Sociolinguistic Variation in American English Adverbial –ly
NWAV 40
Georgetown University
Saturday, October 29, 2011
Allison Shapp and Renee Blake
Department of Linguistics
New York University
Results
Introduction
• What we know about adverbs and variation from earlier historical stages of English
(Tagliamonte and Ito 2002):
a)  Adverbials of manner, the largest group of adverbs, are generally marked with
the suffix –ly. Although, the zero form is also possible.
“I mean you look at life different.”
b) Intensifier use of adverbs that modify adjectives or adverbs are generally zero
marked.
Semantics matters (Tagliamonte
‘If you do absolute perfect, you
and Ito 2002): more -ly in
might get a C.”
abstract or mental sense (“They
take it much more seriously”)
c) Sentential adverbs are generally
than in concrete or physical sense
marked with –ly.
(“I’ve walked upstairs dead
“Honestly, they did.”
quick”).
• We focus on manner adverbs, where the
alternation between -ly and zero generally exists (Opdahl 2000), and for which there
is no difference in meaning (Tagliamonte and Ito 2002)
• The widespread usage of the zero form of the adverbial –ly suffix is associated
specifically with American English (not British English) in terms of both written and
spoken data (Görlach 1991, Liddle 1999, Opdahl 2000); the zero form is the earlier
form
The zero form adverb was present in Early Modern English, replaced by -ly in the 18th
century. In North America, the zero form was sanctioned as “sound”.
a)  The zero adverb is widespread in American English (Wolfram and SchillingEstes 1998)
b)  The zero adverb is geographically and socially diffused (Mencken 1961)
Methodology
Education
The sharpest statistical results are found for education, such that
those with no college education strongly disfavored adverbial
-ly (.28), those who had attended college slightly favored adverbial
-ly (.52), and those with a post-college education strongly favored
adverbial –ly (.70)
Table 1. Goldvarb and Frequency results for adverbial –ly presence
Root Word
• As shown in Table 1, root words «loud», «slow» and «bad» all disfavor
adverbial –ly (.33, .45, .49 respectively), while «horrible» and «poor»
strongly favor it (.75, .96, respectively)
• «horrible» and «poor» are unintended elicitations of «bad», suggesting
ambiguity in questions or lexical choices, which can cause more adverbial -ly
use
• There is a correlation between lexical frequency and -ly usage such that
higher frequency is associated with lower -ly use
Linguistic Background
• The data set is divided into monolingual English speakers (comprising
about 75% of the data), bilinguals with English as a primary language, and
a group of individuals for who English is their second languageMonolingual
• English speakers favored (.52) adverbial -ly the most. Bilingual and ESL
individuals disfavored (.47, .32 respectively) adverbial -ly, with the latter
group disfavoring it the most.
Race/Ethnicity
• Respondents were given this category as an open question. Responses were
categorized into larger categories into which predominant classifications fit
• The elicitations came mostly from Whites (47.8%) and Blacks (33.3%)
• In earlier Goldvarb runs, Black people showed a strong disfavoring of
adverbial -ly, while all other race/ethnic groups showed a favoring effect
Nonetheless, the zero adverb is suggestive of colloquial, informal, or dialectal speech
• To date, there are no studies of American English that correlate the variation of the
dual forms of adverbs of manner with social factors
• In June Jordan’s (1985) “On Call: Political Essays,” she recounts her undergraduate
class creating ‘guidelines’ for AAE based on Alice Walker’s The Color Purple. Rule
#17: Never use the suffix -ly form of an adverb in Black English
• This raises the question, are African American or AAE speakers using adverbial -ly
less than other speakers of American English?
Methods
The Study
THE SURVEY (2008)
The survey (see handout) consists of 20 sentences with the last word of each sentence
missing. The survey is presented to each interviewee orally, and the participants are
asked to fill in each missing word with the first word that comes to mind. There are 9
content sentences, designed to elicit tokens of the three target adverbs that were
easiest to construct elicitation sentences for: slow(ly), bad(ly), and loud(ly), in three
different environments (after so, very and directly after the verb). Two additional
elicited adverbs are included in data analysis: horrib(ly) and poor(ly).
Right quick and right fast are also apparent in American English. However, a 2002 pilot study could not elicit this utterance.
Figure 1 below gives examples of sentences
that were part of the survey:
Table 2. Frequencies for Root Words in dataset
Table 3. Frequencies of actual elicitations from
potential elicitations by Linguistic Environment
Linguistic Environment
• The N's for linguistic environment indicate that the target word was given
less often directly after the verb
• The blanks are more open ended directly following the verb, and
could be filled with other words or phrases. Eg., the sentence “Little
Karen never practices the piano after her lessons. No wonder she
plays ____” is intended to elicit “badly,” but we received answers
that were not adverbs, such as “like a beginner”
• The «so» environment slightly disfavors adverbial –ly (.44) , while
the «very» environment and root word directly following the verb
slightly favors adverbial –ly (.53, .54 respectively)
Age
In the survey, Age is divided into six categories to allow for a wide distribution
Group A: 18-25 years old
Group B: 26-35 years old
Group C: 36-45 years old
Group D: 46-53 years old
Group E: 54-65 years old
Group F: Over 65 years old
In earlier Goldvarb runs, group B, including ages 26-35, showed the greatest
favoring of adverbial -ly
Loud(ly)
I have to cover my ears every time John whistles for a cab! I don’t know how he can whistle so _____.
