FIGURE 1
Matrix, Language Domains
Home use (at home)
Educational use (at school or college)
Professional or work use (at work)
Friends (with friends)
Family (with family)
I am very comfortable using this language
English
Language 2
Figure 2 | ||||||||
Matrix, Information Tasks | ||||||||
| Watching movies | Listening to music | Social media (Facebook, Twitter, Weibo) | Instant messaging (IM) | Reading news and current events | Texting/SMS | Communicating on emails | Phone and voice calls |
I prefer English | ||||||||
I prefer L2 |
The responses were dummy-coded (0 = no attribute, 1 = presence of the attribute).
The variable for language ability was created from the questions below with responses coded in a five-point Likert-type scale (1 = not at all, 5 = very well).
How well do you speak English?
How well do you understand English?
How well do you read English?
How well do you write English?
How well do you speak L2?
How well do you understand L2?
How well do you read L2?
How well do you write L2?
The first of the demographic variables, student status, was determined by the following question:
Are you a high school, undergraduate, or graduate student?
a. I am not a student [disqualify]
b. I am a high school student [disqualify]
c. I am an undergraduate student
d. I am a graduate student
The author used two questions in the survey that asked about immigration status to construct two variables: permanent immigration status and temporary immigration status. The first question was asked in the screening survey: Are you an international student with one of the following visas: F, J, M, A, H1B, or K? (a. Yes b. No) The second question was asked in the main survey:
What is your immigration status?
a. U.S. citizen
b. U.S. citizen by naturalization
c. Permanent resident
d. Not a U.S. citizen
e. Dual citizenship or nationality
f. Deferred Action for Childhood Arrivals (“DACA”)
g. I do not wish to answer this question
h. I don’t know
Respondents who chose a, b, c, or e in the main survey were grouped as permanent, and dummy-coded in SPSS (0 = not permanent, 1 = permanent). Respondents who chose a in the screening survey question and d or f in the main survey question were grouped as temporary status.
Gender, age, age arrived in the United States, years in the United States, and zip code were constructed from the survey questions below.
What is your gender? a. Male b. Female
What year were you born? [numerical text field]
What year did you move to the United States? [numerical text field]
What is your current zip code? [numerical text field]
First-generation college students were identified using the following question: Ideally, what’s your intention for completing a degree? Check all that apply. Nineteen answers were available, and respondents could choose as many as applied; answer L, “I am the first in my family to get a college degree,” was used to identify first-generation college students. The author dummy-coded the variable (0 = not a first-generation college student, 1 = first-generation college student).
TABLE 1 | ||
Focus Group Demographics | ||
Gender | Men | 3 |
Women | 10 | |
Age | 25 to 34 | 4 |
18 to 24 | 7 | |
35 to 44 | 2 | |
Length of time in the US | Median | 3 years |
Mean | 5.2 years | |
Mode | 1 year |
Race and ethnicity categories were taken from adapted from a White House document on reporting race. 48 The question appeared as follows in the survey:
Please indicate your race:
a. Hispanic or Latino (a person of Cuban, Mexican, Puerto Rican, South or Central American, or other Spanish culture or origin, regardless of race)
b. American Indian or Alaska Native (a person having origins in any of the original peoples of North and South America, including Central America, who maintains cultural identification through tribal affiliation or community attachment)
c. Asian (a person having origins in any of the original peoples of the Far East, Southeast Asia, or the Indian Subcontinent, including, for example, Cambodia, China, India, Japan, Korea, Malaysia, Pakistan, the Philippine Islands, Thailand, and Vietnam)
d. Black or African American (a person having origins in any of the black racial groups of Africa)
e. Native Hawaiian or Other Pacific Islander (a person having origins in any of the original peoples of Hawaii, Guam, Samoa, or other Pacific Islands)
f. White (a person having origins in any of the original peoples of Europe)
g. Middle Easterner (a person having origins from the Middle Eastern countries)
h. North African (a person having origins from the North African countries)
i. From multiple races
j. Other (please specify)
The author dummy-coded all the race and ethnicity variables.
Figure 3 shows the binary grouping variables.
FIGURE 3 |
Binary Variables (N = 92) |
|
Finally, reading preferences were created from the matrix in figure 4. The author created variables for language preference for academic and nonacademic reading to determine whether students who are more likely to code-switch prefer a particular language for academic and leisure reading.
FIGURE 4 | ||||
Matrix, Academic vs. Non-academic Reading | ||||
Academic reading (books, e-books, articles, etc.) | Nonacademic reading (books, e-books, articles, etc.) | Internet searching for academic purposes | Internet searching NOT for academic purposes | |
I prefer English | ||||
I prefer L2 |
The data illustrate that students who speak a language besides English are more likely to do so in certain language domains such as home, school, or with friends. The ability of students to speak in English and their second language is also associated with how likely these students are to switch between languages.
Independent samples t-tests (hereafter t-tests) were not conducted for race and ethnicity categories with fewer than 10 respondents, which included Hispanic, American Indian, Native Hawaiian, Middle Easterner, North African, and multiple races. The t-tests showed that the following variables did not have a statistical difference in the mean reported CS score for the multilingual respondents:
The author found no associations using Somers’ d test between reported CS and the following variables:
Table 2 shows the variables where the author found that foreign-born students are very likely to switch to L2 (the non-English language) for information tasks and in linguistic domains such as home, school, and with friends.
TABLE 2 | ||||||
Independent Samples T-Tests | ||||||
DV | Variable Name | IV | N | Mean | Sig | Mean Difference |
Code-switch | Music | I do not prefer L2 for listening to music | 60 | 3.33 | 0.001 | 1.417 |
I prefer L2 for listening to music | 32 | 4.75 | ||||
Code-switch | Academic Reading | I do not prefer L2 for academic reading | 82 | 3.76 | 0.04 | 0.644 |
I prefer L2 for academic reading | 10 | 4.4 | ||||
Code-switch | Social Media | I do not prefer to use L2 for social media | 70 | 3.57 | 0.004 | 1.065 |
I prefer to use L2 for social media | 22 | 4.64 | ||||
Code-switch | Academic Internet Searching | I do not prefer to use L2 to search the Internet for academic purposes | 80 | 3.71 | 0.013 | 0.871 |
I prefer to use L2 to search the Internet for academic purposes | 12 | 4.58 | ||||
Code-switch | Leisurely Internet Searching | I do not prefer to use L2 to search the Internet leisurely, or for fun | 67 | 3.57 | 0.002 | 0.953 |
I prefer to use L2 to search the Internet leisurely, or for fun | 25 | 4.52 | ||||
Code-switch | Movies | I do not prefer to use L2 for watching movies | 58 | 3.19 | 0.001 | 1.722 |
I prefer to use L2 for watching movies | 34 | 4.91 | ||||
Code-switch | Instant Messaging | I do not prefer to use L2 for instant messaging | 68 | 3.54 | 0.002 | 1.081 |
I prefer to use L2 for instant messaging | 24 | 4.63 | ||||
Code-switch | SMS and Texting | I do not prefer to use L2 for SMS | 69 | 3.45 | 0.001 | 1.507 |
I prefer to use L2 for SMS | 23 | 4.96 | ||||
Code-switch | News and Current Events | I do not prefer to use L2 for news and current events | 68 | 3.51 | 0.002 | 1.194 |
I prefer to use L2 for news and current events | 24 | 4.71 | ||||
Code-switch | | I do not prefer to use L2 for communicating on e-mail | 73 | 3.63 | 0.019 | 0.949 |
I prefer to use L2 for communicating on e-mail | 19 | 4.58 | ||||
Code-switch | Phone and Voice Calls | I do not prefer to use L2 on the telephone or for voice calls | 58 | 3.29 | 0.001 | 1.442 |
I prefer to use L2 on the telephone or for voice calls | 34 | 4.74 | ||||
Code-switch | L2 at Home | I do not use L2 at home | 31 | 2.13 | 0.001 | 2.559 |
I use L2 at home | 61 | 4.69 | ||||
Code-switch | L2 at School | I do not use L2 at school | 76 | 3.58 | 0.001 | 1.421 |
I use L2 at school | 16 | 5 | ||||
Code-switch | L2 with Friends | I do not speak L2 with my friends | 51 | 3.02 | 0.001 | –1.81 |
I speak L2 with my friends | 41 | 4.83 | ||||
Code-switch | English with Family | I do not speak English with my family | 41 | 4.07 | 0.002 | 0.446 |
I speak English with my family | 51 | 3.63 | ||||
Code-switch | L2 with Family | I do not speak L2 with my family | 29 | 2.14 | 0.001 | 2.465 |
I speak L2 with my family | 63 | 4.6 | ||||
Code-switch | Comfortable with L2 | I am not comfortable with L2 | 50 | 3.14 | 0.001 | –1.503 |
I am very comfortable with L2 | 42 | 4.64 | ||||
Code-switch | Race and Ethnicity | Not black / African American | 76 | 4.17 | 0.001 | 1.984 |
Black / African American | 16 | 2.19 | ||||
Code-switch | Race and Ethnicity | Not Asian | 54 | 3.3 | 0.001 | –1.283 |
Asian | 38 | 4.58 |
Table 3 shows associations between language ability and CS using Somers’ d test. According to Laerd Statistics, “A value of –1 indicates that all pairs of observations are discordant and a value of +1 indicates that all pairs of observations are concordant.” 49 Table 3 shows moderate to weak effects.
