Quotative Say and the Dynamics of Social Positioning: A Sociolinguistic Inquiry

Cite this article as: Ibrahim, J. (2025). Quotative say and the dynamics of social positioning: A sociolinguistic inquiry. Sokoto Journal of Linguistics and Communication Studies (SOJOLICS), 1(1), 41–57. www.doi.org/10.36349/sojolics.2025.v01i01.007

QUOTATIVE SAY AND THE DYNAMICS OF SOCIAL POSITIONING: A SOCIOLINGUISTIC INQUIRY

By

Jalaludeen Ibrahim, Ph.D

jalaludeen.ibrahim@uib.no; jalalmaradun@gmail.com

Department of Foreign Languages, University of Bergen, Norway

Abstract

This study explores how speakers of Nigerian English (NigE) re-create their own speech and that of others in narrative discourse using the quotative say. The research is based on naturallyoccurring data collected from 180 participants during sociolinguistic interviews conducted in various locations across Nigeria. It investigates how the speakers’ choice of say is constrained by both linguistic factors (such as the content of the quote, the grammatical person of the quotative, and the tense/time reference of the quotative) and social factors (including age, gender, regional origin, and social class). The study adoptsa Variationist Sociolinguistics framework, drawing on Labov (1963, 1966) and Tagliamonte (2012), and employs a mixed-methods approach that combines quantitative and qualitative analyses. For the multivariate analysis of different quotative forms, a statistical program called Rbrul (Johnson, 2009) was utilised. The discourse analytic qualitative method was employed mainly to address the question of how quotative say functions in performed narratives. The findingsestablish that the use of the quotative say is prevalent in NigE. The speakers of this variety of English use this quotative form to frame directly quoted speech without marking dramatic effects, allowing for its application in a wide variety of contexts. The study provides valuable insights into the mechanisms of linguistic change and the function ofsayas a quotative marker in NigE.

Keywords: Narrative Discourse, Nigerian English, Quotative Say, Social Positioning, Variation

1. Introduction

In recent years, there has been an increasing number of research studies on quotatives in different fields and themes. Buchstaller (2014: 49) rightly observes that quotations are commonly multiple-perspective constructions, which contain the perspective of the current speaker as well as that of the reported speaker. She suggests that quotations only need to be “good enough”. This suggestion entails that “good enough” is simply negotiated by the interlocutors themselves as defined by the context. My conceptualisation of quotation in this study is in line with Buchstaller’s (2014: 54) definition of quotation as “a performance whereby speakers re-enact previous behaviours (speech/thought/sound/voice effect and gesture) while assuming the dramatic role of the original source of this reported behaviour”. My interest in this approach lies in her showing how the speaker simultaneously creates interpersonal involvement in producing direct speech using quotative markers.

The quotative system in English consists of a wide array of verbs that function as dialogue introducers. Apart from the more traditional quotatives, such as say and tell, the literature on quotatives has a heavy focus on be like. Other prominent quotative forms include go, which is less transparent, and think, which is a marker of inner reflection. Interestingly, the semantics of each quotative form constrains its use, with each form having a slightly different function. For instance, Romaine and Lange (1991: 235-240) report that while say introduces speech without the contribution of any pragmatic effect, think is mostly used to report internal dialogue, whereas go functions as an option for direct speech and non-lexicalised sounds or gestures. On the other hand, the rise of be like as a viable quotative form upsets the balance amongst the more traditional quotative verbs as be like provides a new choice to speakers, “the partitioning of forms within the system must necessarily re-organise” (Tagliamonte & Hudson, 1999: 57). Moreover, local norms of storytelling and styles of narration differ in many respects, lending themselves to a myriad of pragmatic ramifications that are conveyed using quotatives.

Another quotative form that attracts the attention of sociolinguists is zero quotative, which occurs “where direct speech is with neither a reporting verb nor an attributed speaker” Mathis and Yule (1994: 63). This quotative form is less transparent and is often signaled through performative cues or when there is a clear change in speaker. Thus, the referent is recoverable from the linguistic context, and this is usually signaled with an obvious turn-taking structure or voice quality indicators. In NigE, it is common for speakers to use the quotative sayto re-create their own speech and the speech of other speakers in narrative discourse and free conversations. By way of illustration, consider the use of the quotative sayin examples (1) and (2) below.

(1) She said, “Oh, well I cannot give you more than that”.

(2) The Dean says, “The money is there.”

I said, “I was reluctant.”

I said, “It is not my Ph.D. thesis.”

In (1), using said, the speaker explicitly quotes an utterance made by an assessor. In (2), the speaker narrates how his Dean was encouraging him to publish academic papers in journals without any financial cost implication on his part, since there is money set aside for the publication. The tense choice in narration in (2) highlights how speakers narrate encounters with authority figures to help in redefining the situation. Here, the Dean is the authority figure whose speech is being introduced with the present says, and the non-authority is the teller whose speech is being introduced with the past said. A possible explanation for the shift in tense here is simply to separate speech events, or speakers from one another, especially as the present tense is mostly considered as an evaluative device marking the authority’s past event in the present form.However, the primary focus of this study is to develop a comprehensive account of the acquisition and spread of the quotative say in NigE, which will broaden our understanding of the mechanisms of linguistic change. Thus, the aims of this study are achieved by seeking answers to the following questions: (1) What is the frequency distribution of the different quotative forms in NigE? (2) Who are the principal users of quotative say in NigE? (3) Do factors such as age, gender, regional origin, and social class affect the use of qoutative say in NigE? (4) What are the linguistic constraints that condition the occurrence of quotative say, and how do they interact with extra-linguistic constraints? (5) How does quotative say function in everyday narratives across different social contexts?

