Music And Memory: The Feeling Of Knowing In Pop Tunes

Music and Memory

The Feeling Of Knowing In Pop Tunes

Master Thesis Jessica Akkermans

Jessica Akkermans Dr. M. Sadakata

14-08-2015 University of Amsterdam

Dr. B. Janssen

Index Preface...3 Abstract...5 1. Introduction...6 1.1 Feeling of Knowing...6 1.2 Listening Types...13

1.3 Music and Memory...17

1.4 The current Study ...20

2. Experiment 1 (pilot study)...25

2.1 Background...25

2.2 Method...25

2.2.1 Participants...26

2.2.2 Stimuli and Tasks...26

2.3 Results...26

3. Experiment 2 (main study)...27

3.1 Method...27

3.1.1 Subjects and Materials...27

3.1.2 Design...27

3.1.3 Procedure...28

3.2 Results...30

3.2.1 Correct Response Rates ...30

3.3.2. FOK Analysis ...30

3.3.3 Correlating Listening Strategies...32

3.3.4. Supplementary Results...32 3.3 Discussion...34 4. General Conclusion...37 5. Final Remarks ...41 Acknowledgement...42 References...43

Preface

Almost everyone encounters it. That moment when someone asks you the name of this or that song. You know the answer and are about to answer but for some reason the title, that you are certain you know, cannot seem to pass your lips. You have a feeling you know the answer but cannot reproduce it at that moment. You think if you would have more time or a clue, like the first letter or maybe the artist of the song, you would be able to come up with the answer. What you are experiencing is called The Feeling Of Knowing (FOK). The FOK does not only occur in music but you can also experience this feeling during everyday moments. Movie titles, city names, car brands are just a few topics in which the FOK can occur.

I noticed that I am often in a FOK state and came to understand that I was not the only one who experiences this feeling. Being a student, I often go to bars to do a pub-quiz with my friends. In a pub-quiz people play in teams and are asked different questions on different topics. The topics can range from film and geography to news and music. During the quiz the teams compete against one another by getting the most correct answers. While playing these quizzes it is almost inevitable to get a FOK moment. This gave rise to questions such as; why for one moment we know the answers immediately while for other moments we are annoyed with ourselves because we are sure we know the answer but why are we not able to produce the words? Do some people encounter it more often than others? Is it connected with our intelligence or (when music related) musicality? Are there ways to avoid having a feeling of knowing?

The topics of music and memory and music and emotion have been researched often and new ideas are still brought into the field. Music and emotion gives rise to questions like; why does a song make us cry? Is there an explanation for why we feel happy while listening to a song and sad the next? Why do we sometimes feel that music can help us deal with our emotions? Because emotion is closely linked to memory, this current study used different articles about both subjects. The article Music, Memory And Emotion by Jäncke (2008), became the starting point for the topic of this study. In the article, Jäncke (2008) stated “we are, however, better at recalling the titles of the tunes we listen to (when the tunes are instrumental) than at remembering a melody by simply reading or hearing its title. The opposite pattern occurs when remembering vocals, for which the titles of the songs are much better cues than the melodies.” This was an interesting statement and more research on this topic showed that there were several studies related to this statement in studies about the FOK.

All this became the subject of this master thesis for the master Art Sciences, at the

was on the FOK and the relation between lyrics and melody and peoples listening strategies. It seems that during a song some people are able to sing along with the lyrics in a very short time, even if it is the first time they hear a song. Some are also good in remembering, not just the title or artist of the song but also a great deal of the lyrics. After playing a pub-quiz one night it came to mind that not everyone is able to remember these things about a song. Some people could recognise the melody but were often not able, or maybe not interested, in recalling the title of the song. For this study we developed an experiment to test if there is a difference in the recall and recognition of pop songs between people who listen to lyrics and people who mainly focus on melody. All

participants listened to short song recordings and were presented with titles of songs. In the first trial subjects had to write down the title of the song when hearing the recording and had to sing or hum the melody when presented with the titles. If they did not know the song they had to give a FOK rating for how likely they would know the answer when presented with more information. A second trial followed where participants had to do a forced-choice recognition test in order to check the accuracy of the FOK ratings that participants gave in the first trial.

The main research question was: Does people’s listening strategy influence their feeling of knowing rating patterns?

Abstract

Most research on FOK judgments has been done in relation to language and general knowledge. The relation between music and the FOK, however, has not been studied much. Studies have proven that people are relatively accurate in their FOK predictions. For this study we were interested to see if a distinction could be made between people who mainly listen to melody and people who mainly listen to lyrics and if their listening strategy influences their FOK judgments and recall for pop songs. For this experiment 30 participants were individually examined and were tested in two parts on song recall and recognition. If recall failed, FOK judgment were made. The results indicated a strong main effect of task and marginally no significant main effect of self-indicated listening type The interaction between two factors was significant. Analysis indicated that FOK ratings for melody-write task was significantly higher than title-hum task for some participants. The results were in line with previous findings. People's listening strategy had no effect on their FOK judgments, possible explanations for the findings are discussed.

1. Introduction

It is almost unimaginable to think of life without music. We play and listen to it daily and with techniques such as radio and new media, we are able to choose what kind and when we want to listen to music. It causes physical reactions like shivers and goosebumps (Jäncke, 2008; Ward, 2006) but can also trigger emotional reactions (Jäncke, 2008; Ward, 2006; Ter Bogt, Mulder, Raaijmakers & Gabhainn, 2010). There are different reasons why we listen to music. It can improve our mood, help us cope with stress and can help us create a social identity (Ter Bogt, et al., 2010; Roe, 1985; Zillmann, 2000). This influences the way we listen to music and varies per individual (Roe, 1985). People who listen to punk music for example tend to focus on the lyrics (Roe, 1985). People's intention of listening is connected with our musical memory. Processing the music we hear and the intention we have while listening, can have an influence on what we remember about a song. People focus on different things in a song when hearing it. Some people focus on melody and are able to hum along with a song. Others focus on rhythm and can easily tap along with the rhythm section. While some focus on lyrics and are able to recite parts of the lyrics. After hearing a song we are able to remember it and can usually tell when we heard that particular song, for example how old we were or what stage of life we were in (Lippman & Greenwood, 2012; Stalinski & Schellenberg, 2013; Korenman & Peynircioğlu, 2004). However, sometimes even though we think we know a song, we are not able to reproduce its title. We can remember when we listened to it, how we felt while we heard it, but for some reason the title seems to have slipped our mind. This thesis will discuss different topics related to this temporary inability to recall something. First it will look at the FOK in general and discuss previous studies and theories that have been provided by others. A short side-step will be made to an occurrence that is very similar to the FOK state, the tip of the tongue state. Secondly, I will discuss the topic of music listening and people's reasons for listening to music. Thirdly, a connection will be made between the FOK, music and memory. This is then followed by background information of the experiment of the current study, followed by the main study itself, which looks into the relation of listening strategies and memory recall and the accuracy of FOK judgments. After the current study, the results of the experiment will be discussed and finally, the thesis ends with some final remarks

1.1 Feeling of Knowing

The feeling of knowing, we all experience it. You are listening to a song, you have heard it before and unconsciously you start humming or singing along with the music. Someone notices it and asks what song you are singing. You become aware that you were singing and want to answer the question. But then. Nothing. You cannot seem to produce a name even though you know it is somewhere in your mind, just waiting to be said out loud, at the tip of your tongue. You remember certain things about the song and are sure that you know the title but just not now. This state of knowing something but not being able to produce it is called the Feeling Of Knowing.