TABLE 3 | |||
Somers’ D Test | |||
IV | DV | d | Approximate Significance |
L2 Speaking | Code-switching | 0.553 | 0.001 |
L2 Reading | Code-switching | 0.422 | 0.001 |
English Understanding | Code-switching | –0.305 | 0.003 |
English Reading | Code-switching | –0.299 | 0.004 |
English Writing | Code-switching | –0.237 | 0.013 |
Reported CS is positively associated with higher L2 speaking and reading ability with a moderate effect strength. Additionally, reported CS has an inverse association with the students’ ability to understand, read, and write English. The strength of the inverse association is weak (–0.305, –0.299, –0.237).
In the focus groups, all but one of the participants were bilingual (counting English and Jamaican Creole English as separate languages), and the single monolingual speaker of English spoke British English. There were 10 instances of students reporting code-switching in the focus group transcripts. The author also counted instances of switching from American to British English as a CS. Below are select quotes from the transcripts when participants were asked about how they used their languages. The following languages were discussed in the focus groups: English, Vietnamese, Hindi, French, French Creole (Kreyol), Cantonese, Mandarin, Hakka, Armenian, Bengali (Bangla), British English, Patwa (Jamaican Creole), Portuguese, Spanish, Russian, and Ukrainian. The data collected and described in this section illustrate the complexity of code-switching, the use of different languages for different types of information, the importance of community, and the presence of dialects both in English and non-English languages.
A quote that exemplifies the complexity of code-switching is from a Chinese student whose parents do not speak the same variety of Chinese. The student referred to the Hakka language as a “dialect,” describing her upbringing in a multi-Chinese dialect household, where neither of her parents spoke English very well. She explained that if she doesn’t want her parents to understand what she is saying she switches to English as her secret language. This quote from her is also illustrative of the linguistic diversity of Chinese languages:
“Yeah, when I am in school. Of course, most of the time I speak English in school. When I go home, I speak to my dad, my mom. I talk with my mom in dialect. Because she doesn’t know how to speak Mandarin. My dad speak to me in dialect. I respond back in Mandarin because my dad is not much good. Not much fluent in the dialect. But my dad I can hear and listen. You know, for my sisters, my brothers, I speak in like half English, half dialect, and maybe some has just like all everything mixed one or I don’t want to know what my parents to know what I am talking about, I speak English.”
A female Colombian undergraduate student described which languages she uses for different types of information:
‘‘I do like Spanish literature. When it comes to reading for pleasure, 50/50 for Spanish and English. I read a lot of religion in Spanish. Research, I don’t. And it’s a little limited in Spanish, because my entire college career has been in English. So even when I’ve tried, it’s difficult. I’ve done it, like when I travel back home. Or if I’m trying to read a text that’s in Spanish, and it’s academic, it takes a little bit of work. Because I have the information in my head in English.”
The same student also said:
“I pretty much read what I come across. As long as it’s interested, is interesting, and I’m interested in it. I don’t really have a preference with the news per se. I have always read the Bible in Spanish. Family and friends, and music. It’s like 80% in Spanish, 20% in English. Television is mostly English, just because lack of good stuff (laughter). Seriously. All those telenovelas, get out! So that’s out.”
Academic vocabulary in English and Spanish is learned. This student is obviously a fluent Spanish speaker and reader, but she knows her academic discipline in English only so she cannot readily code-switch to Spanish for academic tasks.
Community plays an important role in CS and, because this male Indian graduate student does not have a community of Bengali speakers, he has limited ability in Bengali. But his comprehension is maintained through preference for Bengali music:
“For research, basically I do everything in English because I have little family and friends here and back home so I can barely speak in Bengali. Other than that, I use English everywhere. For research, I follow the Wall Street Journal and the Economist. I don’t follow Bengali newspapers. Per se about music, I prefer Bengali, and Hindi, and English as well.”
This quote illustrates how a language may not manifest in one ability (like speaking) but can still be used and enjoyed for other kinds of information (music, movies, oral poetry, etc.). And when probed about mixing Hindi and English, the male Bengali student said, “Yeah. It’s Hinglish. Hindi and English. Yeah.”
English, like any language, has dialects. An undergraduate student from Grenada described how British English influences the kinds of information sources she seeks: “Yeah, in Grenada we’re more influenced by the British, so the spelling of our words are along the British standard.” And when probed about her preferences for British English sources, she said:
“Yes, maybe because I’m familiar with it. If I go on the Internet, I would look up bbc.uk. I’m only here three years, so I’ll go back and forth. So it’s not like I’m going to assimilate into the culture. I’m just like an outsider looking in. I feel most comfortable with the British way of news, expressing, so their writing. I don’t know how to explain. It’s so different here. When I did A levels, years ago. It’s just a whole different experience. I started at a community college, BMCC [Borough of Manhattan Community College]. It was like A level all over again, although it was an associate degree. I had to focus on the spelling of the words, how we spell color and you spell color. C-o-l-o-u-r, we spell it different. It was just, no other difficulty really.”
This is an example of how culture influences information seeking. This preference for British sources was verified by the two other Jamaican students who also spent some of their formative education in Jamaica.
This female Bengali undergraduate student uses Bengali exclusively with her family:
“So in school I speak in English. And I live with my aunt. And she works 14 hours a day, so I hardly see her. Whenever I see her for half an hour or an hour, we’re going to talk in Bangla [Bengali], but we don’t like mix stuff. Some of the words are in English. Like we don’t have a word for chair that’s in Bangla. So some of the words has to be in English. But um, most of the cases we don’t mix two languages. And even when I’m talking to my parents I’ll be talking Bangla.”
In response to a probe about making friends in school, this student noticed that the variety of Bengali spoken by the students she meets in Brooklyn contain an unusual accent. This is an example of the linguistic diversity that can exist in the same language between diaspora speakers and recently arrived speakers. Her CS shows that she enjoys and uses her languages for a variety of information consumption:
“I have met like a couple of Bangladeshi people in the school. But as she was saying, they don’t speak Bangla. What I feel they have really weird accent when they speak in Bangla. I feel weird when they speak in Bangla. So they like speak okay, if you speak comfortable speaking in English, I don’t have a problem. And in case of music, I feel like I’m more into Bangla and Hindi music. More than English music. English music is okay,…. If I listen to a song I’ll only listen to like, once or twice. But if I’m gonna listen to music, like the whole day, it has to be Bangla and Hindi. And even in the case of movies or soaps. I watch a lot of English soaps. Like I watched How I Met Your Mother, the whole 9th season. And I watch like Scrubs, and Big Bang Theory. And what’s the other one? I like watching CSI and Breaking Bad. …Yeah sorry, Breaking Bad.”
The following quote from a male Vietnamese undergraduate student shows his sophistication when looking for news about his home country: “I do read Vietnamese news and writings that’s about Vietnamese because I want to know what’s going on at the university. We still have government controlled media. So sometimes the news has bias. I like Vietnamese pop, for my personal life.” This male Chinese undergraduate student, after multiple probes, simply stated that both English and Chinese are important in his life: “I think the language is important, because I prefer the both language in my life for my enjoyment.”
The focus group data clearly show that multilinguals are using languages in different ways to consume and communicate information; they are switching readily in their daily lives before going to school, at school, and after school. These data highlight the complex and fluid ways language and culture influence how multilingual foreign-born students consume information. It’s important to note that within large language categories exist smaller subgroupings or varieties, whether they are varieties of English or Chinese dialects. These language groupings are not homogeneous populations, and care should be taken by academic librarians to acknowledge these differences in both accent and grammaticality. Interactions with foreign-born students should focus on communicating and not correcting.
The survey and data provide evidence of CS in foreign-born students in academic environments like the library at Brooklyn College (“I use L2 at school,” P < .001, MD = 5) and most likely with co-ethnic friends (“I speak L2 with my friends,” P < .001, MD = 1.81). The data show that students who report a second language are more likely to use it for a variety of information tasks like searching the Internet, consuming media, and communicating through social media or similar texting mediums, as shown in table 1. These data show that home is the place that L2 speakers will most likely code-switch ( P < .001, MD = 2.559). They also show that L2 is used for academic purpose like reading ( P < .05, MD = .644), especially for news and current events ( P < .001, MD = 1.194). The moderate association in table 2 between code-switching, L2 speaking, and L2 reading needs to be further investigated. Does it suggest that people who speak and read a language are more engaged in that language? It’s difficult to tell and is further complicated for languages that don’t have a strong print culture (Creoles, for example). The inverse association between English ability and CS is also interesting in that the higher the ability reported in English comprehension, reading, and writing, the lower the likelihood for CS. Brooklyn College library is a multilingual space, as the survey data show, and the language environment of foreign-born students is complex and fluid.
