Evaluation of treatment effects of semantic feature analysis on mild anomia in multiple sclerosis

ABSTRACT

Background

Multiple sclerosis (MS) commonly includes anomia and other communicative deficits that affect communicative participation and quality of life. Anomia treatment in MS is currently unexplored. Owing to the degenerative nature of MS, compensatory treatment might be preferable to restorative treatment. Semantic feature analysis (SFA) has been reported to have a treatment effect in aphasia and traumatic brain injury, and it can also be used as a compensatory word-finding strategy. SFA might therefore be effective on anomia in MS.

Aims

The aim of this study was to evaluate the treatment effects of SFA on mild anomia in MS. Hypotheses were that, if used as a strategy, SFA may improve word-retrieval ability in connected speech, reduce self-perceived word-finding difficulties and increase communicative participation.

Methods & Procedures

Two participants (one with relapsing-remitting MS, one with progressive MS) experiencing anomia and reduced communicative participation took part in this early-phase study with a single-case experimental design (SCED). Outcome measures included accuracy and speed in confrontation naming of treated items, correct information units (CIUs) in a re-telling task, self-reported strategy use in everyday communication, self-reported occurrence of anomia and related nuisance, and self-reported communicative participation. Measurements were carried out before treatment, at the beginning of every third treatment session, once directly after treatment, repeatedly during a maintenance phase, and once at a follow-up session ten weeks after treatment.

Outcomes & Results

No treatment effects on confrontation-naming ability, re-telling ability or self-reported measures were evident in either participant after treatment with SFA. Effects of repeated measures appeared as reduced response time in confrontation naming.

Conclusions

SFA as the sole element of treatment may not be sufficient to obtain treatment effects on mild anomia in MS. Further research is warranted.

KEYWORDS:

• Multiple sclerosis
• mild anomia
• word retrieval
• semantic feature analysis

Introduction

Multiple sclerosis (MS), an inflammatory neurodegenerative disease affecting both the grey and the white matter of the brain and the spinal cord, can lead to various physical and cognitive deficits at any stage of the disease (Maloni, 2018; Reich et al., 2018). MS can be classified either as relapsing-remitting (RR) or as progressive, both of which can be either active or not active (Lublin, 2014). The typical age of onset is 20–40 years and initial symptoms commonly involve motor or sensory functions of the limbs, which may or may not persist or worsen. Cognitive abilities often affected in MS are memory, attention, visual perception, executive functions and processing speed (Benedict et al., 2017; Bobholz & Rao, 2003). Additionally, around 45% of individuals with MS experience motor speech deficits (dysarthria), usually ataxic or spastic, affecting voice quality, articulation, prosody and respiration (Noffs et al., 2018). Further, impaired verbal fluency is commonly reported in individuals with MS. This is often viewed as a general cognitive impairment rather than a pure language deficit (De Dios Pérez, 2017; Renauld et al., 2016), but just as dysarthria may affect the speed, timing and duration of verbal responses (Noffs et al., 2018), performance in such assessment tasks is affected by language abilities as well as other cognitive abilities (Whiteside et al., 2016).

The relationships between cognition and speech, between speech and language or between language and cognition in MS are not fully clarified, but a growing body of evidence suggests that deficits in cognition, speech and language co-exist in many individuals with MS (Carotenuto et al., 2018; De Dios Pérez, 2017; De Dios Pérez et al., 2020; Lethlean & Murdoch, 1997; Noffs et al., 2018; Renauld et al., 2016; Sepulcre et al., 2011). Such deficits have often been described as more severe in progressive MS than in RRMS (Friend et al., 1999; Lethlean & Murdoch, 1997).

When asked, people with MS themselves often report various language difficulties and problems participating in communicative situations (El-Wahsh et al., 2020; Johansson et al., 2021; Klugman & Ross, 2002; Laakso et al., 2000; Tallberg & Bergendal, 2009). In Johansson et al. (2021), 78% of respondents reported at least one symptom related to communication. Language difficulties were reported by 75% in El-Wahsh et al. (2020) and by 63% in Klugman and Ross (2002), where 24% stated that their problems were significant. The most commonly reported language problems are word-retrieval difficulties and difficulties staying on topic when speaking (Johansson et al., 2021). Not being able to find the right word or appropriately specific words, or to rapidly and effectively process verbal information, often affects quality of life (El-Wahsh et al., 2020; Klugman & Ross, 2002). In Klugman and Ross (2002), 62% reported that their communication difficulties had a negative impact on their quality of life. Individuals with MS who experienced language difficulties had significantly lower health-related quality of life than those without self-reported language impairments in El-Wahsh et al. (2020).

While people with MS themselves report experiencing language problems in everyday life, established formal measures often fail to capture those problems. A systematic review of language impairments in MS as reflected by objective measures (Renauld et al., 2016) concluded that such impairments do occur, but not very commonly. Other studies have found language impairments to occur frequently in MS; most commonly word-retrieval difficulties (anomia), but also impairment to various other aspects of language use such as comprehension, lexical access, word definitions, confrontation-naming ability, verbal explanation, verbal reasoning, sentence repetition, narrative ability and pragmatic ability (Arnott et al., 1997; Arrondo et al., 2010; Carotenuto et al., 2018; De Dios Pérez, 2017; De Dios Pérez et al., 2020; Friend et al., 1999; Henry & Beatty, 2006; Klugman & Ross, 2002; Sepulcre et al., 2011).

However, many people with MS perform within the normal range on assessment tasks commonly used to detect impaired language abilities in conditions such as aphasia (De Dios Pérez, 2017; Klugman & Ross, 2002; Laakso et al., 2000; Lethlean & Murdoch, 1997; Renauld et al., 2016; Tallberg & Bergendal, 2009; Yorkston et al., 2014). Compared with the mobility or motor speech impairment that MS often causes, its effects on language and communication may seem more subtle and so may be ignored or go undetected. Part of the reason for the frequent failure to recognise language impairments may also be a lack of sufficiently sensitive assessment methods.