In 1965 graduate student Joseph Hart, at Stanford University, was the first to systematically explore FOK judgments by doing experiments. Hart devised a method that he called the recall-judge-recognition (RJR) paradigm (Hart, 1965; Dunlosky & Metcalfe, 2008). In his research, participants were presented with a variety of general information questions. Questions that were answered correctly were not used for further research. However, when participants were not able not answer a question or gave an incorrect answer, they were asked whether they thought they would be able to choose the correct answer on a multiple-choice task. The accuracy of their prediction corresponded with their correct response in a recognition task, which consisted of a multiple-choice task, that followed after the prediction. So participants first did a recall task, followed with a judgment if they would be able to recognise the correct answers and ended with a recognition task. Therefore calling it the recall-judge-recognition paradigm. Interestingly, the findings showed that participants even though not being able to recall the correct answer, were accurate in their prediction. The recognition accuracy was higher when they predicted that they would recognise it later as compared to when they predicted that they would not know the correct answer. What was surprising about this, at the time, was that people were able to judge if the answer was in their memory even if they could not recall the answer at that moment. How this was possible raised questions and triggered new research to be done on peoples' metacognition (the knowing about knowing), because how was it possible that people did not know the answer but were still able to predict accurately whether they would know the correct answer in the future (Metcalfe & Dunlosky, 2008; Hart, 1965)?

The concept of FOK judgments has since been a topic of interest among several fields of research and has been related with the concept of meta-memory and intuition. Interestingly, the FOK state relates to the knowing of the unknown. It covers our ability to say we know something with the inability to actually produce it (Brown, 1991). The first ideas on the FOK were based on accessing whether certain information is present in the memory and how people are aware of their

knowledge. According to Asher Koriat (2005) the FOK we experience, when asked a question related to memory, is based on a variety of cues. One of these cues is parts of information that are able to be retrieved from the targeted question. Even though the actual answer cannot be recalled, we might be able to remember certain characterisations of the answer, like that it contains the letter K, or that it is a short or long word (Koriat, 2005). Despite not being able to recall the target, people can be very confident that they know the answer and that recovering the answer is imminent (Brown & McNeill, 1966; Brown, 1991).

The RJR paradigm is now a well-known method that is used to find the connection between accuracy of predictions and the actual recognition. In recent years it has become an overall accepted fact that our predictive accuracy is correct and therefore the RJR paradigm has become a reliable testing method. Studies now no longer focus on the correctness of the predictions but rather focus on the understanding of what type of cues people use to get the wanted answer (Dunlosky & Metcalfe, 2008). There are different theories that address the way people are able to correctly predict what they know now and what they will know in the future. According to Metcalfe, Schwartz & Joaquim (1993), both the FOK state and the successful retrieval of answers occurs when there is a high level of cue familiarity. This leads to a strong confidence that the target can be retrieved. The difference between successful retrieval and the FOK is that during successful

retrieval the actual target is found and is accurate but that during the FOK state there is only a conviction that it is represented in the memory but not retrievable at that moment (Zuckerman, Levy, Tibon, Reggev & Maril, 2012). There is reason to believe that people assess their FOK judgments on the basis of familiarity of the question. The more familiar the question is the more confident people are that the target will be recalled at a later moment. Dunlosky and Metcalfe (2008) gave an example of what they called, the domain familiarity. For example, to the question “Who painted The Sunflowers?”, people may not have an immediate answer. But they are able to assess what they know about the question and can thus make judgments on what they know about, for example, the domain of art. When they are relatively familiar in this domain they may be able to attain the answer by narrowing down some answers that they know are incorrect and eliminating the incorrect alternatives in a possible recognition test. Dunlosky and Metcalfe (2008) conclude that strategic multiple-choice decisions are better made in well-known domains as compared to domains in which the person is not familiar. They stated that, “The person may not know who painted a particular painting but may nevertheless have a quite good idea of who did not do so, and such knowledge will help them on the test” (Dunlosky & Metcalfe, 2008, p. 351). Other information that can be cued is the knowledge someone has about the target, but this can still lead to the inability to retrieve the correct answer. For example, someone might remember that there is a letter 'G' in the

name of the answer or in what year the painting was made. This partial information may lead to a high familiarity and lead to a high FOK judgment. The retrieving of such partial information can thus be used to indicate that people will be able to answer the question correctly and is therefore related to high FOK judgments (Dunlosky & Metcalfe, 2008).

Another study on the focus of domains was done by Glenberg and Epstein (1987). In their experiment people were selected based on their expertise in different domains. Subjects were presented with texts that they had to read. The context of the texts was either about their domain of expertise or on another topic. The general finding of the experiment was that people made higher judgments of FOK on parts in the text that were within their own domain of expertise. They

concluded that the knowledge people have about their own domain is one of the cues that influences people when making their predictive judgments (Glenberg & Epstein, 1987).

More research on cue familiarity was done by Reder and her colleague, Ritter (1992). They conducted a series of experiments in which they showed that the familiarity of the cue influences FOK judgments. In one of their experiments they presented participants with math problems. Subjects had to decide whether they knew the answer and wanted to solve it or to let the computer calculate the problem. Before their decision the participants were primed with a cue item.

After each presented question, people had to hit a button if they wanted to retrieve the answer or let it be retrieved by the computer. The findings for this experiment showed that when the cues were primed, participants were more likely to say if they would be able to get the answer, even though the priming might have caused people to retrieve the wrong answer (Reder & Ritter, 1992). Similar to Reder's research, Metcalfe et al. (1993) found that cue priming of verbal pairs had influence on people's FOK judgment but that the actual answer retrieval was not affected. Their results showed that the main factor for influencing the FOK judgments had more to do with the number of times people were presented with the cue as opposed to the retrievability of the actual answer. This had presumably to do with the increasing familiarity (Metcalfe et al. 1993).

All these studies implicated that people use cue-familiarity as one of the ways to make FOK judgments but that also the partial information on the target is important (Metcalfe et al., 1993; Reder & Ritter, 1992; Glenberg & Epstein, 1987; Dunlosky & Metcalfe, 2008). These studies have however not looked at the reason why people experience the FOK states. Reder and Ritter (1992) have suggested that the FOK state people experience indicate them that there is something in their memory that needs to be retrieved. This state provides, through cues and familiarity, information that is used to determine whether they will or will not attempt to retrieve the actual target. Over the last years there has been some progress on understanding the mechanisms that are

been considered part of meta-cognitive judgments and are heuristically based. Heuristic in judgment and decision making is based on systematic biases and can be observed in meta-cognitive

judgments. Sometimes people will be under-confident or over-confident in their judgment making. Research, now tries to find those systematic biases and reasons to improve meta-cognition (Reder & Ritter, 1992).

