text
string | predicted_class
string | confidence
float16 |
|---|---|---|
The quantitative results strongly suggest SASA! is diffusing throughout intervention communities (Table 3 and Fig. 3), with between 69 and 85% of men and women in the sample reporting their friends, neighbours and elders attended SASA! activities. As for partners, 54% of men versus 14% of women report their partner attended, showing significant gendered variation. Large proportions of participants also report talking about SASA! with one or more members of their social network (83% of women and 92% of men) and the majority did so more than once. As for intimate partners, there was also a gendered variation as seen with attendance, with 67% of men versus 58% of women reporting speaking to their partner about SASA!. The results also indicate the diffusion of SASA! in the community beyond the sample: it is not only the participants initiating discussions about SASA!, but friends, neighbours and parents as well. Interestingly, elders are the only group that participants report initiated the conversation the majority of the time (72% of men and 68% of women reported this).Table 3Characteristics of social network participation and communication about SASA!MaleFemalen/N(%)n/N(%)Social Network Members Attending SASA! (as reported by participants) Partner280/520(54%)50/348(14%) Friend463/542(85%)248/342(73%) Neighbour444/534(83%)282/334(84%) Parent66/552(12%)31/342(9%) Elder384/559(69%)245/332(74%) In-law65/543(12%)30/335(9%) Children171/550(31%)188/349(54%)Characteristics of communication with different social network members:Respondent talked to:(N = 512)(N = 289) Both sexes382(75%)142(49%) Same sex only122(24%)140(48%) Opposite sex only8(2%)7(2%)Who initiated talks about SASA!: Network member initiated all talks117/571(21%)96/358(27%) Partner initiated96/376(26%)15/204(7%) Friend initiated175/470(37%)82/246(33%) Neighbour initiated187/389(48%)122/257(48%) Parent initiated30/61(49%)14/32(44%) Elder initiated180/249(72%)108/158(68%) In-law initiated15/54(28%)11/37(30%) Children initiated4/54(7%)38/89(43%)Fig. 3Frequency of social network communication about SASA! | study | 100.0 |
The qualitative data indicate interpersonal communication and materials were the main communication channels through which participants first became aware of and engaged with SASA!. Over half were first exposed when a community activist came to their home and invited them to join an activity or drama:John [CA] mobilised us to come and attend…it even rained on that day but we went and attended...for us we just went because we were mobilised, we did not know what we were going to learn that day (4F).The first time, there was a lady called Mukyala Mukulu (women representative on Local Council). And man called Musomesa are the ones who brought the SASA sensitizations/ activities in our community. We also joined and they taught us. (8 M). | other | 99.9 |
As these quotes illustrate, being personally invited to join—especially by a known community member—was a strong motivating factor for many. For others, their first exposure came through informal discussions with members of their social network. This included casual discussions with other community members who had attended, seen posters or observed activities taking place in the community. And, as noted above, in a few cases couples’ first exposure came following an episode of IPV when the wife sought support at the local council or when a CA intervened. For example, one participant went to the Local Council (LC) to report her husband after a fight over his lack of financial support escalated to physical violence:I went to the LC office and reported him and they called both of us...I wanted this man [her husband] to look after us, not that I reported him so that we could separate. (6F). | other | 99.9 |
Posters played an important role in raising general awareness of SASA! and promoting ongoing attendance. Participants often noted first seeing posters displayed around the community as they came to know about SASA!. The posters continued to have relevance over time as participants continued to review and reflect on the different relationship and family scenarios depicted (detailed later). The ‘loud speaker’ (community public announcement system) was influential in promoting ongoing attendance. Participants frequently spoke of continuing to attend whenever they heard activities announced on the ‘loud speaker,’ whereas their first exposure generally came from the other channels. | other | 99.9 |
60% of women and 95% of men in the quantitative sample reported positive change in their relationship since becoming involved in SASA!. The data suggest interpersonal communication played the strongest role. Men were more likely to report positive changes in their relationship due to SASA! when they attended interactive discussion activities and spoke numerous times (5+) with their partner about SASA! (Table 4). There were also dose response relationships observed for both exposures, with the associations increasing between the low (1-4 times) and high frequency (5+ times) categories. For women, there were strong independent effects (and dose response relationships) for all exposures apart from talking with peers, seeking CA advice and drama attendance (p-value > .05 for these variables) (Table 4). Similar to men, frequent talks with their partner about SASA! had the strongest independent effect on relationship change (aOR 7.08, CI 2.29-21.90). However, women who talked with elders about SASA! were 9.62 times more likely to report relationship change, whereas for men there was no effect. Communication material exposure (aOR 4.3, CI 1.69-10.93) and discussion activities (aOR 3.53, CI 1.46-8.54) also showed strong independent effects on women reporting relationship change.Table 4Association between SASA! exposure and interpersonal communication and reporting positive change in relationshipFemaleMaleVariable(Channel) (N = 354) n % reporting changeOR95% C.I.p-val*aORa95% C.I.p-val* n % reporting changeOR95% C.I.p-val*aORa95% C.I.p-val*SASA! Exposure: Materials/posters< 0.01< 0.010.270.68 0-1 times17/72(24%)1–1–72/79(91%)1–1.00– Few times (2-4)33/83(40%)2.14(1.06-4.30)1.15(0.47-2.78)267/272(96%)2.31(0.86-6.16)1.14(0.36-3.63) Many times (5+)163/199(82%)14.65(7.63-28.14)4.3(1.69-10.93)158/167(95%)1.71(0.61-4.76)0.68(0.18-2.61) Drama/film (mid media)< 0.010.49< 0.010.41 Never48/118(41%)1–1–69/79(87%)1–1.00– Once40/74(54%)1.72(0.95-3.08)0.52(0.21-1.25)144/155(91%)1.9(0.77-4.68)1.61(0.56-4.68) A few (2-4 times)68/99(69%)3.2(1.82-5.61)0.63(0.26-1.55)187/191(98%)6.78(2.06-22.31)3.45(0.76-15.66) Many times (5+)57/63(91%)13.85(5.53-34.69)0.8(0.19-3.44)91/93(98%)6.59(1.40-31.07)2.42(0.37-15.76) Discussion activity (interpersonal comm)< 0.010.02< 0.010.03 Never34/107(32%)1–1–35/43(81%)1–1.00– 1 time47/80(59%)3.06(1.67-5.59)2.16(0.90-5.18)144/155(93%)2.99(1.12-7.99)3.47(1.01-11.92) A few/many times (2+)132/167(79%)8.1(4.66-14.06)3.53(1.46-8.54)312/320(98%)8.91(3.15-25.23)5.77(1.52-21.95) Sought CA advice (interpersonal comm)< 0.010.090.010.83 No147/282(52%)1–1–282/304(93%)1–1.00– Yes66/72(92%)10.1(4.24-24.06)2.65(0.84-8.34)209/214(98%)3.26(1.21-8.75)1.13(0.35-3.71) Multi-channel exposure(N = 352)< 0.01< 0.01(N = 5181)< 0.01< 0.01 Mass media only24/78(31%)1.00–1.00–8/12(67%)1.0–1.00– Low multi-channel exp. (1-4)77/136(57%)2.94(1.63-5.29)3.26(1.73-6.15)238/255(93%)7.00(1.91-25.61)6.17(1.49-25.47) High multi-channel exp. (5+)112/138(81%)9.69(5.10-18.4312.3(6.09-24.85)245/251(98%)20.4(4.80-86.86)15.72(3.22-76.74)Interpersonal Communication about SASA!: Talked to partner< 0.01< 0.01< 0.010.02 never46/147(31%)1–1–130/146(89%)1–1.00– medium (1-4)105/139(76%)6.78(4.03-11.41)3.05(1.53-6.11)202/212(95%)2.49(1.09-5.65)1.34(0.47-3.84) high (5+)62/68(91%)22.69(9.15-56.23)7.08(2.29-21.90)160/160(99%)19.6(2.56-149.53)13.10(1.33-128.54) Talked to peers< 0.010.350.010.26 low (0-2)24/102(24%)1–1–127/142(89%)1–1.00– medium (3-5)32/50(64%)5.78(2.77-12.07)1.78(0.71-4.46)109/113(97%)3.22(1.04-9.99)2.49(0.69-8.97) high (6+)157/202(78%)11.34(6.45-19.95)1(0.41-2.43)255/263(97%)3.76(1.56-9.11)0.94(0.30-3.01) Talked to elders< 0.01< 0.010.120.85 low (0-2)71/190(37%)1–1–268/288(93%)1–1.00– medium (3-5)109/128(85%)9.62(5.44-16.99)4.05(1.85-8.88)185/185(97%)2.23(0.88-5.65)1.21(0.37-3.89) high (6+)33/36(92%)18.44(5.45-62.32)5.7(0.99-32.71)44/45(98%)3.28(0.43-25.09)0.60(0.05-6.85)* Overall p-value estimation based on likelihood ratio test aControlled for age, marital status, education level, SES and other SASA! exposure | study | 100.0 |
Overall for both women and men the data indicate exposure to both materials and activities or dramas/videos had the greatest effect. Men with low (1-4 times) and high (5 or more times) multi-channel exposure were respectively 6.17 and 15.72 times more likely to report relationship change following exposure to SASA! versus those with only materials exposure; and women were 3.26 and 12.3 times more likely to report this (Table 4). | study | 100.0 |
To start, there was a sense, even from participants with less exposure, that SASA! was a visible part of the fabric of the community. This ‘observability’ was evidenced by how participants noted “seeing people talking about it,” referenced those “who are active in it” and “put up posters” and described how “when you come back in the evening they [neighbours/family] will tell you that the SASA! team was here.”(1 M). There was a strong sense of collective engagement in SASA! illustrated in the way participants often discussed their participation as a community, rather than individual endeavour. For example, similar to this participant, when asked about their personal engagement with SASA! many responded using “we” language and spoke about it as a community endeavour: “All the women here attended...we went together”(1F). Both the awareness of SASA! activities and talk among community members stimulated curiosity and motivated attendance. For example, one participant described how his curiosity was piqued after hearing talk about SASA! in the community and led him to attend his first drama:Interviewer: When you heard that there was a SASA! activity, what did you think about that?Participant: “I had to know the meaning of SASA!, [people] would say that there are SASA! dramas there, and then I would ask myself that what is SASA!? Is it a drama? That prompted me …they had even told us that Nandujja [popular traditional dancer] was coming, she was the first one to come. I decided to go and watch her, when she finished then they brought a drama.” (5 M). | other | 99.9 |
Participant: “I had to know the meaning of SASA!, [people] would say that there are SASA! dramas there, and then I would ask myself that what is SASA!? Is it a drama? That prompted me …they had even told us that Nandujja [popular traditional dancer] was coming, she was the first one to come. I decided to go and watch her, when she finished then they brought a drama.” (5 M). | other | 99.94 |
Here the participant demonstrates how SASA!‘s visible presence in the community—with most activities and dramas held in local gathering spaces—stimulated curiosity and encouraged engagement. After attending the drama he, like others, went on to attend different SASA! activities when offered. There was also the perception among many participants that SASA! was appreciated and well received by the wider community:Many people, neighbours, those I work with like the Boda-Boda [motorcycle taxi] riders and even you can see that they like SASA! activities. (8 M). | other | 99.94 |