As mentioned above, anomia is the most common language symptom in MS, with a prevalence of over 60%, both as reported by individuals with MS themselves (El-Wahsh et al., 2020; Johansson et al., 2021; Klugman & Ross, 2002) and as described in studies using formal language measures (Renauld et al., 2016). When comparing confrontation naming abilities in 31 individuals with RRMS to a healthy control group, Kambanaros et al. (2017) found that word retrieval difficulties in MS can affect both nouns and, to an even greater extent, verbs. From a theoretical perspective, anomia may be assumed to occur when one or more underlying word-retrieval sub-processes are disrupted. According to an interactive-activation model of word retrieval (Dell, 1986), constant interaction between semantic, lexical and phonological information entails that, when a stimulus is presented, several word forms with shared semantic and phonological features are activated at the same time. The word form with the highest level of activation is retrieved, but this competing activation can sometimes lead to inaccurate naming (Dell, 1986). Analysis of errors made in confrontation-naming tasks has revealed that, compared with neurotypical subjects, people with MS typically make more semantic errors (for example, word substitutions such as “goat” for the target word “horse”), and give less specific responses (“animal” for “beaver”) or more off-target responses (“worm” for “pretzel”), and that latency or “no verbal response” is fairly common in MS (De Dios Pérez, 2017; De Dios Pérez et al., 2020; Tallberg & Bergendal, 2009).

Efficient word retrieval requires both accuracy and speed in production. In formal tests, such as confrontation-naming tests, anomic difficulties in MS may manifest themselves both as inaccuracy and as response latency (De Dios Pérez et al., 2020). It has been proposed that speed (measured as response time) is more sensitive in detecting mild anomia than accuracy (De Dios Pérez, 2017; De Dios Pérez et al., 2020; Galletta & Goral, 2018). Neurotypical subjects typically respond within a second, and latencies above two seconds are commonly viewed as reflecting some kind of word-retrieval difficulty (De Dios Pérez et al., 2020; Galletta & Goral, 2018; Pashek & Tompkins, 2002).

Various verbal-fluency tasks are commonly used with people with MS (Henry & Beatty, 2006). They are considered to be reliable, sensitive, easily administered and quick neuropsychological assessment tools, and they sometimes – though not always and not fully – distinguish individuals with MS from healthy controls (Henry & Beatty, 2006; Vlaar & Wade, 2003). However, other, more complex tasks may be more sensitive in revealing subtle language difficulties, including those measuring verbal fluency with multiple semantic constraints (for example, requiring words to combine two specified features, such as “square and hard”) (Bauer & Saldert, 2020). Further, the analysis of connected speech may yield more information about language performance than tasks demanding only single-word responses (Fergadiotis & Wright, 2016). For example, producing a narrative requires the use of high-level cognitive and linguistic skills to plan and convey an informative, correct and unambiguous message (Arnott et al., 1997). Whatever the method used, however, it remains a challenge to assess mild or subtle language difficulties in a satisfactory manner (De Dios Pérez, 2017; Laakso et al., 2000; Tallberg & Bergendal, 2009). Moreover, it is important to keep in mind that there can be a considerable discrepancy between assessment scores and a person’s own subjective experience of his or her difficulties (Klugman & Ross, 2002).

Since communication deficits are frequent in MS and may have a negative impact on communicative participation and health-related quality of life, it is important that abilities significant for communication be addressed in assessment and treatment (De Dios Pérez, 2017; De Dios Pérez et al., 2020). The literature on possible treatments for language and communication difficulties in MS is limited. Pharmacological treatments have been reported to have some pro-cognitive effects on phonological fluency (Magnin et al., 2015) as well as on executive functions or other cognitive abilities closely connected to communication skills (DeLuca et al., 2020; DeSousa et al., 2002). However, the field of effective speech and language treatment methods for verbal-production deficits, such as anomia, in MS remains unexplored (De Dios Pérez, 2017; De Dios Pérez et al., 2020). It has been proposed that a compensatory treatment might be preferable to a restorative one, considering the fluctuating or degenerative course of disease in MS (De Dios Pérez, 2017). A strategy automatised at a relatively early stage could presumably be easily adapted if difficulties were to worsen.

One anomia-treatment method that has been proposed for use as a general compensatory word-finding strategy is semantic feature analysis (SFA) (Boyle, 2004). In SFA, the participant is repeatedly asked to name semantic features in a structured way, meaning that SFA can potentially be utilised as a self-cueing strategy. SFA was originally developed as a method for facilitating word retrieval and organising verbal expression in people with traumatic brain injury (TBI) (Massaro & Tompkins, 1994) and in people with aphasia after stroke; it works by promoting activation and strengthening connections in semantic networks (Boyle & Coehlo, 1995). It has been proposed that people with MS have preserved semantic-lexical information but fail to access and retrieve the information stored (Murdoch & Theodoros, 2000; Sepulcre et al., 2011; Tallberg & Bergendal, 2009). If so, their situation is similar to that of people with anomia due to TBI or stroke. Presumably, MS plaques or inflammatory activity in cortical areas important in language processing, or in the white matter paths connecting them, can cause communication difficulties much like the diffuse or focal cortical lesions in TBI or stroke (De Dios Pérez et al., 2020; Friend et al., 1999; Laakso et al., 2000; Lethlean & Murdoch, 1997). SFA has yielded promising treatment effects on anomia in people with aphasia after stroke and in people with TBI (Efstratiadou et al., 2018; Massaro & Tompkins, 1994). In Efstratiadou et al. (2018), 81% of participants with anomic aphasia were observed to have treatment effects from SFA regarding trained items, 50% showed maintenance treatment effects and 17% showed generalisation effects to untrained items. Given these findings, SFA could be expected to have treatment effects in participants with MS who manifest anomic symptoms. However, to the best of our knowledge, the issue of treatment effects from SFA has not yet been explored in individuals with MS.

In summary, communicative deficits, especially word-finding difficulties, are common in MS and, although often subtle and difficult to measure, they may have a negative impact on a person’s communicative participation and quality of life. We currently lack effective evidence-based treatment for anomia in MS. SFA might be suitable for this purpose, functioning as a compensatory strategy and also potentially strengthening connections in the semantic-lexical network. Therefore, the aim of this early-phase single-case experimental-design (SCED) study is to evaluate the treatment effects of SFA on mild anomia in MS. The hypotheses are that SFA treatment, especially if used as a word-finding strategy, may improve word-retrieval ability in connected speech. Further, improved word retrieval may reduce the self-perceived occurrence and nuisance of word-finding difficulties and increase communicative participation.

The specific research questions are:

Is semantic feature analysis treatment in multiple sclerosis effective in:

• Improving confrontation naming of treated items?

• Increasing the proportion of correct information units in a re-telling task?

• Providing those treated with a useful word-finding strategy?

• Decreasing the self-reported occurrence and nuisance of word-finding difficulties?

• Increasing self-reported communicative participation?

