1. INTRODUCTION
⌅ The acquisition of second language (L2) prosody is of crucial
importance for intelligibility and perceptions of foreign accentedness (Flege, Munro, & MacKay, 1995Flege, J., Takagi, N., & Mann, V. (1995). Japanese adults can learn to produce English /ɻ/ and /l/. Language and Speech, 38, 25-55.
; Tajima, Port, & Dalby, 1997Tajima, K., Port, R., & Dalby, J. (1997). Effects of speech timing on intelligibility of foreign accented English. Journal of Phonetics, 25(1), 1-24
; White & Mattys, 2007White, L., & Mattys, S. (2007). Calibrating rhythm: First language and second language studies. Journal of Phonetics, 35(4), 501-522. http://dx.doi.org/10.1016/j.wocn.2007.02.003
). One neglected aspect of L2 prosody is
polysyllabic shortening (PS). PS refers to the phenomenon where vowels,
especially stressed vowels, become shorter the more syllables are added
to the stem, e.g., the vowel in ‘stick’ is longer than the vowel in
‘sticky’ (Lehiste, 1972Lehiste, I. (1972). The timing of utterances and linguistic boundaries. Journal of the Acoustical Society of America, 51(6B), 2018-2024. http://dx.doi.org/10.1121/1.1913062
). This contributes to the perception of a stress-timed rhythm where interstress intervals tend to be regular (Lehiste, 1977Lehiste, I. (1977). Isochrony reconsidered. Journal of Phonetics, 5(3), 253-263. http://dx.doi.org/10.1016/S0095-4470(19)31139-8
; Kim & Cole, 2005Kim, H., & J. Cole (2005). The stress foot as a unit of planned timing: Evidence from shortening in the prosodic phrase. In Proceedings of Interspeech 2005, Lisbon, 2005, pp. 2365-2368. http://dx.doi.org/10.21437/Interspeech.2005-37
; see Turk, 2012Turk A. (2012) The temporal implementation of prosodic structure. In A. Cohn, C. Fougeron, M Huffman (Eds.). The Oxford Handbook of Laboratory Phonology (pp. 242- 253). Oxford, UK: Oxford University Press.
, for more details).
Unlike the well-established PS in English, little is known about it in
Arabic, which does not seem to use it as much as English does. Most
earlier research on the development of the interlanguage (IL) of Arab
learners of English focused on segmental aspects and to a lesser degree
on suprasegmental aspects including syllable structure, stress, and
intonation; however, to the best of our knowledge, no study has tackled
PS in the IL of Arab learners. How PS is used by speakers whose first
language (L1) is syllable-timed (like Arabic) where PS does not seem to
be as evident as in stress-timed languages such as English (see Abu Guba, Mashaqba & Huneety, 2023bAbu Guba, M. N., Mashaqba, B., & Huneety, A. (2023b). Polysyllabic shortening in Modern Standard Arabic. Journal of Semitic Studieshttp://dx.doi.org/10.1093/jss/fgac030
) is not clear. We therefore predict that Arab
speakers of English will not apply the same degree of PS adopted by
English speakers.
The importance of investigating phenomena such
as PS is threefold. First, speech timing contributes to making speech
more understandable (Ordin & Polyanskaya, 2014Ordin, M., Polyanskaya, L. (2014). Development of timing patterns in first and second languages. System, 42, 244-257. http://dx.doi.org/10.1016/j.system.2013.12.004
; Turk & Shattuck-Hufnagel, 2013Turk
A, & Shattuck-Hufnagel S. (2013). What is speech rhythm? A
commentary inspired by Arvaniti & Rodríguez, Krivokapic´, &
Goswami & Leong. Laboratory Phonology, 4, 93-118. http://dx.doi.org/10.1515/lp2013-0005
). Second, it plays a crucial role in the
perception of a foreign accent as lack of durational variation results
in a perception of foreign-accentedness (Polyanskaya, Ordin, & Ulbrich, 2013Polyanskaya,
L., Ordin, M., & Ulbrich, C. (2013). Contribution of timing
patterns into perceived foreign accent. In ). P. Wagner (Ed.), Proceedings of Elektronische Sprachsignalverarbeitung (ESSV) (pp. 197-204). Dresden: TUDpress.
