Compound stress in educated Igbo and Yoruba accents of Nigerian English

Abstract This study examines compound stress in Igbo and Yoruba accents of Nigerian English and juxtaposes their patterns with Standard British English (SBE). Speech recorded from 60 educated Igbo and Yoruba speakers of Nigerian English and three Britons served as data and control respectively. These recordings were analysed perceptually and acoustically. Using simple percentage and chi-square, occurrence of tokens and levels of significance were checked between the compound stress patterns of educated Igbo and Yoruba speakers of English and SBE at 0.05 (p ≤ .05). Optimality theory serves as the theoretical framework. The findings indicate that educated Igbo and Yoruba speakers of Nigerian English, though share certain patterns with SBE, have significant patterns of variations in compound stress assignment. There was a systematic attempt by the Nigerian participants to align left, while SBE aligns right. This showed the propensity of the participants to analogically accentuate the constituent of the compounds that carry significant information. This pattern was supported by the higher pitch and durational values recorded by speakers of educated Igbo and Yoruba accents of Nigerian English. Even in cases where there were similarities in the patterns, the acoustic features vary. This is common to both accents of Nigerian English.


Introduction
The contact of English with the entity now called Nigeria dates back to the 16th century through adventure and commerce.This was later reinforced by the activities of missionaries and colonialists.Following these developments, the language has continued to serve various communicative purposes among the various sociolinguistic groups in Nigeria.Today, because of the widespread nativisation of the language and the multilingual composition of Nigeria, there is evidence that there are geo-tribal and ethnic varieties of the language in Nigeria, which are characterised by ABOUT THE AUTHORS Omotosho Melefa (Faculty Mentor)Department of English and Literary Studies, University of Nigeria, Nsukka, Nigeria Oluwasegun Amoniyan (Principal Investigator) Department of Linguistics, University of Pittsburgh, USA (note: the approval and data collection occurred during my master's degree at the Department of English & Literary Studies, University of Nigeria 2018Nigeria -2020.) .)peculiarities that are induced by the linguistic structures of the speakers' L1 (Jibril, 1982a;Akinjobi, 2006).These peculiarities are evident at the various levels of linguistic analysis-syntactic, phonological, etc.For example, the phonotactic rules of Igbo and Yoruba are different in certain areas from those of English.These areas of differences are believed to largely condition the differences between the suprasegmental features of Nigerian English and Standard British English (SBE).Expectedly, various dimensions of these developments have attracted the attention of scholars.Fajobi (2008Fajobi ( , 2012Fajobi ( , 2013) ) claims that the syllables of Nigerian English are pronounced with tones rather than with stress, such that the stressed syllable in the variety, which is usually articulated with a high tone, may not be the same one that is normally rendered strong in SBE.According to her, these arguments stem from the tonal nature of most Nigerian languages, which is believed to influence the spoken English of the speakers of these languages.However, this claim is arguable since it is not supported with acoustic evidence, which will repudiate extant submissions (supported by acoustic evidence) on how stress is placed in the Nigerian accent of English.Previous studies (e.g., Akinjobi, 2006;Eka, 1985;Gut, 2005;Sunday, 2011;Sunday & Oyatokun, 2016;Udofot, 2003) have provided empirical as well as acoustic evidence on how stress manifests in Nigerian English.This current study aligns with these widespread positions that project stress as a feature of Nigerian English.
The phonotactics of English differs significantly from that of any Nigerian language.English operates a structure that is often summarised as (c 0-3 ) v (c 0-4 ).This means that it allows a complex structure of up to three (3) consonants at the syllable onset and a maximum of four (4) at the coda position.However, for Igbo and Yoruba, Soneye and Oladunjoye (2015) submit that they operate CV, V and N syllable structures.This means that they both allow a maximum of two elements in a syllable (Emenanjo, 2015).These are consonant and vowel (CV).Like English, both (Igbo and Yoruba) allow syllables with zero onsets.Also, Igbo and Yoruba allow syllables with a single syllabic nasal (N).But, the two languages do not allow consonant clusters (e.g., "strength" as "si-ti-re-N-ti"; prompt as "po-N-pi-ti /"pi-ro-mi-ti" pi-ro-mi-ti), long vowels or diphthongs (/ei/as/e/ . ..) (Jowitt, 2019).
Research on the phonological features of English in Nigeria has mainly been on its peculiar segmental features (see Awonusi, 1986;Banjo, 1995, Christopher, 1954;Egbe, 1979;Eka, 1985;Igboanusi, 2006;Jibril, 1982b).This means that so much is left unexplored on the suprasegmental features of Nigerian English.The few works that are available on the suprasegmental features of this variety of English (Gut, 2005, Jowitt 2000;Sunday, 2011;Sunday & Oyatokun, 2016;Udofot, 2003) have focused more on the description of intonation patterns, word stress and rhythmic patterns.Systematic attempt which draws on optimality theory to account for the cognitive processes that characterise the (re)ranking of constraints for compound stress by educated Igbo and Yoruba speakers of English is not yet common.What comes close to this is Melefa and Amoniyan's (2019) and Sunday and Oyatokun's (2016) studies, which account for the cognitive processes that define the patterns of word stress in Nigerian English and educated Igbo English respectively.Even Sunday's (2011) work on the placement of stress on compounds in Nigerian English does not account for the cognitive processes that underlie the assignment of stress to compounds and, of course, not an attempt to compare the patterns that characterise the handling of compound stress by Igbo and Yoruba speakers of English.This implies that there is no adequate insight into the peculiarities that characterise the handling of compound stress by educated Igbo and Yoruba speakers of English in Nigeria.Therefore, the aim of this study is to investigate the handling of compound stress by educated Igbo and Yoruba English speakers with a view to analysing the patterns that characterise the process as well as account for the cognitive processes that underlie the (re)ranking of the constraints of the English language by these speakers in assigning stress to compounds.The patterns of similarities and differences in stress assignment to compounds by Educated Igbo English speakers (EIES) and Educated Yoruba English speakers (EYES) will also be examined and juxtaposed with the SBE.This approach will not only help to discuss the peculiarities in patterns; it will also help to account for the cognitive processes that underpin these patterns.This has the potential of expanding the available knowledge on the phonological features of the English spoken by educated Igbo and Yoruba in Nigeria beyond Igboanusi's (2006) comparative study of segmental features of Igbo and Yoruba English.