Materials and methods

Study design

This early-phase study used a single-case experimental design to explore treatment effects of semantic feature analysis (SFA) in two participants with multiple sclerosis (MS) who experienced anomia. A staggered baseline was followed by a treatment period including 20 individual one-hour sessions, with outcome measures being assessed at the beginning of every third session. Treatment outcome was assessed using repeated measures of accuracy and speed in confrontation naming, information conveyance in a re-telling task and daily self-reported occurrence of anomia and related nuisance (see section on outcome measures below). The participants also reported on whether they used SFA as a strategy when experiencing word-finding difficulties and whether they found it useful for that purpose, as well as on their self-perceived communicative participation. During a maintenance phase after treatment, five assessment sessions were performed, 7–14 days apart. A follow-up session took place ten weeks after the end of treatment. See Figure 1 for an overview of the study procedure.

Figure 1. Overview of the study procedure.

Participants

Two participants took part in the present study. Three participants were initially included but one of them dropped out after ten training sessions. The participants were randomly selected from among seven eligible participants who desired to participate. All participants fulfilled the following criteria: (1) diagnosed with MS; (2) absence of other known neurological illness; (3) no indication of dementia in medical records; (4) subjectively experiencing anomia several times a week; (5) multiple instances of inaccuracy and/or response latency above three seconds in screening tasks of confrontation naming; (6) Swedish as native language; (7) unimpaired or only mildly impaired auditory comprehension; (8) unimpaired or corrected hearing and vision; and (9) absence of dysarthria severely affecting intelligibility.

To obtain a random selection, the seven potential participants were randomly assigned a number from 1 to 7. Following the order of 1 to 7, telephone contact was attempted with the potential participants. Three contact attempts were made with one person on different days and at different times of day before moving on to the next person. Before moving on to the third person, one further attempt was made to contact the first person, and so on. The first three people who answered were offered to take part in the present study. All three accepted. No further attempts were made to contact the remaining potential participants, but they were notified by letter that they would not be included in the study.

Before inclusion in the study, the participants were screened for dysarthria (Hartelius, 2015). Confrontation-naming ability was assessed using 120 pictures of high-frequency items (60 objects and 60 actions) from the Object and Action Naming Battery (OANB) (Druks & Masterson, 2000) and 20 low-frequency items (objects) from the Boston Naming Test (BNT) (Kaplan et al., 1983), chosen from the highest level of difficulty according to Swedish norms (Tallberg, 2005). Phonological and semantic verbal fluency was assessed, along with complex oral semantic fluency. In the phonological (F, A, S) and semantic (verbs and animals) fluency tasks, the participants were asked to name as many words as possible in one minute for each of the features of the task in question. The fluency tasks were administered and scored in accordance with Tallberg et al. (2008). The complex oral semantic fluency task (COSEF) was administered and scored in accordance with Bauer and Saldert (2020), meaning that the participants were asked to name as many words as possible that share two different features (round and flat, and sharp and long) in two minutes each. Further, the participants were asked to estimate how often they experienced word-finding difficulties on an 8-point scale, where 0 = Not at all; 1 = Less than once a week; 2 = At least once a week; 3 = At least three times a week; 4 = At least five times a week; 5 = At least once a day; 6 = Several times a day; and 7 = Every time I try to say something. Additionally, the participants were asked to fill out the Sahlgrenska Academy Self-reported Cognitive Impairment Questionnaire (SASCI-Q) (Eckerström et al., 2013), which contains (i) questions about 29 everyday activities potentially affected by cognitive impairment, (0 = No or rare difficulties and 1 = Frequent difficulties); and (ii) eight questions which compare the current level of cognitive function to previous levels (ten years ago and at the age of 25), where positive differences correspond to self-reported reductions in the level of functioning.

Neither participant in the present study reported any worsening of MS symptoms during the course of the study. Their language difficulties mostly resembled those observed in mild anomic aphasia. Both participants obtained high scores on confrontation-naming tasks in terms of accuracy, but both had several instances of response latency exceeding three seconds (see Table 1). Considering that response latencies above two seconds indicate word-retrieval difficulties (De Dios Pérez, 2017; De Dios Pérez et al., 2020; Galletta & Goral, 2018; Pashek & Tompkins, 2002), latencies above three seconds were considered to be signs of marked word-finding difficulties.

Table 1. Participants

	P1	P2
Age	40	74
Gender	M	F
Education (years)	14	17
Occupation	Student and assistant	Retired analyst and educator
Diagnosis	RRMS not active	Progressive MS active
Years since diagnosis	20	28
Years since symptom onset	27	44
Pharmacological treatment	Natalizumab	None
EDSS^a	6.0	7.0
Ambulation	Cane	Wheelchair
Vision	Glasses	Glasses
Hearing	Unimpaired	Unilateral impaired, corrected
Naming ability
Accuracy
high-frequency items^b (120) low-frequency items^c (20)	113 (94.2%) 14 (70%)	119 (99.2%) 17 (80%)
Response time in seconds (mean)
high-frequency items low-frequency items	2.97 3.77	1.39 3.86
Instances of latency > 3 seconds (out of 140)	55	13
Verbal fluency pre-intervention
Phonological (FAS) Semantic (animals) Semantic (verbs) Complex semantic oral fluency (COSEF)^d	20 14 12 6	32 19 17 13
Self-reports pre-intervention occurrence of anomia (0–7)^e	3	4
communicative participation (CPIB 0–30)^f	19	7
cognitive impairment (SASCI-Q^g):
SASCI-Q^g total^h (0–29)	6	24
SASCI-Q^g difference^i (−96 to +96)	2.5	31

^aExpanded disability status scale (Kurtzke, 1983).

^bFrom The Object and Action Naming Battery (OANB) (Druks & Masterson, 2000).

^cFrom the Boston Naming Test (BNT) (Kaplan et al., 1983), digital pictures courtesy of Pro-Ed.

^dComplex oral semantic fluency task, COSEF (Bauer & Saldert, 2020).

^eOccurrence of anomia scale ranging from 0 = ”Not at all” to 7 = “Every time I try to say something”.

^f10-item General Short Form of the Communicative Participation Item Bank (CPIB) (Baylor et al., 2013); low ratings correspond to a large negative impact on communicative participation.

^gSahlgrenska Academy Self-reported Cognitive Impairment Questionnaire (Eckerström et al., 2013).

^hLow ratings correspond to less severe impairment (Eckerström et al., 2013)

ⁱPositive differences correspond to a reduction of functioning level as compared with earlier in life (Eckerström et al., 2013).