2. Literature Review

The emergence of quotatives can be traced back to the early 1980s when Butters first noted the narrative use ofgo (Butters, 1980) and be like (Butters, 1982) as dialogue introducers in American English. The quotative go was initially documented in the speech of male American speakers born after 1955, and was noted as a hallmark of informal conversation among speakers under the age of thirty-five (Butters, 1980: 305). Butters observed that the quotative gotypically appeared in the present tense, reflecting its role in conveying immediacy and vividness in narration. In a subsequent study, Butters (1982: 149) reported the use ofbe likeas a quotative form particularly suited for expressing internal thoughts or unuttered reactions. Since these foundational observations, research into quotatives has expanded significantly, with numerous studies exploring how quotative forms likego and be likehave diffused across different English-speaking communities and evolved in usage.

Still in the early 1980s, Cheshire (1982) reported the existence of quotative go in a study of non-standard verbs in British English. She sampled the speech of adolescents from a working-class background and found that the quotative go frequently occurred with the non-standard inflection, as in I goes “oh clear off”. Verbs displaying this kind of non-concord are used with a different meaning: “it can be seen that the use of a vernacular verb acts as a lexical constraint on the form of the verb, strongly favouring the non-standard form” (Cheshire, 1982: 43). Here, I goes rather than I go appears as a non-standard quotative marker. Tannen (1986), in her study of oral American narratives, examined the strategiesspeakers used to introduce dialogue in conversational storytelling. She reported that say is the traditional and thus most used quotative in her 18 stories, followed by go and be like (Tannen, 1986: 315-321). She suggested a continuum of quotative usage, ranging from zero quotativescommon in informal speech to graphic verbs typically found in literary narratives. Within this framework, she reported that be like is next to zero on the continuum as its effects depend on the way the dialogue is voiced by the speaker. In contrast, go is identified as an informal feature similar to be like in register, yet similar to say in semantic function (Tannen, 1986: 324).

Besides the literature on quotatives in American English, there are studies on quotatives in other varieties of English. For example, Tagliamonte and Hudson (1998, 1999) investigated the quotative system of British and Canadian youth by analysing 665 quotative tokens from British English narratives and 612 quotative tokens from Canadian English narratives. Since the two papers are very similar, even in their title and analysis, I focused more on reviewing Tagliamonte and Hudson’s more recent paper from 1999. Their multivariate analysis demonstrated that the quotatives be like, go, say, and think are used according to different patterns in the two corpora. The findings revealed that in both British and Canadian English say is the most frequently used quotative, followed by go, be like, and then think in British English and go and zero in Canadian English (Tagliamonte & Hudson, 1999: 157-158). This attests that be like and think are less frequent in Canadian than in British narratives.

Tagliamonte and Hudson (1999: 161) reported that say is favoured by males and be like by females in British English, whereas males favour go and women favour say and think in Canadian English. The quotative go in Canadian English is similar to what Butters (1980) reported: go is associated with the speech of male American speakers. With respect to grammatical person, Tagliamonte and Hudson (1999: 162) reported that quotatives goand say have different patterns in the two varieties as go mainly occurs with third-person subjects and say with first-person subjects in Canadian English, whereas in British English, the results show no preferences for go, but say is favoured in the third-person contexts. Concerning the distribution of the quotative be like, speakers of both British and Canadian English favour be like in the context of first-person subjects. Thus, their results differ from the finding in Ferrara and Bell (1995) that be like is increasingly used with third-person subjects. As for gender differentiation, they claim that gender correlates with the rate of usage in the sense that the further the quotative be like spreads, the more likely it is to differentiate female and male speech (Tagliamonte & Hudson, 1999: 167)

Another interesting finding is that British speakers differ from Canadian speakers in their use of the quotative go as the data demonstrated a correlation with direct speech in British English and a correlation with thoughts in Canadian English. However, the speakers of both varieties are similar in their use of be like to introduce sounds and thoughts and say to introduce direct speech (Tagliamonte & Hudson, 1999: 163). To track the developmental stages of the quotative be like in both British and Canadian English, Tagliamonte and Hudson (1999) compared their results on be like with the findings in Ferarra and Bell (1995). The comparison revealed that the developmental pathways in the two varieties are very similar to American English in many respects. Tagliamonte and Hudson (1999: 168) submitted that, like in American English, the spread of be like into British and Canadian English shows its systematic diffusion to other varieties of English across the globe.

The study by Gardner et al. (2020) is one of the most recent studies in the series of research on English quotatives. They study quotative be like and its co-variants in apparent-time and at two separate points in real-time in two different communities, Toronto, Canada, and York, United Kingdom. Their data was drawn from the Toronto and the York English Archives of informal spoken English and grouped the corpora into two eras of data collection, viz., Era 1 (1996-2004), and Era 2 (2006-2013). The analysis was based on 15,871 quotative tokens (8,797 in Toronto data and 7,074 in York data) extracted from 525 individual speakers (Gardner et al., 2020: 10). Their findings revealed that be like dominates the quotative system in the two communities. Also, in both Toronto and York, go, think,and zero quotatives appear to be relatively stable across age groups, while be like and say are not. Using a subset of data from the study (e.g. Tagliamonte & Hudson, 1999), previous findings suggest that go grows rapidly in use alongside be like. The combination of apparent-time and real-time data indicates that this rapid increase in the use of go was short-lived (Gardner et al., 2020: 11). However, in the present study, I will look at the quotativesayand how it functions in the NigE variety.