Alongside FOK judgments there is also another very similar phenomenon in the domain of meta-cognition that overlaps in many respects with FOK judgments. This phenomenon is called tip of the tongue (TOT) state or judgments. Even though there is a great similarity between the two, TOT judgments focus more on high accessible partial information (Dunlosky & Metcalfe, 2008). The TOT states are judgments that are likely to be retrieved right away after a temporary loss of recall. This in comparison to FOK judgments which focusses on predictions of successful recognition for non recalled things. Both use a similar meta-cognitive process to make these

judgments (Schwartz, 2008). Even though the TOT state also represents a high FOK experience, the TOT state has been researched separately from FOK judgments because the TOT state is often experienced by everyone and is therefore more common among people (Schwartz, 1999).

The tip of the tongue state was first brought to light by William James (1890/1981). He was the first to write about the state. He talked about the recall of a forgotten name, and that the state in which our consciousness turns is unusual. It seems as if there is a gap in our brain, but an intensely active gap nonetheless. It gives us a sort of sense of closeness but we are annoyed by the inability to come up with that name. James wrote that, “a sort of wraith of the name is in it, beckoning us in a given direction, making us at moments tingle with the sense of our closeness. . . . The rhythm of a lost word may be there without a sound to clothe it; or the evanescence sense of something which is the initial vowel or consonant may mock us fitfully, without growing more distinct.’’ (James, 1890/1981: 243–244).

The most used theory for the TOT state is that it reflects partial target access. Various studies have been done to support this theory. For example that people are able to report partial information on the target answer. This could be things like, the number of syllables or the first letter of the word (Schwartz, 2008). Burke et al. (1991) came up with a model that proposes that there is a semantic level of representation of the targeted word that is connected to a articulatory and

phonological level. This level is necessary for the actual retrieval. The two can be unrelated to one another. The model suggests that a person can know the answer without being able to articulate the actual answer. Because the older we get, the more impairments occur in phonological translation, the more TOT states we experience (Dulonsky & Metcalfe, 2008). During the TOT state it is often reported that when being in this state an incorrect word or answer persistently comes to mind. This

incorrect response that keeps coming back is usually called a blocker. This could emerge from the frustration that arises when being in a TOT state. In the blocked TOT state people get more

frustrated because they know the answer that keeps coming to mind is wrong but they cannot seem to get rid of it. Blocked states are also likely to be more difficult to resolve than TOT states that do not evoke a blocker.

According to the blocking hypothesis, blockers cause retrieval difficulty during TOT states. Recent research by Kornell and Metcalfe (2007) predicted that more time would allow participants to forget their blockers, which would enhance TOT resolution. They suggested that people need to take a break from their thinking of the targeted question and focus on something else. This because if your thoughts keep getting blocked by an answer that is not the correct one, you might not be able to break from the block and would not be able to solve the sought-after question. When you allow yourself some time and think of something else it will be easier to overcome the blocker and retrieve the actual answer. Kornell and Metcalfe (2007) tested this by presenting participants with different trivia questions. The people who experienced a TOT state were then retested, they asked whether their state also included a blocker or not. Participants then either had to try to solve the question straight away or had to wait until the end of the experiment and were then asked to solve the problems. In the end the blocked TOT states were resolved with a lower frequency than non blocked TOT states. The overall incubation effect showed that people were able to answer more questions correctly after the delay than on the immediate test. The delay did, however, not help the blocked TOT more than it did for the non-blocked TOT. The delay did make sure the blockers were forgotten and got out of the mind, but there was no difference for the resolution rate. This suggests that the blockers do not really block (Kornell & Metcalfe, 2007; Dulonsky & Metcalfe, 2008). Why people experience TOT states and FOK states has not yet been researched. One motivation to do research on both states could be the irritation and nagging emotional quality both states provide. The frustration can be a motivation for people to keep looking for the answer that they know is somewhere in their mind. Systematic research on the reasons and processes of FOK and TOT judgments is still waiting to be explored further (Dulonsky & Metcalfe, 2008).

In a study by Zuckerman et al. (2012), the researchers wanted to know what cognitive processes are involved during retrieval states and they tried to extend the applicability of previous findings (Zuckerman et al., 2012). In their experiment, participants' EEG responses were measured. To keep people from having to much eye movement they had to watch a computer screen, which showed a coloured circle that changed colour. This was presented while they had to listen to music. Participants were presented with songs and were asked to indicate if they knew the song's title or the lyrics. There were four ways to respond. The first was “I know this song”, which meant that

they could give the title or lyrics of the presented song. The second available answer was “I have a feeling of knowing”, this meant that with more time or a longer fragment of the song they would be able to retrieve the title or lyrics and would also be able to know the correct answer when they would be presented with a forced-choice recognition test. Participants were instructed that when answering they were in a FOK state that they should feel certain they would be able to have the correct answer at a later time. The third answer was similar to the second but was named “Weak Positive Predictions”. This meant that there was no feeling of a successful retrieval of the target but that it might be possible to name the lyrics or title with more time. The last possible answer was “I do not know”, in which people were sure they would not be able, even with more time or

information, to retrieve the target (Zuckerman et al., 2012). Their analysis showed that there was a significant “early left fronto-central difference” (Zuckerman et al., 2012) between the FOK and the “know” state. The FOK also differed from the “don't know” responses. The findings indicated that “neural processes, which may reflect strategy selection, ease of semantic processing, familiarity-related processes, or conflict monitoring, are indicative of the fate of our knowledge judgments long before we actually execute them.” (Zuckerman et al., 2012). This means that the brain is already able to judge whether we know the answer or not, even before we make our decision.

Besides looking at FOK judgments and its cognitive processes a lot of the studies done on the FOK have focussed on FOK judgments and language (e. g. Eakin & Hertzog, 2012;

Hanczakowski et al., 2014; Maril et al., 2003). Less research has been done on the effect of music related to the FOK state and judgments. In an fMRI study (Groussard et al., 2010) on the distinction between musical and verbal semantic memory, a comparison was made between the musical and verbal memory of twenty non-musicians. Participant were presented with a congruence task which involved retrieval and familiarity. In the music condition they had to listen to the beginning of a well-known tune followed by a short silence and a beep tone after which the tune proceeded with either the next part of the tune or a different but familiar melody. Subjects had to indicate whether the ending was congruent with the first part of the piece or not congruent. Another part of the musical task was to discriminate two given sequences of notes that were either the same or had a slight difference. In the verbal task, participant listened to the beginning of a French proverb or a popular saying. This too was followed by a short silence and a beep tone and again participants had to choose if the ending that followed after the beep was either right or wrong. Similar to the musical task there was a task for the verbal condition where subject had to indicate if two meaningless words were the same or if there was a difference between them. During the experiment brain activity was measured, for all tasks, using fMRI scans. The results revealed that the performance

and selection processes were activated in different parts of the brain, which suggests that they are common to both verbal and musical retrievals (Groussard et al., 2010).