According to their self-reports, both participants were bothered by anomia several times a week, and they both experienced reduced communicative participation due to their condition. See Table 1 for demographic data, pre-inclusion results on confrontation-naming and verbal-fluency tasks, and self-reported ratings.

Ethical considerations

Ethical approval for the project was obtained from the Regional Ethical Review Board of Gothenburg, Sweden (case number 506–16). Participants gave their written consent to participate.

Treatment procedure and material

Individual semantic feature analysis (SFA) treatment sessions of approximately one hour were scheduled twice a week for ten weeks. On occasion, treatment occurred once a week or three times a week, as required by the participant. In particular, Participant 1 (P1) missed several appointments, which caused the treatment period to be prolonged. Participant 2 (P2) carried through as planned. The intervention was commenced after an acceptably stable baseline (± 5% variation) had been achieved (five measure points for P1 and seven measure points for P2) using online assessment of accuracy in confrontation naming of a set of 48 untreated words, further described under “Control measures” below.

The treatment sessions followed the SFA protocols for nouns (Boyle & Coehlo, 1995) and verbs (Wambaugh & Ferguson, 2007). Hence the participant was first presented with a chart containing six boxes for object words and six boxes for action words. Each box had a heading indicating the requested feature. For objects, the features were Category, Use, Action, Properties, Location and Association; and for actions they were Subject, Purpose, Body part or tool used, Description, Location and Association. Then a picture was placed at the centre of the chart and the participant was asked, “What is this?” (for objects) or “What is he/she doing?” (for actions). If a participant was unable to name the picture, the treating speech and language therapist (SLT), i.e., the first author, named it and asked the participant to repeat the word. Regardless of whether the participant was able to name the picture correctly, he or she was asked to name the corresponding semantic features in the set order. The agreed-upon features were written in the corresponding boxes by the SLT. When all features had been produced, the participant was asked to summarise, for instance, by saying, “This is a vehicle with four wheels, a steering wheel and a back seat, it rolls and it is used for transport. It can be found in garages and I am thinking of driving to a campsite with our caravan. It is a car.” These summaries were intended to promote the acquisition of a strategy for word retrieval by keeping the participant active and by adding an element of utilising detailed connected speech to the treatment. At the end of each treatment session, the participant was asked to name all the pictures that had been used in that session again.

Each participant chose his or her own treatment material from Papunet’s picture bank (Papunet, 2018). They chose words that they considered relevant and useful in everyday communication, but not necessarily difficult to name. One treatment set, referred to below as “treated items”, comprising ten pictures of objects and ten of actions, was used repeatedly during treatment. Each item in that treatment set was trained five to ten times in total during the treatment period. Additional material from Papunet’s picture bank (Papunet, 2018) was used in the sessions until time was up. That additional material allowed the SFA method to be used on an extensive variety of items. This was important given that the treatment focused on implementing SFA as a general word-retrieval strategy rather than on improving the naming of specific items. All additional items were different from the pictures included in the assessment materials. The treatment items are listed in the Appendix.

Assessment of outcome, procedure and material

Outcome measures included (a) accuracy and speed in confrontation naming of treated items; (b) correct information units in a re-telling task; (c) self-reported use and usefulness of SFA as a communicative strategy; (d) self-reported word-finding difficulties and related nuisance; and (e) self-perceived communicative participation. It was decided not to have separate assessment sessions but to integrate assessment into treatment sessions in order to reduce the number of appointments for the participants. Assessment took 10–20 minutes and the subsequent treatment session took 45–60 minutes. The assessments were video- and audio-recorded for subsequent scoring and further analysis, and transcriptions were made verbatim. Scoring was performed by a member of the research group (certified SLT) who was blind as to the phase to which the recordings and transcriptions pertained. Details on each outcome measure are presented below.

Confrontation-naming ability

Confrontation-naming ability in relation to the treated items was assessed every third treatment session. Each participant used an individual treatment set comprising ten objects and ten actions. All confrontation-naming tasks were presented on a computer screen as separate slide shows with a click sound accompanying each transition from one slide to the next. The slide show automatically changed slides every ten seconds, but the test leader changed slides manually if the participant gave a final answer before the ten seconds were up. No feedback was provided during the assessment. All answers were scored for accuracy and an error analysis was conducted for all incorrect answers. The response time was measured from the click sound to the beginning of each correct answer.

Re-telling ability

It was anticipated that there would be a ceiling effect for accuracy in confrontation-naming ability, given participants’ initial performance on such tasks. For this reason, a measure of word-retrieval ability in connected speech was collected, to capture changes caused by the use of SFA as a communicative strategy. At each assessment point, the participants were asked to describe the plot in animated Lineman cartoons (Cavandoli, 2007; courtesy of Quipos©). In the cartoons, a stick figure (the Lineman) discovers various objects and performs various activities, mostly using nonsense verbal expressions only. To avoid a learning effect from repeated measures, several similar sets of cartoon clips were used. Each set comprised three different episodes of coherent stories, divided into ten-second sequences in order to minimise the memory burden. Every ten seconds, the cartoon was paused and the participant was asked to re-tell as much as possible of the plot to an imaginary naïve listener. Transcriptions of the narratives were scored using analysis in terms of correct information units (CIUs) (Nicholas & Brookshire, 1993). Any words that are intelligible, accurate, relevant and informative about the topic and content are considered to be CIUs. The combination of the three measures of “words per minute”, “correct information units per minute” and “percent correct information units” (%CIU) is considered to reflect the efficiency and informativeness of connected speech (Nicholas & Brookshire, 1993).

Self-reports

The participants were asked for several self-reports. On a weekly basis, starting after the first treatment session, they were asked to report on their use of the strategies trained by answering the question, “Did you use the strategies from the SFA treatment this week?”, where 0 = No, not at all; 1 = Yes, a little, on occasion; 2 = Yes, a few times; 3 = Yes, often; 4 = Yes, on numerous occasions; and X = I did not experience anomia this week and so did not use the strategies. They also answered the question, “Did the strategy work this week?”, where 0 = No, not at all; 1 = Yes, a little, on occasion; 2 = Yes, a few times; 3 = Yes, often; 4 = Yes, fairly consistently; and X = I did not use the strategies this week.

In daily (seven days a week) mobile-phone text messages, the participants reported on the occurrence of word-finding difficulties and the related nuisance by answering two questions: “Did you experience any anomia today?”, where 0 = No, not at all; 1 = Yes, a little, on occasion; 2 = Yes, a few times; 3 = Yes, often; and 4 = Yes, on numerous occasions; and “Were you bothered by the anomia today?”, where 0 = No, not at all; 1 = Yes, somewhat; 2 = Yes, moderately; 3 = Yes, quite a bit; and 4 = Yes, significantly.