).
It has been reported that Jordanian Arab teachers of English were
perceived to speak with a fairly strong foreign accent, despite their
long experience and use of English (e.g., Abu Guba, Mashaqba, Hneety, & Hajeid, 2021Abu
Guba, M. N., Mashaqba, B., Huneety, A. & Hajeid, O. (2021).
Attitudes toward Jordanian Arabic-accented English among native and
non-native speakers of English. ELOPE: English Language Overseas Perspectives and Enquiries, 18(2), 1-29. http://dx.doi.org/10.4312/elope.18.2.9-29
; Abu Guba, Daoud & Jarbou, 2023Abu Guba, M. N., S. Daoud & S. Jarbou. (2023). Foreign accented-speech and perceptions of confidence and intelligence. Journal of Psycholinguistic Research. doi: 10.1007/s10936-023-09940-9.
).
One reason behind this could be the lack of durational variation and PS
in their speech. Third, acquiring English rhythmic patterns constitutes
a major challenge to English learners including those whose L1 is
stress-timed with high durational variability (Ordin, Polyanskaya, & Ulbrich, 2011Ordin, M., Polyanskaya, L., Ulbrich, C. (2011). Acquisition of timing patterns in second language. Proceedings of Interspeech 2011, 1129-1132. http://dx.doi.org/10.21437/Interspeech.2011-336
). This study therefore aims to explore the extent
to which non-native speakers (NNSs) of English whose L1 is not as
stress-timed as English will acquire the mechanisms of PS. The study is
thus intended to be a contribution to the study of IL rhythm. It will
shed light on the development of PS and the extent to which it is
transferable (cf. Ordin & Polyanskaya, 2014Ordin, M., Polyanskaya, L. (2014). Development of timing patterns in first and second languages. System, 42, 244-257. http://dx.doi.org/10.1016/j.system.2013.12.004
). This will enhance our understanding of speech rhythm in general and its interactions in L1 and L2.
The present study will attempt to answer the following questions:
-
How similar is foot duration as produced by English native speakers (NSs) and Arab NNSs? To what extent is foot duration commensurate with the number of segments and syllables among English NSs and NNSs?
-
To what extent do English NSs and NNSs employ PS?
It is expected that foot duration produced by NNSs will be different from that of NSs; the higher the level of the NNSs, the more native-like their performance will be. Also, it is more likely that foot duration among the NNSs will depend on the number of segments and syllables within a foot.
2. BACKGROUND
⌅ Earlier studies on PS in English found that stressed syllables tend to
become shorter the more syllables are added to the stem (e.g., Lehiste, 1972Lehiste, I. (1972). The timing of utterances and linguistic boundaries. Journal of the Acoustical Society of America, 51(6B), 2018-2024. http://dx.doi.org/10.1121/1.1913062
; Port, 1981Port, R. F. (1981). Linguistic timing factors in combination. Journal of the Acoustical Society of America, 69(1), 262-274. http://dx.doi.org/10.1121/1.385347
; White & Turk, 2010White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
). It was also found that PS was more evident in pitch-accented vowels than unaccented ones (e.g., Kim & Cole, 2005Kim, H., & J. Cole (2005). The stress foot as a unit of planned timing: Evidence from shortening in the prosodic phrase. In Proceedings of Interspeech 2005, Lisbon, 2005, pp. 2365-2368. http://dx.doi.org/10.21437/Interspeech.2005-37
; White & Turk, 2010White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
). This type of shortening was generally believed
to be a component of the stress-timed rhythm of English in that it
results in more similar foot duration with regular timing between
interstress intervals (Lehiste, 1972Lehiste, I. (1972). The timing of utterances and linguistic boundaries. Journal of the Acoustical Society of America, 51(6B), 2018-2024. http://dx.doi.org/10.1121/1.1913062
; Port, 1981Port, R. F. (1981). Linguistic timing factors in combination. Journal of the Acoustical Society of America, 69(1), 262-274. http://dx.doi.org/10.1121/1.385347
).