An important detail about the data collected in the survey is that 38 of the 92 respondents self-identified as Asian, and they are more likely than any of the other demographic category surveyed to code-switch ( P < .001, MD = 1.283). Additionally, there were respondents from the Caribbean who did not self-identify with the category of black/African American; they chose the “other” response to write in Caribbean. Similarly, there were Bengali students who did not choose to self-identify with Asian (the author considered, but did not include, a category for South Asian), and several of the respondents wrote in Bengali by selecting the “other” category. It was unfortunate that there were very few Hispanic foreign-born students, since Spanish is widely spoken in New York City.
Furthermore, it’s important to think about how Creole languages do not really appear in more formal academic language domains. For example, there are no calculus books written in Haitian Kreyol as there would be calculus books written in Vietnamese, Mandarin, or Spanish. Another question that these survey data cannot effectively comment on is CS for speakers of other varieties of English, like British English: do those speakers code-switch across varieties of English? For example, would the Trinidadian student from the focus group have reported in the survey or even be conscious of Americanizing her spelling and accent to function in an American English academic environments; is that a form of code-switching? These finite, qualitative distinctions are difficult to tease out in a survey. Fortunately, the focus group data show and identify some of these experiences and processes.
This section discusses two instances when the quantitative findings are complemented by the qualitative findings. There is one instance when the quantitative findings are contradicted by the qualitative data.
Both data sets in this study capture the importance of CS at home. The survey data correlate that home is an important place for language use, and the narrative from the Chinese student who speaks various dialects of Chinese and English with her parents, brothers, and sisters reveals the complexity of her language use. Both data sets show that there is diversity within Chinese and that, as librarians who value linguistic diversity, it is important to be aware that Chinese is not a monolithic language or cultural category. The application to practice is clear: if a library has large groups of Chinese speakers, it will be important to know and identify which cultural and linguistic groups use the academic library.
The second pattern found in both data sets includes the influence of language over the preference for consuming information like academic reading and news. The findings suggest that it is likely that students will code-switch in academic reading contexts and when reading news. The focus group data give us more insight into this information behavior. For example, the data from the Colombian undergraduate student tell us that her code-switching is in very specific domains, like leisurely reading and religious reading, and that she has difficulty reading academic Spanish. This information is important to know for building collections and for supporting students who may want to take their academic experiences to their home country for an internship, for a job, or for pursuing graduate work. In this study we also see how dialect preferences in English influence how Caribbean speakers gravitate toward the British Standard English with consuming news and current events, which is also an example of dialect-switching. These data showing how English dialects can affect information behavior is important for helping librarians understand that, not only is there diversity across languages, there are dialectal differences within major language groups like English, Chinese, Spanish, and others.
The third pattern found in the survey, that students are more likely to use L2 with friends, is contradicted by the example of the female Bengali student who finds that Bengali accents spoken by Brooklyn students are unusual. Again, while students are likely to switch with friends, students who have just arrived may not have friends who speak the same dialect. This is important when distinguishing between immigrant students and international students when it comes to conducting further studies of co-ethnic language use in academic libraries.
Future studies involving code-switching should focus on capturing both quantitative and qualitative data. It’s just as important to understand individual experiences and processes around code-switching, because this helps in interpreting the statistical data. Qualitative data can also reveal missed or new variables that will give the researcher better insight into the relationships between language and information.
This survey and analysis contain some limitations that should be kept in mind when considering the findings. The first limitation is the nature of self-reported usage data, though the author attempted to increase the internal reliability of the data by using focus groups to triangulate the results. The second limitation is related to sampling: the focus groups were very small, and the survey was a convenience sample. Third, income as a variable was collected as a rough estimate using ACS 2014 data by zip code. This is not the best way to capture income data, but it does provide some information. Finally, it should be noted that many of the terms used throughout this article, including race, gender, ethnicity, and immigration status, are social science terms often used in survey research; their use is not meant to give offense or intentionally exclude any groups. There is always some aspect of reductionism that occurs in survey-based research, and the author welcomes feedback on how to make the survey categories more inclusive.
CS is a well-established phenomenon outside of library and information science, and this paper aimed to introduce librarians to this concept and document this behavior in an academic library setting. There are many implications of CS for academic library services, especially in the area of linguistic diversity, standard 6 of ACRL’s Diversity Standards: Cultural Competency for Academic Libraries, which reads: “Librarians and library staff shall support the preservation and promotion of linguistic diversity, and work to foster a climate of inclusion aimed at eliminating discrimination and oppression based on linguistic or other diversities.” 50
The data from this study provide evidence that language influences the information behavior of students in the form of code-switching and dialect switching. More research could be done to investigate information code-switching, which the author has broadly defined as changing languages or dialects for particular information tasks. This kind of research would allow librarians to map language use, language choice, and language preferences of students to actual library collections, services, and resources. Furthermore, this research would be valuable for serving first-generation college students, generation 1.5 students, international students, and immigrant students.
Another area of critical importance is being inclusive of non-English languages in collection development: “collection managers should be attentive to represent the linguistic needs of library constituents, and assure that library resources in print or electronic formats are available, especially to support the academic curricula reflecting all diversity issues, including those of visually disabled constituents.” 51 Increasing the visibility of non-English scholarly sources can be as simple as creating library guides that show students how to access peer-reviewed journals and open access indexes in non-English languages. Engaging with the scholarly literature in non-English languages is particularly important in the social sciences. For example, librarians could create a guide that would allow Spanish-speaking students of urban sociology to engage with and synthesize sociological ideas in Spanish language journals with concepts from English language journals. The practice of incorporating non-English sources into English-language papers is a long scholarly tradition in the humanities. Language access is also an important concept in the diversity standards: “Provide and advocate for the provision of information, reference, referrals, instruction, collection management, and other services in the language appropriate to their constituencies, including the use of interpreters.” As the CS data suggest, our libraries are not monolingual spaces, so making sure that printers can print in different scripts (and in general having technology capable of supporting users’ linguistic preferences), as well as having welcome signage in other languages are steps that libraries can take to make non-English speakers feel more included when using the library as a space for studying or meeting with classmates. Ideally, supporting linguistic diversity in academic libraries would include multilingual staff who could create library instruction and other academic library services that cater to large linguistic populations served by the academic library.
The purpose of this study was to explore language use, language choice, and language preferences in academic libraries, and the author found evidence for code-switching patterns in both qualitative and quantitative data. In the analysis, the author maintains that code-switching patterns are correlated with information tasks and argues that more research could be done on information code-switching to give librarians data on language use and apply those data to library services. For the foreign-born students analyzed in this study, it is clear that their culture and their non-English language represent an active and important part of their identity, information consumption, and academic socialization. In their language choice for information, there is enough statistically significant evidence for information code-switching, when students switch languages for a particular information task. Yet, in most of the academic library literature, these active language communities, their patterns of use, and their preferences have not been the subject of research. These data show that Brooklyn College Library is a rich multilingual space, yet there are only a few studies that discuss multilingualism in academic libraries. What can academic librarians do with these kinds of data?
There are many practical recommendations relating to linguistic diversity, including creating a multilingual-friendly environment. Reference assistance could include offering specialized library instruction or orientations for immigrant students, first-generation college students, and international students. Public services staff could also receive linguistic diversity training that includes information on ESL and EFL populations. Linguistic diversity training might also focus content on creating sensitivity and awareness of patrons who are linguistic minorities (for example, Spanish speakers who speak other Central American languages such as Mayan languages), as well as information about Creoles and pidgins, how accents work, nonwritten languages, and varieties of spoken English that may be relevant to the patron population. Computer labs can be language-friendly, with a variety of keyboard formats and printers available for people who need to print e-mails, share notes, and look up concepts in languages other than English. In the focus group interview with the Colombian undergraduate student, she spoke about wanting to gain some experience working in Colombia; however, she lacked the Spanish academic vocabulary to be competitive. To help students like her, academic librarians could create LibGuides for non-English scholarly sources that include Latindex, for example ( www.latindex.unam.mx/latindex/inicio ). Libraries could focus on hiring multilingual librarians. There is clearly a need for more research in transnationalism, especially in academic libraries that have a high number of foreign-born students. Are these students trying to use their American college degrees and create transnational careers that take advantage of their cultural capital? Are academic libraries spaces currently treated as monolingual rather than multilingual spaces? How does this affect our practice, and how can libraries change to support these students in their information needs? Monica Jacobe, director of the Center for American Language & Culture at The College of New Jersey, speaks about immigration trends in student success:
First-generation college students will no longer be primarily American-born students from working class families. Instead, many more students in that category will be recent immigrants, born all over the world, who completed high school in the U.S. For many schools, they will “look” on paper like domestic applicants, but the support they need will be very different.