(Though their narratives may have been exaggerated if they perceived the interviewer was associated with SASA!). This in turn led to a mutually reinforcing relationship, in that communication about SASA! motivated community members to attend and this generated more discussion. Social network members also played a role by encouraging each other to continue attending activities. As this participant notes, community members often helped spread the word about activities: | other | 99.94 |
What encourages me to go there is the hope that they would bring a new idea…especially the ideas that help us on the things we are working on. And you will find colleagues who will tell you...‘SASA! sessions are going on.’ Because it has helped to create better families now. (3 M). | other | 99.94 |
Within social networks informal conversations about and engagement with SASA! was important not only in motivating attendance, but in enhancing the observability of changes in couples who were involved in SASA!. This is indicated in the way participants frequently reported observing a reduction in IPV in their communities. They may have felt compelled to report changes in their community and, indeed, some accounts did appear superficial with vague, blanket statements that people had changed. However, others, like this participant, shared specific examples of couples they had seen change: | other | 99.9 |
Seeing positive changes in neighbours and friends seemed to increase the perception of SASA! as an effective means of reducing IPV and improving family life:Everyone you talk to, will always tell you that SASA! activities have changed life for the better. (8 M). | other | 99.94 |
A central theme influencing exposure and engagement was SASA!‘s compatibility with participants’ lives and personal challenges. To begin, the close proximity of SASA! to participants’ communities and daily lives made it feel intimate and personal. Several remarked that SASA! was not like other programmes that “decide to stay at the health centres, where they sensitise the people from”(8 M). As this participant illustrates, there was a strong appreciation that SASA! activities came right to them:They have even reached down to the grass root people, instead of people saying that they are going to watch a drama, the drama comes down to them. (5 M). | other | 99.9 |
Second, participants found meaning through their identification and connection with the topics discussed at activities and observed in dramas and videos. They frequently noted activities reflected their own experiences and those around them. One participant explained:[T]he information was good and I think it was like a lesson because we were also going through the same situation. (6F). | other | 99.9 |
For many, seeing the cause and effect of scenarios that reflected their own lives also generated an affective (emotional) response and fostered new understanding. For example, one participant reported being most impacted by the stories depicted in the SASA! videos, explaining:What has affected me most are the videos because they show you the beginning and the end, that if you do this, it shows you what the end result will be. (6 M). | other | 99.9 |
Third, the desire “to learn” was a primary motivation for activity attendance highlighted throughout the narratives. One participant noted she never received any education around being in a relationship and SASA! offered her an opportunity to learn. Here, another participant demonstrates the value many placed on learning and the potential relationship benefits that may follow:I am a person who likes to learn new things. You know when you go for such activities you cannot be the same, even your marriage improves… it is like how we used to go to school, each day we would learn something new, I have learnt how to have a good relationship with my husband. (4F). | other | 99.9 |
I am a person who likes to learn new things. You know when you go for such activities you cannot be the same, even your marriage improves… it is like how we used to go to school, each day we would learn something new, I have learnt how to have a good relationship with my husband. (4F). | other | 99.94 |
This also highlights the importance of participants’ ‘felt need’ for change in their relationship and whether they felt SASA! offered them enough advantages to attend and continue to engage. For example, one couple articulated how their attendance was directly linked to a desire for change in their volatile relationship:I was motivated to come and attend that sensitisation activity about domestic violence...I wanted it to help me because the violence in my relationship was not ending...I thought that if the violence would reduce even in my home, even our relationship would become better. (6 M).What motivated me is that they mobilised us and that they said that they were going to speak about violence and this is what was happening in my home…I had a problem in my home and I had to go and attend. (6F). | other | 99.9 |
I was motivated to come and attend that sensitisation activity about domestic violence...I wanted it to help me because the violence in my relationship was not ending...I thought that if the violence would reduce even in my home, even our relationship would become better. (6 M). | other | 99.94 |
On the flip side some individuals who did not perceive that their relationship needed change because they had no physical IPV, were unmotivated to continue engaging with SASA! more actively—despite ongoing conflict, verbal abuse and/or controlling behaviour. For example, one participant exhibited controlling behavior, barred his wife from working and refused her requests to test for HIV, but noted SASA! was for those experiencing physical violence and not applicable to him:Generally it would have been a good thing but...there are people like me, I personally never fight,... I personally don’t have problems in my relationship that would cause me to go there. Indeed if you had violence in your relationship you would. (1 M). | other | 99.9 |
Other key barriers to engagement were a lack of proximity and incompatibility with participants’ lives. Some reported they did not attend because there were few activities in their area or the times were inconvenient:I attended [only] one because they do not normally come to our community. (7F).I wanted to attend their activities so that I listen to what they teach but I was not able to because I am busy working. But I thought that the next time when they come, I will attend and listen. (3F). | other | 99.9 |
Suggestions were given that activities be held on Sunday or during times of the day when most people have finished work and household chores. However, these reasons may not be the full story, but socially acceptable responses. Participants may have wanted to portray a certain image to the interviewer to avoid, for example, showing a lack of interest in SASA!. | other | 99.94 |
Lastly, while the loud speaker and door-to-door mobilisation were important communication channels in motivating attendance, some reported never hearing activities announced or that CAs failed to return following an initial visit. As these participants demonstrate, the lack of set times and advance notice of activities was a barrier to exposure for some and may impede diffusion in some cases:[I]f you just come one morning and you walk through the community and tell people that come to the activity, you find that people already have their other programmes. (4 M).It is difficult to tell somebody that you should go and participate in SASA! activities. That person will ask you ‘where are they?’ At that time it is difficult to answer that question...because we do not know... You just hear about it in the community that they [CAs] are coming, they [CAs] come and tell us that they are about to start... (5 M). | other | 99.9 |
It is difficult to tell somebody that you should go and participate in SASA! activities. That person will ask you ‘where are they?’ At that time it is difficult to answer that question...because we do not know... You just hear about it in the community that they [CAs] are coming, they [CAs] come and tell us that they are about to start... (5 M). | other | 99.94 |
Community activists played a central role in participants’ change process from their initial knowledge about SASA! ideas through implementation of new behaviours. Their role as change agents appeared to be particularly influential because they were part of community members’ social networks and also often respected ‘opinion leaders’. Several participants, mainly men, indicated pre-existing relationships with their community activists who were “resident[s],” friends, relatives or members of their local council. For example, one participant described how he had always “strongly admired” a CA, noting:So because of that man being part of the SASA! team I wanted to listen and get to know whatever they were discussing. (1 M). | other | 99.9 |
There was an appreciation that CAs were both part of the community – “one of us” – but also had links to outside networks as they “walk with the people from SASA!” and received training. Together this appeared to accord them value in the eyes of participants, legitimising their role and the new ideas they were sharing. This participant illustrates how CAs were able to reach people in casual and intimate ways because they were both respected members of their social network and often familiar with community members’ lives:I saw him [CA] approaching me with a pile of materials. He gathered us together and said to me, ‘I am lucky I have met you because it is you who has married many women.’[teasing tone]...When we gathered he started asking us several questions. During the discussion I started telling him about family problems. In response he told me about the programme [SASA!] and that’s how I started knowing about those programmes. (10 M). | other | 99.9 |
I saw him [CA] approaching me with a pile of materials. He gathered us together and said to me, ‘I am lucky I have met you because it is you who has married many women.’[teasing tone]...When we gathered he started asking us several questions. During the discussion I started telling him about family problems. In response he told me about the programme [SASA!] and that’s how I started knowing about those programmes. (10 M). | other | 99.94 |
There were also cases that illustrated how a previous relationship with a CA or attributes of a CA can be a barrier to change. The most notable example was a participant who reported not being able to take his CA’s messages around SASA! seriously because of the nature of their long time friendship:Joyce [CA] didn’t teach me...because we used to joke a lot and when sometimes she brought a topic [related to SASA!], I would think that she’s still joking so I failed to give her time that way. (9 M). | other | 99.9 |
He also felt he could not go to her for support with his own relationship issues because of ongoing IPV in her relationship: “you cannot ask such a person for support because they are worse off” (9 M). This highlights a challenge surrounding the CA’s role in the community. On one hand observing their change/good relationship facilitated change in others, but on the other hand this can be a barrier when a CA is still experiencing violence in their own relationship, not modelling healthy relationship behaviours or has other attributes or beliefs that are not respected by community members. | other | 99.9 |