Literature review
Assignment of stress to English compounds has received considerable scholarly attention.Earlier studies (e.g., Bolinger, 1972;Ladd, 1984;Sweet, 1892) have claimed that informativity has an influence on the leftward or rightward assignment of stress to compounds in English.They have predicted that assignment of stress to compounds in English, either leftward or rightward, is primarily based on the informativeness of the constituents of the compounds; such that less informative constituents tend to be unstressed.Until recent times, however, empirical evidence for these claims was not provided.In more recent studies, Bell (2012), Plag (2006) and Bell and Plag (2012) provided evidence for strong informativity effects and variability in a large database of compounds taken from the British National Corpus (BNC).Plag and Kunter (2010) found no independent effect of informativity on compound stress in a large database of compounds taken from the Boston University Radio Speech Corpus (Ostendorf et al., 1996).In an attempt to replicate the result of Bell and Plag (2012), Bell and Plag (2013) showed that more informative constituents in noun-noun component (N2) tend to be stressed.The study also raises the critical question of the link between informativity and constituent identity.This link is considered as the greatest predictor of stress in compounds (Arndt-Lappe, 2011;Kunter, 2011;Plag, 2010).However, these studies, like others, have two identifiable problems especially when placed within the context of world Englishes paradigm.Firstly, only a single consideration of informativity was considered, except for Bell and Plag (2012).Secondly, since the studies were carried out on purely L1 contexts, the results cannot be considered as reflecting the assignment of stress to English compounds in the contexts of English as a second language like Nigeria, where there are peculiarities in norms of usage.This raises the question of the need to contextualise these studies in order to find out how stress is assigned to compounds in L2 situations and the factors that predict the assignment.
Within the context of Nigerian English, most works that utilised acoustic approaches to the Nigerian accent of English (e.g.Dyrenko & Fuch, 2018;Gut, 2005;Gut & Milde, 2002;Jamakovic & Fuchs, 2019) have concentrated more on its features at the segmental level, with little attention paid to suprasegmental features like stress.These studies have specifically identified higher pitch levels and longer durations for segments and syllables in Nigerian English which puts it between the speakers' L1 and SBE.This calls for a closer investigation of the sub-varieties of spoken Nigerian English in order to track the influence of the speakers' L1.Scholars (see Anyagwa, 2014;Azidiegwu, 2019;Sunday, 2010Sunday, , 2011;;Sunday & Oyatokun, 2016) have supported the claim that languages in which the position of stress can usually be predicted by a simple rule are said to have fixed stress.For instance, Melefa and Amoniyan (2019) have also shown that word stress in Igbo English (IE) has disparate patterns and largely unpredictable.This position means that the word stress patterns are different from what is obtainable in SBE.
Out of what has been done and statements made on stress in Nigerian English (e.g.Anyagwa, 2014, Jowitt, 1991;Sunday, 2011;Udofot, 2003 etc.), Atoye's (1991) and Kujore's (1985) study have yielded the most far-reaching submissions.Kujore (1985) enumerates the general characteristics of stress placement in Nigerian English to include the tendency for forward stress, the recurrence of forward stress in compounds with final obstruents like firewood, proofread, etc., the tendency for some suffixes to carry stress forward to the syllables preceding them, the tendency for strong clusters to pull stress forward to the preceding syllable and reversal of the order of primary and secondary stress in words such as education, federation, etc.Moreover, Atoye (1991) identifies progressive and regressive stress shift as the defining patterns of word stress in Nigerian English.He argues that though regressive stress shift is a feature of Nigerian English, progressive stress shift is the pattern that distinguishes it from SBE.However, most of these claims on stress in Nigerian English were not always backed with acoustic evidence and field work.For example, Kujore's (1985) study with far-reaching conclusions was not backed with field work or acoustic evidence.This situation emphasises the need to empirically investigate this phenomenon and at the same time provide acoustic data to back up the claims.The current study, which investigated compound stress handling by educated Igbo and Yoruba speakers of Nigerian English, is not only field work based, but the data obtained from the field work were also subjected to empirical and acoustic analyses.