3. Fieldwork and Methodology

The fieldwork was conducted in Nigeria, involving face-to-face sociolinguistic interviews with 180 participants. A mixed sampling approach was employed, combining stratified random sampling (judgement or quota sampling) with snowball sampling based on social networks. To facilitate participant recruitment, the researcher reached out to family members, friends, and former colleagues who assisted in identifying and mobilising potential participants. The interviews took place across a wide range of settings, including recreational centres, bookstores, cafes, restaurants, university campuses, private residences, mosques, churches, grocery stores, and other public and private venues.

The 180 interviews conducted for this study were structured into thematic modules, following the concept of “conversational network” as outlined by Labov (1984). For adolescent and young adult participants, modules included topics such as dreams, family, fights, games, peers, and school. In contrast, middle-aged and older participants were engaged through modules focused on family, marriage, politics, school, and work. The sequence of modules was flexible, allowing transitions to be guided by participants’ interest and willingness to share personal experiences. Elicitation of quotative forms was thus shaped by the natural flow of conversation rather than a predetermined structure. Each interview was recorded for a standard duration of 45 minutes. The data were analysed within the framework of Variationist Sociolinguistics, also known as the Language Variation and Change (LVC) paradigm (Labov, 1963, 1966; Tagliamonte, 2012). This core sociolinguistic theory explains how linguistic variation correlates with social factors such as age, gender, regional origin, and social class, and how these variables reflect ongoing language change across speech communities globally.

The study examines a range of extra-linguistic and linguistic variables. The extra-linguistic factors include key social dimensions such as age, categorised into adolescents (15-20), young adults (21-30), middle-aged adults (31-50), and older adults (51 and above); regional origin (northern and southern Nigeria); gender (male and female); and social class (lower class, middle class, and upper class). The linguistic factors considered are the tense/time reference of the quotative (past tense, present tense, and future time reference), the grammatical person of the quotative (first-person singular, first-person plural, second person, third-person singular, third-person plural, and neuter), and the content of the quote (direct speech, gesture, and thought).Table 1 provides a sampling grid that illustrates the distribution of these extra-linguistic factors across the participant pool.

Table 1: Sampling Grid by age, social class, gender, and regional origin

Age and social class	North		South		Total
	Male	Female	Male	Female
15-20 (lower class)	7	3	7	4	21
15-20 (middle class)	3	8	3	6	20
15-20 (upper class)	1	-	1	3	05
21-30 (lower class)	6	3	3	6	18
21-30 (middle class)	5	7	8	2	22
21-30 (upper class)	-	2	1	3	06
31-50 (lower class)	1	6	3	2	12
31-50 (middle class)	6	1	2	6	15
31-50 (upper class)	4	4	5	3	16
51 and above (lower class)	-	2	2	2	06
51 and above (middle class)	2	1	2	1	06
51 and above (upper class)	10	8	8	7	33
Total	45	45	45	45	180

The study primarily employed a quantitative analytical approach, as the dataset was transformed into numerical values, charts, and graphs to facilitate the examination of variable usage frequency. The statistical analysis was carried out using Rbrul (Johnson, 2009), a specialised software designed for sociolinguistic research, particularly suited for multivariate analysis. This version of Rbrul was obtained from http://cran.r-project.org and was compatible with both Macintosh and Windows operating systems. The analysis drew on a total of 4,053 tokens representing various quotative forms extracted from the dataset.

4. Results

The findings of this study are presented in two main sections. The first section provides a distributional overview of all quotative forms identified in the dataset. The second section presents the results of multivariate analysis focused specifically on the quotative say, with the findings interpreted in relation to comparable research conducted on other English varieties.

4.1 Distributional Analysis

Table 2, visually presented as a bar graph in Figure 1, reports the overall distribution of the different quotative forms across a total of 4,053 tokens. The most frequently occurring quotative form is the traditional say, with 1,620 occurrences (39.9 percent), closely followed by zero, with 1,020 occurrences (25.2 percent). The third most frequently occurring quotative form is be like, with 568 occurrences (14.1 percent). Tell, with 508 occurrences (12.5 percent), is used less than half as often as the zero quotative form. Ask with 159 occurrences (3.9 percent) and think with 54 occurrences (1.3 percent) are used even less. The remaining 25 quotative forms, grouped under the category “other”, collectively contribute just 124 tokens, representing a modest 3.1% of the dataset. This group includes: advise (6), alert (1), announce (1), answer (1), apologise (1), argue (1), call (50), claim (1), complain (5), cry (3), decide (10), explain (1), feel (19), go (3), pray (3), promise (2), reply (2), respond (2), scold (1), scream (1), shout (6), state (1), that is + speaker (1), warn (1), and whisper (1). Given their limited frequency, this category will receive minimal attention in the subsequent analyses.

Table 2: Overall distribution of quotatives in Nigerian English

Quotatives	Number	Percentage (%)
Ask	159	3.9
Be like	568	14.1
Say	1620	39.9
Tell	508	12.5
Think	54	1.3
Zero	1020	25.2
Other	124	3.1
Total	4053	100

 

Figure 1: Overall distribution of quotatives in Nigerian English

4.2 Multivariate Analysis

This section presents the factors that contribute to the use of the quotative say in NigE, analysed through multivariate statistical methods using Rbrul (Johnson, 2009). While Rbrul generates extensive output, this study concentrates on the most relevant aspects, presenting the findings in a clear, step-by-step manner to aid interpretation. Rbrul provides several key statistical outputs used to evaluate model performance and variable effects. One such measure is deviance, which indicates how well the model fits the data; the lower the deviance, the better the fit. It also reports the degree of freedom (DF), representing the number of parameters included in the model, along with the grand mean. Rbrul assesses the influence of each factor group through factor weights. If the factor weights are above 0.5, they favour the application value, while factor weights below 0.5 disfavour it. Rbrul also reports logodds, which are raw regression coefficients that quantify the strength of the relationship between a factor and the dependent variable. If logodds are above 0, there is a positive correlation between the variables and that favours the application value; if they are negative, the correlation is negative and that disfavours the application value. The larger the number, the bigger the effect size. In addition, Rbrul provides the percentage distribution of each variant within individual cells, with higher factor weights generally corresponding to higher percentages of the variants.The program further reports likelihood-ratio Chi-square tests to determine whether an independent variable is significant or not.

