Sub-dialectal Coronal and Non-coronal Assimilation in Yemeni Arabic

This is an investigation of the assimilation process across word boundaries of the four main dialects of Yemeni Arabic. Twenty native speakers of these dialects have been asked to read the collected data, five from each dialect. The reading of the phrases has been transcribed for the analysis. The data under investigation uncovers the fact that assimilation in the four Yemeni dialects can be classified into symmetrical and asymmetrical categories. Symmetrical assimilation does not lead to irregularity. However, asymmetrical assimilation can cause irregularity among the four Yemeni dialects. The asymmetrical category of assimilation is investigated in terms of place, voicing, and emphasis. These irregularities are analogous to earlier typological work undertaken by Lehn (1963), Mohanon (1993), Watson (2002), Zuraiq and Zhang (2006), and Zuraiq and Abu Joudeh (2013). The study provides informative phonetic data useful for the foundation of a cross-dialectal study of assimilation in YA consonants.


Introduction
Coronal (C) sounds as defined by Chomsky and Halle (1968) are consonants produced with the blade of the tongue raised from the neutral position (dental, alveolar, palato-alveolar consonants).Non-coronals are consonants produced with the blade of the tongue in the neutral position (bilabial and velar consonants).This phonological investigation provides an insight into symmetrical and asymmetrical assimilation that is a prototypical feature of four major Yemeni dialects: the San'ani dialect (SD), the Ta'izi dialect (TD), the Tihami dialect (THD), and the Hadhrami dialect (HD).Our analysis is hinged upon the production of native speaker informants of the four dialects.Twenty native speakers (10 males and 10 females) five from each dialect (SD), (TD), (HD), and (THD) have produced the required phrases and their production is recorded.The judgment of whether assimilation has occurred is made by listening carefully to the recording and transcribing the output allophonically.The participants have been asked to read the list of phrases at a normal speaking rate and also repeat the phrase three times.The present work explores the patterns of consonant assimilation in the four dialects.The investigation provides phrases that represent all coronal and non-coronal combinations across word boundaries.Given the fact that the four dialects differ in lexicon, other phrases are constructed to mirror the peculiar phonetic nature of the four dialects.It is worthwhile to note that the present phonological investigation is entirely based on the views of Lehn (1963), Al-Ani (1970), Card (1983), Mohanon (1993), Davis (1995), Jun (1995), Zawaydeh (1999), Watson (2002), Zuraiq and Zhang (2006), and Zuraiq and Abu Joudeh (2013).The views of other phoneticians have been consulted during our literature review.The names of these scholars are listed in the references.However, the views of Zuraiq and Abu Joudeh (2013) have impacted the investigation and the outcome of the present work.

Aim of the Study
This study aims to investigate the phonological process of assimilation: (i) symmetrical assimilation, and (ii) asymmetrical assimilation.The two types will be discussed in terms of coronal and non-coronal consonants.The current study provides an insight into (i) assimilation with regards to the consonant point of contact assimilation, (ii) the different states of the glottis, and (iii) emphasis in the four different types of Yemeni Arabic (YA).Another major goal of the present discussion is to provide wide-ranging data source for the behaviour of assimilation in YA.

Yemeni Dialects
Located in the southern corner of the Arabian Peninsula, Yemen is a country from which the Arabic language is historically known to have sprung.Based on phonetic features, we divide the dialects of Yemen into four major dialects: The SD which covers the northern part of Yemen from Sa'dah to Dhamar.
The TD covers the region of Ta'iz, Ibb, and Aden.The HD covers the western region which includes Mareb, Shabwa, Abyan, Hadhramout, and Al-Mahra.
It goes without saying that within each category a lot of differences can be traced.However, the most important distinctive feature is the similarity on the phonetic level.Behnstedt as cited in Versteegh (2004) differentiates the following main areas: the Tihama dialects; the k-dialects; the South-east Yemenite dialects; the dialects of the central plateau (e.g. the dialect of Sana'a); the dialects of the southern plateau; the dialects of the northern plateau; and the North-east Yemenite dialects.But even this subdivision is not a comprehensive depiction of the entire area there are many mixed zones, and some of the areas will probably have to be subdivided when more data become known.(where in SA) respectively.Other phonemes are pronounced more or less the same with some allophonic variations related to the degree of pharyngealization and allophonic variations.For example, /p/ is produced as [pʼ] before silence, and /tˁ/ is pronounced as a dark [đ] in SD. /θ/ is pronounced as [t] in Aden (a variation of TD) due to the influence of Indian languages./ʁ/ is produced as [q] in Abyan, (a variation of HD), and in Raymah ( a variation of THD).In Yafea and Abyan (HD) and in Raymah and Osaab (THD), /d/ is pronounced with the blade of the tongue against the postalveolar rather than alveolar ridge.