Participants and methodology
The data for this study were collected using a tape recorder.A total of sixty (60) participants, educated Igbo and Yoruba speakers of English chosen from the South-eastern and South-western parts of Nigeria, divided equally by sex and ethnic affiliation were made to read a specially prepared text containing 10 lexical items which include open and closed compounds (see appendix).In all, 30 male and 30 female participants who are aged between 18 and 45 years old and hail from south-eastern states of Abia, Anambra, Enugu, Ebonyi, Imo and south-western states of Ogun, Oyo, Ondo, Lagos and Osun respectively read the text.These participants were undergraduate and postgraduate students who were randomly selected from two leading and oldest Nigerian universities, University of Ibadan, Ibadan for the Yoruba speakers of English and University of Nigeria, Nsukka for the Igbo speakers.The selection of these participants from these two representative universities are considered appropriate because the two universities, which are notable in the two regions of the country, serve as centres where these required participants were readily available.The universities also provided ready access to participants who have not had extended stay outside their respective regions.The selected respondents conformed to educated variety of Nigerian English as defined by Banjo (1996), Udofot (2003Udofot ( , 2004Udofot ( , 2007) ) and Soneye and Oladunjoye (2015).The use of these human participants has the approval of the Faculty of Arts Research Ethics Committee, University of Nigeria, Nsukka.
The recordings were subjected to acoustic, theoretical and statistical analyses.Descriptive statistics of simple percentage using Statistical Package for the Social Sciences (SPSS) was employed to calculate the occurrences of tokens, and chi-square was used to check the level of significance between the compound stress patterns of educated Igbo speakers of English, educated Yoruba speakers of English and SBE at 0.05 (p ≤ .05).The specially prepared texts were also read by three Britons who are MA students in a UK university for the SBE model, which serves as the normative baseline.
Sound Forge, Praat (Boersma & Weenink, 2017) were used for sound files processing and acoustic analysis respectively.For the acoustic analysis, acoustic correlates of stress, pitch (F0), intensity (dB) and duration (ms) were manually set on Praat to determine which syllables of the tested compounds were assigned stress by the participants.Since pitch is the most important acoustic correlate of stress, especially for compounds and phrases, pitch range of 100 Hz was used to determine the stressed syllable for the native control, while it was left at 200 Hz and above for the Nigerian participants.These pitch range differences were necessary because of the established pitch height differences between the two sets of participants.The most common patterns of stress assignment were taken as the norm.Tokens of these common patterns were analysed and converted to simple percentage using the formula below: Results got from the above formula were used to determine the patterns that are prevalent among the participants.The mental processes that govern these prevalent patterns were then discussed using Optimality Theory (OT).These manifest in the form of (re)ranking of universal constraints of English.The compounds analysed using OT were tagged Subject 1-10 in order to achieve a systematic presentation of the compound samples tested (Figure 1).

Theoretical framework
Optimality Theory(OT) developed by Prince and Smolensky (1993) serves as a framework for data analysis.The theory is interested in well-formedness of candidates that satisfy certain constraints (Kager et al., 2004;McCarthy, 2002McCarthy, , 2008;;Prince & Smolensky, 2004).Meanwhile, the language and the aspect of the language under consideration determine the constraints that would be employed to rank the candidates to select the optimal (winning) candidate with an indication of hand ☞ (McCarthy, 2007).
The architecture of OT has three (3) components.CON is the component that defines the set of universal violable constraints.GEN is the component where output candidate parses are composed and generated based on input forms.EVAL is the component that selects an optimal output from the set of alternative candidates, given a language-specific hierarchical ordering of CON.The schema is presented below: Constraints used for compound stress patterns of the participants are: NSR (Nucleus Stress Rule): Stress the rightmost elements in phrases (Cruttenden, 1986, p. 31).
LCPR (Lexical Category Prominence Rule): Operates on simple and compound words.For any pair of sister nodes L where L is a lexical category; then N 2 is strong.For instance, N1 (oil-tanker) in "oil-tanker" driver, driver is strong since the N2 (driver) does not branch into new words (Liberman, 1975, p. 27).

CSR (Compound Stress Rule):
Stresses the penultimate word in phrasal expressions.CSR applies to both (open) compounds and phrases.For instance, in compounds, if it is a noun, the stress will be on the first syllable while in compound adjectives, adverbs, adjectives and verbs; ultimate lexical item bears the prominence.
GER (Grid Euphony Rule): Depicts that every strong lexical item should be followed by a weak position and no weak lexical item should be preceded by more than one weak position.HPR (Hierarchical Prominence Rule): Considers the hierarchical distribution of semantics as inherent in the structure before assigning prominence, however, noun is ranked higher than verbs and adjectives.This connotes that noun is considered first for prominence before any other parts of speech receives prominence.
TS (Tonic syllable): Assign stress to the last content word which carries the central proposition of tone unit.
PAPR (Pitch Accent Prominence Rule): Pitch accent lexical items receive an additional beat to raise its prominence.
HPC (Highest Phrase Condition): Conditions prominence on the lexical item that has highest position in a phrase and compound PPR (Phrasal Prominence Rule): Emphasises prominence on the lexical item that is most meaningful in phrases and compounds.TGA (Text-to-Grid Alignment): Relates to the lexical structure either TGA (L), TGA (M) or TGA (R) (Selkirk, 1984).

