How the Titles of Popular Songs have Changed over the Last 60 Years

Billboard magazine has been keeping track of the 100 hottest (most popular) songs of the year since 1958. Lists of the Hot 100 titles from 1960 to 2019 (6001 titles) were used to study the way in which popular song titles changed over time. Based on significant polynomial regression trends and significant results from a discriminant function analysis, it is concluded that there were three main phases in titles (early, middle, and late) and that these phases differ in predictable manners in terms of stylistic features such as length, abstraction, activity, and the use of the word “love”. Early phase titles are longer, more concrete, more passive, and they do not use the word “love” often; middle phase titles are of medium length, more abstract, of medium activation, and use the word “love” frequently. Titles of the last phase are shorter, more concrete, more active, and do not often employ the word love. A possible factor contributing to these differences is the rise in popularity of rock and roll and hip-hop respectively and their different periods of ascendency.


Introduction
This article describes how titles of popular songs (represented by the songs on the Billboard Hot 100 charts) have changed across time from 1960 to 2019. Titles are analyzed in terms of their length, their pleasantness, activation, and imagery, their use of particular words ("I", "love", "you", and "the"), their use of elision (e.g., rockin'), and the length of the words in them. Graphs plotting changes across time and a discriminant function analysis indicate the presence of three mutually distinguishable phases in title style, with the earliest 20 years (1960)(1961)(1962)(1963)(1964)(1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979) being represented by one type of title, the middle 20 scientific articles because of their power to attract readers (Luscombe, Rijkers, & Sloof, 2018). The craft of forming "titles that sell" is discussed by Beall (2009, chapter 8). Beall concludes that popular song titles should be three of four words long, that they should be unique, that they should be found within the song itself, and that they should be expressed in the second rather than the first or third person, i.e., they should speak about "you" rather than "I" or "she/he".

Why would we Expect Titles to Change across Time?
It is probably not time itself but factors associated with time that are responsible for changes in the titles of popular songs. One such factor is the early predominance and later "death" of rock and roll (Note 3) (Note 4). In the 1950s and 1960s rock and roll was omnipresent in popular music (although country songs, instrumentals, and ballads hailing back to the 1940s were also popular). Many critics and analysts agree that by 2010 rock and roll was supposedly "dead", and that it had been replaced by other popular genres such as disco, pop, metal, grunge, and new metal. All of these genres arrived on the scene during the 60 years studied, as did hip-hop. Hip-hop is often equated with rap and is related to rhythm and blues. All of these styles of music, and rock and roll as well, have their roots in the music of the Mississippi Delta early in the 20 th century and they are informed by the cultural traditions of various groups of immigrants (Star & Waterman, 2007, chapters 8, 9, 11). Important social changes took place between 1960 and 2019: the movement toward desegregation, the Vietnam, Gulf, and Afghanistan wars, and several economic crises (including Black Monday in 1987 and the financial crisis of 2007-2009). It was argued above that popular songs (and their titles) are a society-wide phenomenon. With all of society changing around them, not to mention the necessity for the "fashions" of a period to be distinct from those of previous periods, it would be extremely surprising if the titles of popular songs remained consistent. Although Sacco (2009a, 2009b) were studying lyrics rather than titles, they concluded that popular songs mirrored the issues of the days in which they were popular. The research described below aims to outline how the titles of popular songs have changed over the last 60 years; the author discusses one possible contributing factor to the changes observed.