Table 3 reports the results of a multivariate analysis that tested the social and linguistic constraints operating on the use of say in NigE. The Table shows that the effect of age is not statistically significant (p value = 0.0853) as a factor conditioning the use of say, with the old speakers (FW 0.609) favouring say, and the adolescents (FW 0.479), the middle-aged (FW 0.462), and the young adults (FW 0.448) disfavouring it. Social class is not statistically significant (p value = 0.196) as a factor conditioning the use of say since it is used in relatively equal proportions among the lower class (FW 0.551), the middle class (FW 0.519), and the upper class (FW 0.442). Although the lower class and the middle class favour the use of say, while the upper class disfavours it.The Table further illustrates that the effect of sex is not statistically significant (p value = 0.2), with males (FW 0.528) favouring the use of say and females (FW 0.472) disfavouring it. This finding differs from the findings in Buchstaller and D’Arcy (2009), who report sex as a significant factor conditioning the use of say among older and younger speakers for both British and New Zealand English. My finding also differs from those of Tagliamonte and Hudson (1999), who report sex as a statistically significant factor conditioning the use of say in favour of females in Canadian English and in favour of males in British English. Regional origin too is not statistically significant (p value = 0.188) as a factor conditioning the use of say, with the north (FW 0.528) favouring it and the south (FW 0.472) disfavouring it.

Table 3: Contribution of social and linguistic factors on the use of say inNigerian English

Total number of tokens 4053
Deviance 4615.013
Df 18
Grand mean 0.4
Factors	Logodds	Tokens (N)	Proportion of application value	Factor weight
Age	P value = 0.0853
Older adults	0.444	973	0.419	0.609
Adolescents	-0.084	1124	0.383	0.479
Middle aged	-0.151	993	0.394	0.462
Young adults	-0.209	963	0.406	0.448
Social class	P value = 0.196
Lower class	0.2010	1318	0.398	0.551
Middle class	0.0404	1451	0.414	0.519
Upper class	-0.2414	1284	0.386	0.442
Sex	P value = 0.2
Males	0.111	1991	0.428	0.528
Females	-0.111	2062	0.372	0.472
Regional origin	P value = 0.188
North	0.113	1897	0.431	0.528
South	-0.113	2156	0.372	0.472
Grammatical person	P value = 0.533
Second person	0.2474	54	0.315	0.562
Third-person plural	0.1360	441	0.392	0.534
Third-person singular	0.0670	1534	0.407	0.517
First-person singular	-0.0314	1863	0.402	0.492
First-person plural	-0.1790	96	0.375	0.455
Neuter	-0.2400	65	0.323	0.443
Tense/time reference	P value = 2.43e-102
Future	0.853	2204	0.526	0.701
Past	0.254	407	0.403	0.563
Present	-1.107	1442	0.206	0.248
Content of the quote	P value = 2.64e-13
Speech	5.245	3959	0.408	0.995
Gesture	4.451	27	0.222	0.988
Thought	-9.696	67	0.001	<.001
Speaker	Random

Turning to linguistic constraints, the effect of grammatical person is not statistically significant (p value = 0.533) as a factor conditioning the use of say in NigE. This factor is favoured with second person (FW 0.562), third-person plural (FW 0.534), and third-person singular (FW 0.517) contexts, whereas it is disfavoured with first-person singular (FW 0.492), first-person plural (FW 0.455), and neuter (FW 0.443) contexts. My results for grammatical person differ from the findings in Buchstaller and D’Arcy (2009) who report grammatical person as a significant factor conditioning the use of say in favour of third-person subjects in both British and New Zealand English data for older speakers. As for younger speakers, the effect is weak on first-person subjects for both British and New Zealand English. As for the tense/time reference of the quotative, the effect is statistically significant (p value = 2.43e-102) and the strongest favouring effect is future time reference (WF 0.701), favoured slightly with the past tense (FW 0.563), and disfavoured with the present tense (FW 0.248). This finding corroborates the findings in Buchstaller and D’Arcy (2009) who report that tense/time reference is a significant factor conditioning the use of say among older and younger speakers for both British and New Zealand English. My results for the content of the quote show that the effect is strongly significant (p. value = 2.64e-13), with say strongly favoured with direct speech (FW 0.995). This finding corroborates the findings in Buchstaller and D’Arcy (2009) who report the content of the quote as a significant factor conditioning the use of say among older and younger speakers in both British and American English data. It is also similar to the finding in Tagliamonte and Hudson (1999), who report that the content of the quote is a significant factor conditioning the use of say with direct speech in both British and Canadian English. In contrast, my finding for the content of the quote differs from Blyth et al. (1990), who report say as neutral with no effect.

4.2.1 Effect of age and sex on the use of say

Table 4 and Figure 2 present results for cross-tabulation of age and sex to determine their effect on the use of say in my data. The results demonstrate that the adolescent males (N = 241, 14.8 percent) are the most frequent users of say, closely followed by the middle-aged males (N = 216, 13.4 percent). While males in the old age group (N = 207, 12.7 percent) slightly lead females (N = 201, 12.4 percent) in the same age group, the young adult females (N = 202, 12.5 percent) favour the use of say over the young adult males (N = 189, 11.7 percent). The chi-square analysis demonstrates that the interaction test between age and sex on the use of say is not statistically significant at p > .05, with adolescent males leading. 