Result and discussion
For the 10 compounds tested, mealtime, backfire, post office, passbook, proofread, firewood, foul play, sitting room, bush meat and halfway, the assignment of stress generally revealed similar patterns among educated Igbo and Yoruba participants, as well as some closeness to the SBE, albeit with major patterns of variation which were evident in the acoustic features.As observed from the data, the patterns of variation were not strictly defined by the underlying sociolinguistic variables such as sex, ethnicity, education, etc.They rather showed possible evidence of unstructured mastery of SBE compound stress patterns.Overall, all the compounds except foul play, sitting room, and halfway, had similar stress assignment patterns with the SBE.The results also showed a general tendency by the Nigerian participants to align left irrespective of the compound involved.
The trend showed that this tendency is anchored on the principle of analogy and informativity of constituents of the compounds (Bell & Plag, 2012, 2013), as the results showed that the participants operated from a fixed cognitive domain which ingrains leftward assignment.Analysis of tokens of the compounds is presented in what follows.

Stress pattern of mealtime
Analysis of the first compound, mealtime, showed some similarities in patterns of stress assignment which cut across EIES and EYES.The data showed that 90% of EIES and 93.3% of EYES respondents assigned stress to the first part of the compound, meal.The percentage of the participants that assigned stress to the first part of the compound, ˈmealtime is 90% and 93.3% respectively for EIES and EYES (see Table 1).However, 6.7% (2 participants) and 3.3% (1 participant) of the participants, which comprise male and female respondents assigned the primary stress to the second component and both items respectively, realising mealTIME and MEALTIME.
From the three patterns of stress assignment by the participants, three candidate sets were got for the OT analysis, which means that EIES and EYES realised MEALtime, mealTIME and MEALTIME for this test item.Despite these variable patterns, the most common compound stress pattern for meal time among the participants is MEAL time.This is evident in the intensity and pitch measurements for meal (see Table 2), which is higher than those for time.As realised by EIES, the intensity for meal is 83.77dB, pitch is 272.2Hz while time has 70.85 dB for intensity and 245.2 Hz for pitch.However, EYES had a range of 80.39 dB as the intensity for meal and 158 Hz for pitch.This means that the average EIES and EYES had a range of 80 to 90 dB as the value for sound intensity to show the compound stress pattern for mealtime, which makes it higher than the second lexical item time.These patterns of stress assignment to this compound supports earlier submissions (e.g.Sunday, 2011;Udofot, 2003).
From the OT tableau above, the ranking of these constraints by the respondents is: HPC>>PPA>>RHR.This means that HPC dominates PPA, which also dominates RHR.So, for these respondents, the optimal candidate is MEALtime.The tableau indicates that the assignment of stress to mealtimeby educated Igbo and Yoruba speakers of Nigerian English interacts with violable constraints, where Candidate (i) did not incur any violation unlike Candidate (ii) and Candidate (iii) which incurred fatal violation for HPC) and partial violation (PPA and LHR).Comparison of the values for EIES and EYES using ANOVA is done in Table 2 to provide evidence for the uniformity of the pattern of stress placement for the item.The analysis of variance showed that there is no significant difference in the assignment of stress to mealtimeby EIES and EYES.Table 2 shows that the level of significance (0.759) was higher than ≤0.05.
In OT, these three patterns are considered as the possible input sets for the respondents.For these inputs, the constraints HPC, PPA and RHR, which revealed the cognitive processes that produced the inputs, were used to rank them thus:

Stress pattern of backfire
Results got for backfire are similar to those for mealtime.Assignment of stress to back was done by 8 (26%) of EIES and (12) 40% by EYES.Stress was assigned to fire by 66.7% of EIES and 60% of EYES.This connotes that Candidate (ii) has larger percentage of users amongst respondents.However, it is significant to note that while educated Igbo speakers of English have three possible patterns of stress assignment to backfire, namely BACKfire, backFIRE and BACKFIRE, educated Yoruba speakers of English have two variant realisations of the item: BACKfire and backFIRE (see Table 3 for details).
The three patterns of stress assignment in Table 4 provide three candidate sets for the OT analysis.Tableau for Subject 2 showed that CSR and NSR, which mandate prominence were violated by candidate (i) and (iii).The fatal violation of these constraints as shown in the tableau prevents these candidates from emerging as the optimal candidate.CSR is a markedness constraint and it is higher ranked than LHR, which is a faithfulness and alignment constraint.What seems to be common pattern among EIES and EYES is that lexical items that carry weightier meanings are rendered louder than those that are not.So, these speakers tend to foreground lexical dominance of an item in a given situation.The data revealed no significant difference between the compound stress pattern of backfire by EIES (66.7%) and EYES (60%).This portends a harmonised pattern of stress placement for EIES and EYES.This harmonised usage of this pattern, backfire, despite differences in linguistic background, is in line with previous studies (see Atoye, 1991;Udofot, 2003, etc.) who have argued that developments such as this have the potential of becoming a precursor to standard spoken Nigerian English (Figure 2(a,b)).