And finally, many who engaged with SASA! and took on board the new ideas and behaviours (and some who did not) reported sharing what they had learned with others and, in some cases, becoming change agents in their own right. This was not dependent on extensive exposure or change, as diffusing SASA! messages was reported by participants with minimal exposure as well as those who had not applied much or any of the learning in their own life. Those with less exposure reported sharing ideas they had picked up from posters or seeing an activity once, or referring people to SASA! after hearing about it from others. Those with more exposure exhibited a deeper motivation to tell others about SASA! so they could also experience the benefits they enjoyed. For example, one participant, who experienced profound changes in his relationship due to SASA!, reported actively passing his learning on to others in the community:“[SASA!] has helped to create better families now...we have actively participated to the extent that if you get like five people, we help them...so that they also learn and train others and this has increased the number of peaceful homes here.” (3 M). | other | 99.9 |
The external visibility of change in couples also appeared to make them an attraction, with some more active participants assuming a role similar to a CA within the community (and beyond in some cases). This new status was meaningful for them and reinforced their own changes and desire to continue engaging with SASA! and sharing their learning to help others. | other | 99.94 |
The findings indicate materials and mid media channels generated awareness and knowledge, while the concurrent influence from interpersonal communication with CAs and different social network members more frequently facilitated changes in behaviour or ‘adoption’. The most influential attributes of the intervention were its observability, compatibility and the relative advantage it provided to community members. Broadly, exposure through multiple communication channels was most effective in facilitating change. Those with low and high ‘multi-channel’ SASA! exposure had much higher odds of reporting relationship change than those with exposure to mass media materials only. In addition, the dose-response relationship observed suggests those with more ‘multi-channel’ intervention exposure were more likely to experience relationship change. While it is difficult to ground the findings in the context of other IPV studies given the dearth of research examining diffusion, ‘what works’ reviews of IPV intervention evidence have also found programmes which combine mass media messaging and community mobilisation with more interpersonal engagement (i.e. interactive group activities and individual counselling) are more effective in generating behaviour change [11, 40]. | review | 99.9 |
Dramas and videos (mid media channels) appeared to generate identification among participants and understanding of the causes and effects of IPV. This is evidenced by the narratives on realistic storylines facilitating identification while also modelling alternative perspectives and behaviours. This underscores the importance of developing program content that closely reflects the lives of people in the specific context. The concurrent influences of interpersonal communication with CAs and community members then appeared to give the media messages credibility, facilitating favourable attitudes towards them and encouraging behaviour change. Consistent with many diffusion studies [20, 25, 41], this concurrent influence was the most influential factor in the uptake of new ideas and behaviours. Our quantitative data showed interpersonal communication channel exposures alone (i.e. discussion activities with a CA, seeking CA advice, and talk about SASA! with partner, peers and elders) were associated with relationship change after controlling for the effect of all exposures, whereas materials and mid media exposure were not. | study | 100.0 |
The findings also suggest talk about SASA! among peers may raise awareness and motivate attendance, whereas discussions with elders and partners may be more influential in changing their behaviour in their relationships. Talking with elders (among women) and one’s partner about SASA! showed independent effects with relationship change, whereas talking with peers did not. Diffusion research indicates those who are more similar to an individual are more influential in persuading them to adopt new ideas and behaviours . Thus, it is interesting that talk with peers about SASA! did not have an independent effect, while talk with elders, who are in theory less similar, did. This may be due to the respected role elders have in Ugandan society regarding relationship guidance. This finding may also be attributed to ‘ssengas’ or paternal aunts who traditionally provide Baganda women guidance on relationships as some were sensitised as part of the intervention. While there is currently a growing call within IPV prevention to focus efforts on younger populations , these findings support the simultaneous engagement of older generations, at least in culturally similar settings. During formative research and intervention design or adaptation it may be helpful for practitioners to, 1) examine who advises on and influences relationships in the given context (e.g. peers, elders, religious leaders, local leaders, schools) and ensure the intervention engages them as well as youth, 2) consider how to deliberately promote talking between partners about their relationship and what they are learning. | study | 98.9 |
The intervention’s engagement of community members as change agents (i.e. community activists) appeared particularly impactful, because the ideas were emerging from trusted and known friends and neighbours rather than people from outside the community. The qualitative data indicates that CAs’ influence stemmed from the multiple roles they embodied within the community’s social network as community member (“one of us”), opinion leader and change agent. In our broader qualitative evaluation we found CAs viewed their role as a vocation irrespective of formal programme support. It was a means through which they gained a valued identity, particularly for those who had not completed their formal education or achieved anything of note in their community. These findings are in agreement with diffusion theory and studies that found identifying opinion leaders and using them as change agents can increase diffusion of health promotion interventions at the community level [20, 41, 44]. However, similar to our findings, others have found the identification of actual opinion leaders was essential to their effectiveness . Valente and Pumpuang studied methods for identifying opinion leaders and found programs using multiple methods such as expert identification and peer nominations were most likely to secure successful opinion leaders for promoting behaviour change [20, 41, 44]. | study | 99.94 |
Overall, our findings point to the value of using diffusion of innovations theory in the evaluation of IPV interventions. Intervention research often examines the effect of exposure to different aspects of the intervention on the intended outcomes, but stops there. Our findings demonstrate the vital role played by CAs as well as the importance of the intervention’s attributes of observability, compatibility and relative advantage—whereas relative findings for trialability and complexity did not emerge in the qualitative data. These insights made an important contribution to our understanding of how the intervention worked to influence behaviour change and may have been missed if the study was guided by an individual behaviour change model alone (e.g., transtheoretical model ) as they do not account for community level intervention and social network factors. The study on the Stepping Stones noted earlier illuminates this point: through examining diffusion the study found the messages were not spreading to the wider community, illuminating an important limitation that can be used to inform future interventions . As Kippax and Stephenson contends we must design research to elucidate the ways individuals engage with interventions/messages and capture the mechanisms of change in order to find out what worked to improve interventions . Diffusion of innovations theory should be given strong consideration both when developing and researching community mobilisation interventions like SASA! which are designed to diffuse through community social networks and change agents. | study | 99.9 |
The study had a number of limitations. First, using data collected post-intervention introduces the potential for increased social desirability bias, especially when relying on self-reported attitudes and behaviours specifically promoted by the intervention. Participants may have exaggerated the impact of SASA!, out of a desire to please the investigators and present the programme in a positive light. While the qualitative data were interrogated across cases for the level of detail provided verses general statements of change—and triangulated with partner account where possible—we cannot exclude the potential influence of social desirability bias in their responses. Second, given the dynamic context, other factors may also influence changes in relationships such as: ssengas not sensitized by SASA! [42, 48] and messaging from HIV prevention campaigns, religious groups and other anti-violence initiatives in Kampala on trust, love and morality [48, 49]. | study | 99.94 |
Third, the single interview design was a limitation. Collecting data at multiple time points through longitudinal or pre/post interviews both places less reliance on recall and allows the researcher to observe across interviews the consistency in participant’s accounts of changes experienced, increasing validity. Participant’s ability to recall the ways they engaged with the intervention—sometimes over several years—likely impacted the data. While a longitudinal design would have been preferable, it was not deemed feasible due to the mobile characteristics of the study communities, resource constraints and other factors related to the RCT design. The qualitative study helped offset the recall challenge some as it was designed to capture the sequence of events and relationship changes over time using the participatory timeline tool. And, fourth, to avoid contamination in the RCT mass media (e.g. radio, TV) was not used and our conceptual framework had to be altered to reflect this. Poster exposure proved difficult to classify in the quantitative analysis as they were both displayed in the community and used during some discussion activities. This made it less clear cut what was communication materials/media exposure versus interpersonal communication exposure. While this required the quantitative results to be interpreted with caution, the qualitative data helped to tease this out more. | study | 100.0 |
This study contributes to our knowledge of IPV prevention in a low-income, urban East African context. Specifically, it highlights how using a community mobilisation approach that includes content that reflects peoples’ lives and direct leadership of local change agents can facilitate diffusion and powerful collective change processes in communities. Overall, this study makes clear the value of applying diffusion of innovations theory using mixed methods to illuminate the how as part of complex public health intervention evaluations. | study | 99.94 |