Additionally, the participants were asked once before the intervention (A1), once after the intervention (A2) and once at the follow-up ten weeks after the intervention (A3) to score how much their condition interfered with communicative participation in the ten-item General Short Form of the Communicative Participation Item Bank (CPIB) (Baylor et al., 2013), on a scale where 3 = Not at all; 2 = A little; 1 = Quite a bit; and 0 = Very much.

Control measures

In addition to the outcome measures, various control measures were assessed to control for possible confounding factors affecting results.

Two control tasks requiring various executive functions and verbal responses were administered in every assessment session. (1) The number-repetition task from CELF-4 (Semel et al., 2003), assessing working memory, was used to detect any major changes in overall performance due to participants’ state of health or to worsening of their disease. (2) The verbal-fluency tasks (FAS, Animals, Verbs and COSEF) described above under “Participants” were used to control for effects from repeated measures.

Further, two sets of items from the same picture bank as the treated items (Papunet, 2018) were used as controls to identify possible effects from repeated measures. (1) One picture set for each participant, consisting of ten objects and ten actions, matched as far as possible in terms of frequency of occurrence, word length and semantic properties to the twenty items used in treatment, was assessed at A1, A2 and A3. These items were untrained and the participants were not exposed to the pictures on any other occasion. (2) One set of 48 untrained items (24 objects and 24 actions) was assessed as frequently as the treated items but not used in treatment. In assessment, the set of treated items (ten objects and ten actions) was presented first, then the 24 untreated objects and finally the 24 untreated actions.

Statistical analysis

Treatment effects were analysed using the percentage of non-overlapping data (PND) by dividing the number of post-baseline data points above the highest point (or below the lowest point, for response time) in the baseline phase with the total number of post-baseline data points (Scruggs et al., 1987). Treatment can be considered very effective when PND ≥ 90%, effective when PND = 71%–89% and ineffective or questionable when PND ≤ 70% (Scruggs et al., 1988).

Results

No clinically significant changes were evident in either participant for any of the outcome measures (confrontation-naming ability, re-telling ability and self-reports) after treatment with SFA. Both participants were able to participate in the treatment without any apparent difficulties. After only a couple of sessions, they were able to complete the tasks quite independently. They typically suggested relevant semantic features, for both objects and actions. However, none of them seemed to reflect on the strategy itself, and they did not spontaneously use the strategy in a noticeable fashion.

Scores on the control task of number repetition were quite stable throughout the study, indicating that no major changes occurred in overall performance, state of health or disease severity. Effects from repeated measures were present in confrontation-naming response times as well as in the verbal-fluency control tasks. Each outcome will be presented separately below.

Confrontation-naming ability

Both participants had a consistently high level of accuracy in the confrontation-naming tasks. Only a few errors were made by each participant. An error analysis revealed semantic paraphasias and circumlocutions in all cases except for one visual misinterpretation, and one instance of non-response. This is well in line with previous findings on type of errors produced in MS (De Dios Pérez, 2017; De Dios Pérez et al., 2020; Tallberg & Bergendal, 2009). Figure 2 shows the results for the repeatedly measured treated items alongside those for the control task involving repeatedly measured untreated items. PND for accuracy in treated items was 0% for both participants, indicating that no treatment effects on accuracy in confrontation naming was evident. By contrast, accuracy for treated items decreased somewhat during treatment in the case of P1. This was due to a few paraphrasing answers, such as “time for coffee” for “clock”. Note that, because only twenty treated items were used, a couple of incorrect answers resulted in a 10% difference. Hence due caution should be exercised when interpreting the results. For P2, variability in accuracy stabilised during treatment at consistent top scores for treated and untreated items alike.

Figure 2. Accuracy and speed in repeatedly measured confrontation naming of treated and untreated items.

Response times varied greatly for both participants, for both objects and actions, as is evident in the broad ranges. Response time did not decrease significantly after treatment, but it did decrease progressively for P2, starting during the baseline phase. This decrease in response time was more evident for repeatedly measured items (both treated and untreated) than for the untreated control measures which were not repeatedly measured; see Figure 2 and Table 2. P1 had a PND of 0% for response time in the treated items while the corresponding value for P2 was 69%.

Table 2. Accuracy and speed in the confrontation-naming tasks

	P1			P2
	A1 Baseline	A2 Directly after	A3 Follow-up	A1 Baseline	A2 Directly after	A3 Follow-up
Treated items total (0–20)	19 (95%)	16 (80%)	18 (90%)	17 (85%)	20 (100%)	19 (95%)
Response time mean (range)	2.05 (1.0–6.2)	2.32 (0.8–6.2)	1.97 (0.9–5.2)	2.78 (0.9–7.7)	1.78 (0.7–6.3)	1.47 (0.8–6.2)
SD	1.32	1.69	1.00	2.22	1.39	1.34
Actions (0–10)	10 (100%)	8 (80%)	9 (90%)	8 (80%)	10 (100%)	10 (100%)
Response time mean (range)	1.97 (1.2–4.1)	3.13 (1.1–6.2)	2.24 (1.3–5.2)	3.65 (1.1–7.7)	1.53 (0.9–3.2)	1.04 (0.8–1.7)
SD	0.90	2.02	1.21	2.50	0.89	0.31
Objects (0–10)	9 (90%)	8 (80%)	9 (90%)	9 (90%)	10 (100%)	9 (90%)
Response time mean (range)	2.14 (1.0–6.2)	1.51 (0.8–2.6)	1.7 (0.9–3.2)	2.01 (0.9–6.3)	2.02 (0.7–6.3)	1.94 (0.8–6.2)
SD	1.72	0.73	0.72	1.73	1.77	1.86
Untreated control items total (0–20)	18 (90%)	19 (95%)	19 (95%)	18 (90%)	20 (100%)	19 (95%)
Response time mean (range)	2.81 (0.7–7.5)	3.17 (0.9–7.8)	3.33 (1.1–9.4)	1.61 (0.7–4.8)	1.48 (0.9–4.0)	1.63 (0.7–8.3)
SD	1.83	1.81	2.34	1.06	1.03	1.71
Actions (0–10)	9 (90%)	9 (90%)	9 (90%)	8 (80%)	10 (100%)	9 (90%)
Response time mean (range)	3.33 (0.9–7.5)	3.71 (1.4–6.3)	3.31 (1.2–9.4)	2.2 (1.0–4.8)	1.9 (0.7–1.8)	1.16 (0.7–1.8)
SD	2.20	1.43	2.81	1.28	0.37	0.39
Objects (0–10)	9 (90%)	10 (100%)	10 (100%)	10 (100%)	10 (100%)	10 (100%)
Response time mean (range)	2.28 (0.7–5.0)	2.69 (0.9–7.8)	3.34 (1.1–7.3)	1.14 (0.7–2.5)	1.05 (0.7–4.0)	2.05 (0.8–8.3)
SD	1.28	2.04	1.99	0.53	0.87	2.30
Untreated repeatedly measured control items total (0–48)	43 (90%)	45 (94%)	45 (94%)	43 (90%)	47 (98%)	48 (100%)
Response time mean (range)	2.41 (0.7–8.2)	2.36 (0.8–6.5)	2.48 (0.9–8.2)	1.97 (0.7–9.4)	1.16 (0.5–4.6)	1.12 (0.6–3.6)
SD	1.99	1.49	1.64	1.61	0.65	0.54
Actions (0–24)	21 (88%)	22 (92%)	22 (92%)	20 (83%)	24 (100%)	24 (100%)
Response time mean (range)	2.97 (1.1–7.8)	2.59 (0.8–5.7)	2.33 (0.9–8.2)	2.32 (1.0–9.4)	1.28 (0.6–4.6)	0.93 (0.6–1.7)
SD	2.13	1.60	1.72	2.06	0.85	0.25
Objects (0–24)	22 (92%)	23 (96%)	23 (96%)	23 (96%)	23 (96%)	24 (100%)
Response time mean (range)	1.87 (0.7–8.2)	2.14 (1.0–6.5)	2.62 (1.0–6.3)	1.67 (0.7–4.0)	1.03 (0.5–1.6)	1.31 (0.6–3.6)
SD	1.72	1.37	1.59	1.04	0.31	0.67