However, other researchers argued that
vowel shortening could also be related to several factors including
word-initial/final lengthening, accentual lengthening, and demarcating
prosodic boundaries (Beckman & Edwards, 1990Beckman,
M. E., & Edwards, J. (1990). Lengthenings and shortenings and the
nature of prosodic constituency. In J. Kingston, & M. E. Beckman
(Eds.), Papers in Laboratory Phonology I: Between the Grammar and the Physics of Speech (pp. 152-178). Cambridge: Cambridge University Press.
; Turk & White, 1999Turk, A. E., & White, L. (1999). Structural influences on accentual lengthening in English. Journal of Phonetics, 27(2), 171-206. http://dx.doi.org/10.1006/jpho.1999.0093
; Turk & Shattuck-Hufnagel, 2000Turk, A. E., & Shattuck-Hufnagel, S. (2000). Word-boundary-related duration patterns in English. Journal of Phonetics, 28(4), 397-440. http://dx.doi.org/10.1006/jpho.2000.0123
; White & Turk, 2010White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
). Syllables in word-initial position tend to be
longer than their counterparts in non-initial position. For example, the
diphthong in the monosyllabic word ‘choir’ will be longer than that in
‘acquire’, all other things being equal, due to the effects of initial
lengthening. Similarly, final lengthening could increase the duration of
the vowel in the monosyllabic word ‘tune’, when it is word-final, but
not in the disyllable ‘tuna’, as the long vowel will not be in
word-final position and therefore less affected by final lengthening (Turk & Shattuck-Hufnagel, 2000Turk, A. E., & Shattuck-Hufnagel, S. (2000). Word-boundary-related duration patterns in English. Journal of Phonetics, 28(4), 397-440. http://dx.doi.org/10.1006/jpho.2000.0123
).
Accentual lengthening, where segments receiving a pitch-accent/phrasal stress undergo lengthening (White & Turk, 2010White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
), is greater in a monosyllable than in a disyllable, as in ‘knee’ and ‘kneecap’ (Turk & White, 1999Turk, A. E., & White, L. (1999). Structural influences on accentual lengthening in English. Journal of Phonetics, 27(2), 171-206. http://dx.doi.org/10.1006/jpho.1999.0093
). Furthermore, Turk and Shattuck-Hufnagel (2000)Turk, A. E., & Shattuck-Hufnagel, S. (2000). Word-boundary-related duration patterns in English. Journal of Phonetics, 28(4), 397-440. http://dx.doi.org/10.1006/jpho.2000.0123
reported that PS was greater in pitch-accented words, an indication that accentual lengthening and PS are interrelated. White and Turk (2010)White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
suggested that PS observed in polysyllabic words
might result from different degrees of accentual lengthening where it is
highest in monosyllables but attenuates in polysyllabic words. A third
possible reason behind PS is related to signaling word boundaries.
Evidence for using PS as a cue to prosodic boundaries comes from
languages with fixed word-initial stress such as Estonian and Finnish.
In these languages PS tends to be inexistent because prosodic boundaries
are already signaled by primary stress, and therefore there is no need
to use vowel length to demarcate word edges (Suomi, 2007Suomi, K. (2007). On the tonal and temporal domains of accent in Finnish. Journal of Phonetics, 35(1), 40-55. http://dx.doi.org/10.1016/j.wocn.2005.12.001
).
Although these factors suggest that PS can be related to aspects other than stress-timing, they do not refute the contention that PS can also be related to stress-timing tendencies. The purpose of this study is not to explain why this shortening happens; rather it aims to find the extent to which Jordanian Arab NNSs of English apply vowel shortening in their ILs.