What is assimilation?
Assimilation is one type of phonological processes such as dissimilation, deletion, epenthesis, metathesis and vowel reduction (Dobrovolsky and Katamba (1996).For Crystal (1991:65), assimilation is the "influence exercised by one sound segment upon the articulation of another, so that the sounds become more alike, or identical".In the view of Abercrombie (1980), juxtapositional assimilations involves "changes in pronunciation which take place under certain circumstances at the ends and the beginnings of words (changes at word 'boundaries, that is to say) when these words occur in connected speech, or in compounds'.For example, the word "is" and the word [ʃi:] is pronounced [ʃi:]; but when these two words come together in the phrase (is she), they are often pronounced [iʃʃi:] rather than [iz ʃi:]" (ibid:136).Conversely, phonological assimilation which is a connected speech feature bordering on another juxtaposed feature in a segment in less careful speech has been examined as a purely linguistic phenomenon (Kohler 1991  Assimilation is not compulsory in many languages, including Arabic: that is to say a speaker may, if he chooses, avoid making them.However, this phonological process usually happens in connected speech by speakers without paying attention to their articulation.When they are made, however, they have the impact, regardless if they are regressive or progressive, of producing some economy of effort in the utterance of a sequence of words.The results of assimilation are to reduce the number, or the extent, of the movements and adjustments that the speech-producing organs have to perform in the transition from one word to the next.Assimilation saves effort by means of three different types of changes in the sequence of speech-producing movements.These different types of changes include: (i) the state of the glottis and movement of articulators, (ii) velic action, and movement of articulators, and (iii) pharyngealization and movement of articulators.

Categories of assimilation
There are two major categories of assimilation: (i) symmetrical and (ii) asymmetrical.These are discussed below.

/n/ cases
Assimilation in YA is an attested phonological process.YA is marked by the above three types of assimilation.However, assimilation which involves the movement of articulation is more common than the other types: (1) /man rabbuk/ → /marrabbuk/ (who's your Lord?) /man li:/ → /mal li:/ (who's my supporter?) The examples in (1) show that the sound /n/ loses its nasality and totally assimilates with the adjacent segment.Consider the following examples: (2) /man qallak/ → /maŋqallak/ (who told you?) The examples in (2) illustrate that the /n/ loses its nasality partially and becomes velar [ŋ] in the first example while it becomes interdental [n̪ ] in the second example.However, the uvular /q/ becomes more advanced and is influenced by the sound /ŋ/ in the first example, while the sound /θ/ becomes partially nasalized as a result of the adjacent /n/.Assimilation which involves successive movements of two different articulators is replaced by the movement of one articulator only.
Below is a discussion of assimilation of the sound /n/ in connection with the three different types mentioned above:

(i) Assimilation concerning the state of the glottis and movement of articulators
Let us consider the following data in (3): IJALEL 4(5):234-248, 2015

237
(3) Table 2. Assimilation of /n/ to coronals Based on the production of all informants of the four dialects, we can produce a phonological rule for the alveolar in complete assimilation with the dental sound, namely: when the voiced /n/ is followed by the voiceless /θ/, it loses its nasality, becomes voiceless, and completely assimilates with the dental sound.
Another phonological rule can be deduced from our informants about the alveolar /n/, namely: when the voiced /n/ is followed by the voiceless /f/, it loses its nasality, becomes voiceless, and completely assimilates with the labiodental sound.
Our informants have led us to a third phonological rule for the alveolar in complete assimilation with the postalveolar /ʃ/, namely: when the voiced /n/ is followed by the voiceless /ʃ/, it loses its nasality, becomes voiceless, and completely.

(ii) Assimilation concerning the velic action and movement of articulators
Let us consider the following data in (4): (4) Table 3. Assimilation of /n/ to dorsals A phonological rule can be deduced from our informants about the alveolar /n/, namely: when the voiced /n/ is followed by the voiced /g/, it completely assimilates and becomes velar.
Based on the production of all informants of the four dialects, we can introduce a phonological rule for the alveolar in complete assimilation with the voiceless velar sound, namely: when the voiced /n/ is followed by the voiceless /k/, it According to our informants, a third phonological rule can be presented for the alveolar /n/ in complete assimilation with the uvular /q/, namely: when the voiced /n/ is followed by the voiceless /q/, it becomes voiceless, and completely assimilates with the uvular sound.

(iii) Assimilation concerning pharyngealization and movement of articulators
Let us consider the following data in ( 5): (5) Table 4. Based on the production of all informants of the four dialects, we can produce a phonological rule for the alveolar /n/ in partial assimilation with the pharyngealized /tˁ/ sound, namely: when the voiced /n/ is followed by the voiceless /tˁ/, it becomes voiceless pharyngealized, and partially assimilates with the adjacent sound.
Another phonological rule can be deduced from our informants about the alveolar /n/, namely: when the voiced /n/ is followed by the pharyngealized voiceless / ðˁ/, pharyngealized, and partially assimilates with the adjacent sound.
Our informants have led us to a third phonological rule for the alveolar in partial assimilation with the alveolar /sˁ/, namely: when the voiced /n/ is followed by the voiceless /sˁ/, it becomes voiceless, and partially assimilates with the adjacent sound.