The verbal aspect of music has also been related to the FOK state. Other researches on the FOK and music have looked to connections between the FOK in relation to songs. People often experience a FOK state when listening to songs. We have the feeling we know the song without knowing the title or lyrics of the song. Professor Zehra Peynircioğlu of the department of

Psychology at the American University in Washington DC is one of the few who has done multiple experiments and research on the FOK in relation to music (Peynircioğlu, 1998, 2004, 2008, 2011). In her initial research (1998) Peynircioğlu et al. concluded that people are better at recalling titles of instrumental tunes when they are cued with the melody than they are with remembering a melody when reading or hearing its title. However the opposite occurs with vocal music. Here the titles of songs are better cues than melodies (Peynircioğlu et al. 1998). It could be that this asymmetry is due to the fact that people listen differently to instrumental music then they do to lyrical music. There are different reasons as to why people listen to music but there are also different ways to listen to it, the topic of music listening is explored below.

1.2 Listening Types

“Listening to music is a cognitive process and depends on a variety of experiences that are inter-related with emotional and cognitive processes in the brain” (Ter bogt et al., 2010). There are different reasons as to why people listen to music. Most people appreciate it because it enriches and invigorates their life (Ter Bogt et al., 2010). In their study, Ter Bogt et al. tried to identify listener groups in terms of musical involvement and music use. They did this to be able to establish a typology of listeners. After extensive research they concluded that one of the biggest reasons for music listening is that it helps with the creation of atmosphere. It is a social type of musical use, which is used cross-cultural and has been done over centuries long. Evidence shows that music has been used within the human culture within ceremonies or festivities, be it for comfort, to energise or to invite people to sing and dance (Mithen, 2005 as cited in Ter Bogt et al., 2010, p. 2). Nowadays, technological developments enable us to listen to music wherever and whenever we want. Listening to music has never been so easy. Because it is so easy to pick our own music a large variation between the types of music people listen to arises and the active

involvement with music varies per individual. Listeners can now use music for different reasons and have different motives for listening to it. In various literature written about music listening (e.g.

Ter Bogt, et al., 2010; Roe, 1985; Zillmann, 1988, 2000; Madsen & Geringer, 1990) several important uses of music are listed. The most important reason for music listening is mood

management or enhancement and a way to avoid boredom (Ter Bogt et al., 2010). Using music to pass the time and to listen to it while doing something tedious may help making the boring things seem less boring and make it more acceptable. This reason was derived from Zillmann's (1988) theory of mood management. The theory states that people strategically, either consciously or not, select ways to use media that will help them feel and give them pleasant sensations. This

achievement of an emotional state can also be applied on a negative emotion if they believe that this could lead to a positive emotional state on the long term (Zillmann, 2000).

A second reason concerns the way people cope with emotional problems. For example, listening to music when one is feeling sad or lonely can be eased by the choice of music. People can also listen to loud music when dealing with anger or confusion. It can be a tool for coping with emotional distress. Thirdly, listeners are able to attain knowledge of the world through music and it can offer elements that can be useful in the process of self-identity. According to Ter Bogt et al. (2010), “Listeners are influenced by lyrics and ideas expressed by artists and also more subtly through identification with images of an artist or band.” A fourth reason is that music enables the listener to identify themselves in social society (Ter Bogt et al., 2010).

In their research (Ter Bogt et al., 2010), the authors sought to make a typology for music listeners that was grounded on the level of musical importance and the type of music people use for mood enhancement, coping with stress, identity construction and social identity. The typology was also tested in relation to the emotional response that participants indicated to feel and the personal background characteristics of each individual. They were able to distinguish three groups. One group was labelled 'high-involved'. The second and largest group was called 'medium-involved' and the third group was the 'low-involved' group. The overall finding was that there was a connection between the importance of music and using it for different purposes. They concluded that when people are moved by music, they use it to enhance their mood and often for coping with negative emotions (Ter Bogt et al., 2010).

A Swedish study done by Keith Roe (1985), also looked at motivations and listening patterns for music. He extracted three common factors. The first consisted, similar to that of Ter Bogt et al. (2010), of motivation involved with creating atmosphere and controlling mood. The second was silence filling and passing the time. And the third consisted of the attention to lyrics. His results showed that there was a small difference in gender when looking at motivation. Girls were slightly more motivated to use music to listen to lyrics. Other results showed that there was a positive correlation with lyrics and generation orientation. This could suggest that when we know

the lyrics of a song it can enhance exchange in peer group conversations. It seemed that the extent to which people are motivated to listen to a type of music is related to the degree of which that music is fulfilling their needs and wants. In punk music the strongest correlation was with the attention to lyrics, which could be a result of people relating to the artist or feeling emotionally understood when listening to these lyrics (Roe, 1985).

In Ter Bogt (2010) and in Roe's (1985) research it seemed that musical or personal background had no influence in the results. To see if there were differences for people with different musical background Madsen & Geringer (1990) conducted an experiment that was designed to investigate patterns of music listening among music and non-music students. They did this with regard to four primary elements of music namely, rhythm, dynamics, timbre and melody. The research tried to determine whether there was a consistency in listeners with regard to the patterns of these musical elements and whether there was a difference between the listening pattern of musicians and non-musicians. There were 60 participants that were described as non-musicians and 60 as musicians. All subjects had to listen to 10 orchestral excerpts. Results showed that musicians do listen differently than non-musicians. Musicians focus more of their listening time on melody as compared to rhythm and dynamics and spend the least of their time focussing on timbre. Meanwhile non-musicians focus on dynamics and secondly on melody and timbre (Madsen & Geringer, 1990). Their research was only conducted with instrumental music and raises question about the influence of lyrical music.

The influence of lyrics in music originated from the growing interest of scholars in studying the relation between music and language. For studying this relation, songs are used as the main source because they are built out of a combination of text and melody. Because they are so integrated with each other, studies on memory for songs often disregard the two components and use it as one entity. However, there seems to be a distinction between music and language components of song in perception, memory and in singing (Peretz, Radeau & Arguin, 2004). In their study, Peretz, Radeau and Arguin tried to understand how the song components of text and melody become so integrated through words combined with notes. They tried to provide a method for future studies on the connections between melody and lyrics in song memory. The method consisted of presenting short song fragments to participants which they had to recognise afterward together with similar fragments. The research was conducted using an adaptation of a classical experimental technique using auditory priming. This was done to find a relation between text and tune in song memory. The results showed that the priming technique was effective for memory organisation for songs. It also gave new information on the way melody and text are related (Peretz, Radeau & Arguin, 2004).