Very few studies investigated PS in the IL of English NNSs. Krivokapic (2013)Krivokapic, J. (2013). Rhythm and convergence between speakers of American and Indian English. Laboratory Phonology, 4 (1), 9-65. http://dx.doi.org/10.1515/lp-2013-0003
comparing four American NSs with four Indian
speakers of English as an L2, found that the Indian speakers, who used
English all their lives, employed PS in an English nativelike manner. No
sufficient information about the participants was available to know the
degree of their bilingual status, which casts doubt on the validity of
the results. Dealing with bilinguals rather than foreign language
speakers, Gibson and Summers (2018)Gibson, T. A., & Summers, C. (2018). Polysyllabic shortening in speakers exposed to two languages. Bilingualism: Language & Cognition, 21(3), 471-478. http://dx.doi.org/10.1017/S1366728918000068
compared the use of PS in English and Spanish by a
group of English monolinguals and a group of balanced bilinguals in
English and Spanish repeating nonwords in the two languages. They found
that the English monolinguals did not implement English PS when
repeating the Spanish nonwords, while the bilingual group used a degree
of PS that was appropriate with the language in question. Both groups
produced more PS in English than in Spanish, which is a syllable-timed
language. To account for the lack of transfer of English timing patterns
into Spanish (which was expected as the English monolinguals’ level in
Spanish was limited, as adults), the researchers argued that this
unanticipated finding could relate to the group’s early exposure to
Spanish (around age 4), an exposure that could have given them the
ability to acquire motor plans that might have prevented transfer of PS
from L1 to L2. This does not seem to be the case. It seems that PS,
which is a characteristic of stress-timed rhythms, does not transfer to
L1s with syllable-timed rhythms at beginning levels as it is a marked
feature and L2 learners start with a syllable-timed rhythm regardless of
their L1 rhythm class, most probably due to lack of adequate
articulatory control, which improves as L2 proficiency increases (Ordin & Polyanskaya, 2014Ordin, M., Polyanskaya, L. (2014). Development of timing patterns in first and second languages. System, 42, 244-257. http://dx.doi.org/10.1016/j.system.2013.12.004
).
In another study tackling PS among bilingual children, Gibson & Bernales (2019)Gibson, T. A. & Bernales, C. (2019). Polysyllabic shortening in Spanish-English bilingual children. International Journal of Bilingualism, 24(2), 437-454. http://dx.doi.org/10.1177/1367006919846426
compared PS in Spanish and English bilingual
children with that in monolingual Spanish and English ones. They found
that both groups implemented PS similarly although Spanish is
syllable-timed, while English is stress-timed. They argued that lack of
differences between the groups could be attributed to a universal
phonetic constraint whereby speakers use one puff of air to produce a
string of segments and these segments tend to be shorter the more
segments or syllables there are (Quené, 2008, p. 1109)Quené, H. (2008). Multilevel modeling of between-speaker and within-speaker variation in spontaneous speech tempo. Journal of the Acoustical Society of America, 123(2), 1104-1113. http://dx.doi.org/10.1121/1.2821762
.
Studies dealing with the acquisition of
polysyllabic shortening by Arabic speakers of English are lacking.
Moreover, previous studies dealing with timing patterns in Arabic and
the ILs of Arabic speakers of English are rather limited. Abu Guba, Mashaqba, Jarbou & Al-Haj Eid (2023c)Abu
Guba, M. N., Mashaqba, B., Jarbou, S. & Al-Haj Eid, O. (2023c).
Production of vowel reduction among Jordanian Arab learners of English:
Acoustic investigation. Poznan Journal of Contemporary Linguistics. https://doi.org/10.1515/psicl-2022-2011.
compared the degree of vowel reduction as produced by Jordanian Arab
speakers of English and English NSs. They found that the Jordanian NNSs
of English, even the advanced ones, failed to produce English reduced
vowels in a native-like manner. Although the Jordanian NNSs were aware
that the reduced vowels were unstressed, they did not reduce the
duration of the reduced vowels, which were significantly much longer
than those produced by the English NSs. These findings seem to suggest
that Jordanian Arabic speakers do not use durational variation in their
speech as much as English NSs do. This seems to be related to the
mechanisms of Arabic speech rhythm, which is clearly different from that
of English (Abu Guba, Fareh & Yagi, 2023aAbu Guba, M. N., Fareh, S. & Yagi, S. (2023a). Arabic and English speech rhythms: A contrastive review. International Journal of Arabic and English Studies. Volume 23, Issue 1, Pages 183 - 202. doi: 10.33806/ijaes2000.23.1.10
). Put differently, it could be the case that the
syllable-timed Jordanian Arabic speech rhythm results in less variation
in timing patterns where vowel duration tends to be similar in feet
regardless of their segmental make-up.