Stress Patterns of Post Office, passbook, proofread, firewood, foul play, sitting room and Bush meat
Analysis of some other compounds also revealed similar patterns of stress placements amongst the respondents (see Table 5).For the item, post office, 73.35% of the participants had a pattern that is similar to SBE.The population that stressed the first items of the compounds as indicated in the tableaux (post office, passbook, proofread and firewood) is 77.5% for EYES and 69.15% for EIES.The other patterns are 19.2% for EYES and 23.33% for EIES.The assignment of prominence to the second item cuts across both genders.No peculiar gender induced variation was observed in the data.
Placement of stress on bush meat, as shown in Table 5, also showed significantly similar patterns among EIES and EYES.For instance, EIES produced two candidates, BUSH meat and bush MEAT, while EYES also have two candidates in the articulation of bush meat: BUSH meat and BUSH MEAT, but the larger percentage assigned prominence to BUSH.This aligns with the findings of Bell andPlag (2012, 2013).It appears the participants followed informativeness and constituent identity to assign stress to the left.From Table 5, a total of 83.3% (EIES) and 96.7% (EYES) of the participants assigned stress to the first lexical item of the compound (bush meat).
The participants generally satisfy Left-Headed-Rule by assigning prominence to the left of the compounds as evident in the test items above, irrespectively of their intricacies.These patterns of stress assignment are generally emblematic of the handling of stress in compounds by the participants, albeit there were some differences in patterns among respondents.Even in cases where stress should be shifted to the second item in compounds e.g.halfway, foul play, most of the participants clearly followed the pattern that seem to have been entrenched in their cognitive domain.These general patterns are similar to Sunday's (2010) submission that speakers of Nigerian English assign prominence to the left lexical items in phrases such as past question, many detractors, those calls, etc.These patterns are, however, different from those reported for SBE and American English in Bell (2012) and Bell and Plag (2012).They are also different from German and Dutch as reported by Truckenbrodt (2016).

Stress pattern of halfway
For halfway, 100% of EIES and 83.3% of EYES assigned prominence to half.Table 6 shows a uniform pattern of stress assignment for EIES unlike EYES who stressed the first and second lexical items respectively.Following Left-Headed-Rule, the pattern of left alignment is also evident here.Assignment of stress to the compound for both EIES and EYES aligns left, which is different from SBE where alignment is to the right.
From Table 6, it is evident that only 10% of EYES participants aligned right (RHR) in the placement of stress on halfway.This means that most of these participants aligned left (LHR), a pattern which is different from SBE's right alignment (RHR).So, the pattern that is peculiar to EIES and EYES is HALFway, which is evidently the optimal candidate for these participants.
Figure Figure 3 presents a summary of compound stress patterns that are common to EIES and EYES.
From Figure Figure 3, it is evident that 70.45% of EIES and 80.95% of EYES stressed the first lexical item of compounds.EIES (21.9%) and EYES (16.2%) assigned stress to the second lexical    The findings of this study suggest an interaction between the EIES/EYES and their indigenous languages (Igbo and Yoruba).For example, Igbo and Yoruba have three basic tonal categories: High (H), Low (L) and Mid (M).These three acute accents metamorphose into HH, HL, LH, LL, or HM in complex sequences for two or more syllables in utterances (Bamgbose, 1966, Clark, 1990;Connell & Ladd, 1990;Laniran, 1992;Liberman et al., 1992).These studies reveal that Igbo and Yoruba are tonal languages and share similarities though not without variations.The observed variation manifested in the realisations of the Igbo and Yoruba participants.The present study did not discover absolute similarities.For example, Clark identified a strong correlation between HH and loudness with duration, but a weak relationship existed between LL and loudness with duration.That is, each of these accents has varied loudness and duration.However, Amoniyan (2023) discovered that Igbo and Yoruba English varieties have loudness as an important cue for prominence variation.Similarly, this present study shows that Igbo English participants transferred loudness from Igbo to English.It attests to the inevitable consequence of language contact, which is evident in the results.No wonder EIES assigned prominence to the left (LHR) or right (RHR), and loudness is the significant cue that showed the difference between stressed and unstressed syllables of compounds (see Table 2, Figure Figure 2).This observation aligns with Amoniyan (2023) and Jolayemi (2006Jolayemi ( , 2008)), but disagrees with the submission of Peng and Ann (2001), Ko (2013) and Kozasa (2004), who assert that duration is a stress cue in native and non-native of English.With these results, we opine that loudness and duration, or either, is a stress cue in this non-native environment and that in a multilingual context such as this, language contact effects are expected.Another noticeable transfer is discovered in the stress patterns of both lexical items in compounds.This feature does not characterize native English, but EIES and EYES manifest it because it is characteristic of the prominent sequence of two or more syllables in their L1 (see Table 5, Tables 2, -9).Also, regardless of the magnitude of this feature, it suggests that the feature delineates the varieties under study and implicates language variation in non-native language ecology.Among English native speakers, such a feature (as stress clash) is resolved by inserting a weak syllable or weakening either of the members of compounds to become HL, LH, or HLH to satisfy trochaic condition.This study shows that the L1 patterns of these speakers of Nigerian English exert influence on their accent of English in this non-native milieu.
In addition, while Igbo and Yoruba both belong to the Niger-Congo language family, they (actually) belong to different branches within that family.Igbo is classified as a member of the Volta-Niger branch, specifically the Volta-Cross subgroup, while Yoruba is classified as a member of the Defoid branch, specifically the Yoruboid subgroup (Stewart, 1971).One important aspect of language variation within a language family is the development of distinct features and patterns over time.Stress patterns, which refer to the prominence or emphasis given to certain syllables in a word, can vary significantly between different languages, even within the same language family.The specific stress patterns in Igbo and Yoruba have likely developed independently over time as the languages evolved separately.While they may share some linguistic characteristics due to their common Niger-Congo ancestry, other historical, cultural, and geographical factors have contributed to their distinct stress patterns.It is important to note that stress patterns can be highly complex and are influenced by various factors such as syllable structure, tone, and vowel length.Different languages within the same language family can exhibit diverse stress patterns due to these factors and other language-specific developments.Therefore, although Igbo and Yoruba are related languages, the variation observed in their stress assignment to English compounds is premised on the natural linguistic evolution and divergence occurring over time within language families.
In addition to the natural linguistic evolution and divergence contributing to the distinct stress patterns in Igbo and Yoruba, several other factors can influence stress patterns.The two languages have different phonological and morphological systems, which can affect stress patterns.The phonological structure of a language, including the inventory of sounds and syllable patterns, can determine where stress falls within a word.Differences in syllable structure and the presence of specific consonant or vowel combinations in each language can lead to variations in stress placement.Both languages have tonal systems where the pitch contour of a syllable can change its meaning.Tone and stress are often interrelated, and the interaction between the two languages can influence stress patterns.Different tonal patterns and rules within Igbo and Yoruba can impact stress assignment and syllable prominence.Also, historical and cultural factors can play a role in shaping stress patterns.Igbo and Yoruba have distinct histories, cultural practices, and influences from neighbouring languages and ethnic groups.These factors can contribute to unique stress patterns as languages interact and adapt to their environment over time.We deem it fit to highlight that language contact with other languages through trade, colonization, or migration can introduce new stress patterns or influence existing ones.It could have impacted their respective systems if Igbo and Yoruba had come into contact with other languages with different stress patterns.Languages evolve and develop over time through the speech and interaction patterns of individuals within a community.Language change can occur through natural processes such as phonetic shifts, analogical changes, or lexical borrowing.These processes can influence stress patterns, as individuals may modify stress placement in certain words or borrow stress patterns from other languages.
In conclusion, the distinct stress patterns in Igbo and Yoruba result from a combination of linguistic evolution, phonological and morphological differences, tonal systems, historical and cultural influences (Awonusi, 1986), language contact, and individual language development.These factors have contributed to the development of unique stress patterns within each of these two language groups despite their common Niger-Congo ancestry.