Bad(ly)
My neighbor’s kid never gets punished by his parents, not even when he behaves very _______.
Slow(ly)
Tom broke his pinky last month and is still recovering. That’s why he’s typing ______.
Filler
I want to go to the store to buy some milk. I have to remember to bring some _________.
Figure 1. Sample of survey questions
The social variables of age, education, race/ethnicity, sex, and linguistic
background are recorded. In addition, occupation and birthplace are recorded,
and will be categorized and coded during another research phase.
Figure 3. Goldvarb Results for adverbial –ly Presence by Age
Age: 18-25 (A), 26-35 (B), 36-45 (C), 46-53 (D), 54-65 (E), 65+ (F)
Education: No College (N), College (C), Post (P)
Linguistic Background: Monoling (M), Biling (B), ESL (2)
Race/Ethnicity: White (W), Black (B), Asian (A), Latino (L), Mixed (M), Other (O)
Sex: Male (M), Female (F)
• In collapsed age groups, group B favors adverbial –ly (.59), with the
youngest aged group A (18-25 years old) slightly favoring it (.52) , and
all other groups C-E (36 years and older) combined very slightly
disfavoring adverbial –ly (.49)
Figure 4. Goldvarb Results for Adverbial –ly by Race/Ethnicity
• The results hold when individuals identified as Black are compared to
all other race/ethnic groups, with the the Black group disfavoring
adverbial –ly (.39), and all others favoring it (.56) .
Sex
There is no difference in adverbial usage by the two sexes.
Race Matters: Recoding and Reanalysis
Black Group
• With the Black group isolated, only 3 of the 6 factor groups are
significant
• Root words generally pattern the same, with «bad», now slightly
favoring adverbial -ly (.53)
• The effect of Education is similar to the larger dataset
• For Linguistic Background, it is the monolingual English speakers that
slightly disfavor (.49) and the the ESL speakers that disfavor adverbial
-ly (.34), whereas the bilingual speakers favor it (.65). We should use
caution here, since 92% of the tokens are from monolinguals
White Group
• In the isolated White group, Sex becomes significant compared to the
larger dataset
• In the Sex category, White women slightly favor adverbial -ly usage
(.52) and White men disfavor it (.48)
• The pattern among White speakers with regards to Age is more similar to
the overall dataset than the Black group, except for Group E (54-65 years
old), which has switched from disfavoring to slightly favoring adverbial
–ly use (.52)
Figure 2. Social variables coded for data
DATA COLLECTION
The data used in this analysis were collected in 2008 by an undergraduate
linguistics class consisting of approximately 50 students. Each student was
trained to survey 18 people, and was assigned a quota for age, sex, and race
in order to insure a corpus balanced for those demographics. Students were
instructed to have the majority of surveys conducted with native speakers of
English. This resulted in approximately 900 speakers, and a total of 6,625
usable tokens to be analyzed.
METHODOLOGICAL ISSUES
1.  Undergraduates with limited research experience collect the data
2.  Data must be collected face-to-face to limit the elicitation time
3.  Cannot control for demographics as New York is a diverse metropolis
ANALYSIS
1.  Multivariate analysis using Goldvarb X. The dependant variable is
whether or not the -ly morpheme is present in the token, with the
presence of -ly taken as the application value
2.  Two linguistic variables (word root and lexical environment) coded
3.  Five social variables (age, education, linguistic background, race, and
sex) coded
Discussion
• According to Tagliamonte and Ito (2002:259), zero adverbs are more frequent in North
America than in Britain due to the fact that it started as robust, was sanctioned by Noah
Webster, and was allowed to continue to “develop unchecked by social pressure in a
receptive environment”
• However, there have been no variationist studies on North American English to
address whether zero adverbs are undergoing linguistic change towards an increase in
frequency
• Our findings confirm that based on the adverbs examined, zero adverbs ARE more
frequent in North American English. Tagliamonte and Ito (2002) found the -ly form to
represent 85% of their data. For American English, we find the -ly form representing
63.6% of our data. This high frequency of -ly lends perspective to claims of high or
increasing zero adverb in North American English (cf., Wolfram and Schilling-Estes
1998)
• As noted by Opdahl (2000a, 2000b), our results confirm that lexical items are factors
that affect the rate of zero adverb use. However, the research is limited by the fact that
the questionnaire is designed to elicit certain forms. In the future more adverbials will
be added for further clarity; nonetheless, lexical distribution of adverbs and lexical
environment play a role in the the rate of zero adverb use
• In the American English dataset, the zero adverb is a socio-symbolic
resource such that:
1)  It is least used by educated people
2)  It is least used by those entering and anchoring themselves within
the workplace, and hence within the linguistic marketplace
3)  It is attributed to race/ethnicity to the extent that it is used most
by African Americans
• The socio-symbolic nature of the variant was also found in other studies:
1)  Attributed to style (Van Draat 1910; Patridge 1969)
2)  Attributed to class (Macaulay 1995)
3)  Attributed to education and sex (Tagliamonte and Ito 2002)
• Future directions include adding more lexical items, looking at spoken
data, improving the questionnaire and analyzing the additional data
collected.
*Special thanks to Gregory Guy and Amy Wong
*For References: See Handout