Many models of speech production recognize, despite their differences, that there are three main stages of speech planning: the formation of the concept to be conveyed; the retrieval of phonological representations; and the articulation of the planned speech (Stemberger, 1985; Dell, 1988; Roelofs, 1997; Levelt et al., 1999). One of the debates in the last two decades is what constitutes as the primary planning unit in cross-linguistic online speech production during the phonological encoding stage. The present study aims to shed further light on the issue by investigating whether the primary processing unit is universal and to what extent it is determined by one's language background. We explore this question with bilinguals speaking two different types of languages (Germanic and Sinitic). Further, with the use of ERPs, we investigate the neural activation pattern of the phoneme underlying the presence or absence of behavioral effects which existing debates on primary processing unit have mainly relied upon. In the following, we first introduce the existing literature on Germanic languages which suggested the phoneme to be universally activated in all languages, which is followed by a description of behavioral research in Sinitic languages that challenges this universality view and suggests that the planning unit is language-specific. We then discuss (1) how understanding processing mechanisms of bilinguals with different language backgrounds may shed light on the issue, and (2) how ERPs would give us more insight into the activation pattern of segmental information. | study | 99.9 |
In West-Germanic languages, the phoneme has been found to serve as the primary unit of phonological planning. Initial evidence comes from speech error analyses indicating a large portion of phoneme-sized insertion, deletion, and substitution errors in English (Shattuck-Hufnagel and Klatt, 1979). Further evidence comes from experimental paradigms which reported faster responses in conditions with phoneme onset overlap than no overlap conditions. For example, in the picture-word interference paradigm participants named pictures faster when they were superimposed by a word that matched with the first phoneme of the picture (e.g., Schriefers et al., 1990; Meyer and Schriefers, 1991; Damian and Martin, 1999; Starreveld, 2000). Also in other production tasks, like the implicit priming paradigm (Meyer, 1991; Roelofs, 1999, 2003) and the color-object picture naming task (Damian and Dumay, 2007, 2009), overlap of the onset phoneme facilitated speech latencies. Phoneme facilitation was also found during reading aloud tasks that used priming (e.g., Forster and Davis, 1991; Kinoshita, 2000; Schiller, 2004, 2008; Timmer and Schiller, 2012; for a review see Timmer and Schiller, 2014). Taken together, the earlier consensus in the literature was that the phoneme serves as the primary unit of phonological encoding during on-line speech production. The prevalence of evidence for initial phoneme encoding in native speakers of West-Germanic languages, such as Dutch and English, has led to the assumption that the initial phoneme activation during phonological encoding is a universal feature. Dominant models of speech production include this assumption (Roelofs, 1997; Levelt et al., 1999). | review | 99.9 |
Several studies investigating Sinitic languages have challenged the universality of phoneme as a functional unit of planning. Natural slips of the tongue in Mandarin Chinese rarely cause phoneme-sized errors (Chen, 2000). The same experimental paradigms used in Western languages discussed above did not find evidence of phoneme onset facilitation in varieties of Sinitic languages. For example, no facilitation was found in the implicit priming paradigm (Chen et al., 2002 for Mandarin, Chen et al., 2002; O'Seaghdha et al., 2010; Cantonese: Wong et al., 2012), the picture-word interference paradigm (Wong and Chen, 2008 for Cantonese), the primed word naming paradigm (Chen and Li, 2011 for Mandarin), the primed picture-naming task (Yu et al., 2014 for Mandarin), or the color-object picture naming task (Qu et al., 2012 for Mandarin). More specifically, in the latter paradigm, participants were asked to name the color and object of a picture drawing. No facilitation was present when the first phoneme of the color and object matched. O'Seaghdha et al. (2010) therefore states that Germanic and Sinitic languages differ in the proximate unit for initial phonological processing. Phonological encoding during lexical retrieval in languages such as Mandarin, a Sinitic language, seems to begin with syllable retrieval, which constitutes as the initial functional unit of phonological encoding, which subsequently is followed by retrieval of individual phonemes within the syllable frame for articulation. In languages such as Japanese, this initially retrieved syllabic unit is called a mora (e.g., Kureta et al., 2006; Verdonschot et al., 2011). In contrast to Mandarin and Japanese, in languages such as English, a Germanic language, the phoneme is retrieved as the primary functional unit. | review | 55.8 |
The dissociation between syllable and phoneme facilitation also seems to be supported by the different scripts and phonotactics of Sinitic vs. West-Germanic languages. In Germanic languages, the phoneme can serve as a more efficient functional unit of processing. Given that Sinitic languages are syllabic and re-syllabification is far less common than in Germanic languages, it is probably more feasible to access the entire syllable in Sinitic languages, as an efficient functional unit during phonological encoding. Thus, the primary functional unit of phonological encoding has been suggested to be language specific (O'Seaghdha et al., 2010). | study | 99.94 |
It is important to note that speech perception studies have revealed an interestingly dynamic role that a processing unit can play in spoken word recognition (Cutler et al., 2001). Studies on bilinguals have shown that whether the processing unit in their dominant language may be used during the processing of the other language is contingent upon the language-specific characteristics of the languages involved (Cutler et al., 1989, 1992). For example, English-French bilinguals were observed to use their English processing strategy when listening to French. However, French-English bilinguals used different strategies depending on the language spoken because the French strategy would have been inefficient during English processing. | study | 97.6 |
Within the speech production literature, the debate between the language-universal vs. language-specific functional unit of phonological encoding can also benefit from further research on bilingual speakers, especially bilingual speakers of languages which have been proposed to have different functional units of phonological encoding. In a masked priming experiment, native Mandarin speakers who were highly proficient in English, showed phoneme onset priming in their production of English (Verdonschot et al., 2013). For highly proficient Japanese-English bilinguals the same pattern was revealed, showing sensitivity to the phoneme onset in English (Ida et al., 2015; Nakayama et al., 2016). This suggests that native speakers of a syllabic/moraic language can employ different processing units depending on the language used (i.e., a phoneme-sized unit for phonological encoding in English). Interestingly, while the Japanese-English bilinguals adhered to their native mora-based processing strategy in Japanese (Ida et al., 2015), Mandarin-English bilinguals showed phoneme onset priming even in their production of Chinese (although contingent upon syllable structure sharing between the prime and target) instead of only syllable priming (Verdonschot et al., 2013). One question that arises is what might have modulated the adaptive behavior of bilingual speakers. The question may be adjudicated with more insights from neurophysiological studies. | study | 99.9 |
ERP studies with bilinguals from the same population as discussed above, namely Mandarin-English bilinguals with high-level proficiency in English, reported significant neural responses to segmental repetition in Mandarin Chinese, despite the lack of segmental priming effect in response time (Qu et al., 2012; Yu et al., 2014). Specifically, Qu and colleagues found that initial segment repetition in a color-object picture naming task elicited more positive ERPs in the posterior regions during the 200–300 ms time window and more negative ERPs in the anterior regions during the 300–400 ms time window after picture onset, relative to no-repetition trials. Yu and colleagues reported that overlapping phonemes in a picture-naming priming task in both the initial and non-initial position evoked more positive ERPs in the 180–300 ms interval throughout the whole scalp as well as more negative ERPs in the mid-anterior regions in the 350–450 ms interval. These ERP components are claimed to be in agreement with previous ERP studies on overt speech production, as shown in results of meta analyses of phonological encoding and internal monitoring by Indefrey and Levelt (2004) and Indefrey (2011). | study | 99.9 |
fMRI evidence with Mandarin speakers also argued for the distinctive neural representations of phonemes and syllables (Yu et al., 2015; see also findings in Peeva et al., 2010 on distinct activation patterns for phoneme and syllable in French). For phoneme activation both studies show activation of the pallidum and putamen. Yu shows additional activation of the STg region, which seems specifically activated for Chinese languages (Fu et al., 2002). While we note that the interpretation of cognitive processes based on neurophysiological observations has to be taken with caution (Munding et al., 2016) the above mentioned studies seem to suggest that there is a potentially more important role of the segment in speech encoding for Sinitic languages than previously suggested, which raises the question of the mechanism that explains the lack/presence of phoneme repetition priming effect in the behavioral response time data. | study | 99.94 |
Qu et al. (2012) proposed an account that maintains segment as a functional unit of planning even in Sinitic languages which involves overriding phonological activation by a monitoring process (hereafter the Monitoring Account). This was endorsed by Yu et al. (2014). The effect during the 180–300 ms time window found by Yu et al. (2014) could be related to the P2, which reflects lexical access (e.g., Indefrey and Levelt, 2004; Hirschfeld et al., 2008; Costa et al., 2009; Strijkers et al., 2010; Aristei et al., 2011). The P2 is, for example, manipulated by cognate status with greater positivity for non-cognates than cognates (Strijkers et al., 2010 and descriptively reported in Christoffels et al., 2007). Due to the clear phonological overlap for cognates, both representations are strongly co-activated. A feedback loop sends phonological activation back to phonologically linked lexical representations, but this does not happen for unrelated words or non-cognates (e.g., Dell, 1986; Levelt et al., 1998; Costa et al., 2005). Taken together, smaller P2 seems to reflect easier retrieval of phonologically related words in general. The second component found by Qu et al. (2012) and Yu et al. (2014) could reflect self-monitoring, a process that cancels out the facilitations created in the P2 which leads to the lack of behavioral segmental onset facilitation. | study | 99.94 |
