of gestures, 49.5 % of the harmful IFs are expressed with head movements, mostly turning (16.5 %), raising (15.5 %) and nodding heads (12.6 %). Although constant eye contacts appear more often (42.7 %) in terms of older adults with AD’ eye movements, more than a half (56.3 %) of the harmful Ifs are expressed with no and occasional eye contact. The frequency of hand movements is 70.9 %, and most speakers (74.8 %) tend to sit upright when expressing harmful IFs. In terms of prosodic features, fast pace, stress, and even tone are rather clustered in Figure 3, and these are also the most frequently appeared features in the harmful IF of older adults with AD. For example, rapid speech pace appears in 95.1 % of the harmful IFs and even tone 87.4 %.

5.2.5 Case Study (I-Interviewer, P-Participant)

1. I: dan shi ni jia mei mei //hai shi jing chang lai kan ni de //dui ba

   1. But your sister often visits your, right?

2. P: (2s) wo bu yao ta lai //(3s) wo bu yao ta lai //wo yao wo zi ji lao gong lai ya

   1. (2s) I asked her not to come (3s) I don’t want her to come. I want my husband to come.

3. I: o //dan shi ni bu shi li hun le ma

   1. Oh, but didn’t you divorce?

4. P: ((Immediately head up and cast a glance at the older adults beside her in the same room))

   1. he ni shuo guo //jie mei jian bu zhu zai yi kuai de

   2. Told you that sisters don’t live together.

5. P: ni zen me you

   1. How could you

6. P: shuo zhe hua //bu shi shuo guo le ma // jie mei jian bu zhu zai yi kuai de

   1. say that! I had told you! Sisters don’t live together!

7. P: (6s) mei jia ren dou bu zhu zai yi kuai de //zhu yao shi kan fu mu de //you bu shi kan ni de

   1. (6s) Every one doesn’t live together. They mainly come to visit parents, not you.

The participant LMJ (short for Li Mingjia) is a 69-year-old female with AD and a MoCA-B^[3] score of 8. She speaks slowly with a low pitch and repeatedly keeps mute during the interview. According to the entire discourse content and the participant’s background information, the study learns that LMJ faces some family conflicts for divorce, having no kid, and a poor relationship with her little brother and sister. The logical coherence is poor in her pragmatic communication.

1. Emotions

   In this case, LMJ’s emotional state is negative.

   The background emotion is spry, depending on her physical and mental conditions. The primary emotion is raged. The social emotion is resentful. Her negative emotional state appears suddenly and indicates that the AD patient LMJ fails to control her emotions, even though the state aligns with the IF of complaining.

   In the conversation, when the interviewer performs the IF of asking – how could her husband visit her after divorce, LMJ does not use the IF of hui’da (answering) but angrily blames the interviewer for saying that (Line 4–7). It meets the STFE-Match Assumption but lacking logical coherence to align with the forward-and-backward interdependency (FBID).

2. Gestures

   In this example, LMJ performs the gestures of heads-down, frowning and staring (see Figure 4). When hearing the interviewer’s question about whether she has divorced, LMJ immediately heads up and glances at her roommate (see Figure 4) and then expresses the IF of bao’yuan (complaining). The reason may be LMJ’s unwillingness to let her roommate know the interviewer’s repeated content about her divorce.

3. Prosodic features

   Upon conducting a spectrogram analysis with Praat, it becomes evident that LMJ’s speech is characterized by low pitch, slow pacing and even and falling tones.

4. P: he ni shuo guo //jie mei jian bu zhu zai yi kuai de

   Notably, there is a sudden increase in pitch and pace when she articulates the complaining IF.

   The case above demonstrates that cognitive impairment abnormalizes the emotions and attitudes of older adults with AD, making their pragmatic communication peculiar.

Figure 4:

AD-69-F-LMJ’s facial expressions.

5.3 A Comparative Analysis of Multimodal Features in the Expression of Harmful Illocutionary Force Conducted Between the Older Adults with Normal Aging and Alzheimer’s Disease

These instances of IF expressed by people with AD and cognitively healthy older adults are transformed into binary variables and inputted into SPSS as dummy variables for a Chi-square test to explore the differences in the multimodal resources between the two groups.

5.4 Summary

According to statistical data (see Tables 3 and 4 in Appendix), both AD group and healthy group show strong negative primary emotion when expressing harmful IFs, with multimodal features such as sitting upright, stress and even tone occurring frequently. However, as shown in Table 2 participants diagnosed with AD often exhibit less consistent eye contact during interviews than their healthy counterparts. In terms of facial expressions, older adults with AD were more prone to show negative facial expression such as sadness and anger, whereas healthy controls tended to put on neutral faces, expressing tranquillity and calmness. Furthermore, compared to healthy controls, AD older adults exhibited substantially fewer head shaking and forward tilting such multimodal features. When it comes to the prosodic traits observed in elderly individuals with AD, the occurrence of low pitch and pauses is less frequent. Conversely, a rising intonation and a fast speech rate, which are indicative of heightened emotions such as excitement and anger, are more prevalent.

6.1 AD Older Adults Present a Certain Degree of Emotional Apathy and Are Troubled by Affective Disorders and Infelicity Compared with Cognitively Healthy Older Adults

The analysis in Chapter 5 shows that when expressing certain type of IF, older adults in good cognitive health are typically able to convey the corresponding emotions. Statistical data (see Appendix 2 and 3) indicates that the percentage of negative social emotions is notably reduced in the AD group, with other-directed social emotions at 21.4 % and self-directed social emotions at 39.8 %, compared to cognitively healthy older adults where these figures stand at 29.2 % and 41.6 % respectively. In contrast, the AD group exhibits a much higher frequency (35.9 %) of neutral social emotions. This suggests that older adults with AD did not meet the felicity condition when performing speech acts, because they were unable to display the accompanying emotions when expressing IFs with clear emotional inclinations (Austin 1962). According to the classical Speech Act Theory, emotion is the constitutive condition of illocutionary acts and correlating to its appropriateness.