This study will further investigate timing patterns by examining the extent to which Jordanian Arabic speakers use PS in their ILs. This will shed more light on Arabic speech rhythm and its mechanisms and on the nature of PS in general. Is PS a phonetic universal and to what extent is it related to the speech rhythm of a language?
3. METHODS
⌅3.1. Participants
⌅Three groups of participants, with no known speech or hearing disorders, took part in this study. Group 1 (n =10) comprised American English native speakers living in the United Arab Emirates at the time of recoding; all of them used General American English to record the sentences. Group 2 (n = 10) represented advanced Jordanian speakers of English; all of them received a degree in English language and literature (8 BA and 2 MA) from a university in Jordan. All of them studied in Arabic medium schools, and none had lived outside Jordan for more than a month. Group 3 (n = 10) consisted of intermediate Jordanian speakers of English. They were sophomore students studying English language at a Jordanian university at the time of recording. Again, all of them studied in Arabic medium schools and none lived outside Jordan for more than a month. For both groups of NNSs, they learned English in public Jordanian schools through formal instruction by Jordanian non-native speakers of English. None had travelled to an English-speaking country, and none had been taught by native speakers of English. Note that none of the NNS groups had received any phonological training; they only did a three-hour course in phonetics that focused on learning the IPA phonetic symbols. More details are provided in Table 1. Note further that none of the participants was aware of the purposes of the study.
Mean age and range | Mean exposure to English | Gender | |
---|---|---|---|
NS Group | 30 (23-36) | NA | 8 females and 2 males |
Advanced Group | 32 (26-38) | 22 years | 8 females and 2 males |
Intermediate Group | 20 (19-23) | 14 years | 8 females and 2 males |
3.2. Tools
⌅ Participants were requested to read three sets of words, given in (1),
in the carrier sentence “I say ___ twice” three times (the middle one
was analyzed). This is to minimize the effects of accentual lengthening,
and preclude the effects of initial lengthening, or final lengthening
(see White & Turk, 2010White, L., & Turk, A. E. (2010). English words on the Procrustean bed: Polysyllabic shortening reconsidered. Journal of Phonetics, 38(3), 459-471. http://dx.doi.org/10.1016/j.wocn.2010.05.002
; Turk, 2012Turk A. (2012) The temporal implementation of prosodic structure. In A. Cohn, C. Fougeron, M Huffman (Eds.). The Oxford Handbook of Laboratory Phonology (pp. 242- 253). Oxford, UK: Oxford University Press.
). In this context, the target word is expected to bear a nuclear pitch accent as it has the only new information (Port, 1981Port, R. F. (1981). Linguistic timing factors in combination. Journal of the Acoustical Society of America, 69(1), 262-274. http://dx.doi.org/10.1121/1.385347
). In a couple of instances, the nuclear pitch
accent did not fall on the target word and the participants were
instructed to repeat that, without bringing this to their attention.
Trisyllabic words were chosen carefully. Words with a potential stress shift as per Arabic stress rules (see Abu Guba, 2018Abu Guba, M. N. (2018). Stress assignment in polysyllabic words in Levantine Arabic: An Optimality-theoretic analysis. Lingua Posnaniensis, 60(2), 7-24.
,
for details on stress rules in Arabic) were excluded. It has been
observed that Jordanian learners of English tend to shift stress
rightward to closed peninitial syllables, as in ‘meaningful’, with
stress incorrectly assigned to the second syllable. Moreover, no
four-syllabic words with stress on the initial syllable are attested in
Jordanian Arabic; therefore, such four-syllable English words were
excluded because it would be impossible to control for possible stress
shift by Jordanian speakers. For example, in a word like ‘meaningfully’,
stress would shift to the second syllable according to Arabic stress
rules and so many Arab learners would stress the second syllable in
their ILs.