There are also alternative accounts explaining these observations. Roelofs (2015) proposed an account that recognizes the universal role of segmental planning and explains away the null effect of segment repetition priming in languages like Mandarin by assuming that segmental activation was hidden by the parallel selection of the other segments of the first syllable during the planning of the actual response (hereafter the Concurrent Retrieval Account). Another account, from a very different perspective, is to attribute the lack of RT evidence to the lack of intentional orientation toward segments in Mandarin Chinese and attribute the observed phoneme repetition ERP effects as an index of phonological connectivity rather than functional engagement of segments in preparation for production. This approach was proposed by O'Seaghdha et al. (2013) (hereafter the Connectivity Account), but was refuted by Qu et al. (2012) as neuro-physiologically infeasible. It is important to note that the participants in Qu et al. (2012) and Yu et al. (2014) are university students, who learned Chinese mainly via the Latin alphabet-based Pinyin system. Nowadays, they also mainly use the Pinyin input method on digital systems to type the logographic Chinese characters, which potentially could have boosted their sensitivity to phonemic representations of Chinese characters. | study | 99.0 |
More empirical data, especially those that tap both into the behavioral and neural patterns of phonological encoding, are crucially needed to resolve the debates. Therefore, the current study aims to extend the body of literature by examining Dutch-Cantonese bilinguals. The segment-retrieval hypothesis has been consistently supported in studies with Dutch-speaking individuals. On the contrary, Cantonese speakers have consistently shown a lack of initial phoneme onset priming, despite evidence that sub-syllabic units, such as consonant-vowel (CV) and rhyme, may serve as possible processing units (Wong and Chen, 2008, 2009; Wong et al., 2012). Specifically, we were interested in whether balanced Dutch-Cantonese bilinguals would demonstrate initial segment-based serial planning during speech production in their syllabic language (Cantonese) in the behavioral data. Furthermore, we were interested in how such an influence manifests itself in the ERP neural response patterns. In doing so, we aimed to bring in new data that may shed light on the existing debates concerning the universality of segment as a functional unit of speech encoding, and the possible mechanism that explains the mismatch of phoneme repetition priming effect between the ERP and behavioral responses. | study | 99.94 |
We used a naming task in which participants were asked to name the color of colored line drawings of objects in Cantonese1. The relationship between color and object was manipulated in such a way that the onset of the color and object was phonologically related or -unrelated (see Figure 1). We expected to see not only ERP differences (i.e., evidence of segment as a functional unit of phonological processing), but also behavioral facilitation (i.e., evidence of segment being the proximate unit of processing) as has been found in Dutch. Furthermore, we were interested in how ERP evidence on initial segment priming may differ from or confirm findings in Qu et al. (2012) and Yu et al. (2014). | study | 100.0 |
Example of a colored object in the two experimental conditions: phoneme match and phoneme mismatch. The Cantonese character, English translation, and jyutping (i.e., Romanized system for Cantonese). The picture has been adjusted from Severens et al. (2005). | other | 99.75 |
The data from 18 bilingual speakers of Dutch and Cantonese Chinese (four females; average age = 23.9; SD = 3.37) were used in the analyses. Out of the 23 subjects who participated in the experiment the data from five participants were rejected due to technical problems (n = 3), extremely slow responses (above 2 SDs of the group mean; n = 1), or being left handed (n = 1). All participants have normal or corrected-to-normal vision. None of them were color-blind or had a history of neurological impairments or language disorders. The Dutch-Contonese bilinguals were from the Netherlands and the experiment was conducted in the Netherlands as well. All participants were proficient in both Cantonese and Dutch. Cantonese was learned at home and at Cantonese Saturday school. For 14 participants, Cantonese was the mother tongue for both parents and for four participants one of the parents spoke Cantonese as a mother tongue. On average participants went to Cantonese school for 9.5 years (SD = 2.92). Dutch was learned at school as all participants were born in the Netherlands or moved there before the age of school. They attended Dutch schools and followed the same curriculum as other Dutch children. These characteristics of their language experience essentially render the participants as balanced bilinguals, as they are proficient early speakers of both Cantonese and Dutch. See Table 1 for an overview of their language proficiency as rated by a self-rated proficiency questionnaire adapted from Christoffels et al. (2007). | study | 100.0 |
Forty-eight black-and-white line drawings of objects were selected from various databases including Snodgrass Vanderwart (Snodgrass and Vanderwart, 1980), Els Severens (Severens et al., 2005), and Alario picture set (Alario and Ferrand, 1999). The color paired with the line drawing was not its canonical color (e.g., tomato was not paired with a red line). Each line drawing was presented in two of eight colors (red, orange, yellow, green, blue, purple, gray, and black) to create two conditions: (1) the first phoneme of the color and line drawing match in Cantonese (e.g., 藍駱駝, /laam4/ /lok3to4/, “blue camel;” phoneme-match), and (2) the first phoneme mismatches in Cantonese (e.g., 紅駱駝, /hung4/ /lok3to4/, “red camel;” phoneme-mismatch; see Figure 1). The color names in Cantonese were monosyllabic, and all picture names were disyllabic. Tonal mismatch between the color name and first syllable of the picture name was attempted. Due to the small number of depictable pictures and the phonological matching between color and object in Cantonese but not Dutch, there are some pairs which had tonal overlap, which, nevertheless are comparably distributed in the match condition (14 pairs) and the mismatch condition (10 pairs). Further, there are seven pairs that overlap in the second phomeme between the color and the noun (i.e., five match pairs and two mismatch pairs). Note that when the pairs with phoneme match on the second phoneme are the removed the results are the same (see results). All the stimuli are represented in the Appendix. | study | 100.0 |
The study was approved by the ethics review board at Leiden University. Participants first signed an informed consent form and filled out a self-rated language proficiency questionnaire. They were tested individually in a quiet room seated ~90 cm from the computer screen. The experiment was controlled by the software package E-Prime 2. Speech production onset was measured though an integrated voice-key (microphone). | study | 99.94 |
The experiment consisted of three parts: (1) learning phase (48 trials), (2) practice phase (48 trials), and (3) experimental phase (96 trials). Each phase was preceded by eight practice trials. During the learning phase, eight color patches, followed by 56 drawings of objects, were presented with their corresponding Cantonese characters. Both the color patches and drawings were presented in random order. The participant made sure they knew the names and the experimenter pressed a button to continue to the next color patch/drawing. During the practice phase, the same color patches and drawings were presented one by one in random order. Participants were asked to name them as fast and accurately as possible. During the experimental phase, the drawings were presented one by one in one of eight colors. Participants only named the color of color-object drawings as first used by Navarrete and Costa (2005). | study | 99.94 |
Each trial in the experimental phase consisted of a fixation-cross (400–700 ms), followed by a color-drawing that disappeared once the participant initiated a verbal response or after a time-out of 3,000 ms, after which a blank screen was presented for 2,000 ms. All pictures were 10 × 10 cm and centered on the screen. | study | 99.7 |
During the test phase each drawing was presented twice, once in the color where the phoneme matches the drawing and once in the mismatching color. All drawings appeared in each of two created blocks. In each block, half of the pictures were presented in the phoneme-match and the other half in the phoneme-mismatch condition. Blocks and trials were randomized over participants. | study | 98.3 |
The electroencephalogram (EEG) signals were sampled at 512 Hz and continuously recorded using 32 Ag/AgCl electrodes distributed according to the extended International 10–20 system. Two electrodes of the flat type (above and below the left eye) recorded the eye-blinks. Another two electrodes (external canthi of each eye) recorded horizontal eye-movements. The EEG signal was re-referenced to the mastoids (left and right; baseline). | study | 99.94 |
For the EEG analysis, epochs of 600 ms with an additional 200 ms pre-stimulus baseline were created. The EEG signal was filtered with a high-pass filter of 0.01 Hz/24 dB and a low-pass filter of 40 Hz/24 dB. Ocular artifacts were corrected using the Gratton et al. (1983) algorithm. Non-ocular artifacts were removed based on the following criteria: trials with amplitudes below −200 μV, above +200 μV, or made a voltage step of 100 μV within 200 ms. The ERP grand averages were time-locked to the onset of the target word and calculated separately for each of the two conditions over all participants. | study | 100.0 |
To avoid any a priori bias with respect to choosing time windows and localization for ERP analyses, which allows for vast number of comparisons, a multivariate statistical tool called partial least squares (PLS) was used (McIntosh et al., 1996; Lobaugh et al., 2001; Krishnan et al., 2011). All ERP data is submitted to PLS by importing microvolts for every ms from stimulus presentation to 450 ms for each electrode. This is done for all participants and conditions. Singular value decomposition (SVD) identifies a set of latent variables (LVs), that correspond to the strongest ERP patterns in the data based on our experimental contrast/conditions (match vs. mismatch). An LV explains how much of the covariance was explained by our experimental contrast. To visualize the LVs, the salience is computed and represented as design scores and salience maps. Design scores code the effects of the LV between −1 and 1 (see top part of Figure 2) and salience plots show where this effect is present spatiotemporally (bottom part od Figure 2). In the salience maps, the dots show for which electrodes and temporal windows the LV is explaining the covariance. Thus, the relation between the experimental design contrasts (represented by the LV) and the spatiotemporal pattern of ERP amplitude changes is represented by the electrode saliences in Figure 2. The estimate of obtaining a singular value by chance (similar to a p-value) was computed by 1,000 permutations. The reliability of electrode saliences at each time point was assessed by 200 bootstrap re-samplings, which applies random sampling with replacement. For examples of how PLS can be applied to EEG data, see Lobaugh et al., 2001; Hay et al., 2002; Düzel et al., 2003; Itier et al., 2004; Grundy and Shedden, 2014 and for an example for speech production ERP data see, Christoffels et al. (2016). For more detailed explanation of applications and formulas see McIntosh and Lobaugh (2004). In short, PLS analyses allowed us to narrow the time windows and locations of experimental effects in order to perform conventional ERP statistics. | study | 99.94 |