Peretz, Radeau and Arguin (2004) concluded that a song is built in a certain order, the components of songs are not always sequential but they do have an order in presenting information, for example, song beginnings. The beginning of a song is easier to recognise than other parts of the song. This applies to both text and melody. When you are presented with a few notes from a different part of the song, it is very difficult to recognise it. The first notes or words of a song, however, are more reliable triggers for the memory. The reason for this could be that beginnings are ordered differently in memory and that they act as guide mark for the whole song. In the experiment lyrics were much easier to recognise than melody. So there seems to be a asymmetrical pattern that the connection between the lyrics of a song and memory are made before there is a connection between the melody and memory (Peretz, Radeau & Arguin, 2004). In their initial experiment the targets that were presented did not correspond to the song beginnings so the advantage of lyrics could not be related to the fact that the name of a song generally corresponds with the beginning of the text. In general, lyrics seem easier to recognise. The advantage of text could be connected with the fact that syllables are a better indication of words than tones are of familiar tunes. On average, it takes about six notes to recognise a familiar tune, while it can take only two syllables to recognise a word (Peretz, Radeau & Arguin, 2004). Another reason why lyrics make it easier to recognise a song is that lyrics play an important role in songs. As mentioned before, there are different reasons and ways to listen to music. This is also the case when we look at lyrical music. Listeners can refer to the lyrics to see if there is an emotional message in it and if they can relate to that emotion (Masakata, 2007). Lyrical music uses several elements to create an emotionally engaging

experience. These are elements such as, lyrics, melody, dynamics, instrumentation and other aspects that are all combined to keep the listener interested (Nichols, Morris, Basu & Raphael, 2009). While you would think, using Zillmann's (1988) mood management theory, that when feeling romantically lonely or emotionally down, people would listen to happy and uplifting sounding music with a positive message to brighten their mood. It was Zillmann (2000) who through experiments found that people who are, for example, feeling romantically lonely choose to listen to music with “love-lamenting” lyrics rather than music with “love-celebrating” lyrics (Zillmann, 2000). He suggests that the reason why people choose this type of music is because they get a feeling of understanding with the singer. In support of this, adolescents in one study (Saarikallio & Erkkilä, 2007) indicated that when listening to music where the lyrics described problems that were similar to their own, could help them feel less alone in their problems and feel understood. They took the songs as a source of comfort (Zillmann, 2000; Saarikallio & Erkkilä, 2007).

A comparative analyses done by Lippman & Greenwood (2012) led to more support that lyrics can give a voice to the listeners experiences. In their study participants were asked to select

one piece of music that they considered personally valuable and had to elaborate on the reasons why their song was of significance to them. In their experiment, the majority of the participants claimed that part of the reason why their song was of any significance was because of the lyrics and the way they believed it applied to their lives. Other participants indicated that their piece helped them with feeling less alone. They said that the lyrics were a great part of this help and that the lyrics could be applied to their lives. This could mean that lyrics are a big component, in music, in accomplishing emotional validation (Lippman & Greenwood, 2012). Even though many researches have shown that lyrics have the ability to give voice to feelings and experiences of people (Lippmann & Greenwood, 2012; Saarikallio & Erkkilä, 2007), other research has shown that under certain conditions lyrics are not the most important reason for people to listen to music (Roe, 1985).

As there are different views on the importance of lyrics and its relation to people's reason to listen to music with lyrics, it seems that song recall and recognition is easier when focussing on the lyrics. Melody and lyrics in songs are imminently linked. It is for example not easy to recite the lyrics of a song without invoking the melody and it is also not easy to sing the melody without invoking the lyrics (Peynircioğlu & Rabinovitz, 2008).

1.3 Music and Memory

Hearing music can be associated with emotions, and music is known to activate the entire limbic system, which is involved in processing of emotions and works on the formation and controlling of memory (Jäncke, 2008). Besides evoking emotions, music can also evoke autobiographical

memories. There are many different types of memory. The first distinction in types of memory can be made between short-term and long-term memory. These types of memory suggest that they evoke memory for different periods of time. It is however, a misconception that short-term only applies to things that happened a few days ago in comparison with long-term which is said to be memory of things that happens a long time ago. Short-term memory is, in fact, defined as memory for information that is currently in the mind and there is a limited amount of information that can be held in the short-term memory. Moreover, long-term memory refers to information that is stored somewhere in the brain but is not present or accessible at that moment and needs to be retrieved. There is no limit to the amount of information that can be held in the brain for the long-term memory (Ward, 2006).

Long-term memory can consist of different components. Discrimination can be made between memory that is consciously accessible, called explicit memory, and memory that is not, labeled implicit memory (Ward, 2006). Explicit memory can again be split into two types of memory. One of them is the episodic memory. Episodic memory represents our memory that is

made out of experiences and specific life events in time. Episodic memories are highly personal, they include all the memories of the events in our lives. It is the memory of autobiographical events, like times, places, associated emotions and other knowledge you encountered in your life that can be recalled from personal experience. Another form of explicit memory is semantic memory. This type of memory consists of a more structured record of facts, meanings, concepts and knowledge about the outside world that we have acquired. It refers to general factual knowledge, and is shared with others. It is not related to personal experience or place and time in which they were

encountered. Many basic concepts in life like intelligence, compassion and humor are part of the semantic memory (Ward, 2006). Musical semantic memory allows us to be in a high FOK state when we listen to music. It gives the listener the ability to hum or sing along with the melody and can eventually lead to be able to retrieve the title, composer or performer of a song (Platel &

Eustache, 2000). Autobiographical information can also be associated with musical melodies. When we hear music, talk about music or find music that reminds us of the past it can be associated with our autobiographical memory and can often evoke a strong FOK (Jäncke, 2008).

A topic of interest is to what extent people are aware of their own memories. Flavell (1971) has termed a person's knowledge of his own memory the meta-memory. There are two different types of meta-memory. The first are certain facts that people could know about memory, for example that it is easier to remember short lists rather than long or that adults are better at

memorising things compared to young children. Even that people can assess what items are in their memory and which are not. The second type of meta-memory can be termed memory monitoring (Hart 1965; Flavell, 1971). So meta-memory is knowledge and awareness of your own memory. This also includes the contents and processes of your own memory. FOK judgments are a big part of the meta-memory. It implies what you remember and how accurate you are of your memories. Research has shown that for FOK judgments people are often fairly accurate (Nelson, 1984). For example, you have an uncle who lives in another country and you realise you forgot what city, but you’re 90% sure it starts with an M. Meta-memory includes the strategies we use to help us remember that particular city where the uncle currently lives.

What strategies people use to make these judgments have been the focus of many research. Koriat and Goldsmith (1996) argued that all memory output is related to meta-memory, at least partially, and that any attribution on memory can be considered meta-memory (Koriat & Goldsmith, 1996). Meta-cognition is an important factor for the process of ones own memories. Meta-cognition refers to our own knowledge, regulation, and control of our cognitive system. If we do not know what information is available in our knowledge, it could be difficult to make decisions on what we

want to learn and moreover, can learn. If our meta-memory judgments are not accurate, studying would be difficult because we could not perceive what we already know or not (Thiede et al., 2003).