All the recordings were made in a quiet place using
an LG professional recorder at a 44k sampling rate. Upon completing all
the recordings, the researcher measured the duration of the long vowel
/i:/ in all the words using Praat 1.4.9 (Boersma & Weenink, 2015Boersma. P, & Weenink, D. (2015). Praat: Doing Phonetics by Computer [ Computer program]. Version retrieved July 2019 from http://www.praat.org.
).
This was done auditorily by listening to the sounds and visually by
inspecting the spectrograms and waveforms. The beginning of the vowel
was taken to be the beginning of formant structure and the periodic
waveform (Kent & Read, 2002Kent, R., & Read, C. (2002). The Acoustic Analysis of Speech. 2nd edition. Delmar: Cengage Learning.
).
Furthermore, the whole duration of each word was measured to find out
the duration of each foot and to calculate the percentage of the long
vowel in the whole word. In words ending in /d/, the burst phase of the
/d/ was taken as the right boundary of the word, i.e., to the beginning
of the hold phase of the stop /t/ in the word twice. For the words
ending in a vowel, the end of striations was taken as the end of the
word, and for words ending in /s/, the end of the noise of the sibilant
was taken as the end of the word (Kent & Read, 2002Kent, R., & Read, C. (2002). The Acoustic Analysis of Speech. 2nd edition. Delmar: Cengage Learning.
).
A random subset of the recordings was measured by a Jordanian Arab
phonetician and no discrepancy was found between the two transcribers.
4. RESULTS AND DISCUSSION
⌅Results show that there were clear differences between the NSs and the NNSs, but negligible differences between the two NNSs’ groups. In this section, we first compare the duration of feet across the three groups, and then we compare the vowel duration in the three types of words as produced by the three groups.
4.1. Foot duration
⌅ The NSs exhibited less variation in foot length than did the NNSs, with
feet becoming longer the more segments were added to the foot (Figures 1 and 2).
NSs’ feet in monosyllabic words were 367.3 ms; in disyllabic words,
402; and in trisyllabic words, 553 ms. That is, disyllabic feet were 35
ms longer than monosyllabic feet and trisyllabic feet were 186 ms longer
than monosyllabic ones. This runs against the isochrony view of feet in
English (e.g., Abercrombie, 1967Abercrombie, D. (1967). Elements of General Phonetics. Edinburgh: Edinburgh University Press.
).
Rather, these results lend support to the view that similar foot
duration in English, which is a stress-timed language, is a tendency
such that syllables do not have the same duration, with stressed
syllables being longer than unstressed ones, and compression of
syllables applying the more syllables there are in a foot (see Roach, 1982Roach, P. (1982). On the distinction between “stress-timed” and “syllable-timed” languages. In D. Crystal (Ed.), Linguistic Controversies (pp. 73-79). London: Arnold.
; Dauer, 1983Dauer, R. M. (1983). Stress-timing and syllable-timing reanalyzed. Journal of Phonetics, 11(1), 51-62. http://dx.doi.org/10.1016/S0095-4470(19)30776-4
).
Foot duration among the NNSs exhibited more variation as shown in figure 1. Although monosyllabic feet were close to the ones produced by the NSs, disyllabic and trisyllabic feet were considerably longer across the two groups of NNSs. Disyllabic and trisyllabic feet produced by the advanced group were about 75 and 91 ms, respectively, longer than those of NSs. Similarly, the intermediate group produced disyllabic and trisyllabic feet that were 88 and 108 ms, respectively, longer than feet produced by the NSs. Differences between the advanced and the intermediate groups were much smaller: 13 ms between disyllabic feet and 17 ms between trisyllabic feet. A mixed model ANOVA comparing foot duration across the three groups with the number of syllables as a fixed factor and word as a random factor, controlling for speaker and gender revealed a main effect for the interaction of group and number of syllables (F(4, 4.256) = 27.264, p = .003, ηp 2= 96, power = 1). Bonferroni post-hoc tests revealed that the differences between all the groups were significant; p was < .001 for all the differences between the NS and NNS groups while it stood at .033 for the difference between the advanced and the intermediate groups.