Table 4: Cross-tabulation of age and say on the use of say

	Females		Males		Total
	N	%	N	%	N	%
Adolescents	189	11.7	241	14.8	430	26.5
Young adults	202	12.5	189	11.7	391	24.2
Middle aged	175	10.8	216	13.4	391	24.2
Older adults	201	12.4	207	12.7	408	25.1
Total	767	47.4	853	52.6	1620	100

χ² (3): 6.558, p> .05

Figure 2: Interaction between age and sex on the use of say

4.2.2 Effect of age and regional origin on the use of say

As Table 5 and Figure 3 reveal, say has the same frequency of use for the southern adolescents and the northern middle-aged group (N = 223, 13.7 percent each), which turn out to be the highest users. Among the young adults, the southerners favour the use of say (N = 204, 12.6 percent) over the northerners (N = 187, 11.5 percent). Similarly, the southern old age group (N = 207, 12.7 percent) leads the northern old age group (N = 201, 12.4 percent). The chi-square analysis reports that the interaction test between age and regional origin on the use of say is not statistically significant at p> .001, with the southern adolescents and the northern middle-aged group leading with equal frequency.

Table 5: Cross-tabulation of age and regional origin on the use of say

	North		South		Total
	N	%	N	%	N	%
Adolescents	207	12.7	223	13.7	430	26.4
Young adults	187	11.5	204	12.6	391	24.1
Middle aged	223	13.7	168	10.7	391	24.4
Older adults	201	12.4	207	12.7	408	25.1
Total	818	50.3	802	49.7	1620	100

χ² (3): 9.003, p> .001

Figure 3: Interaction between age and regional origin on the use of say

4.2.3 Effect of age and social class on the use of say

As for age and social class, the results for cross-tabulation are given in Table 6. As expected, the Table shows that the upper-class old speakers (N = 302, 18.6 percent) are the most frequent users of say, with a wide gap to the lower-class old (N = 86, 5.3 percent) and the middle-class old (N = 20, 1.3 percent). The middle-class young adults (N = 218, 13.4 percent) are next to the upper-class old in frequency, leading the lower-class young adults (N = 138, 8.5 percent) and the upper-class young adults (N = 35, 2.2 percent). With respect to adolescents, the lower-class adolescents (N = 199, 12.3 percent) favour the use of say more than the middle-class adolescents (N = 191, 11.8 percent) and the upper-class adolescents (N = 40, 2.5 percent). In the middle-aged group, the middle class (N = 172, 10.6 percent) leads over the upper class (N = 118, 7.3 percent) and the lower class (N = 101, 6.2 percent). In Figure 4, the chi-square analysis demonstrates that the interaction test between age and social class on the use of say proves to be strongly significant at p< .001, with the upper-class old speakers leading.

Table 6: Cross-tabulation of age and social class on the use of say

	Lower class		Middle class		Upper class		Total
	N	%	N	%	N	%	N	%
Adolescents	199	12.3	191	11.8	40	2.5	430	26.6
Young adults	138	8.5	218	13.4	35	2.2	391	24.1
Middle aged	101	6.2	172	10.6	118	7.3	391	24.1
Older adults	86	5.3	20	1.3	302	18.6	408	25.2
Total	524	32.3	601	37.1	495	30.6	1620	100

χ² (6): 615.465, p< .001

Figure 4: Interaction between age and social class on the use of say

4.2.4 Effect of age and the content of the quote on the use of say

Table 7 and Figure 5 reveal that the adolescents (N = 428, 26.4 percent) lead in the use of say with direct speech. Similarly, the old age speakers (N = 406, 26 percent), the middle-aged speakers (with N = 390, 24.1 percent), and the young adults (with N = 390, 24.1) all favour the use of say with direct speech over gesture. In the case of thought, the results show that say does not express thought in NigE since there is no instance of this in the data. The chi-square analysis shows that the interaction test between age and the content of the quote on the use of say is not significant at p> .05, with all age groups favouring direct speech.

Table7: Cross-tabulation of age and the content of the quote on the use of say

	Direct speech		Gesture		Thought		Total
	N	%	N	%	N	%	N	%
Adolescents	428	26.4	2	0.1	0	0	430	26.5
Young adults	390	24.1	1	0.1	0	0	391	24.2
Middle aged	390	24.1	1	0.1	0	0	391	24.2
Older adults	406	25	2	0.1	0	0	408	25.1
Total	1614	99.6	6	0.4	0	0	1620	100

χ² (6): 0.542, p> .05

 

Figure 5: Interaction between age and the content of the quote on the use of say

4.2.5 Effect of age and grammatical person on the use of say

In Table 8, the results demonstrate that say is most favoured by the adolescents in first-person singular contexts (N = 203, 12.5 percent), closely followed by the young adults in first-person singular contexts (N = 195, 12 percent). While the young adults favour the use of say in first-person plural contexts (N = 10, 0.6 percent), the adolescents favour the use of say in second-person contexts (N = 6, 0.4 percent). For the third-person singular, the middle-aged group (N = 161, 9.9 percent) slightly leads the adolescents (N = 160, 9.8 percent), the old age group (N = 157, 9.7 percent), and the young adults (N = 146, 9 percent). The old age group (N = 58, 3.6 percent) leads in the use of say in third-person plural contexts over adolescents (N = 49, 3.1 percent), the young adults (N = 37, 2.3 percent), and the middle-aged group (N = 29, 1.8 percent). With respect to neuter contexts, the old and the middle-aged groups have the same frequency (N = 8, 0.5 percent each), leading the adolescents (N = 4, 0.3 percent), and the young adults (N = 1, 0.1 percent). Figure 6 presents the interaction test between age and grammatical person of the quotative on the use of say, and the chi-square analysis proves that the interaction is not statistically significant at p> .05, with the adolescents favouring the use of say in first-person singular contexts. Like my findings for be like, the findings here suggest that adolescents prefer to use say to report themselves in first-person singular contexts. My findings here differ from the findings in Buchstaller and D’Arcy (2009), who report significant interaction between age and grammatical person, with say favoured by younger speakers in first-person contexts, while their older speakers favour say in third-person contexts.