How can academic librarians imagine a shift from monocultural and monolingual approaches to multilingual approaches? And what services will need to be rethought? These are the kinds of questions additional studies on language use and libraries can answer. CS is just one conceptual tool from sociolinguistics that has very practical applications in our work with international and immigrant students.
1. Barbara E. Bullock and Almeida Jacqueline Toribio, “Themes in the Study of Code-Switching,” in The Cambridge Handbook of Linguistic Code-Switching , eds. Barbara E. Bullock and Almeida Jacqueline Toribio, Cambridge Handbooks in Linguistics (Cambridge; New York: Cambridge University Press, 2009), 1–19.
2. Jan-Peter Blom and John J. Gumperz, “Social Meaning in Linguistic Structure: Code-Switching in Norway,” in Directions in Sociolinguistics: The Ethnography of Communication , eds. John J. Gumperz and Dell Hymes (New York: Blackwell, 1986), 409.
3. Ibid., 411.
4. Ibid.
5. Carol Myers-Scotton, “The Rise of Codeswitching as a Research Topic,” in Social Motivations for Codeswitching: Evidence from Africa (Oxford: Clarendon Press, 1995), 57.
6. Joshua Fishman, “Domains and the Relationship between Micro- and Macro-Sociolinguistics,” in Directions in Sociolinguistics: The Ethnography of Communication , eds. John J. Gumperz and Dell Hymes (New York: Blackwell, 1986), 441.
7. Ibid., 442–43.
8. L.B. Breitborde, “Levels of Analysis in Sociolinguistic Explanation: Bilingual Code Switching, Social Relations, and Domain Theory,” International Journal of the Sociology of Language 1983, no. 39 (Jan. 1983): 19.
9. Fishman, “Domains and the Relationship between Micro- and Macro-Sociolinguistics,” 447.
10. Ibid., 448.
11. Alejandro Portes and Rubén Rumbaut, “Children of Immigrants Longitudinal Study (CILS), 1991–2006,” Children of Immigrants Longitudinal Study (CILS), 1991–2006 (ICPSR 20520) , 2009, available online at www.icpsr.umich.edu/icpsrweb/RCMD/studies/20520 [accessed 24 March 2015].
12. Breitborde, “Levels of Analysis in Sociolinguistic Explanation,” 5.
13. Shana Poplack, “Sometimes I’ll Start a Sentence in Spanish Y Termino En Español: Toward a Typology of Codeswitching,” Linguistics 18, no. 7/8 (1980): 581–618, doi:10.1515/ling.1980.18.7-8.581 .
14. Ibid., 583.
15. Ibid., 584.
16. Ibid., 600.
17. Ibid., 610.
18. Jannis Androutsopoulos, “Language and the Three Spheres of Hip Hop,” in Global Linguistic Flows: Hip Hop Cultures, Youth Identities, and the Politics of Language , eds. H. Sammy Alim, Awad Ibrahim, and Alastair Pennycook (New York: Routledge, 2009), 54–58.
19. Elena Seoane, “World Englishes Today,” in World Englishes: New Theoretical and Methodological Considerations , eds. Elena Seoane and Cristina Suárez Gómez, Varieties of English around the World G57 (Amsterdam; Philadelphia: John Benjamins Publishing Company, 2016), 1.
20. Marta Fairclough and Flavia Belpoliti, “Emerging Literacy in Spanish among Hispanic Heritage Language University Students in the USA: A Pilot Study,” International Journal of Bilingual Education and Bilingualism 19, no. 2 (Mar. 3, 2016): 185–201, doi:10.1080/13670050.2015.1037718 .
21. Ibid., 186.
22. Ibid., 189.
23. Magdalena Malechová, “Multilingualism as a Sociolinguistic Contact Phenomenon with Regard to Current Forms of Multilingual Communication Code-Switching as One of the Contemporary Communication Trends,” Višejezičnost Kao Sociolingvistički Fenomen Kontakta Imajući U Vidu Suvremene Oblike Višejezične Komunikacije Mijenjanje Kodova Kao Suvremeni Komunikacijski Trend 49, no. 1/2 (July 2016): 86–93.
24. Ibid., 91.
25. Bettina Kümmerling-Meibauer, “Code-Switching in Multilingual Picturebooks,” Bookbird: A Journal of International Children’s Literature (Johns Hopkins University Press) 51, no. 3 (July 2013): 12–21.
26. Ibid., 19.
27. Amanda B. Click, Claire Walker Wiley, and Meggan Houlihan, “The Internationalization of the Academic Library: A Systematic Review of 25 Years of Literature on International Students,” doi:10.5860/crl.v78i3.16591 .
28. Nadia Caidi, Danielle Allard, and Lisa Quirke, “Information Practices of Immigrants,” Annual Review of Information Science and Technology 44, no. 1 (Jan. 1, 2010): 515, doi:10.1002/aris.2010.1440440118 .
29. Ignacio J. Ferrer-Vinent, “For English, Press 1: International Students’ Language Preference at the Reference Desk,” Reference Librarian 51, no. 3 (Sept. 2010): 189–201, doi:10.1080/02763871003800429 .
30. Karen Bordonaro, “Is Library Database Searching a Language Learning Activity?” College & Research Libraries 71, no. 3 (May 2010): 273–84.
31. Sara Luly and Holger Lenz, “Language in Context: A Model of Language Oriented Library Instruction,” Journal of Academic Librarianship 41, no. 2 (Mar. 2015): 140–48, doi:10.1016/j.acalib.2015.01.001 .
32. Frans Albarillo, “Is the Library’s Online Orientation Program Effective with English Language Learners?” College & Research Libraries , 78, no. 5 (July 2017): 656–59, doi:10.5860/crl.78.5.652 .
33. Sonia Smith, “Library Instruction for Romanized Hebrew,” Journal of Academic Librarianship 41, no. 2 (Mar. 2015): 197–200, doi:10.1016/j.acalib.2014.08.003 .
34. Ibid., 199.
35. Molavi Fereshteh, “Main Issues in Cataloging Persian Language Materials in North America,” Cataloging & Classification Quarterly 43, no. 2 (Dec. 8, 2006): 77–82, doi:10.1300/J104v43n02_06 .
36. Kim SungKyung, “Romanization in Cataloging of Korean Materials,” Cataloging & Classification Quarterly 43, no. 2 (Dec. 8, 2006): 53–76, doi:10.1300/J104v43n02_05 .
37. Hikaru Nakano, “Non-Roman Language Cataloging in Bulk: A Case Study of Japanese Language Materials,” Cataloging & Classification Quarterly 55, no. 2 (Feb. 2017): 75–88, doi:10.1080/01639374.2016.1250853 .
38. Albarillo, “Is the Library’s Online Orientation Program Effective with English Language Learners?” 3.
39. Xiang Li, Kevin McDowell, and Xiaotong Wang, “Building Bridges: Outreach to International Students via Vernacular Language Videos,” Reference Services Review 44, no. 3 (July 2016): 325, doi:10.1108/RSR-10-2015-0044 .
40. Misa Mi and Yingting Zhang, “Culturally Competent Library Services and Related Factors among Health Sciences Librarians: An Exploratory Study,” Journal of the Medical Library Association 105, no. 2 (Apr. 2017): 135, doi:10.5195/jmla.2017.203 .
41. Ibid., 136.
42. Ibid., 133.
43. Institutional Research and Data Analysis at Brooklyn College, City University of New York, “Fall 2014 Final Enrollment Reports, Enrollment Table 21 Country of Birth,” available online at www.brooklyn.cuny.edu/bc/offices/avpbandp/ipra/enrollment/F14/Enrollment-Table21.pdf [accessed 20 November 2016].
44. United States Census Bureau, “American FactFinder,” available online at http://factfinder.census.gov/faces/nav/jsf/pages/index.xhtml [accessed 20 November 2016].
45. Laerd Statistics, “Independent-Samples T-Test Using SPSS Statistics,” Statistical Tutorials and Software Guides (2015), available online at https://statistics.laerd.com [accessed 21 November 2016].
46. Laerd Statistics, “Somers’ D Using SPSS Statistics,” Statistical Tutorials and Software Guides (2016), available online at https://statistics.laerd.com [accessed 21 November 2016].
47. Arlene Fink, How to Conduct Surveys: A Step-by-Step Guide (Los Angeles: SAGE, 2013), 45.
48. Executive Office of the President, Office of Management and Budget, “Revisions to the Standards for the Classification of Federal Data on Race and Ethnicity,” The White House, available online at https://www.whitehouse.gov/node/15626 [accessed 21 November 2016].
49. Laerd Statistics, “Somers’ D Using SPSS Statistics.”
50. Association of College and Research Libraries, “Diversity Standards: Cultural Competency for Academic Libraries” (2012), available online at www.ala.org/acrl/standards/diversity [accessed 30 August 2016].
51. Racial and Ethnic Diversity Committee Members, “Diversity Standards: Cultural Competency for Academic Libraries (2012),” Association of College and Research Libraries, available online at www.ala.org/acrl/standards/diversity [accessed 21 November 2016].