Method
Song titles were downloaded from the various Wikipedia pages listing the Hot 100 songs for each of 60 years . For example, the 1970 list (Note 5) was the source for that year. There were 6001 titles in all (1969 included a tie for 100 th place) and these titles contained a total of 19,067 words. Each title was described in terms of the total number of words in it, and the average length of its words.
To assess the emotional impact of words in the titles, each word for every title was scored by being matched to the Dictionary of Affect (Whissell, 2009), and values for matched words were imported into the data set. The matching rate for the Dictionary was 86.4%, somewhat lower than the rate of 90% typically expected for everyday English texts (Whissell, 2009). This was likely because the texts under study were titles rather than being conversations or expositions, and because of abbreviation and elision www.scholink.org/ojs/index.php/sll Studies in Linguistics and Literature Vol. 5, No. 4, 2021 143 Published by SCHOLINK INC.
The Dictionary has three scoring dimensions, pleasantness, activation, and imagery. Scores for these dimensions are based on earlier participants' ratings of the words in a context-free task. The presence of many pleasant words such as "love", "sweet", or "darling" in a title would lead to a higher score for pleasantness. The presence of many active words such as "party", "play" and "lover" would lead to a higher score for activation, and the presence of concrete words such as "baby", "hands" and "girls" to a high score for imagery (because the referents of these words are easily pictured or imagined). The   Titles were scored in terms of how often they included each of the words "I", "love" and "you" partly because many popular songs have been love-songs. In his 1976 hit which appeared on the Hot 100 for that year, Paul McCartney referred to popular songs as "Silly Love Songs" and repeated the phrase "I love you" several times in the chorus. The singer concluded that in 1976 the world had not yet had enough of love songs, and that the genre had an enduring popularity. In addition, the presence of the word "the" was counted in each title: the presence of this word indicated a more complex, sentence-like structure than its absence. Finally, the presence of elision (as, for example, with the replacement of a "g" by an apostrophe in "rockin'") was noted. Frequent employment of elision indicated the use of less formal and more unsophisticated language within titles. Table 1 provides an example of the scoring procedures of the research using three Hot 100 titles labeled A, B, and C. The first of these (which includes "raindrops") is the least pleasant and the third (which contains the word "wonder") is the most pleasant. The first two titles are moderately active and the third is moderately passive. Title A, which contains the words "raindrops" and "head" is the easiest to picture and has the highest imagery score while Title C, which does not contain many words that promote a clear mental picture, is the hardest to envision and therefore the most abstract. None of the titles contains the words "I" or "love" but one title (C) contains the word "you". Two of the titles contain the word "the" (B and C), while only one contains an elision (A). Titles A and B are both six words long while title C is four words long, and the average length of words within titles is similar for B and C while A contains longer words.

What do Popular Song Titles Look Like as a Whole?
Emotionally scored words in popular song titles (N=16,474) were more pleasant ( .52, t=32.78, p<.001, d=.28). Titles were therefore more positive, punchier, and easier to envision than everyday English, which makes sense in view of the functions of titles as handles, summaries, and hooks that aim to attract attention to a song.
The average title was 3.18 words long (within the ideal range suggested by Beall, 2009), and mean word length was 4.57 letters. The word "love" was in 2.4% of titles, the word "I" in 1.7% and the word "you" in 2.6%. Seventeen percent of titles included an elision and 12% include the word "the". Beall's (2009) advice to writers to focus on the second rather than the first person and to have three or four words in a title were observed in most cases.

Predicting Changing Trends from the Powers of Time
Standardized time and its squared and cubed powers were employed in stepwise linear regressions to predict changes across time for each of the nine variables outlined in Table 1. Pleasantness was excluded from this analysis as it did not correlate significantly with any of the powers of time, and in fact contributed nothing to prediction when it was included. Cases were means for each variable for each of the years 1960-2019, so the equations had 3 possible predictors and 60 cases. The regressions were successful with R ranging from .32 (for active language) to .89 (for title length), with a median of .65 (which characterized the function for the use of the word "the"). The functions are depicted in Figure 1 and described in more detail in Table 2. Seven of the nine functions included significant quadratic terms. This implies that the modeled curves were taking one direction early in the period and that they reversed direction later. For example, this happened with word length, which first decreased and then increased. Alternatively, the beginning, middle, and end phases evinced different trends. This was the case with the use of elision, which decreased, leveled off, and then decreased again.

Figure 1. Changes in Titles across Time According to Modeled Data from Polynomial Functions
Note: The first two functions represent multiple variables, and the function labeled "Asterisk" applies to the employment of elision. Functions for "I", "you", word length, and imagery were quadratic; the function for "the" was linear; the remaining functions are combinations of linear, quadratic, and cubic trends ( Table 2).
A visual inspection will reveal that changes in the modeled curves are centered about the years 1980-99.
It was therefore possible to divide the 60-year time span into three equal phases: early (1960-1979), middle (1980-1999), and late (2000-2019). variables in each phase. A three-group stepwise discriminant function was conducted to predict phase (early, middle, or late) from the nine measures. Prediction was extremely successful with 98% correct prediction (one error), and a canonical correlation of .92 based on three predictive variables: word length, title length, and imagery. The first function discriminated the early phase from the late phase.
Standardized discriminant function coefficients were .58 for imagery, .59 for word length, and 1.41 for title length. High scores on this function were characteristic of early titles and low scores of late titles.
The second function had a high standardized weight for imagery (.96) with weights <.1 for the other two variables. Low scores on this function were distinctive for the middle phase. According to the highest coefficients, titles of the early phase were longer, those of the late phase shorter, and those of the middle phase more abstract. Other differences among groups, e.g. that for the use of the word "love" (Table 2), were not needed for prediction. The single misclassified year (1984) belonged to the middle phase but had titles with imagery scores that matched the mean for the early phase (62.7), which led to its being miss-classified.