Table 8: Cross-tabulation of age and grammatical person on the use of say

	Adolescents		Young adults		Middle aged		Older adults		Total
	N	%	N	%	N	%	N	%	N	%
First (S)	203	12.5	195	12	180	11.1	171	10.5	749	46.1
First (P)	8	0.5	10	0.6	9	0.5	9	0.6	36	2.2
Second	6	0.4	2	0.1	4	0.3	5	0.3	17	1.1
Third (S)	160	9.8	146	9	161	9.9	157	9.7	624	38.4
Third (P)	49	3.1	37	2.3	29	1.8	58	3.6	173	10.8
Neuter	4	0.3	1	0.1	8	0.5	8	0.5	21	1.4
Total	430	26.6	391	24.1	391	24.1	408	25.2	1620	100

χ² (15): 22.172, p> .05

Figure 6: Interaction between age and grammatical person on the use of say

4.2.6 Effect of sex and regional origin on the use of say

The Table below reports results for cross-tabulation of sex and regional origin to determine their effect on the use of say, and it reveals that the northern males (N = 472, 29.2 percent) use say more frequently than the northern females (N = 346, 21.4 percent). On the other hand, the southern females (N = 421, 25.9 percent) lead the southern males (N = 381, 23.5) in the use of say. In Figure 7, the chi-square analysis demonstrates that the interaction test between sex and regional origin on the use of say is strongly significant at p< .001, with northern males leading in the use of say.

Table 9: Cross-tabulation of sex and regional origin on the use of say

	North		South		Total
	N	%	N	%	N	%
Females	346	21.4	421	25.9	767	47.3
Males	472	29.2	381	23.5	853	52.7
Total	818	50.6	802	49.4	1620	100

χ² (1): 16.884, p< .001

Figure 7: Interaction between sex and regional origin on the use of say

4.2.7 Effect of sex and social class on the use of say

Turning to sex and social class, the results in Table 10 show that the middle-class males (N = 318, 19.6 percent) are the most frequent users of say, closely followed by the middle-class females (N = 283, 17.5 percent). Similarly, the lower-class males (with N = 271, 16.7 percent) lead in the use of say over the lower-class females (N = 253, 15.6 percent), whereas the upper-class males (N = 264, 16.3 percent) lead over the upper-class females (N = 231, 14.3 percent). Figure 8 illustrates the interaction test between sex and social class on the use of say, and the chi-square analysis demonstrates that there is no significant interaction at p> .05, with males leading in all three classes. This finding differs from the findings in Buchstaller and D’Arcy (2009), who report significant interaction between sex and social class for older speakers in British English data, with working-class women leading in the use of say.

Table 10: Cross-tabulation of sex and social class on the use of say

	Lower class		Middle class		Upper class		Total
	N	%	N	%	N	%	N	%
Females	253	15.6	283	17.5	231	14.3	767	47.4
Males	271	16.7	318	19.6	264	16.3	853	52.6
Total	524	32.3	601	37.1	495	30.6	1620	100

χ² (2): 0.29, p> .05

Figure 8: Interaction between sex and social class on the use of say

4.2.8 Effect of sex and tense/time reference on the use of say

According to Table 11 and Figure 9, say occurs most frequently in the past tense with males (N = 635, 39.2 percent) leading females (N = 524, 32.4 percent). While females (N = 164, 10.1 percent) lead males (N = 133, 8.2 percent) in the use of say with the present tense, males (N = 85, 5.3 percent) lead females (N = 79, 4.8 percent) in the use of say with the future time reference. The chi-square analysis reveals that the interaction test between sex and tense/time reference of the quotative on the use of say is statistically significant at p< .05, with males leading in the past tense.

Table 11: Cross-tabulation of sex and tense/time reference on the use of say

	Past		Present		Future		Total
	N	%	N	%	N	%	N	%
Females	524	32.4	164	10.1	79	4.8	767	47.3
Males	635	39.2	133	8.2	85	5.3	853	52.7
Total	1159	71.6	297	18.3	164	10.1	1620	100

χ² (2): 9.545, p< .05

Figure 9: Interaction between sex and tense/time reference on the use of say

4.2.9 Effect of social class and the content of the quote on the use of say

In Table 12 and Figure 10, the results demonstrate that the middle class (N = 600, 37 percent) uses say with direct speech most frequently. Similarly, both the lower class (N = 523, 32.3 percent) and the upper class (N = 491, 30.3 percent) favour the use of say with direct speech. For gesture, the upper class (N = 4, 0.2 percent) leads in the use of say over the lower class and the middle class (N = 1, 0.1 percent each). In the case of thought, the results show that say does not express thought in NE since there is no instance of this in the data. The interaction test between social class and the content of the quote on the use of say proves to be statistically significant at p< .001, with the middle class leading in favour of direct speech.