* Frans Albarillo is Assistant Professor/Reference & Instruction at Brooklyn College, City University of New York; e-mail: [email protected] . ©2018 Frans Albarillo, Attribution-NonCommercial ( http://creativecommons.org/licenses/by-nc/4.0/ ) CC BY-NC.
Contact ACRL for article usage statistics from 2010-April 2017.
2024 |
January: 616 |
February: 493 |
March: 736 |
April: 666 |
May: 590 |
June: 320 |
July: 347 |
August: 401 |
September: 288 |
2023 |
January: 495 |
February: 495 |
March: 786 |
April: 437 |
May: 708 |
June: 396 |
July: 196 |
August: 183 |
September: 379 |
October: 281 |
November: 365 |
December: 291 |
2022 |
January: 675 |
February: 663 |
March: 810 |
April: 670 |
May: 712 |
June: 392 |
July: 295 |
August: 207 |
September: 341 |
October: 665 |
November: 648 |
December: 475 |
2021 |
January: 697 |
February: 734 |
March: 1060 |
April: 945 |
May: 882 |
June: 677 |
July: 421 |
August: 440 |
September: 679 |
October: 855 |
November: 840 |
December: 657 |
2020 |
January: 815 |
February: 1058 |
March: 1094 |
April: 744 |
May: 776 |
June: 660 |
July: 582 |
August: 481 |
September: 707 |
October: 770 |
November: 787 |
December: 779 |
2019 |
January: 94 |
February: 164 |
March: 261 |
April: 667 |
May: 293 |
June: 242 |
July: 392 |
August: 346 |
September: 618 |
October: 804 |
November: 833 |
December: 682 |
2018 |
January: 14 |
February: 11 |
March: 6 |
April: 8 |
May: 5 |
June: 21 |
July: 582 |
August: 249 |
September: 88 |
October: 82 |
November: 62 |
December: 63 |
2017 |
April: 0 |
May: 0 |
June: 0 |
July: 0 |
August: 0 |
September: 0 |
October: 146 |
November: 62 |
December: 29 |
© 2024 Association of College and Research Libraries , a division of the American Library Association
Print ISSN: 0010-0870 | Online ISSN: 2150-6701
ALA Privacy Policy
ISSN: 2150-6701
"Pórtate bien; esta gente son blanca" is a line I'd often hear my mom say when I was growing up. It translates to: "Behave yourself; these people are white."
I was introduced to code-switching at a relatively young age — significantly younger than a lot of my inner-city Latine peers. I learned the skill before I even attended school because my parents were desperately trying to assimilate back into American culture after attending college in the Dominican Republic.
At the time, in the early '90s, my parents weren't even aware that there was an actual term to describe what they were learning as two college-educated Dominican immigrants trying to achieve the American dream. They didn't know that trying to mask their Latin American accents and attempting to reduce the volume at which they spoke was actually a form of code-switching. Back then, the only folks using that term were academics. But that's what it was. My parents were teaching my younger siblings and me how to speak, behave, dress, and carry ourselves in predominantly white spaces. They were teaching us an inherent survival tool.
What does it mean to code-switch, exactly? For many marginalized communities, particularly communities of color, it means assimilating to white-centric ideas around what it means to be "well-behaved" or "professional." It's shifting the language you use or the way you physically present or express yourself to make others more comfortable.
Don't I deserve to be treated fairly and respectfully regardless of my background or the color of my skin?
By the time I was in college, code-switching was so natural to me that I wasn't even always cognizant of when I was doing it. I learned how to do it everywhere, from school to jobs and even in friend groups where I was the only person representing both Latinidad and Blackness. Of course, I knew the spaces where I could authentically be myself — often those occupied by Black and Latine folks — but I had learned how to instantly tone myself down in spaces where the folks in the room were mostly white. I soon began to consider my code-switching to be a superpower. And in many ways, it can be. But I also learned that it could come at a serious psychological cost.
My perception of code-switching almost immediately changed when I started working for Latin media startups. My direct reports were all Latina women, and they often had either Latin American, Miami, or New York accents. Many of them were extremely assertive when they spoke — and unapologetically spoke loudly and in Spanglish. Not to mention everyone wore whatever they wanted: long acrylic nails, red lipstick, oversized hoop earrings, heels, and crop tops. There was no official dress code.
Going to work felt like getting together with family. To be clear, it wasn't perfect. Latine families can be toxic, too. But I always felt safe enough to be my most authentic self, and I've carried that into every role I've taken throughout my career. It made me realize that before that point, whenever I found myself in a white-dominated space, I was adjusting everything about myself, from my style of speech to how I express myself, in the hopes that I would be treated fairly and respectfully. Then it hit me: at what cost? At the cost of honoring myself? And don't I deserve to be treated fairly and respectfully regardless of my background or the color of my skin? It taught me that being myself and being able to embrace all the things that make me uniquely Johanna, including my Latina identity and Dominican heritage, has absolutely nothing to do with professionalism.
Coming to this realization early on in my career made me realize how much energy it takes for one to code-switch for eight hours a day, five days a week, and in some cases more. It became crystal clear to me how that can easily come in the way of one's creativity, confidence, and even productivity.
Research shows that code-switching often occurs in spaces where negative stereotypes of marginalized communities run high , and it is often the folks who subscribe to these stereotypes who lead and dictate what is considered "appropriate" or "professional behavior." I am proud to be living in a day and age when Latines not only have been unapologetically doing away with the code-switching — and the hiding, masking, and toning down of their Latinidad — but are also thriving as a result of it. Some of our biggest stars are perfect examples of that. Bad Bunny is the No. 1 artist in the world, singing Latin trap and reggaetón music while refusing to do it in English. Sofía Vergara proved she's an excellent actress in some ways because of her heavy Colombian accent with a recent Emmy nomination for her outstanding performance in "Griselda."
That's why there's no better time than this Latine Heritage Month to delve into how this generation of Latines are thriving and finding success from simply being their most authentic selves, regardless of the spaces they occupy. Through opinion pieces, profiles, and personal essays, we're exploring how Latines across industries drastically transformed their careers and their lives when they started to embrace themselves fully. You can start reading here .
For every Latine reader who comes across this package, I hope it reminds you that you should behave however you want!
Con amor, Johanna Ferreira, Juntos content director
Johanna Ferreira is the content director for PS Juntos. With more than 10 years of experience, Johanna focuses on how intersectional identities are a central part of Latine culture. Previously, she spent close to three years as the deputy editor at HipLatina, and she has freelanced for numerous outlets including Refinery29, Oprah magazine, Allure, InStyle, and Well+Good. She has also moderated and spoken on numerous panels on Latine identity.
Advances in machine learning and deep learning research are reshaping our technology. Machine learning and deep learning have accomplished various astounding feats, and key research articles have resulted in technical advances used by billions of people. The research in this sector is advancing at a breakneck pace and assisting you to keep up. Here is a collection of the most important scientific study papers in machine learning.
The authors of this work examined why ACGAN training becomes unstable as the number of classes in the dataset grows. The researchers revealed that the unstable training occurs due to a gradient explosion problem caused by the unboundedness of the input feature vectors and the classifier’s poor classification capabilities during the early training stage. The researchers presented the Data-to-Data Cross-Entropy loss (D2D-CE) and the Rebooted Auxiliary Classifier Generative Adversarial Network to alleviate the instability and reinforce ACGAN (ReACGAN). Additionally, extensive tests of ReACGAN demonstrate that it is resistant to hyperparameter selection and is compatible with a variety of architectures and differentiable augmentations.
This article is ranked #1 on CIFAR-10 for Conditional Image Generation.
For the research paper, read here .
For code, see here .
The authors presented a straightforward and computationally fast unsupervised strategy for learning dense spacetime representations from unlabeled films in this study. The approach demonstrates rapid convergence of training and a high degree of data efficiency. Furthermore, the researchers obtain VOS accuracy superior to previous results despite employing a fraction of the previously necessary training data. The researchers acknowledge that the research findings may be utilised maliciously, such as for unlawful surveillance, and that they are excited to investigate how this skill might be used to better learn a broader spectrum of invariances by exploiting larger temporal windows in movies with complex (ego-)motion, which is more prone to disocclusions.
This study is ranked #1 on DAVIS 2017 for Unsupervised Video Object Segmentation (val).
The authors offer an atlas-based technique for producing unsupervised temporally consistent surface reconstructions by requiring a point on the canonical shape representation to translate to metrically consistent 3D locations on the reconstructed surfaces. Finally, the researchers envisage a plethora of potential applications for the method. For example, by substituting an image-based loss for the Chamfer distance, one may apply the method to RGB video sequences, which the researchers feel will spur development in video-based 3D reconstruction.
This article is ranked #1 on ANIM in the category of Surface Reconstruction.
The researchers propose a revolutionary interactive architecture called EdgeFlow that uses user interaction data without resorting to post-processing or iterative optimisation. The suggested technique achieves state-of-the-art performance on common benchmarks due to its coarse-to-fine network design. Additionally, the researchers create an effective interactive segmentation tool that enables the user to improve the segmentation result through flexible options incrementally.