PLS results. The top part of the figure represents the design scores for the LV (y-axis). The bottom part of the figure indicates a PLS electrode saliency map showing the reliability of LV for the match vs. mismatch comparison. The x-axis represents time in milliseconds (0–450) and the y-axis represents electrode salience (i.e., reliability of the LV). | study | 99.94 |
The independent factor Phoneme condition (match vs. mismatch) was first examined with whole-brain PLS analysis within the interval. The LV suggests that the phoneme match and mismatch trials were processed differently (see Figure 2) and accounted for 100% of the variance, as this design only has one latent variable, p < 0.05. The electrode saliences, reflecting confidence intervals for salience across time points and electrodes, revealed that this effect was most reliable within the 125–175, 200–300, and the 300–400 ms time windows throughout the left temporal hemisphere (electrodes F7, FC5, T7, C3, CP5, and P7). Based on correspondence with PLS, we chose an analogous location in the right hemisphere for subsequent componential analysis. These time-windows were analyzed by a classic statistical ERP analyses with Phoneme condition (match vs. mismatch) and Hemisphere (left: F7, FC5, T7, C3, CP5, P7 vs. right: F8, FC6, T8, C4, CP6, P8) as independent variables and ERPs measured in μV as the dependent variable. The Greenhouse–Geisser correction was applied to all repeated measures to correct for possible violations of sphericity. Note that based on visual inspection of Figure 3, the 0–100 ms time window suggests possible differences, but neither PLS nor classic ERP analysis revealed significant effects of this time window. | study | 100.0 |
Averaged stimulus-locked ERP waveforms for the phoneme match (solid line; e.g., 藍駱駝 /laam4/ /lok3to4/, “blue camel”) and phoneme mismatch condition (dashed line; e.g., 紅駱駝 /hung4/ /lok3to4/, “red camel”) for each hemisphere including all electrodes used in the statistical analysis. A 20 Hz filter was applied for the clarity of the waveforms. | study | 99.94 |
The independent factor Phoneme condition (match vs. mismatch) with the dependent variable, RTs, were submitted to a repeated-measures ANOVA, which revealed that the colors of phonologically related color-object pairs (e.g., 藍駱駝, /laam4/ /lok3to4/, “blue camel;” 784 ms; SE = 30.14) were named 20 ms faster compared to phonologically unrelated pairs [e.g., 紅駱駝, /hung4/ /lok3to4/, “red camel;” 804 ms; SE = 32.27; F(1, 17) = 7.12, p = 0.016]2. | study | 100.0 |
Trials that included incorrect responses or electrophysiological artifacts were removed from the analysis. For the phoneme match condition 11.34% of the data was removed, leaving 88.66% in the analysis. For the phoneme mismatch condition 10.42% of the data was removed, leaving 89.58% in the analysis. | study | 99.94 |
The repeated-measures ANOVA revealed a main effect of Phoneme condition [F(1, 17) = 5.35, MSe = 22.24, p < 0.05] that interacted with Localization [F(1, 17) = 6.64, MSe = 1.08, p < 0.05]. The phoneme mismatch condition revealed greater positive amplitudes than the phoneme match condition throughout the left-hemisphere [F(1, 17) = 8.84, MSe = 10.44, p < 0.01; phoneme match: μV = 3.10; SE = 0.71 vs. phoneme mismatch: μV = 4.41; SE = 0.71] but not the right hemisphere [F(1, 17) = 2.63, MSe = 12.88, ns; phoneme match: μV = 2.98; SE = 0.78 vs. phoneme mismatch: μV = 3.78; SE = 0.83; see Figure 3]. | study | 100.0 |
The repeated-measures ANOVA revealed a main effect of Phoneme condition [F(1, 17) = 4.82, MSe = 30.14, p < 0.05] that interacted with Localization [F(1, 17) = 5.15, MSe = 1.67, p < 0.05]. The phoneme match condition revealed greater negative amplitudes than the phoneme mismatch condition throughout the left-hemisphere [F(1, 17) = 8.62, MSe = 13.03, p < 0.01; phoneme match: μV = 5.37; SE = 0.97 vs. phoneme mismatch: μV = 6.81; SE = 0.95] but not the right hemisphere [F(1, 17) = 2.22, MSe = 18.78, ns; phoneme match: μV = 5.67; SE = 1.14 vs. phoneme mismatch: μV = 6.55; SE = 1.17; see Figure 3]. | study | 100.0 |
The repeated-measures ANOVA revealed a main effect of Phoneme condition [F(1, 17) = 4.45, MSe = 28.36, p = 0.05] that interacted with Localization [F(1, 17) = 9.73, MSe = 1.49, p < 0.01]. The phoneme match condition revealed greater negative amplitudes than the phoneme mismatch condition throughout the left-hemisphere [F(1, 17) = 9.63, MSe = 11.74, p < 0.01; phoneme match: μV = 7.11; SE = 1.02 vs. phoneme mismatch: μV = 8.55; SE = 1.01] but not the right hemisphere [F(1, 17) = 1.52, MSe = 18.11, ns; phoneme match: μV = 7.19; SE = 1.08 vs. phoneme mismatch: μV = 7.91; SE = 1.17; see Figure 3]. | study | 100.0 |
The present study investigated segment-based serial planning mechanism during syllabic language (Cantonese) speech planning by Dutch-Cantonese bilinguals. Our results show a behavioral facilitation for phoneme onset sharing in Cantonese, a syllabic language, for Dutch-Cantonese bilinguals. This is in contrast to behavioral results with monolingual Cantonese speakers where segmental onset facilitation is normally absent (Wong and Chen, 2008; Wong et al., 2012). Our finding is probably due to the native proficiency of our participants in Dutch, a segment-based language in which phoneme onset effects are found (Schriefers et al., 1990; Meyer, 1991; Meyer and Schriefers, 1991). Their experience with Dutch facilitates segment production in Cantonese. Thus, our bilinguals seemed to be able to utilize their segment-oriented production strategy used in Germanic languages and apply it to their syllabic language, Cantonese. This is in line with a previous study showing that Mandarin-English bilinguals with a good proficiency in English can demonstrate initial segmental facilitation during reading aloud in Mandarin under specific conditions (e.g., same tonal pattern). However, Ida et al. (2015) were not able to replicate this effect with proficient Japanese-English bilinguals in Japanese. Thus, there is mixed evidence that the unit size of English as a second language can exhibit influence on that of L1 speech processing. The present study is the first to show phoneme onset facilitation in Cantonese speech production for native speakers of a segment-based language. This may suggest that the primary processing unit in a specific language is dependent on the speakers's general language background. Within spoken word recognition bilinguals have also shown to use the unit of lexical access from their dominant language in their second language if it is an efficient processing strategy for that language (Cutler et al., 1989, 1992). | study | 100.0 |
With regard to the ERP neural responses, our bilingual DC speakers showed earlier activation of ERP components compared to previous studies with bilingual Mandarin-English speakers. The bilinguals in the present study are native speakers of a Germanic language, while in previous studies second language learners of a Germanic language were investigated (Qu et al., 2012; Yu et al., 2014). Thus, differences in proficiency may explain a divergence of activation in the ERPs. Specifically, both Yu et al. (2014) and Qu et al. (2012) reported the first significant differences between phoneme match and mismatch conditions from ~190 ms after picture presentation. In the present study, however, phonological processing was first revealed during the 125–175 ms time window. Phoneme mismatch trials induced greater positive amplitudes compared to the phoneme match condition. This component could reflect a P2 with easier retrieval of phonologically related words in general. While a meta-analysis of neurophysiological studies has demonstrated a pattern of processing steps of among others lexical access and phonological encoding during speech production (Indefrey, 2011), it has to be noted that the link between cognitive operations and neurophysiological observations has to be taken with caution (Munding et al., 2016). | study | 99.94 |
While Qu et al. (2012) only analyzed components after ~190 ms in Figure 1C of their article, visual inspection of the figure suggests that an earlier P2 is present around ~125 ms after picture onset with a slightly greater positivity for the phoneme mismatch than match condition in the anterior regions. This is in line with the results in the current study, although it needs to be confirmed with further analysis of this P2 component. The trend in Qu et al. (2012) and the results of the present study suggest a primary role for segmental planning early on during lexical access for highly proficient Dutch speakers in the syllabic language of Cantonese. A potential though not robust effect of segment overlap on P2 in Qu's study (visually also smaller than that in the present study) could be attributed to the experience of her participants with Pinyin as well as English which are both segment-based and could have introduced certain degrees of enhanced sensitivity to the phoneme. | study | 100.0 |
Yu et al. (2014), however, revealed a later onset of the P2 component, around 180 ms after picture onset, with the opposite pattern of conditions (i.e., greater positivity for the phoneme match than mismatch condition). This pattern is probably due to the fact that they used a different paradigm than color only naming in the present study, or color-object naming, used in Qu et al.'s (2012) study. During their experiment, disyllabic pictures were named one after the other, while the phonological onset relationship was manipulated between consecutive pictures. Therefore, the temporal distance between the prime and target picture presentation includes the inter-trial-interval (ITI; 600 ms) and fixation cross (500 ms). The presentation of the phonological prime at least 1,100 ms before the target created a different situation from the previously mentioned studies where the prime (color) and target (object) are presented at the same time. Therefore, it is plausible that the early segmental effect in the present study is introduced during lexical access (P2), as the bilinguals adopted the segmental encoding which they acquired by speaking a segment-oriented language (i.e., Dutch). | study | 100.0 |
The P2 is followed by a negative component between 200 and 300 ms with greater negativities for the phoneme match compared to the phoneme mismatch condition. This component, commonly named N2 (or descriptively named N3 by Strijkers et al., 2010), reflects phonological encoding during language production. During this phonological encoding stage speakers continuously monitor whether their phonological output is correct. Multiple lexical or phonological representations are available during this process. Therefore, conflict can arise from co-activation of phonological representation of color and object. In order to correctly produce an utterance the conflict must be resolved. This conflict could possibly be resolved through inhibition. This is similar to effects found with cognates vs. non-cognates (Christoffels et al., 2007; Strijkers et al., 2010; and descriptively reported in Verhoef et al., 2009) and during a word interference paradigm (Hoshino and Thierry, 2011). When producing a lexical item other lexical items or phonological representations are activated as well and this has to be resolved. Despite the conflicts, all of the above mentioned paradigms show behavioral facilitation for phonological matching conditions. The N2 effect has also been related to general response inhibition (e.g., Pfefferbaum et al., 1985; Jodo and Kayama, 1992; Thorpe et al., 1996) as well as to response conflict monitoring (e.g., Nieuwenhuis et al., 2003; Donkers and Van Boxtel, 2004). Thus, our results can be taken as evidence that the phonological overlap between color and object causes inhibition or response conflict monitoring during the N2 component, in line with the existing literature. | study | 100.0 |