This shows that the NNSs produced much longer feet the more segments were added, which suggests that their IL is more syllable-timed than stress-timed, and they do not seem to use PS as much as the English NSs do (discussed in the following subsection).
4.2. Polysyllabic shortening
⌅In this subsection, we first compare the duration of the long vowel across the three groups and then we compare the percentages of the long vowel in the whole word to give a clearer picture and provide more evidence on PS across the three groups.
4.2.1. Long vowel duration
⌅Figures 3 and 4 give details on the duration of long vowels in monosyllabic, disyllabic and trisyllabic words. Overall, the vowel becomes shorter the longer the word is across the three groups; however, clear differences were attested among the NSs and the NNSs in the realization of long vowels in polysyllabic words.
Mean vowel length in monosyllabic words was very close across the three groups (around 150 ms). However, clear differences among the three groups in producing disyllabic and trisyllabic feet were attested. NSs produced shorter vowels in disyllabic feet (100 ms) and even shorter ones in trisyllabic feet (87 ms). The long vowel was 50 ms (33%) shorter in disyllables than in monosyllables and 13 ms (13%) shorter in trisyllables than in disyllables. This shows that NSs apply polysyllabic shortening, which is in harmony with the literature on PS in English (cf. Section 1).
By contrast, the NNSs did not shorten their vowels considerably in longer words with less than a 7-ms difference between vowels in monosyllabic feet and polysyllabic feet across both groups. Moreover, the vowels in polysyllabic words across the two NNS’ groups were much longer than those produced by the NSs (about 42 ms longer in disyllabic words and 52 ms longer in trisyllabic words), although the differences between the three groups in producing the long vowels in monosyllabic words were too small. The differences between the advanced and the intermediate groups were rather small: the long vowel in disyllables produced by the intermediate group was 12 ms longer than that produced by the advanced group, and the long vowel in trisyllabic feet produced by the intermediate group was 6 ms longer than that realized by the advanced group.
A mixed model ANOVA comparing vowel length across the three groups with the number of syllables as a fixed factor and word as a random factor, controlling for speaker and gender, revealed a main effect only for the interaction of group and number of syllables (F(4, 3.583) = 24.245, P = .007, ηp 2= 96, power = .98). Bonferroni post-hoc tests revealed that the differences between the NS group on the one hand, and the advanced and the intermediate groups on the other hand, were significant (p < .001). However, the differences between the advanced and the intermediate groups were not significant (p = .151).
4.2.2. Percentages of the long vowel in the whole word
⌅To control for tempo effects, we compare the percentages of the long vowel in the different words in this section. This is necessary as the differences in the previous section might be affected negatively by speaking rates that can be different across and within speakers. Figures 5 and 6 below show that the mean percentages of the long vowel in the whole word in monosyllabic words across the three groups were very close (around 40%). However, clear differences between the NS group and the other NNS groups with respect to the percentages in disyllabic and trisyllabic feet were attested, but very small differences between the NNS groups were attested. Although the advanced and the intermediate groups reduced the percentages of their vowels in disyllabic and trisyllabic words, their percentages were still higher than those of the NS group. They were 5% and 7% higher in disyllabic words, respectively, and about 7% higher in trisyllabic words for both groups. The differences between the advanced and the intermediate groups were less than 2%.
A mixed model ANOVA comparing the vowel percentages across the three groups with the number of syllables as a fixed factor and word as a random factor, controlling for speaker and gender, revealed a main effect for group (F(2, 8.329) = 5.228, P = .034, ηp2= 56, power = .67), and for the number of syllables (F(2, .823) = 655.821, P = .048, ηp2= 99, power = .72), but not for the interaction between groups and syllables (p= .615). This is because the differences between the three groups with respect to monosyllabic words were marginal, and the differences between the advanced and the intermediate groups were also very small. Bonferroni post hoc tests revealed that only the differences between the NS group on the one hand, and the other two NNS groups, on the other hand, were significant, while the differences between the two NNS groups were not significant (p= .494).