Comparison of EIES and EYES patterns with SBE
Generally, the patterns of stress assignment to compounds by EIES and EYES share some similarities with SBE; however, there are significant areas of differences.The participants' patterns are governed by certain peculiarities which do not characterise SBE.In assigning stress to compounds, irrespective of its class, more than 60% of EIES and EYES generally assign stress using LHR (Left Head Rule), while less than 30% assign prominence that satisfies RHR where it is required.This pattern is evident in the assignment of stress to foul 'play, sitting 'room, half 'way, etc. SBE follows right alignment for the assignment of stress to these compounds.This was evident in the patterns of assignment by the three Britons who served as control.
The placement of stress on compounds like mealtime, post office, passbook, proofread, firewood, bush meat, and backfire showed some similarities with SBE, albeit there were differences in acoustic properties.For the articulation of "mealtime", 90% of EIES and 93.3% of EYES respondents assigned stress to meal, the first element of the compound, which is in consonance with the SBE.However, 3(5%) of the participants placed stress on time (which is the second item), while 2(3.3%) of the participants assigned stress to the two syllables which revealed possible interference from Igbo and Yoruba languages' tonal structure where every syllable receives prominence.SBE models which serve as the control variable have duration of 0.7 seconds for the production of mealtime, where meal has 0.33 seconds with 70.81 dB and 114.5 Hz and time has 0.36 seconds with 65.28 dB and 103.3 Hz.The measurement of intensity for the SBE for meal and time showed that meal has 70.81 dB which is higher than the intensity for time which is 65.28 dB.
However, EIES and EYES who also assigned stress to the first item of the group did not have the same duration, intensity and pitch.The average EIES and EYES have 0.59 seconds with 78.2 dB and 252.6 Hz for the articulation of meal while time has 0.46 seconds with 72.15 dB and 247 Hz, making a total of 1.0 second for the compound (mealtime).Acoustic evidence showed that EIES and EYES, especially the Yoruba speakers of English deliberately prolonged/m/, realising it as syllabic consonant; however, for stress patterns, pitch and intensity are considered as valuable indicators of prosodic elements that determine stress placement.The compound "mealtime" has two syllables as "meal" for CVC and "time" for CVC where each of the syllables has a vowel either/i:/having F 1 (419.59)and F 2 (2122.07);or/ai/as 712.96 and 1741.74respectively (Figure 4).This is because diphthongs are quantitatively considered heavier than monophthongs (front vowels) and are ranked higher on sonority scale than monophthongs.The variant handling of these by the participants could have contributed to the values that were read for them.
However, the inconsistencies with English stress patterns as well as intricacies of compound stress placement which were attested to in previous studies (e.g.Bell & Plag, 2012;Gussenhoven, 2004;Kunter, 2011;Sunday, 2010;Sunday & Oyatokun, 2016), which are fixed and free depending on the lexical items that are being considered, could have imposed certain limitations on respondents' performance.For example, English has left and right head rule for stress placement, unlike French that has right-head-rule.This complexity could pose problems for L2 handlers of English.Sometimes, just as observed by Atoye  (1991), these categories of users of the language evolve certain discretionary strategies to cope with these complexities.That is why OT works with violable constraints that explain the cognitive processes that were followed by language users.Both EIES and EYES seemed to have advanced left alignment as a discretionary strategy to cope with the intricacies of English compound stress assignment.
Safe for the acoustic values, the peculiarities that characterise the handling of "foul play" by the British Nigerian participants, which are shown in Figures 5 and 6, may not have been noticeable.The realisation by the Nigerian respondents, which is captured in Figure Figure 6, showed a shift from the patterns in SBE.This shift is shown not only in the longer duration that was expended by EIES and EYES but also in assigning stress to the leftmost item.While the pattern for SBE showed adherence to RHR (Right Head Rule), EIES and EYES utilised Left-Head-Rule as the strategy for assigning stress to the compound.
The differences in these patterns are also observable in the acoustic values (see Table 7).For instance, the control variable spent 70 ms on the compound, while educated Igbo and Yoruba speakers of English spent an average of 90 ms, indicating that the control was faster and spent less time.
In Table 8, it is also evident that vowel quality and consonant configuration for EIES and EYES are higher than the average for the control.This shows that an average EIES and EYES had heavier and louder voice quality than the native speakers who spoke calmly with ease.However, a more elaborate study may be required to describe the articulatory properties of these two categories of participants in order to justify this claim.
In Table 9, the first word "bush" was given more prominence by educated Igbo and Yoruba speakers of English, as 90% of the participants assigned stress to the first lexical item, which implies for EIES and EYES that the pattern emphasizes bush and not meat.This pattern is close to   the SBE.Also, 8.3% of the participants assigned stress to the second lexical item (meat) which is different from SBE.From the analysis, majority of the participants stressed BUSH as against meat.
In addition, it is believed that exposure and educational attainment could have reduced the impact of the participants' L1 considerably, though 1 (1.7%) of the participants stressed both lexical items.This showed that SBE and other L1 varieties of English (e.g., American English) might have influenced the spoken English of EIES and EYES.
Placement of stress on these items, post office, pass book, proof read and fire wood showed some similarities as well as variations.About 73.35% of the participants had placement that is similar to what is obtainable in SBE.The percentage of the respondents that stressed the first items of these compounds, (as indicated on the tableau for post office, pass book, proof read and fire wood), is 73.35%; while 18.9% of the participants assigned prominence to the second items (book, read and wood).