The last time window, 300–400 ms, coincides with that of Qu et al. (2012) and Yu et al. (2014) who also investigated segment sized phonological activation during speech production in syllabic languages. In their monitoring account they suggest this component could reflect self-monitoring. A higher cognitive load is suggested to be present for the phoneme-related condition, because they can cause speech errors, and is suggested to cancel out the small segment facilitation effect in the behavioral data where no segmental-priming effect is observed (Qu et al., 2012; Yu et al., 2014). | study | 100.0 |
To summarize, the present study revealed behavioral segmental onset facilitation as well as early facilitation of repeated segment-sized phonemes during lexical access, reflected in the P2 component. This is followed by two negative components where overlapping phonology causes inhibition and additional self-monitoring. This raises the question of how onset overlap can be named faster than mismatched cases while introducing more response conflict during the later ERP components, especially given the findings in research with manual responses that have demonstrated delayed responses after conflict (e.g., Tillman and Wiens, 2011). There seem to be two important differences between manual and speech response. First, speech responses are not as diverse as manual responses. During manual responses there are usually two completely different responses (e.g., left vs. right hand response), while during speech production, response options involve similar vocal tracts and articulators. Second, the motor responses of speech develop differently over time than manual responses. Facilitative relationship of the two phonologically related representations can be observed earlier in time. Only later the two representations start diverging (Acheson et al., 2012). This could explain why the later competition (demonstrated in the ERPs) in our data does not have the same impact on speech onset latencies as on the speed of manual responses in the case for non-linguistic tasks on the speed of manual responses (e.g., Tillman and Wiens, 2011). | study | 100.0 |
Further, the P2 present in the current experiment also showed a visual trend in Qu et al.'s (2012) study but with a seemingly smaller differences of P2 compared to the current study. It could therefore be that early activation of the segment was not strong enough for the speakers of Mandarin in their study to show behavioral facilitation, probably due to the lower level of their proficiency in a segment-oriented language (i.e., English). In contrast, our bilingual DC speakers grew up in the Netherlands and so, their Dutch has reached native/near native proficiency. Thus, while our study lends support to the proposal in Qu et al. (2012) that segment can serve as a functional unit of planning even in Sinitic languages such as Cantonese, we believe that the lack of phonological activation (as indexed by the reaction time) in their study is not likely explained by the Monitoring Account. With regard to the debate between the Concurrent Retrieval Account (Roelofs, 2015) and the Connectivity Account (O'Seaghdha et al., 2013), neither would have predicted the behavioral phoneme onset facilitation effect in our bilingual speakers. Further experimental evidence is needed to test the extent of their feasibility. | study | 100.0 |
To conclude, the behavioral data, together with the P2, provided evidence that Dutch-Cantonese bilinguals used the segment as the primary unit of planning during Cantonese speech planning. The results of the present study are compared to previous studies. However, a direct comparison with the present study was not possible due to task differences and participants' different language experiences. For example, we reported earlier ERP evidence during the lexical accessing stage for segmental processing than previous research with unbalanced bilinguals, who are less proficient in an alphabetic language (Qu et al., 2012; Yu et al., 2014). Our results have been interpreted and discussed within the existing literature and theories. In future research, a more direct comparison is preferable. Specifically, our proposal that proficiency of bilingual speakers in one language can modulate the adaptive behavior of processing unit in another certainly needs corroborative evidence from studies that tap directly into the effect of proficiency level on phonological encoding, preferably with planned comparisons of processing patterns in different languages within the same bilingual populations. In this way, further insights into the cross-linguistic functional unit(s) of phonological planning can be obtained. Future endeavors should also explore whether a timing difference can be shown between segmental and syllabic activation within and between Germanic and syllabic languages. To conclude, the language background of bilinguals can have a significant impact on the primary processing unit during speech production in a specific language. | study | 99.94 |
KT and YC: substantial contributions to the conception and design of the manuscript: interpretation of data for the manuscript; drafting the manuscript and revising it critically; Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. | other | 99.94 |
The first years of a child's life are characterized by constant biological and psychosocial changes, which lead to major acquisitions in the motor, social-affective, and cognitive domains.1 During this period, the central nervous system (CNS) is constantly changing, myelination and synaptic organization reach the peak at 2 years of age, favoring the learning processes.2 , 3 However, the CNS is not the only factor responsible for motor development, it is also related to the musculoskeletal development and cardiorespiratory fitness, all influenced by stimuli and environmental factors.4 - 7 There is evidence that low socioeconomic status8 - 11 and family and school environment of poor quality can negatively influence the development of healthy children,8 , 10 , 12 , 13 while favorable environmental conditions, such as adequate stimuli, higher maternal education and socioeconomic status seem to positively influence children's motor and cognitive development.8 , 14 - 16 | review | 99.75 |
As for environmental factors, it is known that, since the late 1970s, women have become part of the labor market and required a place to leave their children during work hours. So, children began to spend much of their day in a different environment. However, in Brazil, with the approval of a national Law in 1996 (Lei Nacional de Diretrizes e Bases da Educação Nacional-LDB), the day care centers no longer have a social welfare aspect, but an educational character and are responsible for the “integral development of children up to 6-years old, in their physical, psychological, intellectual and social aspects, complementing the action of the family and of the community”(Art. 29 LDB). Rossetti-Ferreira et al.17 highlight that children insertion in day care centers offers a possibility of additional stimuli, as they interact with other children and caregivers. However, this benefit is directly related to the quality of the child care provided. | other | 98.56 |
Under this new perspective, concern for the environment as a delineator factor of development in the early years of life has led some researchers to question the influence of the school environment as a space for children development,2 , 8 , 18 - 21 since the lived experiences of those years are related to the cognitive22 , 23 and motor5 , 24 development in subsequent years, many children have the day care center as their main source of stimulation and interaction. In this context, it was found that day care centers with adequate equipment, good quality in the care and teaching methodology had positive influence on children development.2 , 15 , 25 However, there are few studies assessing the school environment influence on fine motor skills of children at 2 years of age, period in which they begin to develop greater independence in daily activities and ability to use their hands functionally. Studies evaluating the fine motor performance among frequenters of Brazilian nurseries attributed the poor performance of children to the lack of quality of day care centers in Brazil.2 , 18 , 26 However, this conclusion was based on studies that evaluated only the quality of day care centers in Brazil, without relating their impact on children's performance. | review | 99.7 |
In a previous study that assessed the performance of children attending public and private day care centers, all with level B of socioeconomic status and aged 0 to 3 years, children attending public day care centers showed cognitive and fine motor performance lower than those in private day care centers. These results was attributed to possible structural and pedagogical differences between the types of day care centers, as the children showed no changes in organic systems and were of the same socioeconomic level. This study, however, did not evaluate the quality of the day care environment.27 | study | 99.94 |
To the best of our knowledge, there are no studies evaluating school environment quality and assessing its influence on motor performance of children. In an attempt to fill this gap in the literature, the aim of this study was to investigate the impact of extrinsic factors, represented by the quality and length of stay in day care, level of maternal education, and family socioeconomic status, on fine motor development of children aged two years. Considering that several studies indicate the influence of environment on the development of biologically healthy children,18 , 28 our hypothesis is that the higher the level of maternal education and the family socioeconomic status and the better the quality of day care, better will be the fine motor development of children at 2 years of age. It is also expected that children early enrolled in day care centers would have a better fine motor performance, since at 2 years of age the most motivated activities at home are related to gross motor skills and, in day care centers, fine motor skills are encouraged.27 | study | 99.94 |
After surveying the Child Education Municipal Centers (CEMEIS) and private day care centers of a medium-sized city in the State of São Paulo, it was found a population of 570 children aged 2 years, 470 (83%) attending the municipal network and 100 (17%) the private schools. These children are distributed in 40 day care centers: 22 (55%) in municipal institutions and 18 (45%) in private day care centers. After sample size calculation for finite populations, with two-tailed alpha of 5% and 95% confidence interval, a sample of 80 children was estimated. This calculation does not require the definition of an outcome. | study | 99.9 |