Mean response time, for correct answers only, presented in seconds.

Re-telling ability

The mean percentage of correct information units (%CIU) at baseline in the re-telling task was 50% for P1 and 56% for P2; see Figure 3. In a still unpublished study using the same Lineman material as in the present study, neurotypical subjects obtained a mean score of 72% and manifested a range of 52–90% (De Geer et al., unpublished) .

Figure 3. On the left: number of words per minute and number of correct information units (CIUs) per minute in the re-telling task. On the right: percentage of CIUs (%CIU) in the re-telling task.

The participants’ re-telling ability was quite stable throughout the present study. No clinically significant post-intervention change was evident either in words per minute or in CIUs per minute. P1’s mean performance at baseline was 113.1 words/minute and 57.6 CIUs/minute, as against 116.6 words/minute (PND 58%) and 58.3 CIUs/minute (PND 33%) after the intervention. P2’s mean performance at baseline was 156.3 words/minute and 87.1 CIUs/minute, as against 165.3 words/minute (PND 25%) and 89.4 CIUs/minute (PND 0%) after the intervention. Similarly, the percentage of correct information units (%CIU) did not change significantly; PND values were 8% for P1 and 0% for P2. See Figure 3.

Self-reports

Self-reported strategy use and usefulness

Both participants reported scarce use of the strategy of specifying semantic features, learnt during the SFA treatment. Moreover, they experienced limited benefit from using the strategy when they had word-finding difficulties. The weekly reports from both participants varied between 1 (Yes, a little, on occasion) and 2 (Yes, a few times) for both use and usefulness. P1 reported 0 (no use) in the first week and hence no rating for usefulness. Likewise, P2 reported no use in the second week. See Figure 4. Anecdotally, the participants spontaneously reported that they had experienced themselves as more structured when speaking after the treatment, i.e., better able to convey the most important and relevant facts in a given situation.

Figure 4. Self-reports. On top: means and range values per week of the daily ratings of occurrence of anomia and related nuisance (scale: 0 “Not at all” to 4 “On numerous occasions”/“Significantly”). At bottom: weekly self-reported use and usefulness of the SFA strategy (scale: 0 “Not at all” to 4 “On numerous occasions”/“Fairly consistently”), collected in assessment sessions after the baseline (B), i.e., in the intervention phase (I), directly after the intervention (After), in the maintenance phase (M) and at the follow-up session.

Self-reported occurrence of word-finding difficulties and related nuisance

The participants were asked to rate whether they had experienced any anomia, using daily mobile-phone text messages, on a scale where 0 = No, not at all; 1 = Yes, a little, on occasion; 2 = Yes, a few times; 3 = Yes, often; and 4 = Yes, on numerous occasions. They also rated whether they had been bothered by anomia on that day (0 = No, not at all; 1 = Yes, somewhat; 2 = Yes, moderately; 3 = Yes, quite a bit; and 4 = Yes, significantly). In Figure 4, the contents of the daily reports are presented as weekly mean values and ranges for the rated occurrences of anomia and anomia-related nuisance. As can be seen there, the self-reported occurrence of anomia varied only slightly throughout the study (PND 0%), for both participants, and so did their self-reported anomia-related nuisance.

P1 almost every day reported anomia occurring “a little, on occasion” (1 on the scale) or “a few times a day” (2). He never reported “not at all” (0), but once on a day in week 18 he reported frequent anomia (3). Similarly, he typically reported being “somewhat” (1) or “moderately” (2) bothered by the anomia every day, except for “quite a bit” (3) on one day in each of weeks 6, 15 and 25. However, it should be noted that on each of those days he reported only anomia occurring “on occasion” (1). Similarly, on the day when he reported frequent anomia (3), he reported only being “somewhat” bothered by it (1).

P2’s reports on occurrence and nuisance were somewhat disparate to begin with but conformed better to each other from week 10 onwards. P2 reported anomia occurring “a few times a day” (2 on the scale) every day except for twelve days when she reported anomia occurring “a little, on occasion” (1). Once, in the first week, she reported frequent anomia (3). During that first week, her rated anomia-related nuisance varied between “not at all” bothered (0 on the scale; one day only) and “moderately” bothered (2). Throughout the remaining period, she reported being either “somewhat” (1) or “moderately” (2) bothered by the anomia.