Table 12: Cross-tabulation of social class and content of the quote on the use of say

	Direct speech		Gesture		Thought		Total
	N	%	N	%	N	%	N	%
Lower class	523	32.3	1	0.1	0	0	424	32.4
Middle class	600	37	1	0.1	0	0	601	37.1
Upper class	491	30.3	4	0.2	0	0	495	30.5
Total	1614	96.6	6	0.4	0	0	1620	100

χ² (4): 27.397, p< .001

Figure 10: Interaction between social class and content of the quote on the use of say

4.2.10 Effect of social class and tense/time reference on the use of say

As for the results for cross-tabulation of social class and tense/time reference of the quotative, Table 13 and Figure 11 report that say occurs most frequently in the past tense with the middle class (N = 436, 26.9 percent) leading the upper class (N = 362, 22.4 percent) and the lower class (N = 361, 22.3 percent). While the lower class (N = 130, 8 percent) leads in the use of say with the present tense over the middle class (N = 92, 5.7 percent) and the upper class (N = 75, 4.6 percent), the middle class (N = 73, 4.5 percent) leads in the use of say with the future time reference over the upper class (N = 58, 3.6 percent) and the lower class (N = 33, 2 percent). The chi-square analysis illustrates that the interaction between social class and tense/time reference of the quotative on the use of say is strongly significant at p < .001, with the middle class leading in favour of the past tense use.

Table 13: Cross-tabulation of social class and tense/time reference on the use of say

	Past		Present		Future		Total
	N	%	N	%	N	%	N	%
Lower class	361	22.3	130	8	33	2	524	32.3
Middle class	436	26.9	92	5.7	73	4.5	601	37.1
Upper class	362	22.4	75	4.6	58	3.6	495	30.6
Total	1159	71.6	297	18.3	164	10.1	1620	100

χ² (4): 29.739, p< .001

Figure 11: Interaction between social class and tense/time reference on the use of say

5. Discussion

Crucially, despite the frequent use of be like, the presence of say as a traditional quotative has not decreased among NigE speakers, and this provides a piece of evidence that be like is still undergoing diffusion into NigE. In particular, the use of say in this study yields no wide generational variation in that the distributional analysis shows that say is used in relatively equal proportions across all four age groups, viz., the adolescents (N = 430, 26.5 percent), the young adults (N = 391, 24.2 percent), the middle aged (N = 391, 24.2 percent), and the old age group (N = 408, 25.8 percent). On the other hand, the multivariate analysis reports that only the old speakers (FW 0.609) favour the use of say, whereas the adolescents (FW 0.479), the middle aged (FW 0.462), and the young adults (FW 0.448) disfavour it. Consequently, the effect of age is not statistically significant as a factor conditioning the use of say in NigE. The same pattern is observed for tell and zero, i.e. there is no wide generational variation across all four age groups, and the effect of age on the use of tell and zero is not statistically significant. With respect to sex, it is observed that say is favoured by males over females, but the effect is not statistically significant. The same constraint hierarchy is observed for zero, i.e. males favour it over females, and the effect of sex is not statistically significant. While females favour the use of tell over males, the effect is equally not statistically significant. These findings demonstrate that the use of quotatives differentiates the sexes in NigE, but how they exhibit the difference varies considerably. Furthermore, the analyses have revealed that, unlike be like, which has a strong effect on the regional origin, the quotatives say, tell, and zero are not significantly constrained by regional origin. While the north favours the use of say and tell, the south favours the use of zero, but with no significant effect. In the case of social class, the effect is not statistically significant as a factor constraining the use of say, whereas the effect is statistically significant in the use of tell and zero in NigE.

Looking at the discourse-pragmatic functions of the quotative say, this markerprimarily introduces quotations that have been explicitly said. Thus, this quotative form indicates that whatever follows is a representation of what was really said as it frames directly quoted speech. According to Romaine and Lange (1991: 235), the traditional say reports speech without the contribution of any pragmatic effect,which allows its use in a wide variety of contexts. They also describe the quotative say as a default verb of reporting, “the one the speaker chooses when there is no particular reason to choose another verb” (Romaine & Lange, 1991: 242). This suggests that,say, as an unmarked choice is consistent with its general discourse function, especially as it is rarely used with thought or internal dialogue.

What is intriguing and regular with the quotative say in my data is tense choice in narration,especially between authority figures and non-authority figures. People narrate encounters with authority figures to help in redefining the situation. Authority figures are known to speak with the voice of perceived wisdom or with public voices such as the voice of the law and adult morality. Arguably, authority stories can be considered as attempts to redefine oneself by way of manipulating one’s “footing” (Goffman, 1981), and this is a clear example of what Tannen (1986) calls “constructed dialogue”.^[1] In conversations involving figures of authority, storytellers often alternate between tenses to mark status relations. It is usually the authority figure whose talk is introduced in the present tense, whereas the non-authority figure gets introduced in the past. Consider example (3).

(3) So when the teacher came on Monday, he say, “You, stand up, did you do your assignment that I asked you to do?”

I now responded, “Sir, I do not know”.

He say, “Ehn, you do not know?”

I said, “Yes, I do not know.”

He went to the next person, “You, stand up, what about the assignment that I gave you to do, can you recite it?”

The other one said, “No, sir, I cannot, I do not know anything, I do not know it.”

He say, “Ahh, you too?”

“Yes”

“Okay, you people will see my true colour today.”

In (3), the speaker narrates a story from her primary school days when she and some of her class members refused to do their assignments and their teacher queried them. The authority here is the teacher and the teller, along with other reportee, who is non-authority. The authority’s speech is introduced with the present say and the non-authority with the past said. The speaker is a member of the lower class who only attained primary education, and the English she speaks is marked with errors. A noticeable error here is in marking the third-person singular, where the speaker keeps saying “say”, omitting the ‘s’. However, there are instances in my data where speakers are not consistent with the he says/I said pattern of tense alternation between authority, and non-authority as illustrated in (4).