One function of the meta-memory that has been extensively investigated, is the positive prediction judgment. Positive prediction judgments happen when something cannot be recalled, and this judgment relates to the prediction people make about their ability to recall the answer at a later state, for example, FOK judgments (Metcalfe, Schwartz, & Joaquim, 1993). These judgments can be explained by several theories. One that has already been mentioned is the theory of Hart (1965). He proposed that a cognitive monitoring mechanism makes us feel if the target information is available and thus if we are able to recall it or not. As a result there is a predictive output in the form of positive predictions. Other views include the cue-familiarity theory and the accessibility account (Metcalfe, Schwartz, & Joaquim, 1993; Reder, 1987). Koriat and Levy-Sadot (2001) proposed a combination of these two theories, suggesting there is an congruence between the cue-familiarity and accessibility accounts. Their theory states that cue-familiarity is assessed first and is later followed, when the familiarity is high enough, by accessibility mechanisms. These then lead to further memory search, looking for the targeted answer (Koriat & Levy-Sadot, 2001; Zuckerman, 2012) So, when we have a higher familiarity we are better at retrieving information on wether we might know the answer or not. For many years researchers were interested in associations between emotions and memory. In the early 1900s, it was reported that people are better at remembering positive information rather than unpleasant information (Stalinski & Schellenberg, 2013). Others claimed that general mood was related to memory, with happy participants being less likely to recall negative experiences (Stalinski & Schellenberg, 2013). Recent studies have shown similar findings. Much research has been done on positive emotions and memory. For example, the effect of positive words and the improvement of memory, or that positive stimuli are better remembered than neutral stimuli (Stalinski & Schellenberg, 2013).

Mood also affects the kind of information that people are likely to remember. Information learned while in a positive or negative mood is more likely to be recalled than in a 'neutral' mood. Similarly, music in major keys (which are happy sounding) is better remembered than music in minor keys (which are sad sounding) when listeners are in a positive mood, but the opposite occurs when listeners are in a negative mood (Houston & Haddock, 2007). Thus, memory is influenced by what is being processed and how it is processed can be dependable on emotion (Stalinski &

Schellenberg, 2013). Memory and music appreciation are often related. To test the relation of memory and music appreciation, Stalinski and Schellenberg (2013) presented a set of experiments to test whether liking predicts memory for music. The majority of people enjoys listening to music and different people like to listen to different kind of music. The experiment used a variety of

different types of musical fragments so that everyone would appreciate at least one type of music. (Stalinski & Schellenberg, 2013). The findings showed that there is a direct association between emotional reactions and memory recognition. It suggests that there is a relationship between musica appreciation and memory and that the liking of a song is associated with better memory and vice versa. In the case of music liking, it showed that memory benefits only occur with liked music excerpts and not with disliked music (Stalinski & Schellenberg, 2013).

Other researchers have explored the role of familiarity and the role of music liking on memory performance. Many different types of materials have been used for example, studies using words, text, pictures or in Stalinski and Schellenbergs (2013) case, music. Using musical materials, Bartlett, Halpern, and Dowling (1995) showed that episodic memory for well-known tunes was better than that for newly heard tunes. In general, it has been widely accepted that the higher the familiarity for the materials, the easier they are able to be recalled. The role of familiarity in the domain of meta-memory is less clear. The studies on cue response and FOK ratings have shown that they are influenced by the given cues or by the familiarity someone has on the general topic as posed in the question (think back to the “Sunflower” question and the knowledge someone has in the domain of art) (Reder & Ritter,1992; Korenman & Peynircioğlu, 2004).

So far it seems that higher familiarity with cue words, does seem to increase FOK ratings. However high familiarity with the target words itself seems to have no effect or a lower rating on FOK ratings. In a another study with musical materials, Peynircioğlu, Tekcan, Wagner, Baxter & Shaffer (1998) showed that the FOK ratings for melodies and the titles differed depending on whether the music had lyrics. In both cases, melody and title, the predictions for melody recognition were more accurate than they were for title recognition (Peyniriogly et al., 1998; Korenman & Peynircioğlu, 2004). The link between lyric en music suggests that music, similar to text, is encoded in our semantic memory (Jäncke, 2008). Memory for text and melody of songs are intimately related (Peynircioğlu et al., 1998; Herholz, 2012).

1.4 The current Study

Studies on the FOK in music showed that melodies and titles were easier to remember than lyrics when cued by one another. The finding that melodies are easier to remember than lyrics was shown by studies by Peynircioğlu et al. (1998) and Sousou (1997). However, there are also studies that show that participants tended to recognise lyrics better than melody (Peretz et al., 2004).

In their research Peynircioğlu, Rabinovitz and Thompson (2008) instead of looking at episodic-memory for new songs tried to measure semantic episodic-memory for semi well-known songs. Different

than other studies they focused on recall and not recognition. Even though lyric recognition is easier than melody recognition, when participants were instructed to learn a song and asked to reproduce it rather than recognising it, melodies seemed to be remembered easier than lyrics. Lyrics were however, better cues for melodies and than vice versa and lyrics were better cues for titles than vice versa. A reason for this could be that there is a difference in hierarchy between the three. It could be that there is more exposure of melodies and titles then there is of lyrics. And therefore make them more memorable than lyrics, but lyrics are better cues for melody and title than vice versa

(Peynircioğlu, Rabinovitz & Thompson, 2008). In the non-lyrical music domain, the FOK for melodies and titles was explored by Peynircioğlu et al. (1998). They found that there was a higher FOK rating for titles than for melodies. However the accuracy of these ratings was higher for melodies than for titles. This was only the case for instrumental music. With songs, with lyrics not presented, there was a higher FOK for melodies and the accuracy was only marginally greater. In both instrumental music and songs the melodies were cued by the title and vice versa. In their research Rabinovitz and Peynircioğlu (2011) tried to find three possible reasons for these differences in FOK ratings between songs and instrumental music (Rabinovitz & Peynircioğlu, 2011).

One of the main differences between songs and instrumental music is that songs exist of melody and text, namely, lyrics. This provides for an extra aid for the memory. Another difference is that melodies and titles can be learned separately, knowing the title of the song can be seen as of no importance. Lyrics, however, are usually learned together with the melody. Even if the lyrics themselves are not remembered, they are better cues than titles for obtaining melody recall

(Peynircioğlu, Rabinovitz & Thompson, 2008). The bond between lyrics and melodies can induce verbal mediation in songs for melodies and titles which is seen in the FOK differences. The differences in FOK ratings can also be a result of the familiarity of materials. It is possible that songs in general are more familiar to people than instrumental music (Rabinovitz & Peynircioğlu, 2011).