Self-reported communicative participation

Initially, both participants reported restricted communicative participation due to their condition; see Table 3. In particular, P2 experienced a large negative impact in most communicative situations listed in the CPIB (Baylor et al., 2013). P1’s and P2’s reports on communicative participation varied only slightly between A1 (baseline), A2 (post-intervention) and A3 (follow-up), indicating that there were no treatment effects. Compared with A1, P1 reported a slightly larger negative impact on communicative participation, down by 4 points (13%) at A2 and down by 3 points (10%) at A3. P2 reported a somewhat smaller negative impact at A2 (up by 3 points or 10%) but a larger negative impact at A3 (down by 3 points or 10%) compared with her rating at A1.

Table 3. Self-reported communicative participation

		P1			P2
		A1 Baseline	A2 Directly after	A3 Follow-up	A1 Baseline	A2 Directly after	A3 Follow-up
CPIB^a total^b (0–30)		19	15	16	7	10	4
theta^b (−3.0 to +3.0)		0.15	−0.33	−0.22	−1.22	−0.89	−1.6
T-score		51.5	46.7	47.8	37.8	41.1	34

^a10-item General Short Form of the Communicative Participation Item Bank (CPIB) (Baylor et al., 2013).

^bHigher ratings correspond to a less negative impact on communicative participation (Baylor et al., 2013).

Control tasks

Scores on the number-repetition control task remained stable throughout the study, indicating that no major changes occurred in overall performance, state of health or disease severity. See Figure 5.

Figure 5. Scores on the number-repetition control task.

As can be seen in Figure 6, verbal-fluency scores increased slightly throughout the study, starting during the baseline phase for P2. P1 reached the Swedish norms for COSEF (19.46 ± 5.55) (Bauer & Saldert, 2020) but not those for FAS (47.1 ± 13.3), Animals (26.0 ± 5.3) or Verbs (21.9 ± 6.4) (Tallberg et al., 2008). P2 reached and exceeded the Swedish norms for all verbal-fluency tasks (FAS 48.2 ± 9.0, Animals 20.4 ± 4.0, Verbs 20.8 ± 5.4 and COSEF 16.73 ± 3.33). Note that the norms are stratified for gender, age and level of education, meaning that they are different for P1 and P2.

Figure 6. Number of correct answers in the verbal-fluency control tasks.

Discussion

People with MS commonly experience anomia, which may both reduce their communicative participation and lower their quality of life even when the symptoms are considered mild in an SLT assessment (El-Wahsh et al., 2020; Klugman & Ross, 2002; Laakso et al., 2000). We currently lack knowledge of effective evidence-based methods for treating anomia in MS. This study therefore aimed to explore the efficacy of one possible treatment method, semantic feature analysis (SFA), using both objectively assessed and self-reported measures. No treatment effects were apparent in confrontation-naming ability, re-telling ability or self-reports. There are several possible reasons for the lack of measurable changes in those outcome measures.

First of all, it is highly likely that the degree of anomia influences the treatment effects. The participants in the present study experienced anomia in their everyday life, and related nuisance, to a degree which compelled them to take part in the study. During the course of the study, the participants reported occurrence of anomia every day, at least occasionally or a few times a day, and reported being somewhat or moderately bothered by it. Where anomia occurs only a few times a day even at baseline, any treatment effect may be very difficult to detect owing to floor effects. The scale used in the present study may not have been sensitive enough to capture small changes in the occurrence of anomia and related nuisance that may have occurred.

Even so, experiencing word-finding difficulties on a daily basis may undoubtedly have a negative impact on a person’s everyday communication. As previously mentioned, the difficulties objectively observed may differ greatly from those subjectively experienced (Klugman & Ross, 2002). This highlights the need to use patient-reported outcomes in addition to formal assessment (El-Wahsh et al., 2020).

Further, the difficulties experienced are undoubtedly influenced by many factors, including individual personality traits, coping strategies, social support and communicative demands, as well as disease-related pain, fatigue and physical ability (Yorkston et al., 2014). When it comes to disease types, it should be noted that – in contrast to some previous findings (Friend et al., 1999; Lethlean & Murdoch, 1997) – P1, who had RRMS, manifested more detectable difficulties in most assessment tasks than P2, who had progressive MS. On the other hand, P1 consistently reported fewer difficulties and fewer negative consequences of the difficulties than P2 did. In particular, P2 reported a considerable cognitive impairment and a highly negative impact on communicative participation, although she achieved top scores for confrontation naming and was within the neurotypical norms for verbal fluency.

These findings support those of previous studies to the effect that commonly used assessment tasks may be insufficient for detecting subtle language difficulties (De Dios Pérez, 2017; Laakso et al., 2000; Lethlean & Murdoch, 1997; Renauld et al., 2016; Tallberg & Bergendal, 2009; Yorkston et al., 2014).

Overall, the participants’ difficulties as observed in the assessment of their ability to retrieve single words were quite subtle. Their scores were around or above 90% accuracy even at baseline. As expected, the differences between baseline and post-treatment in raw scores for accuracy in naming ability in all confrontation-naming tasks were minimal for both participants. Although the SFA method has proved effective in people with mild to moderate aphasia and in people with TBI (Efstratiadou et al., 2018), it might not be effective when it comes to further improving confrontation-naming accuracy in such mild or subtle anomia as manifested by the present participants.

Response time in confrontation-naming tasks has been found to be more sensitive than accuracy when it comes to detecting mild anomia (De Dios Pérez, 2017; De Dios Pérez et al., 2020; Galletta & Goral, 2018). In the present study, response times did not decrease significantly after treatment. However, P2’s response time did decrease progressively during the course of the repeated exposures in assessment. This indicates an effect of repeated measures, but not necessarily a treatment effect, as it was seen for untrained items as well. Similarly, the participants’ performance on the verbal-fluency control tasks also suggests an effect of familiarisation with the tasks.

Besides the tasks assessing the retrieval of single words, there were also outcome measures aiming to capture treatment effects in connected speech, based on analysis of correct information units (Nicholas & Brookshire, 1993). In future studies, it would be interesting to include connected-speech samples from the participants’ actual everyday communication, covering a range of communicative situations. Further, additional self-reported measures, such as in-depth interviews, would probably be required to capture the variation in difficulties experienced by people with MS in various communicative contexts. In the present study, the discrepancy observed between the self-reports on the occurrence of anomia and those on the nuisance experienced from those occurrences may reflect differences in communicative demands, communicative abilities and comfortableness in various situations involving everyday communication. In the present study, the task of re-telling a cartoon was intended to resemble an everyday communicative situation, such as re-telling a story or an event. Although that task may not have sufficient ecological validity to detect generalisation effects across all connected speech, it would be adequate to detect any changes within the same re-telling task. Both P1 and P2 scored lower than the reference group. Although P2’s %CIU score on the re-telling task was close to the lower end of the reference range, it was far from the average reference score. This might indicate that the re-telling task was more sensitive in detecting mild anomia than the confrontation-naming tasks. Nevertheless, no treatment effects in the form of an increase in %CIU or CIUs/minute were present in either P1 or P2. This might imply that, contrary to intentions, SFA did not work as a strategy.