The "Death" of Rock and Roll? A Post Hoc Theory
The final proposed explanation for changes in song titles is related to an issue raised earlier -the supposed death of rock and roll-and this explanation will be treated analytically. Is rock and roll really dead? Many critics had reached this conclusion (Note 6). However, data from Google Ngram (Note 7) (which reviews the words employed in millions of books) and Google Trends (which reports on the frequency of different computer searches on Google) suggest that rock and roll is far from dead. Like the Billboard charts, these sources access the behaviors of millions of individuals on an ongoing basis, so they are valid sources for information as to popularity. The three-word phrase "rock and roll" appeared 75 times per million words in Google Ngram for 2019. "Rock and roll" was also used frequently in searches on Google, as attested to by data from Google Trends.
Not dead, therefore. Although it "lives" rock and roll is no longer in ascendancy. It is not the main music of today, in spite of its continued popularity, reflected in its many mentions and associated searches. It has been overtaken in both of these metrics, by hip-hop. Hip-hop, which was not even on the horizon in 1960, achieved 94 mentions per million words in 2019 according to Google Ngram ( Figure 2). Its use in Google searches was almost three times as high as that of "rock and roll" for the year October 2019-October 2020. Within the time period studied , rock and roll grew in popularity, but hip-hop also grew (though somewhat later in time). This is reflected in the rising curves for mentions of both music genres in Figure 2. A rough measure of the ascendancy of rock and roll in music could be estimated by subtracting the mentions per million for hip-hop from those for rock and roll (this answers the question "How much more popular is rock and roll than hip-hop"). Rock and roll is in ascendancy when its mentions exceed those for hip-hop. It loses this position when mentions for  in 1960 to 20 in 1970 to 27 in 1980 to 43 in 1990, 44 in 2000, -6 in 2010, and -19 in 2019. There is a gigantic drop in ascendancy score between 2000 and 2019 (the beginning and end of the last phase).
Ascendancy data for rock and roll are, in fact, quadratic first rising, then staying high, then falling. A second indicator, total popularity could be represented as the total number of mentions for both types of music, or by the sum of the two mentions at each point, which rises steadily from 16 in 1960 to 169 in 2019. Many of the effects associated with year and its powers are strongly correlated with the ascendancy of rock and roll and the joint popularity of the two genres.

Conclusions
In one approach to understanding changes in popular song titles over a sixty-year range, we can envision the sixty years as being underpinned by three phases. The first is an early phase, with rock and roll rising and ascendant, the second a middle phase, with rock and roll still ascendant and hip-hop rising, and the third a late phase, with rock and roll losing its ascendancy to hip-hop, although both remained popular. While recognizing that causation has not been demonstrated in this article (merely correlation) and furthermore that there are other potential causes for the differences observed, and other popular music genres besides rock and roll and hip hop, we can cautiously describe the three phases as follows: Phase One (1960)(1961)(1962)(1963)(1964)(1965)(1966)(1967)(1968)(1969)(1970)(1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979) Emergence and Ascendency of Rock and Roll. Popular song titles in this phase are more emotionally passive; the words in them are more concrete; titles are longer and the words within them are longer; there is a relatively lower use of the words "I", "love", "you" and a relatively www.scholink.org/ojs/index.php/sll Studies in Linguistics and Literature Vol. 5, No. 4, 2021 149 Published by SCHOLINK INC.
higher use of the word "the"; elision is employed more frequently.
Further research could investigate the characteristics of song titles in the different Hot 100 lists representing different genres in order to clarify the contribution of genre to title style. If differences between titles of popular hip-hop songs and other popular songs are in the same direction as those noted above, the role of genre in affecting song titles will be supported.