The main features of this master thesis have been derived from several previous studies by Peynircioğlu and others. The research in 1998 by Peynircioğlu et al. was one of the first reported researches on the FOK with music stimuli. The method had been derived from Hart (1965) which has been repeated many times on FOK judgments, either in relation to music or other topics, which makes this a reliable method to test the FOK on pop tunes. In the research (1998) participants had to recall melodies and titles of musical pieces. If their recall failed subjects were asked to give a FOK rating, thereafter a recognition test followed. Their research consisted of two experiments. The first experiment used stimuli based on instrumental music. The second experiment used song stimuli

whose lyrics were not presented during the experiment. In this first research on the FOK in music the authors tested 60 students. After a pilot study they choose snippets of highly familiar

instrumental pieces and these were separated into two conditions; 'name that tune' condition and 'hum that tune'. In the 'name that tune' condition subjects were presented with a melody and were asked to provide the title. In the 'hum that tune' condition they were presented with the title and were asked to provide the melody. The recall test was combined with the FOK task. After each snippet or title, participants either gave the title or sung a melody or they gave a FOK rating on scale between 1 and 6. For the second part of the experiment there was a forced-choice task, subjects received a sheet of paper with the titles and had to indicate which snippet represented the right answer after listening to the set of three snippets. For the title recognition they received a sheet with alternative titles and had to circle the right answer after hearing the targeted snippet.

For the second experiment the same procedure was followed with the exception of different stimuli. The snippets were replaced with melodies from songs with lyrics and were also selected by extensive piloting. The snippets were picked so that the titles as part of the lyrics were never presented. The songs were not presented in the original recording as opposed to the instrumental music from the first experiment but were recorded on a keyboard. The result of their research showed that with instrumental music, people were better at recalling titles when they heard a melody as compared to recalling a melody when seeing its title. This was reversed for songs, here melodies were recalled more often when seeing a title than titles were recalled when hearing the melody. They conclude that words in song may have a mediating effect between memory for title and memory for melody. The verbal title may cue the verbal text which can then cue the melody, which is more effective than the melody cuing the text and the text cuing the title. However because the two experiments could not be compared, due to between-experiment manipulation, the

comparison between memory or meta-memory for music with and without lyrics is speculative (Peynircioğlu et al., 1998).

To follow up this experiment Rabinovitz and Peynircioğlu (2008) used the same method but added the lyrics component because this had been suggested to work as a mediating factor. In their first experiment they looked at the highness of FOK judgment accuracy in comparison with cues for recall. In their second experiment they looked at the effectiveness of lyrics and melody cues in memory and meta-memory. The two experiments were made to show that verbal, titles and lyrics, and non-verbal, melodies, of songs interact in a systematic way in provoking our memories for the songs and influence the judgment about the memories (Peynircioğlu & Rabinovitz, 2008)

In 2011 the initial study (1998) was again replicated by Peynircioğlu, Rabinovitz and Thomson. They conducted several different experiments, constantly changing one small part of the experiment

to see if this had any influence on the results. They did this to explore the effects of the variations in familiarity and the use of lyrics or not to see if differences in FOK ratings could be observed. In Experiment 1 they tried to rule out a simple explanation that was based on the different strategies of listening for the type of material. The only change they made from the initial experiment was to combine both songs and instrumental pieces and show them in the same but mixed list. In Experiment 2, Songs and instrumental music were presented in both a single melody line and complete orchestral formats. This to make sure that both types of music had an equal sonic complexity and to minimise the differences that occurred in recognisability and the musical

information that could influence the familiarity and music processing. For the songs they also tested if it made a difference to have the title cue as part of the not presented lyrics, so they could look at verbal mediation. In Experiment 3, only songs were presented. Some were presented with lyrics and others had lyrics but these were not presented. This was done to look at the role of lyrics in more detail. Experiments 4 and 5 tested the episodic memory. New musical pieces were composed and were also assigned with lyrics or not, so that the familiarity was kept constant and could be manipulated by repeating the presentation of the new melodies. Experiments 6 and 7, focussed on the semantic domain. They varied with both familiarity and lyrics. In Experiment 6, they focused on familiarity for one musical genre. The stimuli consisted of TV tunes, some with lyrics and others without. Within each group, there were tunes with both high and low familiarity. They tried to test the influence of familiarity on FOK ratings. In Experiment 7, participants were tested by dividing them into two groups. The first was made of people who were more instrumental classical listeners and the second were participants who were more interested, and thus presumably more familiar, in pop songs (Peynircioğlu, Rabinovitz & Thompson, 2011).

The experiments mentioned above, showed that a lot of information is available for studies on music and FOK ratings. The experiment for the current study is made out of a combination of several of the experiments done by Peynircioğlu, Rabinovitz & Thompson (2011). Similar to the procedure in experiment 7 the participants were divided into two types. But instead of looking at their familiarity domain or interest, we wanted to look at the type of listening strategy they use (focussing on melody or lyrics). The type of music had been collected from semi-familiar pop tunes. The stimuli were snippets from original recordings, with the lyrics presented, from the top 2000 list of 2012 (NPO2, 2012) which could be classified under the genre popular music. The songs were based on semi-familiarity to make sure there were enough FOK judgments available. The particular focus of this experiment was weather people's listening strategies will influence the way they judge their FOK. To know how to indicate what type of listener people are, participants were

presented with a questionnaire where they were asked to give a self indication on their listening strategy.

The hypothesis of the current study is based on the notion that the FOK is generally accurate. This is simply due to a basic quality of memory (Koriat, 2005). The information that comes to mind is much more likely to be correct. Therefore, the overall accessibility of information is generally a prediction of correct memory (Koriat, 2005). The current study hypothesises that answers after a FOK rating are overall fairly accurate but that a distinction can be made between listening strategies. It hypothesises that people who generally focus on lyrics are better at recalling a title of a song than people who mainly focus on melody. Besides having a higher recall rate it is hypothesised that lyric listeners will also have a more accurate FOK rating. So when indicating their FOK, lyric listeners will have a higher correctness level to their actual answer in the second trial than melody listeners. Overall lyric listeners will score better in the FOK experiment than melody listener. This hypothesis was derived from previous studies in which participants were more convinced that they would recognise melodies better when given the titles as cues than recognising titles when melodies would be given as cues. It was suggested that it could have been the lyrics that played a mediating role in songs and that the lyrics were influencing the participants' FOK

judgments and that title-lyric-melody mediation was more effective than the melody-lyric-title mediation (Peynircioğlu et al., 1998). Research by Peretz, Radeau and Arguin (2004) found that there is an advantage of text. This advantage could connected with the fact that syllables are a better indication of words than tones are of familiar tunes. The advantage of text becomes clear when we see that on average, it takes about six notes to recognise a familiar tune, while it can take only two syllables to recognise a word (Peretz, Radeau & Arguin, 2004). If there is an advantage of text it would mean that it would have been easier for participants to recognise a song when hearing the lyrics as compared to instrumental music. This could mean that people who focus on lyrics would score better in a recall and recognition test. So people who indicate that they spend more of their listening time on lyrical music than melody should take advantage out of this general notion that lyrics make it easier to recall a song and that there is an advantage of having text in a song. In addition to this hypothesis the experiment will look at the self indicated listening type and the implicit score of types of listening.