(4) The pastor said, “No, you people are not calling our child.”

Then our principal said, “That is the result if you like continue with it; if you do not like it, then let us stop this thing.”

Then the pastor said, “It is better to stop it.”

I said, “We cannot continue this match.”

In (4), the speaker narrates a presentation of awards event in a secondary school dominated by Christians. The first two best students who received awards during the event are Muslims. One of the pastors in attendance was surprised by how Muslims would perform better than Christians in a school dominated by Christians. The pastor who had a particular brilliant student in mind drew the attention of the principal, “no, you people are not calling our child”. The principal, who is also a Christian, explained the outcome of the results and highlighted that there was nothing he could do, only if the pastor wanted the event to stop. In another turn, the speaker reporting himself maintains the same tense form as the pastor and the principal, who are both authority figures. The lack of tense shift here suggests that the speaker is unable to manipulate footings effectively in his narration. One important thing to note is that authority figures are not considered as authors of their speech, but rather representatives of the voices of authority.

The authority figures in most of the stories are not people with names. They are ‘the judge’, ‘the nurse’, this guy’, my teacher’. There are thus two senses in which authority figures in stories are not the authors of their words. In the first place, it is the teller who is the author of the story, and in the second place, it is the public that is presented as the real author of the words authority figures speak.

 (Johnstone, 1987: 49)

This suggests that authority stories are simply conventionalised public authorship conveyed through the individual authorship of the teller. Thus, this explains yet another way of understanding the pattern of tense alternation in dialogue introducers, the quotative say in this case. Meanwhile, in the examples with thehesays/I said pattern, the authority figures are presented as speaking with the voice of public authority, whereas in the example that does not follow the so-called he says/I said pattern, the authority figures are presented as individual authors of their speech.

6. Conclusion

Taken together, this study has demonstrated that there is a wide range of options for marking quotative expressions in Nigerian English (NigE), with the quotative sayfunctioning as a default reporting verb. It is primarily used to introduce explicitly spoken quotes without adding any pragmatic nuance, making it suitable for diverse contexts. The study has provided the sociolinguistic profile of the quotative say,Including the distributional analyses and multivariate analyses of both linguistic and extra-linguistic factors describing how they constrain the use of say among the speakers of NigE. The unmarked nature of the quotative say aligns with its general role in discourse, particularly when compared to other quotatives that convey internal thoughts. Notably, in narratives involving figures, the use of say reveals a patterned alternation in tense: present tense for authoritative speech and past tense for non-authoritative voices. This tense distinction reflects underlying status dynamics and serves as a strategy for storytellers to negotiate their social positioning. Overall, this study has provided an initial description of the patterns of use of say in NigE. Future research could explorewhether and how the use of this qoutative form evolves over time.

References:

Buchstaller, I. (2014). Quotatives: New trends and sociolinguistic implications. Oxford: Wiley-Blackwell.

Buchstaller, I. and D’Arcy, A. (2009). Localized globalization: A multi-local, multi-variateinvestigation of quotative be like. Journal of Sociolinguistics 13(3): 291-331. 

Butters, R. R. (1980). Narrative go ‘say’. American Speech 55: 304-307.

Butters, R. R. (1982). Editor’s Note. American Speech 57(2): 149.

Cheshire, J. (1982). Variation in an English dialect: A sociolinguistic study: Cambridge: CUP.

Ferrara, K. and Bell, B. (1995). Sociolinguistic variation and discourse function of constructeddialogue introducers: The case of be + like. American Speech 70(3). 265-290.

Gardner, M., Denis, D., Brook, M. and Tagliamonte, S. A. (2020). Be like and constant rateeffect: the bottom to the top of the S-curve. English Languageand Linguistics, 1-44.

Goffman, E. (1981). Forms of Talk. Philadelphia: University of Pennsylvania Press.

Johnson, D. E. (2009). Getting off the GoldVarb Standard: Introducing Rbrul for Mixed-EffectsVariable Rule Analysis. Language and Linguistics Compass, 3(1), 359-383.

Johnstone, B. (1987). “He says… so I said”: verb tense alternation and narrativedepictions ofauthority in American English. Linguistics 25: 33-52.

Labov, W. (1963). The social motivation of a sound change. Word 19: 273-309.

Labov, W. (1966). The Social Stratification of English in New York City.Washington, DC: Centrefor Applied Linguistics.

Labov, W. (1984). Field methods of the project on linguistic change and variation. InLanguagein Use, J. Baugh and J. Sherzer (eds), 28-53.

Mathis, T. and Yule, G. (1994). Zero Quotatives. Discourse Processes 18(1): 63-76.

Romaine, S. and Lange, D. (1991). The use of like as a marker of reported speech and thought: Acase of grammaticalisation in progress. American Speech 66(3): 227-279.

Tagliamonte, S. A. (2012). Variationist sociolinguistics: Change, observation, interpretation.Malden, MA: Willey-Blackwell.

Tagliamonte, S. A. and Hudson, R. (1999). Be like et al. beyond America: The quotative system in British and Canadian youth. Journal of Sociolinguistics 3(2): 147-172.

Tannen, D. (1986). Introducing constructed dialogue in Greek and American conversational andliterary narratives. In Florian Coulmas (ed.), Direct andindirect speech, 311-332. NewYork: Mouton de Gruyter.

---

[1]According to Goffman (1981: 128), footing is defined as “the alignment we take up to ourselves and others present as expressed in the way we manage the production or reception of an utterance”. In other words, it is how we project ourselves as speakers when we emerge in interactions.