Since the treatment failed to improve word-retrieval ability, and since it was not used as a strategy on occasions of anomia, no effects could be expected on connected speech or self-reported communicative participation either. It has been argued that, in a progressive disease such as MS, a compensatory strategy might be useful (De Dios Pérez, 2017). It could be that the strategy of naming semantic features was not useful for these participants, at this stage of their disease. Another possibility is that the treatment provided in the present study was insufficient to enable the participants to adopt and automatise a strategy. In fact, neither of them had any difficulties participating in the treatment, in completing the tasks or in suggesting relevant semantic features. Asking the participants to summarise the specific “on-target” features in treatment was expected to contribute to the use of a learnt strategy and to promote effective information conveyance in connected speech. However, the element of summarisation may not have been sufficient to promote participants’ adoption of SFA as a strategy when experiencing word-finding difficulties, either in the re-telling task or in everyday communication. It is possible that their anomia was not prominent enough to make them consider using a strategy, or that the occurrences were not frequent enough to give them sufficient opportunities for strategy use. The participants reported using the strategy occasionally or a few times a week. The sending of daily text messages reporting on word-finding difficulties was expected to serve as a constant reminder, but this evidently did not promote strategy use. Anecdotally, both participants spontaneously reported feeling benefits from the treatment. They mainly reported experiencing more structure when speaking, i.e., being able to sort out irrelevant details and focusing on conveying the most important facts. These reports are well in line with the original aim of SFA, namely to promote the organisation of verbal expression in people with traumatic brain injury (Massaro & Tompkins, 1994). As already discussed, however, this was not evident in terms of change over time in scores on the re-telling task or in self-reports on the occurrence of anomia and related nuisance, on SFA strategy use or on communicative participation.

In sum, the findings from the present study indicate that SFA might not be a suitable treatment method for subtle anomia in MS, at least not on its own. In future research, it would be interesting to further explore treatment more specifically targeting strategy use, possibly incorporating the SFA method but focusing on actively practising strategy use in actual communication settings.

Limitations

Some limitations to the present study must be considered. First, this is a phase I–II trial exploring whether or not SFA could be considered appropriate as a treatment method for subtle anomia in MS. Hence there is a need for additional studies to further explore the efficacy of SFA as well as other methods on anomia in MS. Second, findings from a single-subject experimental-design study such as this cannot unreservedly be generalised onto a larger population. Third, the self-report scales for anomia and strategy use are not well established and have not been tested for reliability, meaning that all results should be considered with caution.

Conclusions

SFA as the sole element of treatment content may not be sufficient to yield treatment effects on subtle anomia in MS. There is still a need for effective treatment methods for mild anomia in MS that may yield generalisation effects on connected speech and communicative participation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Swedish Research Council for Health, Working Life and Welfare (FORTE) [2015-00503] and the Swedish Research Council (VR) [2016-01275].

Appendix

Unique treatment items used by one or both participants

Table

Nouns			Verbs
Swedish	English		Swedish	English
antenn	aerial		analysera	to analyse
batterier	batteries		banta	to diet
brandsläckare	fire extinguisher		basta	to have a sauna bath
brygga	(small) jetty		blanda	to blend/mix
brödrost	toaster		byta (lampa)	to replace (a light bulb)
busskur/busshållplats	bus shelter/bus stop		dirigera (trafik)	to direct (traffic)
cementblandare	cement mixer		diska	to wash the dishes
champinjon	field mushroom		fickparkera	to parallel park
croissant	croissant		fika	to have coffee
delfin	dolphin		flirta	to flirt
diskmaskin	dishwasher		frankera	to put a stamp on
fingerborg	thimble		fråga	to ask
fjärrkontroll	remote control		fuska	to cheat
grill	barbecue		fylla på	to fill up
hushållspapper	kitchen paper		följa efter	to follow
husvagn	caravan		förtrolla	to enchant
häftapparat	stapler		gala	to crow
iPad	iPad		ge	to give
isberg	iceberg		gro	to sprout
kaffebryggare	coffee maker		gymnastisera	to do gymnastics
kikare	binoculars		gå på (konsert)	to attend (a concert)
kylskåp	refrigerator		hacka	to break (rocks)
laddare	charger		hinna med (bussen)	to catch (the bus)
laptop	laptop computer		hugga	to chop
limstift	glue stick		hypnotisera	to hypnotise
maskindisktablett	dishwasher tablet		häfta (ihop)	to staple
mobiltelefon	mobile phone		hänga (kläder)	to hang (clothes)
mikrovågsugn	microwave oven		importera	to import
mikroskop	microscope		jama	to miaow
skridskor	skates		kakla	to tile
slalompjäxor	ski boots		klättra	to climb
spritkök	camping stove		knäcka (nötter)	to crack (nuts)
spårvagn	tram		krocka	to crash
studentmössa	student’s cap		krympa	to shrink
telefon	telephone		krypa	to crawl
truck	forklift		köa	to queue
vin	wine		lacka (naglarna)	to varnish (fingernails)
väggklocka	wall clock		landa	to land
älg	elk		lassa	to load
			livrädda	to rescue
			lugga	to pull someone’s hair
			lyfta	to take off (of aircraft)
			minnas	to remember
			missa (bussen)	to miss (the bus)
			misshandla	to beat up
			musicera	to play an instrument
			mäta	to measure
			prata	to talk
			promenera	to stroll
			rida	to ride (a horse)
			riva/demolera	to demolish
			ro	to row (a boat)
			rymma	to escape
			servera	to serve (at table)
			skoja	to joke
			skotta	to shovel
			skvätta	to splash
			slicka	to lick
			slåss	to fight
			snubbla	to stumble
			sola (sig)	to sunbathe
			stämpla	to stamp
			tappa upp (öl)	to pour (beer) from a tap
			trilla	to fall
			vaccinera	to vaccinate
			vänta	to wait
			värpa	to lay (an egg)

These are all unique treatment items. There are two reasons why there are more actions than objects. First, the two participants were more likely to choose the same items for objects than for actions. Second, the action items took slightly less time to work through.