2. Experiment 1 (pilot study)

2.1 Background

The FOK research by Peynircioğlu et al. (1998) included a pilot experiment to attain 144 snippets of highly familiar instrumental pieces. They recruited 20 students and asked to indicate how familiar a presented melody was on a scale from 1 to 10. They chose the pieces that received mean ratings between 3 and 8. This was done so that otherwise participants were either not familiar with the piece at all (rating 1 or 2) or too familiar with the piece (rating 9 or 10). Because the current study is a follow up of the Peynircioğlu et al. (1998) we initially tried to see if we were able to use the same stimuli. However the original stimuli were not available. Therefore the study by

Burgoyne, Bountouridis, Van Balen & Henkjan Honing (2013) was used for extracting stimuli. The study (2013), tried to find what makes a song catchy. In order to answer this question, they set up an experiment in the form of a application game named “Hooked”. The aim of this game was to test songs for familiarity. This was done using the game where people were presented with fragments of the song. They had to answer if they knew the song. After each fragment the music would stop for 6 seconds and would continue either in the congruent order or at a different point in the song.

Participants were asked to indicate if the continuation of the song was correct or not. The game used different fragments of the same song and the response rate would eventually lead to the most familiar part of the song. There were two versions of this app, one for Dutch natives and another for English natives. The researchers used music from the Dutch Top 2000 list from the year 2012 (NPO2, 2012). For British version the game consisted of songs from the British pop chart (Burgoyne et al., 2013). The stimuli used in the “Hooked” research was available for the current study and the authors were willing to provide a list of 460 songs, including the start times and familiarity ratings of each song. Therefore, songs from the Top 2000 list from 2012 were used for this experiment (NPO2, 2012).

2.2 Method

For the current study a selection of 120 semi-familiar songs was made out of the list provided by Burgoyne et al. (2013). To determine which songs were the best suitable for this experiment, a pilot study was done to select the songs with the highest semi-familiarity rating.

2.2.1 Participants

The pilot study consisted of one short experiment. In this experiment, five participants with ages ranging from 22 to 28 years old were tested. No distinction was made between musically trained and non-musically trained participants. Each participant had a different musical background varying from professional musician to no musical experience whatsoever.

2.2.2 Stimuli and Tasks

The stimuli for the pilot study included 120 song fragments from 2012's top 2000 list (NPO2, 2012). We selected 120 songs that appeared to be the most semi-familiar. These songs were first cut into snippets of around 4 to 8 seconds. The snippets always contained lyrics. The title of the song (when present in the lyrics) was never presented.

The experiment was conducted in a quiet and private room. Participants had to listen to the 120 fragments and after each snippet they were asked if they knew the song or not. If they knew the song, they were instructed to give the title of the song and had about 30 seconds to provide the answer. If they were not able to recall the title they were asked to indicate their FOK on a scale from 1 to 7, so how likely (1 being not likely and 7 being very likely) would they know the title of the song if more information, for example the name of the artist or the first letter of the title, would be presented to them.

2.3 Results

The answers for each participant were notated in excel and at the end of the pilot study the correct response rates were compared. The songs that were indicated as known by all participants were either omitted and changed to a different song or the snippet was changed to a different part of the song. The songs that were not recognised by any of the participants were removed and replaced by another song. After these changes were made the new snippets were tested again on three new participants with a different musical background in order to confirm the songs were semi-familiar.

3. Experiment 2 (main study)

3.1 Method

3.1.1 Subjects and Materials

30 students between the age of 18 to 30 years old participated in this experiment. There were 16 males and 14 females, 8 of which indicated to have had 4 years or more formal theoretical music training and 12 had 6 or more years of formal training on a musical instrument including voice. Out of the thirty people 7 had had no formal theoretical training and of which 6 also did not play any instrument.

The stimuli consisted of the 120 song fragments that were selected in the pilot study. The songs were originally selected from the top 2000 from the year 2012 (NPO2, 2012). The 120 songs were selected for semi familiarity and were cut into snippets with a mean duration of 5 seconds. The snippets always contained lyrics but the title of the song (when present in the lyrics) was never presented. Out of the 120 fragments 40 were used as target snippets, 20 of which were used for the melody-title condition and 20 for the title-melody condition. The remaining 80 songs were used for the second part of the experiment in which subjects had to do a forced-choice task. The targets were paired with the other songs so that they were relatively similar to the target snippet looking at both title and melody similarities. For example, for the title-melody condition the subject was presented with the title “Sweet Home Alabama” and had to choose from three fragments that were paired, for example in this case, “Fly Away” and “All Summer Long”. The same procedure was used to test the melody-title condition but focussed on title similarity.

3.1.2 Design

The design of this study was derived from the study by Peynircioğlu et al. (1998). But the focus of this experiment to look at the influence of using different listening strategies in relation to the FOK rating. The experiment consisted of two parts (recall phase and recognition phase) and had two conditions. The first was the melody-title condition. In this condition subjects were presented with 20 short song fragments and were asked to write down the title of the song. If they were not able to recall the title they were asked to indicate on a scale from 1 (not very likely) to 7 (very likely) how likely it was that given more information they would be able to recall the song at a later moment. To measure the degree of FOK, participants had to indicate their FOK rating by using the numeric buttons on the keyboard. The second condition was the title-melody condition. In this condition

participants were presented with the title of a song and were asked to indicate if they knew the song. If they did they were asked to sing or hum the song. This served as proof for the researcher to conclude that the subject really knew the song and not just said he or she did. The humming was recorded for five seconds and similar to the first condition if the song was not recalled a FOK rating from 1 to 7 had to be given.

In the second part of the experiment (recognition phase) the same fragments as before were used but now combined with a forced-choice task. For the melody-title condition the subject listened the target fragment and had to choose between three titles. For the title-melody condition this was reversed and participants were presented with the title and had to choose between three fragments. The target fragments were always the same and all participants performed all two conditions in the two phases. The experiment was programmed and carried out by software PsychoPy2 (v1.81).

3.1.3 Procedure

The participants were tested individually due to the singing or humming in the title-hum task during the recall phase. In order to ease any possible discomfort when singing or humming was involved, participants were assured that the quality of the result was not important and that this form of testing was only done to see if the subject really knew the song.

Before taking part in the experiment participants were given an information booklet, had to sign a form of informed consent for the ethical commission of the University of Amsterdam and were asked to state that they had normal hearing and normal vision. After their consent the participants were provided with stereo headphones and sat in a quiet room in front of a computer. After they had been given time to read the instructions on the opening screen, any remaining questions were answered and the experiment began. The experiment was conducted using PsychoPy (v1.81) and took approximately 30 minutes per participant. They were instructed through the experiment to listen carefully and after each snippet they were asked: “Do you know this song?”. For the first task (melody-write) the right answer had to be written down on a provided answering sheet. The first task consisted of 20 melodies. After the 20 melodies, they were presented with 20 titles (title-hum task). After each given title a recording of 5 seconds would play. They either had to sing or hum the song or were asked to say “I do not know the tune” for the non-recalled songs. For both melody and title condition the FOK judgments of non-recalled songs had to be indicated. After each song fragment or title subjects rated their FOK on a 7-point scale that indicated whether they would be able to recall the right answer when presented with more information at a later time (1 being “not likely, 7 being “very likely”). This rating was responded by typing in the number on the keyboard.