Research has demonstrated that older adults with Alzheimer’s Disease (AD) often suffer from an emotional recognition disorder, i.e., the inability to infer the mental states of others, process information, navigate social contexts, and adjust their behaviour appropriately. Such deficits often lead to adverse outcomes such as social dysfunction and difficulties in grasping the nuances of interpersonal dynamics (Zhang et al. 2020). As AD progresses, afflicted individuals encounter a significant risk, ranging from 35 % to 85 %, of developing neuropsychiatric issues including apathy, depression, anxiety, and agitation. Moreover, emotional apathy is the most predominant symptom in dementia patients, surpassing other conditions (Zhao et al. 2016). Studies additionally point out that older adults with AD particularly struggle in identifying negative emotions like sadness, fear, and anger (Moreira et al. 2021; Mccade et al. 2013), while the recognition of happiness and neutral expressions is less problematic.

It follows that emotional apathy in seniors with AD may substantially contribute to the production of infelicitous illocutionary forces (IFs) in their speech acts. Compared with cognitively healthy older adults, older adults with AD manifest a certain degree of emotional impairment, which may manifest as a reduction or absence of emotion, reduced spontaneous emotion, and a lack of emotional response to positive or negative stimuli. These factors lead to the inconsistency in the emotional state of cognitively impaired older adults when expressing harmful IFs with significant emotional tendency, resulting in infelicitous IFs. Analyses have indicated that, in contrast to cognitively healthy seniors, those with AD exhibit deficiencies in negative emotions when enacting harmful IFs, leading to instances of infelicity.

6.2 AD Older Adults Are Inconsistent in Their Emotions and Facial Expressions when Expressing IFs

The comparative analysis elucidates a paradoxical phenomenon among the older adults suffering from AD: despite their capacity to produce diverse facial expressions, they exhibit a notable degree of emotional apathy. This incongruence between affection and its outward manifestation is symptomatic of a divergence from the STFE-Match Assumption proposed by Gu (2013b). This principle asserts that a ‘whole person’, the agent of real-life discourse activities, has dynamic sounds, emotions, and gestures. If the prosody, facial expressions, movements and other features of external expressions can faithfully reflect the speaker’s inner thoughts and emotions, the speaker follows the STFE-match principle; otherwise, the speaker violates the principle. Deviations may arise from either deliberate rhetorical strategy or from pragmatic impairment. The latter is notably exemplified by the disconnect between emotional intent and its corresponding expressions elucidated in AD patients, who predominantly project a neutral emotion when engaging in harmful IF. This pragmatic decline therefore complicates the interpretation for interlocutors.

6.3 AD Older Adults Are Less Capable of Employing Resources Such as Gestures Compared with Cognitively Healthy Older Adults

Prior research has established that communication inherently incorporates multiple modalities, thereby rendering illocutionary forces (IFs) multimodal and IFIDs multidimensional. Indeed, the harmonious interplay of speech, gestures, and emotions is indispensable for expressing IFs and for recipients’ comprehension of speakers’ intent. It is rather common in clinical studies to examine the importance of non-verbal behaviours for communication. Some studies have shown that people with dementia can demonstrate verbal and gestural disorders associated with their semantic memory deficits. This is in accordance with our findings.

The above analysis shows that cognitively healthy older adults can better employ gestures in their speech. However, there is an observable decline in the use of multimodal features that typically associate with the four established classes of IFs. Conversely, AD patients tend to exhibit more frequent occurrence of non-normative features. Additionally, the quantitative study discloses that elderly individuals with AD utilize hand gestures less frequently, with a rate of 70.9 %, compared to the 73.5 % observed in their cognitively healthy peers. This manifests AD older adults’ deficit in employing bodily modes in communication. This is consonant with the literature, which documents that cognitively compromised individuals struggle to incorporate hand movements in their communication. They produce more ambiguous gestures, fewer metaphorical gestures than gestures referring to specific content, and fewer two-handed gestures expressing complex concepts (Glosser, Wiley, and Barnoski 1998).

Nonetheless, the communicative behavior of AD patients is not homogeneous. In fact, people with neurodegenerative diseases, including aphasia and AD, are known to ‘compensate’ in verbal communication (Perniss 2018). Hamilton (2008) examined the communication of older adults with dementia in terms of gestures, expressions and tones, and concluded that even those with severe dementia retained some capability to convey their pragmatic intents non-verbally, melding speech with prosody, facial expressions, and gestures.

6.4 AD Older Adults Demonstrate Insufficient Ability to Employ Prosody-Assisted Verbal Expression

The prosodic features of AD older adults are also different from cognitively healthy ones, mainly in low scale, high scale, falling tone, convex tune, stress, and pause, etc. A comparative analysis of the four classes of IFs in the two groups reveals that the prosodic features closely related to the corresponding IFs are less frequently observed in AD patients’ speech. For instance, in the harmful class, the speakers’ emotions are often negative, and are usually filled with helplessness, frustration, and grievance. Correspondingly, they shall speak faintly or at a low pitch. However, in the harmful class, moderate and low pitch only account for 49.5 % in AD older adults, which is much lower than the 73.5 % of cognitively healthy older adults. As mentioned, although even tones are the most common tonal pattern (87.4 %), they are less likely to reflect specific emotions; whereas falling tones, which do reflect negative emotions, are less dominant (43.7 %) in the harmful IFs of AD older adults. This corresponds to the emotional apathy of AD older adults, and reflects their deficiency to resort to prosodic resources.