This paper is ranked #1 on Interactive Segmentation on PASCAL VOC
The authors of this work examined whether it is possible to transfer the success of task-agnostic web-scale pre-training in natural language processing to another domain. The findings indicate that adopting this formula resulted in the emergence of similar behaviours in the field of computer vision, and the authors examine the social ramifications of this line of research. CLIP models learn to accomplish a range of tasks during pre-training to optimise their training objective. Using natural language prompting, CLIP can then use this task learning to enable zero-shot transfer to many existing datasets. When applied at a large scale, this technique can compete with task-specific supervised models, while there is still much space for improvement.
This research is ranked #1 on Zero-Shot Transfer Image Classification on SUN
The researchers in this article conduct a thorough examination of the features of convolutions and transformers, resulting in a principled approach for combining them into a new family of models dubbed CoAtNet. Extensive experiments demonstrate that CoAtNet combines the advantages of ConvNets and Transformers, achieving state-of-the-art performance across a range of data sizes and compute budgets. Take note that this article is currently concentrating on ImageNet classification for model construction. However, the researchers believe their approach is relevant to a broader range of applications, such as object detection and semantic segmentation.
This paper is ranked #1 on Image Classification on ImageNet (using extra training data).
The authors of this article suggest the SwinIR image restoration model, which is based on the Swin Transformer . The model comprises three modules: shallow feature extraction, deep feature extraction, and human-recognition reconstruction. For deep feature extraction, the researchers employ a stack of residual Swin Transformer blocks (RSTB), each formed of Swin Transformer layers, a convolution layer, and a residual connection.
This research article is ranked #1 on Image Super-Resolution on Manga109 – 4x upscaling.
Ways to incorporate historical data are still unclear: initialising reward estimates with historical samples can suffer from bogus and imbalanced data coverage, leading to computational and storage issues—particularly in continuous action spaces. The paper addresses the obstacles by proposing ‘Artificial Replay’, an algorithm to incorporate historical data into any arbitrary base bandit algorithm.
Read the full paper here .
Author(s) – Sean R. Sinclair et al.
The paper proposes an algorithm in which the meta-learner teaches itself to overcome the meta-optimisation challenge. The algorithm focuses on meta-learning with gradients, which guarantees performance improvements. Furthermore, the paper also looks at how bootstrapping opens up possibilities.
Read the full paper here .
Author(s) – Sebastian Flennerhag et al.
The research describes the LaMDA system which caused chaos in AI this summer when a former Google engineer claimed that it had shown signs of sentience. LaMDA is a family of large language models for dialogue applications based on Transformer architecture. The interesting feature of the model is its fine-tuning with human-annotated data and the possibility of consulting external sources. This is a very interesting model family, which we might encounter in many applications we use daily.
Author(s) – Yujia Li et al.
Systems can help programmers become more productive. The following research addresses the problems with incorporating innovations in AI into these systems. AlphaCode is a system that creates solutions for problems that require deeper reasoning.
Author(s) – Tian Dong et al.
The paper focuses on Privacy Preserving Machine Learning, specifically deducting the leakage of sensitive data in machine learning. It puts forth one of the first propositions of using dataset condensation techniques to preserve the data efficiency during model training and furnish membership privacy.
Author(s) – Léo Grinsztajn, Edouard Oyallon and Gaël Varoquaux
The research answers why deep learning models still find it hard to compete on tabular data compared to tree-based models. It is shown that MLP-like architectures are more sensitive to uninformative features in data compared to their tree-based counterparts.
Author(s) – Samuel Daulton et al.
The paper proposes ‘MORBO’, a scalable method for multiple-objective BO as it performs better than that of high-dimensional search spaces. MORBO significantly improves the sample efficiency and, where existing BO algorithms fail, MORBO provides improved sample efficiencies over the current approach.
Author(s) – Yann LeCun
The research offers a vision about how to progress towards general AI. The study combines several concepts: a configurable predictive world model, behaviour driven through intrinsic motivation, and hierarchical joint embedding architectures trained with self-supervised
learning.
Author(s) – Shreshth Tuli, Giuliano Casale and Nicholas R. Jennings
This is a specialised paper applying transformer architecture to the problem of unsupervised anomaly detection in multivariate time series. Many architectures which were successful in other fields are, at some point, also being applied to time series. The research shows improved performance on some known data sets.
Author(s) – Zhiqi Bu et al.
In the paper, researchers study the problem of differentially private (DP) fine-tuning of large pre-trained models—a recent privacy-preserving approach suitable for solving downstream tasks with sensitive data. Existing work has demonstrated that high accuracy is possible under strong privacy constraints yet requires significant computational overhead or modifications to the network architecture.
Usually, increasing model size when pretraining natural language representations often result in improved performance on downstream tasks, but the training times become longer. To address these problems, the authors in their work presented two parameter-reduction techniques to lower memory consumption and increase the training speed of BERT. The authors also used a self-supervised loss that focuses on modelling inter-sentence coherence and consistently helped downstream tasks with multi-sentence inputs. According to results, this model established new state-of-the-art results on the GLUE, RACE, and squad benchmarks while having fewer parameters compared to BERT-large.
Check the paper here .
Microsoft Research, along with the University of Washington and the University of California, in this paper, introduced a model-agnostic and task agnostic methodology for testing NLP models known as CheckList. This is also the winner of the best paper award at the ACL conference this year. It included a matrix of general linguistic capabilities and test types that facilitate comprehensive test ideation, as well as a software tool to generate a large and diverse number of test cases quickly.
Linformer is a Transformer architecture for tackling the self-attention bottleneck in Transformers. It reduces self-attention to an O(n) operation in both space- and time complexity. It is a new self-attention mechanism which allows the researchers to compute the contextual mapping in linear time and memory complexity with respect to the sequence length.
Read more about the paper here .
Plug and Play Language Models ( PPLM ) are a combination of pre-trained language models with one or more simple attribute classifiers. This, in turn, assists in text generation without any further training. According to the authors, model samples demonstrated control over sentiment styles, and extensive automated and human-annotated evaluations showed attribute alignment and fluency.
The researchers at Google, in this paper , introduced Reformer. This work showcased that the architecture of a Transformer can be executed efficiently on long sequences and with small memory. The authors believe that the ability to handle long sequences opens the way for the use of the Reformer on many generative tasks. In addition to generating very long coherent text, the Reformer can bring the power of Transformer models to other domains like time-series forecasting, music, image and video generation.
The irony here is that one of the popular language models, Transformers have been made to do computer vision tasks. In this paper , the authors claimed that the vision transformer could go toe-to-toe with the state-of-the-art models on image recognition benchmarks, reaching accuracies as high as 88.36% on ImageNet and 94.55% on CIFAR-100. For this, the vision transformer receives input as a one-dimensional sequence of token embeddings. The image is then reshaped into a sequence of flattened 2D patches. The transformers in this work use constant widths through all of its layers.
Winner of the CVPR best paper award, in this work, the authors proposed a method to learn 3D deformable object categories from raw single-view images, without external supervision. This method uses an autoencoder that factored each input image into depth, albedo, viewpoint and illumination. The authors showcased that reasoning about illumination can be used to exploit the underlying object symmetry even if the appearance is not symmetric due to shading.
In this paper, OpenAI researchers examined whether similar models can learn useful representations for images. For this, the researchers trained a sequence Transformer to auto-regressively predict pixels, without incorporating knowledge of the 2D input structure. Despite training on low-resolution ImageNet without labels, the researchers found that a GPT-2 scale model learns strong image representations as measured by linear probing, fine-tuning, and low-data classification. On CIFAR-10, it achieved 96.3% accuracy with a linear probe, outperforming a supervised Wide ResNet, and 99.0% accuracy with full fine-tuning and matching the top supervised pre-trained models. An even larger model, trained on a mixture of ImageNet and web images, is competitive with self-supervised benchmarks on ImageNet, achieving 72.0% top-1 accuracy on a linear probe of their features.
In this paper, the authors provided a high-level introduction to deep RL , discussed some of its initial applications to neuroscience, and surveyed its wider implications for research on brain and behaviour and concluded with a list of opportunities for next-stage research. Although DeepRL seems to be promising, the authors wrote that it is still a work in progress and its implications in neuroscience should be looked at as a great opportunity. For instance, deep RL provides an agent-based framework for studying the way that reward shapes representation, and how representation, in turn, shapes learning and decision making — two issues which together span a large swath of what is most central to neuroscience.
Why humans doing certain things are often linked to dopamine , a hormone that acts as the reward system (think: the likes on your Instagram page). So, keeping this fact in hindsight, DeepMind with the help of Harvard labs, analysed dopamine cells in mice and recorded how the mice received rewards while they learned a task. They then checked these recordings for consistency in the activity of the dopamine neurons with standard temporal difference algorithms. This paper proposed an account of dopamine-based reinforcement learning inspired by recent artificial intelligence research on distributional reinforcement learning. The authors hypothesised that the brain represents possible future rewards not as a single mean but as a probability distribution, effectively representing multiple future outcomes simultaneously and in parallel.