From Table 9, the vowel quality and quantity of the item that bears primary stress is greater than the second item, which serves as secondary stress (see; Akinjobi, 2006;Anyagwa, 2014;Gimson, 1975;Onuigbo, 1996;Roach 1991).
The results also revealed that educated Igbo and Yoruba speakers of English assign stress to the first elements of compounds.For instance, from the analysis, 73.35% of the participants stressed the first lexical item of these compounds, which aligns with SBE.Unlike Kujore's (1985) position on delayed stress in Nigerian English, 73.35% of the participants assigned stress to the first elements of compounds.Of course, Sunday (2010) revealed that Kujore's position was without empirical evidence.Melefa and Amoniyan (2019) showed that educated Igbo speakers of English assign prominence to the left-most syllable at the word stress level.
The handling of stress assignment to the nominal head compound, sitting room also revealed similarities and differences between stress assignment by EIES, EYES and SBE.While EIES and EYES generally assigned stress to the item following Left-Head-Rule, that is, stressing the leftmost item, sitting with varying percentages, SBE participants stressed room, which showed the adoption of Right-Head-Rule.The patterns for EIES and EYES are shown in Table 10.The general strategy adopted by EIES and EYES to cope with the complexities of stress assignment to compounds is left alignment (Left-Head-Rule).This implies that the first lexical item of a compound, which occurs at the left edge is assigned stress predominantly by the participants.The consistency with this pattern, even when Right-Head-Rule should have been applied, showed that the participants studied generally process compound stress differently from the SBE.This finding is contrary to Kujore's (1985) claim on forward stress in compounds by Nigerian English speakers, of course, which was not backed with empirical evidence.
In acoustic analysis, pitch contour of a sound is the curve that tracks the perceived pitch of the sound over time.The analysis revealed that the Nigerian participants rendered the initial lexical item of the compound (sitting room) higher than the later lexical item which is evident from the varied acoustic parameters.This is different from the SBE participants who assigned stress to the second item, room.For instance, the F0_Hz for the SBE participants showed 78.54 for sitting and 100.9 Hz for room, respectively.However, for the Nigerian participant, 291.5 Hz was recorded for sitting and 199.6 Hz for room respectively.The implication of this is that SBE and Nigerian participants have varying patterns of stress assignment to this compound.Theoretically, this study shows that OT constraints for compounds do not have universal application to EIES and EYES.It reveals that contextual factors may shape the functionality of OT constraints.In OT, constraints for compounds have two universal hypotheses for stress patterns (Booij, 1995;Prince & Smolensky, 1993).First, closed compounds have strong stress on the first lexical item summarized as [A] [B], [A] is strong where A and B are lexical items.The second hypothesis holds that nominal and verbal compounds have default stress on their initial constituent.On the other hand, adjectival and prepositional compounds are stressed on their right-hand constituents (Backhuys, 1989;McCarthy, 2007;Visch, 1989).Compound stress patterns in EIES and EYES showed that the two assumptions for compounds which apply universally to native speakers of English do not have universal application to non-native English varieties, as shown in EIES and EYES data analysed in this study.For example, the first hypothesis of nominal and verbal compound stress emphasized stress assignment on the first lexical item, but tokens such as meal time, post office, pass book, proof read and fire wood realized by the participants of this study do not align with this principle of universal constraints application, as EIES and EYES assign prominence to either left or right constituents as evident in the varied percentages.This further shows that the assumption of OT for nominal and verbal compound stress is not universal to non-native English speakers (EIES and EYES).Also, the second hypothesis of compound stress among native speakers expects the second lexical item of compounds to receive prominence if the compounds are adjectival or prepositional.This assumption is not also applicable to these participants as percentages of stress patterns for bush meat vary between the first and the second constituent in the varieties.This study has shown that OT universal constraints that govern the assignment of stress to compounds, either nominal, adjectival or verbal compounds do not apply to EIES and EYES.This study, therefore, foregrounds the need to reconsider compound stress rules and constraints that apply to them in non-native English environments.The principle of constraints universality in OT does not apply to the data studied.
In this study, an attempt was made to reexamine the principle of constraint ranking and its universal application to phonological data.The data showed that the cognitive process that shaped the participants' constraint ordering is the principle of constraint reranking.The observed cognitive processes showed the participants reranked the candidates in the input for the optimal candidates to emerge.Unlike in English where CSR is ranked higher, EIES and EYES ranked HPC higher instead.This variation is assumed due to language contact and the integration or fusion of the L1/L2 systems.The English often has the order of constraints as CSR>>NSR>>LHR (LHR/RHR, depending on the word class), but EIES and EYES reranked the constraints as HPC>>PPA>>LHR.The use of LHR or RHR relies on the word classes in English, but EIES and EYES prefer LHR more than its counterpart regardless (see Table 5).However, the constraints reranking in EIES and EYES are not without internal variation between the varieties (e.g., candidate 3 in Subject 1, 2, 3c − 9c).This further suggests that prominence variability is inevitable among multilingual non-native English speakers.That said, this paper has widened the frontiers of OT in accounting for varieties of English other than the native speakers, which originators of OT might not have possibly conceived in the OT architecture.