Children of both sexes, born at term (40.1±2 weeks), appropriate weight for gestational age (3.2±0.5kg), with Apgar score>7 in the first and fifth minutes of life, and within the age of two years (23±3 months) were included. At the evaluation time, all children presented with normal weight percentile and appropriate height for chronological age, according to the World Health Organization,29 and attending full-time day care centers for at least six months (±14.8 months). Children with neurological disorders, genetic syndromes or congenital malformations were excluded. Parents or guardians have authorized their participation and gave written informed consent. Children were excluded from the study if, during the evaluation, they were crying or showing irritation that could prevent the application of tests or any health condition that could hinder the evaluation. | study | 99.94 |
After approval by the Institutional Review Board of the Centro Universitário Central Paulista (No 31/2011), all full-time day care centers of a city in the interior of São Paulo who served children aged two years were invited to participate. Among the 13 public day care centers, three were excluded due to lack of space for motor performance evaluation. Among the 28 private day care centers invited, only nine agreed to participate. There is no way of knowing whether there are significant differences between the evaluated day care centers and the ones who refused to participate, as the researchers were prevented from entering the non-participating centers. In the participating day care centers, all children aged two years were invited and evaluated all children whose parents or guardians gave WIC. | study | 99.94 |
Data collection consisted of physical evaluation and application of a questionnaire to parents to gather birth data, health conditions, age enrollment in day care, and maternal education. For assessment of the family's socioeconomic status, the ABEP questionnaire (Associação Brasileira de Empresas de Pesquisa/Brazilian Market Research Association) was used for economic classification.30 The questionnaire uses characteristics of the home (number of TV, radio, bathroom, car, housemaid, washing machine, DVD, refrigerator, freezer, and education level of the head of the household) to quantify the socioeconomic level at A, B, C, D or E.30 | study | 99.94 |
The children were called in their rooms and sent to a place determined by the day care direction for motor performance evaluation and anthropometric measurements. It is noteworthy that the researcher observed the hours of sleep, feeding, and bathing proposed by the institutions, and if a child was crying and refused to do the activities, he/she was immediately conducted to his/her room. | other | 99.9 |
BSITD-III is a reliable developmental rating scale, which is validated for children 0-42 months31 to measure cognitive, motor, language, social-emotional, and adaptive behavior. BSITD-III application occurs according to the child's age, with the first evaluation task corresponding to the age. In the present study, only the fine motor domain was evaluated. The tasks were scored by assigning 1 when the child was able to do the activity, attend certain requirements in the scale manual.31 The task was not scored (score 0) when the child did not perform or performed improperly the activity. When starting the test, the child was asked to perform three consecutive activities, otherwise, the researcher returned to the entry corresponding to the previous age, until the child perform correctly three consecutive activities. The assessment was stopped when the child did not perform five consecutive activities. The final score was transformed into a standard score ranging from 1 to 19 points, according to the tables contained in the manual.31 According to the BSITD-III, a mean of 10 and a standard deviation of 3 are considered. However, for the present study, the score is reclassified into 5 levels: below the average (<7); medium-low (7-9); medium (10); medium-high (11-13); and above average (>13). For BSITD-III application, three researchers were trained and obtained an inter-observer agreement rate of 98%. | study | 100.0 |
ITERS-R was developed by Harms et al.32 to assess day care centers and is a valid and reliable measurement tool of physical quality and human resources in these institutions. It consists of 39 items grouped into seven subscales to evaluate the school environment: space and furnishings, personal care routines, language and literacy, learning activities, interaction, program structure, and adult needs. | other | 99.9 |
Each item score varies from 1 to 7: inappropriate (1-2); minimum (3-4); good (5-7). The score for each subscale and the overall average of these scores were calculated. For scale application, three researchers were trained and obtained an inter-observer agreement rate of 96%. | study | 68.5 |
For data analysis, fine motor performance was divided into five categories according to the BSITD-III manual, as mentioned above.31 However, we found no result below average, and four categories were assessed. Data regarding the day care center quality were also reclassified into two categories: good quality (score≥5) and poor quality (score<5). Data were processed using the SPSS Statistics 17 software (IBM Corporation, New York, United States of America). For normality test of fine motor performance, the Kolmogorov-Smirnov test was used, which indicated the non-normality of the data. Spearman's test was used to assess the relationship between fine motor development of children and extrinsic factors: quality and length of stay in day care centers, maternal education and socioeconomic status of the family. For correlation analysis, we used Dancey and Reidy classification33: rs=0.10-0.30 is a weak association; rs=0.40-0.60 is a moderate association; and rs=0.70-1 is a strong association, “s” refers to Spearman's test. A 5% significance level was adopted. | study | 100.0 |
Table 2 shows the results of the correlation between fine motor performance in tasks and the BSITD-III standardized score, maternal education, ABEP classification, day care time, and the school environment characteristics. Positive and weak correlations were found between fine motor activities and day care time, maternal education and fine motor performance (rs=0.247; p=0.017), maternal education and the ITERS-R space and furniture items (rs=0.327; p=0.001), routine personal care (rs=0.352; p=0.001), language and understanding (rs=0.294; p=0.004), activities (rs=0.464; p=0.000), interaction (rs=0.253; p=0.015), program structure (rs=0.381; p=0.000), parents and staff (rs=0.464; p=0.000), and overall quality (rs=0.294; p=0.004). | study | 100.0 |
The BSITD proposed activities for this age group involve stacking blocks, drawing simple shapes, shape sorter, build three-dimensional figures with blocks, and cut paper.31 Children attending day care for a longer period of time had higher scores in these activities. Children stay in day care is still quite controversial in the literature. A study11 indicates that a more extended time in day care is harmful to children's development, particularly among children attending early care. However, Votruba-Drzaletal34 report that it is harmful only when there is low quality of care. The relationship found in the present study shows that although such tasks may have been stimulated at home, the extended time in day care favored the acquisition of several important skills for fine motor development and school learning. | study | 99.94 |
In this study, we found a positive relationship between maternal education and the environmental quality of day care centers attended by children; it was shown that mothers with higher education choose higher quality day care for their children. Similarly, Sylva et al.14 reported that mothers with higher levels of education choose school environments that are richer in stimuli for child development, as they understand the day care center as an important environment for development, and not just a child-care place during their working time. Regarding the socioeconomic factor, there was no correlation with the child's performance, although it is a risk factor well described in the literature.8 , 11 , 28 It is believed that this lack of correlation is due to the small amount of children in socioeconomic class A. This prevented the representation of higher class children in this study. | study | 100.0 |
There was no association between the overall quality of day care center measured using ITERS-R and fine motor development of children. The small number of children attending good quality day care may have contributed to these results, as 76% of the children attending good quality day care had above average performance, while that percentage was 58.8% in the population attending poor quality daycare. Authors who identify the school environment as a risk factor for child development reported routines with predominance of activities dedicated to nutrition and hygiene.2 , 8 , 19 , 21 However, when considering the quality items of classrooms using the ITERS-R, only the item oral language and understanding were related to fine motor performance. This item assesses the opportunity for children to express themselves in the classroom, how teachers foster children communication and how they strive to be understood. Appropriate language stimulation enables the interaction of children in the school environment, favors their stimulation and creates links between them, the teacher, and classmates, which promotes child development in general13 , 28 and appears to be linked to fine motor performance of the assessed children. Another aspect related to language is the influence of the ability of children to communicate and understand tasks during the fine motor performance assessments according to Bayley, as during the assessment they needed to understand the commands of investigators and repeat the proposed activities. Children who could better understand the proposed task could do it more easily, while others may have failed to do the activities for failing to understand what was proposed. According to the Embodiment cognition theory,35 children build the representation of objects through their physical interaction with them and, in this process, multiple factors interact, including brain and body, context and prior learning. Thus, it is believed that children in day care centers who received better language stimulation performed better in the evaluation because they understand and know better the proposed tasks. | study | 99.94 |
It is believed that the other items assessed in day care centers have no correlation with fine motor performance due to the small number of good quality day care centers, particularly in the item related to the quality of different pedagogical activities developed in day care centers, such as music lessons and fine and gross motor activities, among others. This is due mainly to the difficulty of access to private day care centers, as few agreed to participate in the study due to the assessment time and difficulty of reconciling the activities in day care with the BSITD application. | study | 99.94 |
Another important factor to highlight is that the school environment influence is often reported on cognitive development. Thus, the issues assessed with the use of ITERS-R may not have addressed important points for fine motor development of children, as the scale generally approaches aspects related to cognitive stimulation in children, such as interaction with teachers, language stimulation, and established routines, among others. Therefore, it can be considered that there is lack of instruments to evaluate the school environment focusing on motor development of children and scales that are more representative of motor performance are required. | other | 99.6 |