In this paper, the authors bridged natural language processing (NLP) and reinforcement learning (RL). They examined both recurrent LSTM models and large-scale Transformer models for NLP and discrete-action space tasks for RL. The results suggested that the lottery ticket hypothesis is not restricted to supervised learning of natural images, but rather represents a broader phenomenon in deep neural networks.
In this paper, the authors explored if the reward function itself can be a good locus of learned knowledge. They proposed a scalable framework for learning useful intrinsic reward functions across multiple lifetimes of experience and showed that it is feasible to learn and capture knowledge about long-term exploration and exploitation into a reward function.
The progress of AutoML has largely focused on the architecture of neural networks, where it has relied on sophisticated expert-designed layers as building blocks, or similarly restrictive search spaces. In this paper , the authors showed that AutoML could go further with AutoML Zero, that automatically discovers complete machine learning algorithms just using basic mathematical operations as building blocks. The researchers demonstrated this by introducing a novel framework that significantly reduced human bias through a generic search space.
Batch normalization is an essential component of meta-learning pipelines. However, there are several challenges. So, in this paper, the authors evaluated a range of approaches to batch normalization for meta-learning scenarios and developed a novel approach — TaskNorm. Experiments demonstrated that the choice of batch normalization has a dramatic effect on both classification accuracy and training time for both gradient-based and gradient-free meta-learning approaches. The TaskNorm has been found to be consistently improving the performance.
Meta-learning algorithms need meta-training tasks to be mutually exclusive, such that no single model can solve all of the tasks at once. In this paper, the authors designed a meta-regularisation objective using information theory that successfully uses data from non-mutually-exclusive tasks to efficiently adapt to novel tasks.
Model Agnostic Meta-Learning (MAML) consists of optimisation loops, from which the inner loop can efficiently learn new tasks. In this paper, the authors demonstrated that feature reuse is the dominant factor and led to ANIL (Almost No Inner Loop) algorithm — a simplification of MAML where the inner loop is removed for all but the (task-specific) head of the underlying neural network.
This paper proposed attempts to reinterpret a standard discriminative classifier as an energy-based model. In this setting, wrote the authors, the standard class probabilities can be easily computed. They demonstrated that energy-based training of the joint distribution improves calibration, robustness, handout-of-distribution detection while also enabling the proposed model to generate samples rivalling the quality of recent GAN approaches. This work improves upon the recently proposed techniques for scaling up the training of energy-based models. It has also been the first to achieve performance rivalling the state-of-the-art in both generative and discriminative learning within one hybrid model.
This paper investigated the commonly assumed notion that neural networks cannot be recovered from its outputs, as they depend on its parameters in a highly nonlinear way. The authors claimed that by observing only its output, one could identify the architecture, weights, and biases of an unknown deep ReLU network. By dissecting the set of region boundaries into components associated with particular neurons, the researchers showed that it is possible to recover the weights of neurons and their arrangement within the network.
Authors: Suyash Mahajan, Salma Shaikh, Jash Vora, Gunjan Kandhari, Rutuja Pawar,
Abstract: The paper embark on predicting the outcomes of Indian Premier League (IPL) cricket match using a supervised learning approach from a team composition perspective. The study suggests that the relative team strength between the competing teams forms a distinctive feature for predicting the winner. Modeling the team strength boils down to modeling individual player‘s batting and bowling performances, forming the basis of our approach.
Research Methodology: In this paper, two methodologies have been used. MySQL database is used for storing data whereas Java for the GUI. The algorithm used is Clustering Algorithm for prediction. The steps followed are as
For the research paper, read here
Author : Roshan Tavhare
Institute : University of Mumbai
Abstract : The main idea behind this project is to develop a nonintrusive system which can detect fatigue of any human and can issue a timely warning. Drivers who do not take regular breaks when driving long distances run a high risk of becoming drowsy a state which they often fail to recognize early enough.
Research Methodology : A training set of labeled facial landmarks on an image. These images are manually labeled, specifying specific (x, y) -coordinates of regions surrounding each facial structure.
Authors: Chinmaya Mishra Praveen Kumar and Reddy Kumar Moda, Syed Saqib Bukhari and Andreas Dengel
Institute: German Research Center for Artificial Intelligence (DFKI), Kaiserslautern, Germany
Abstract: In this paper, the researchers explore various text data augmentation techniques in text space and word embedding space. They studied the effect of various augmented datasets on the efficiency of different deep learning models for relation classification in text.
Research Methodology: The researchers implemented five text data augmentation techniques (Similar word, synonyms, interpolation, extrapolation and random noise method) and explored the ways in which we could preserve the grammatical and the contextual structures of the sentences while generating new sentences automatically using data augmentation techniques.
Author : Prateek Kaushik
Institute : G D Goenka University, Gurugram
Abstract : This research paper described a personalised smart health monitoring device using wireless sensors and the latest technology.
Research Methodology: Machine learning and Deep Learning techniques are discussed which works as a catalyst to improve the performance of any health monitor system such supervised machine learning algorithms, unsupervised machine learning algorithms, auto-encoder, convolutional neural network and restricted boltzmann machine .
Author : A.Pavithra, C.Anandhakumar and V.Nithin Meenashisundharam
Institute : Sree Saraswathi Thyagaraja College,
Abstract : This article we discuss about Big data on IoT and how it is interrelated to each other along with the necessity of implementing Big data with IoT and its benefits, job market
Research Methodology : Machine learning, Deep Learning, and Artificial Intelligence are key technologies that are used to provide value-added applications along with IoT and big data in addition to being used in a stand-alone mod.
Author: Stephen Merity
In this work of art, the Harvard grad author, Stephen “Smerity” Merity, investigated the current state of NLP, the models being used and other alternate approaches. In this process, he tears down the conventional methods from top to bottom, including etymology.
The author also voices the need for a Moore’s Law for machine learning that encourages a minicomputer future while also announcing his plans on rebuilding the codebase from the ground up both as an educational tool for others and as a strong platform for future work in academia and industry.
Authors: Mingxing Tan and Quoc V. Le
In this work, the authors propose a compound scaling method that tells when to increase or decrease depth, height and resolution of a certain network.
Convolutional Neural Networks(CNNs) are at the heart of many machine vision applications.
EfficientNets are believed to superpass state-of-the-art accuracy with up to 10x better efficiency (smaller and faster).
Authors: Mikhail Belkin, Daniel Hsu, Siyuan Ma, Soumik Mandal
In this paper , an attempt has been made to reconcile classical understanding and modern practice within a unified performance curve.
The “double descent” curve overtakes the classic U-shaped bias-variance trade-off curve by showing how increasing model capacity beyond the point of interpolation results in improved performance.
Authors: Jonathan Frankle, Michael Carbin
Neural network pruning techniques can reduce the parameter counts of trained networks by over 90%, decreasing storage requirements and improving computational performance of inference without compromising accuracy.
The authors find that a standard pruning technique naturally uncovers subnetworks whose initializations made them capable of training effectively. Based on these results, they introduce the “lottery ticket hypothesis:”
Authors: Francois Chollet
This work summarizes and critically assesses the definitions of intelligence and evaluation approaches, while making apparent the historical conceptions of intelligence that have implicitly guided them.
The author, also the creator of keras, introduces a formal definition of intelligence based on Algorithmic Information Theory and using this definition, he also proposes a set of guidelines for what a general AI benchmark should look like.
Authors: Sawan Kumar, Sharmistha Jat, Karan Saxena and Partha Talukdar
Word Sense Disambiguation (WSD) is a longstanding but open problem in Natural Language Processing (NLP). Current supervised WSD methods treat senses as discrete labels and also resort to predicting the Most-Frequent-Sense (MFS) for words unseen during training.
The researchers from IISc Bangalore in collaboration with Carnegie Mellon University propose Extended WSD Incorporating Sense Embeddings (EWISE), a supervised model to perform WSD by predicting over a continuous sense embedding space as opposed to a discrete label space.
Authors: Shaojie Bai, J. Zico Kolter and Vladlen Koltun
Motivated by the observation that the hidden layers of many existing deep sequence models converge towards some fixed point, the researchers at Carnegie Mellon University present a new approach to modeling sequential data through deep equilibrium model (DEQ) models.
Using this approach, training and prediction in these networks require only constant memory, regardless of the effective “depth” of the network.
Authors: Robert G, Patricia R, Claudio M, Matthias Bethge, Felix A. W and Wieland B
Convolutional Neural Networks (CNNs) are commonly thought to recognise objects by learning increasingly complex representations of object shapes. The authors in this paper , evaluate CNNs and human observers on images with a texture-shape cue conflict. They show that ImageNet-trained CNNs are strongly biased towards recognising textures rather than shapes, which is in stark contrast to human behavioural evidence.