Conclusion
The aim of this study was to examine the patterns of compound stress handling in educated Igbo and Yoruba accents of Nigerian English and to juxtapose both with the SBE.The handling of compound stress by EIES and EYES in comparison with the patterns in SBE reveals patterns of similarities and differences in stress assignment to compounds.The similarities are conditioned by the normative role of the SBE in universities in Nigeria, while the differences are possible indexes of L1 influence and linguistic transfer.Although the patterns of differences observed were basically epiphenomenal in that they were not directly triggered by sociolinguistic or ethnolinguistic variables such as sex and ethnicity, they had widespread distribution among EIES and EYES.The widespread variations in the assignment of stress to compounds by the participants and the SBE is evident in the realisation of three regular patterns of stress assignment to each of the 10 compounds that were tested.There were certain defining features of stress assignment to compounds by the participants.Although there were areas of similarities between stress assignment to compounds by EIES, EYES

Figure 1 .
Figure 1.Pie chart showing the distribution of patterns of variation for mealtime.
Prominence Alignment): Focuses on the lexical item that should receive prominence.RNR (Right-node-raising): Stress the lexical item at the right hand of the lexical configurations.LHR (Left Head Rule): It emphasises that stress should be at the leftmost position.RHR (Right Head Rule): It emphasises that stress should be at the rightmost position.

Figure 3 .
Figure 3. Bar chart showing the distribution of patterns of stress for compounds by (EIES and EYES).

Figure 4 .
Figure 4. Spectral slide of British control's articulation of mealtime.

doi.org/10.1080/23311983.2023.2279345 items
of compounds.Also, 6.13% of EIES and 2.84% of EYES speakers stress the two lexical items in the open compounds.The results revealed that stressed assignment to compounds by the participants tilt more towards a predetermined pattern of left alignment (Left-Headed-Rule) irrespective of the intricacies of the specific compounds that are involved.