Taking all the results
together, we can clearly see that the NNS groups do not use PS as much
as the NS group do, an indication that the NNSs’ IL tends to be less
stress-timed, and the NNSs have not acquired the stress-timed rhythm of
the English language. This is in line with Abu Guba et al.’s (2023b)Abu Guba, M. N., Mashaqba, B., & Huneety, A. (2023b). Polysyllabic shortening in Modern Standard Arabic. Journal of Semitic Studieshttp://dx.doi.org/10.1093/jss/fgac030
finding that the degree of PS in Modern Standard
Arabic (the formal variety in Arabic) as produced by Jordanian speakers
is not comparable to that in English; Jordanian speakers produced vowels
in disyllabic words and trisyllabic words with similar durations, but
the vowels were shorter than in monosyllabic words.
Moreover, the present study’s findings do not agree with Gibson and Bernales (2019)Gibson, T. A. & Bernales, C. (2019). Polysyllabic shortening in Spanish-English bilingual children. International Journal of Bilingualism, 24(2), 437-454. http://dx.doi.org/10.1177/1367006919846426
, who argued that PS could arise from a universal
phonetic constraint that requires a faster articulation rate the more
syllables are added and that it is not strongly related to language
proficiency. The slightly better performance of the advanced group
(although statistically not significant except for foot duration) seems
to suggest that more experience with the language may have a positive
effect on the acquisition of timing properties of the English rhythm.
This is in line with the observation that L2 learners seem to adjust
their L1 motor plans in acquiring L2 phonology (Flege, Takagi & Mann, 1995Flege, J., Takagi, N., & Mann, V. (1995). Japanese adults can learn to produce English /ɻ/ and /l/. Language and Speech, 38, 25-55.
).
However, it seems the advanced group in this study did not acquire
English timing patterns adequately to block L1 transfer. This suggests
that such prosodic aspects are difficult to master, and they need more
efforts and time to acquire (cf. Abu Guba, 2021Abu Guba, M. N. (2021). Gemination within English loanwords in Ammani Arabic: An Optimality-theoretic analysis. Journal of Linguistics, 57(1), 1-38. http://dx.doi.org/10.1017/S0022226720000183
, and Abu Guba et al., 2023cAbu
Guba, M. N., Mashaqba, B., Jarbou, S. & Al-Haj Eid, O. (2023c).
Production of vowel reduction among Jordanian Arab learners of English:
Acoustic investigation. Poznan Journal of Contemporary Linguistics. https://doi.org/10.1515/psicl-2022-2011.
,
who reported that Jordanian-Arabic advanced speakers of English
performed better than intermediate speakers producing English prosodic
aspects but failed to perform in a near-native manner).
5. CONCLUSION
⌅It has been shown that the English NSs tend to employ PS more than do the two groups of NNSs. However, these results show that isochronous foot duration and polysyllabic shortening are a tendency in English speech timing, rather than a fundamental process. The performance of the advanced group was slightly better than that of the intermediate group, but the differences between the two NNS groups were not statistically significant except for the differences in foot duration. It seems that the NNSs still use the mechanisms of their L1 rhythm (which is less stress-timed than that of English (cf. Section 2). A future study that explores PS in native Jordanian colloquial Arabic words is recommended to find out the extent to which L1 transfer plays a role in acquiring this feature.
Findings suggest that acquiring the timing patterns of the stress-timed English rhythm is challenging to Arabic-speaking NNSs. It seems that the NNSs need to be made explicitly aware of these timing properties as a prerequisite to reduce the effect of L1 transfer and ultimately acquire these timing patterns. It is likely that explicit teaching and training on these aspects could be helpful here, which is left for future research. English language learners and teachers should pay more attention to timing patterns, which would help boost intelligibility and reduce foreign accentedness. They should work on mastering English durational variation quite early especially because Arabic is rhythmically different from English