Authors: Marc G. B , Will D , Robert D , Adrien A T , Pablo S C , Nicolas Le R , Dale S, Tor L, Clare L
The authors propose a new perspective on representation learning in reinforcement learning
based on geometric properties of the space of value functions. This work shows that adversarial value functions exhibit interesting structure, and are good auxiliary tasks when learning a representation of an environment. The authors believe this work to open up the possibility of automatically generating auxiliary tasks in deep reinforcement learning.
Authors: Adam Gaier & David Ha
In this work , the authors explore whether neural network architectures alone, without learning any weight parameters, can encode solutions for a given task. In this paper, they propose a search method for neural network architectures that can already perform a task without any explicit weight training.
Authors: Prajit Ramachandran, Niki P, Ashish Vaswani,Irwan Bello Anselm Levskaya, Jonathon S
In this work, the Google researchers verified that content-based interactions can serve the vision models . The proposed stand-alone local self-attention layer achieves competitive predictive performance on ImageNet classification and COCO object detection tasks while requiring fewer parameters and floating-point operations than the corresponding convolution baselines. Results show that attention is especially effective in the later parts of the network.
Authors: Mario Lucic, Michael Tschannen, Marvin Ritter, Xiaohua Z, Olivier B and Sylvain Gelly
Modern-day models can produce high quality, close to reality when fed with a vast quantity of labelled data. To solve this large data dependency, researchers from Google released this work , to demonstrate how one can benefit from recent work on self- and semi-supervised learning to outperform the state of the art on both unsupervised ImageNet synthesis, as well as in the conditional setting.
The proposed approach is able to match the sample quality of the current state-of-the-art conditional model BigGAN on ImageNet using only 10% of the labels and outperform it using 20% of the labels.
Authors: Zhenzhong Lan, Mingda Chen, Sebastian Goodman, Kevin G, Piyush Sharma and Radu S
The authors present two parameter-reduction techniques to lower memory consumption and increase the training speed of BERT and to address the challenges posed by increasing model size and GPU/TPU memory limitations, longer training times, and unexpected model degradation
As a result, this proposed model establishes new state-of-the-art results on the GLUE, RACE, and SQuAD benchmarks while having fewer parameters compared to BERT-large.
Author: Taesung Park, Ming-Yu Liu, Ting-Chun Wang and Jun-Yan Zhu
Nvidia in collaboration with UC Berkeley and MIT proposed a model which has a spatially-adaptive normalization layer for synthesizing photorealistic images given an input semantic layout.
This model retained visual fidelity and alignment with challenging input layouts while allowing the user to control both semantic and style.
📣 Want to advertise in AIM? Book here
Biggest ai stories, delivered to your inbox every week..
Discover how Cypher 2024 expands to the USA, bridging AI innovation gaps and tackling the challenges of enterprise AI adoption
© Analytics India Magazine Pvt Ltd & AIM Media House LLC 2024
The Belamy, our weekly Newsletter is a rage. Just enter your email below.
Discover the world's research
IMAGES
VIDEO
COMMENTS
This paper highlights and reviews code-switching in studies of which the scope is its usage in popular culture, with the aims to explore the most frequently discussed issues in the realm of code ...
This article reviews recent empirical studies and provides corpus evidence that codeswitching is a flexible and creative strategy for bilingual language use. It proposes that codeswitching serves as a toolkit for optimizing performance in cooperative communication, and is subject to linguistic and cognitive constraints.
This paper reviews the state of the art in code-switching research and explores three controversial issues: switching vs. borrowing, grammaticality, and variability and uniformity. It provides examples of code-switching in Welsh and Spanish, and argues for empirical methods and community norms to study bilingual speech.
Furthermore, research assessing code-switching attitudes has employed direct data collection methods, such as surveys and questionnaires (e.g. Dewaele & Wei, 2014b ... here. Following Clément and Noels (1992; see also Clément et al., 1993; Damji et al., 1996), our focus in the present paper is the actual cultural identity of the individual, ...
Code-switching is a common phenomenon that bilinguals engage in, using words from two languages within a single discourse. This study examines how code-switching behavior affects a child's linguistic competency in English and Mandarin, and finds that it is a sign of linguistic competence rather than confusion or incompetence.
The trend topics by author keywords of research on code switching have been revealed . The analysis reveals interesting patterns in the trajectory of code-switching research over time. One notable finding is the continuous attention toward code switching in aphasia (Jomaa et al., 2022; Lee & Faroqi-Shah, 2021). This persistent interest suggests ...
Code-switching is a linguistic phenomenon often associated with the architecture of discourse varieties. A good number of studies in the bilingual and multilingual contexts have zoomed in on the use of code-switching primarily analysing its roles and functions in varied discourse settings. This paper highlights and reviews code-switching in studies of which the scope is its usage in popular ...
A critical overview of the linguistic and sociolinguistic research on code-switching (CS), a phenomenon of mixing two or more languages in speech or writing. The paper reviews empirical studies, theoretical perspectives, and methodological issues in CS, with a focus on bilingual classroom interaction.
Switching between languages, or codeswitching, is a cognitive ability that multilinguals can perform with ease. This study investigates whether codeswitching during sentence reading affects early access to meaning, as indexed by the robust brain response called the N400. We hypothesize that the brain prioritizes the meaning of the word during comprehension with codeswitching costs emerging at ...
%0 Conference Proceedings %T The Decades Progress on Code-Switching Research in NLP: A Systematic Survey on Trends and Challenges %A Winata, Genta %A Aji, Alham Fikri %A Yong, Zheng Xin %A Solorio, Thamar %Y Rogers, Anna %Y Boyd-Graber, Jordan %Y Okazaki, Naoaki %S Findings of the Association for Computational Linguistics: ACL 2023 %D 2023 %8 July %I Association for Computational ...
The next distinction to be made is that between code-switching and borrowing—the subject of a longstanding debate in code-switching research. ... GLUECoS: An evaluation benchmark for code-switched NLP. Paper presented at 58th Annual Meeting of the Association for Computational Linguistics, Online. July 5-10; Stroudsburg: Association for ...
Code-switching, the alternation of languages within a conversation or utterance, is a common communicative phenomenon that occurs in multilingual communities across the world. This survey reviews computational approaches for code-switched Speech and Natural Language Processing. We motivate why processing code-switched text and speech is essential for building intelligent agents and systems ...
Code-Switching, a common phenomenon in written text and conversation, has been studied over decades by the natural language processing (NLP) research community. Initially, code-switching is intensively explored by leveraging linguistic theories and, currently, more machine-learning oriented approaches to develop models. We introduce a comprehensive systematic survey on code-switching research ...
Code switching: Linguistic. Code-switching (CS) refers to the mixing, by bilinguals. (or multilinguals), of two or more languages in. discourse, often with no change of interlocutor or. topic ...
This editorial introduces a collection of articles on code-switching (CS) and language switching (LS), two phenomena of bilingual language use. It discusses the linguistic, psycholinguistic, and neurophysiological approaches to CS and LS, and the relationship with cognitive control.
This paper provides a critical review of the state of the art in code-switching research being conducted in linguistics. Three issues of theoretical and practical importance are explored: (a) code ...
Code-switching is an active research area that has significant implications for academic libraries. Using data from focus groups and a survey tool, this paper examines language preferences of foreign-born students for particular information tasks. ... The main finding of this paper is that students' culture and language represent an active ...
Gumperz (1982 b) defined code-switching. as "the juxtaposition within the same speech exchange of passages of speech belonging to two. different grammatical systems or subsystems" (p. 59). The emphasis is on the two grammatical. systems of one language, although most people refer to code-switching as the mixed use of.
This paper reviews the literature on code switching in sociology, linguistic anthropology, and sociolinguistics, and suggests a definition of the term for sociocultural analysis. It argues that code switching is a practice of selecting or altering linguistic elements to contextualize talk in interaction, and that it has social and cultural functions and meanings.
Classroom code-switching refers to the alternating use of more than one linguistic code in the classroom by any of the classroom participants. This chapter provides a review of the historical ...
use of code switching; and Dayag's analysis on print advertisements (as cited in Martin, 2014). Code switching is definitely being utilized in various domains of Philippine society. In Philippine classroom discourse, code switching which is also known as pedagogic code switching or classroom code switching, has become a tricky issue
This paper revisits fundamental concepts of Bilingualism in the Philippines, code switching and the recently implemented Mother Tongue Based-Multilingual Education. After which, studies related to ...
Research shows that code-switching often occurs in spaces where negative stereotypes of marginalized communities run high, and it is often the folks who subscribe to these stereotypes who lead and ...
Research Methodology: Machine learning, Deep Learning, and Artificial Intelligence are key technologies that are used to provide value-added applications along with IoT and big data in addition to being used in a stand-alone mod. For the research paper, read here. Single Headed Attention RNN: Stop Thinking With Your Head . Author: Stephen Merity
Abstract and Figures. This study determined the influence of code switching to the academic performance of students in English. A total of 40 incoming Grade 10 students participated in this study ...