/al/ cases
(i) the sun /al/ Grammatically, the /al/ is a definite article.Phonetically, however, the /al/ is of two types.The first is called the sun /al/.The sun /al/ means the /l/ is subject to complete assimilation with the coronals.If the /al/ is followed by an interdental, alveolar, or a prepalatal sound, it completely assimilates with the adjacent sound.Also, the /l/ is dropped and the next consonant is geminated (doubled).For more details on gemination in TD, see Aldubai (2015).It is worthwhile to note that the /al/ must always be in syllable initial position.In other words, the /l/ is in complete assimilation with the coronal consonants in all the Yemeni dialects.
Let us consider the following data in (6):  The /l/ completely assimilates when followed by interdental, dental, alveolar, or postalveolar (coronal) sounds, whether these sounds are voiced or voiceless, provided that the /l/ is in the initial syllable position and receives primary stress.The assimilation causes gemination, i.e., doubling the consonant that replaces the /l/.

(ii) the moon /l/
The second is called the moon /al/.The moon /al/ means the /l/ is not subject to assimilation with the noncoronals.
The moon /l/ does not feature in all the dialects of the Yemen.However, according to the production of our informants from the inhabitants of the Sabir mountain which belongs to the Ta'iz province, the moon /l/ is in complete assimilation when it is followed by non-coronal sounds (except for the palatal sound) whether it is bilabial, alveolar, postalveolar, velar, uvular, pharyngeal, or glottal.In other words, the speakers of the Sabir area do not distinguish between the sun and the moon /al/; they only produce the moon /al/ except for the palatal.This is illustrated by the following data: The moon /l/ assimilation is an attested phonological process around the city of Ta'iz.This process can be attributed to the fact that the speakers of this dialect oversimplify the rule and assimilate all the consonants whether they are coronals or non-coronals.The only exception of this rule is when /l/ is followed by the palatal sound /j/ such as /aljawm/ and /aljaman/.In this case, the /l/ is not in assimilation with /j/ due to the fact that the /j/ is an adjacent semivowel to /l/.In terms of articulation, this makes it difficult to move from a vowel /a/ to a semivowel /j/.
The data in (8) below is a summary of all the four dialects with regards to the sun and moon /l/: (8) Table (7) displays data that show the definite article /al/ when it is followed by coronal or noncoronal sounds.This is in line with the analysis put forward by Kenstowicz (1994).However, in the THD, the sound /l/ is transformed to /m/ in the sun and moon /al/ when it is followed by any sound without exception.Changing /l/ into /m/ is not only specific to the people of Tihamah but it also features in the dialect of the tribe of Hashid to the north of Sana'a.Based on our YA informants, we can introduce a phonological rule, namely: the /t/ gets deleted when preceded by /n/ and followed by /d/, /θ/, /ð/.This is explained in the following examples:

Asymmetrical assimilation types
Asymmetrical assimilation is a phonological process where the dialects differ from one to another with regards to assimilation.In symmetrical assimilation, all the informants of the four dialects have made unanimous production of assimilation.However, in asymmetrical assimilation, some dialect informants assimilate while others do not.
The asymmetrical assimilation can be divided into (i) place assimilation, (ii) voicing assimilation, and (iii) emphasis assimilation.The three types are discussed in terms of coronals (C) and non-coronals (NC).

non-coronal & coronals (NC&C)
According to our informants, nasal and non-nasal bilabials and labiodentals non-coronals do not assimilate to coronals neither progressively nor regressively in SD, TD, HD, and THD as shown in (10)

non-coronals & non-coronals (NC & NC)
Place assimilation occurs when the two adjacent consonants are both non-coronals in TD and in THD.However, it does not occur in HD but in SD, a vowel epenthesis occurs across word boundaries as demonstrated by (12) below: ( The data in (13) show that a vowel is inserted between /ɡ/ and /ʁ/ in SD.The vowel epenthesis prevents the two sounds to assimilate.In TD and THD, however, a complete assimilation takes place where /q/ is changed to /ʁ/.
There is no assimilation in HD as the /q/ is pronounced as a uvular /G/.Similarly, /ʁ/ does not assimilate to the voiceless velar according to the production of all the four dialect informants as explained by ( 13): (13) /balliʁ kamaal/ [balliʁ kamaal] 'inform Kamal' However, some informants of THD and HD have changed /ʁ/ to /q/ as in the following example:
These phonological facts will be explained below.Examples in (24) below show that when two coronals disagree on sonorancy, no assimilation takes place.This applies to the four dialects: In (25), we can notice that when the adjacent coronals share sonorancy, place assimilation occurs, making the two coronals identical.Voiced alveolar plosive is in complete assimilation when followed by the voiced dental fricative as has been observed in the four dialects: Within stridents, alveolars assimilate to postalveolars, and vice versa.This has featured in all the four dialects.It is worthwhile to note that in TD and THD, the /ʤ/ is not a segment in the phonemic system of these two dialects.However, in SD, a vowel is inserted between the emphatic /sˁ/ and /ʤ/ which blocks the assimilation as illustrated in (32)  The /χ/ undergoes regressive assimilation and becomes voiced /ʁ/ as a result of the influence of the adjacent voiced /ʁ/ in all the four dialects.

coronals & non-coronals (C & NC)
Based on the production of our informants, voice assimilation between coronals and non-coronals does not occur in the four dialects.

non-coronal and coronals (NC & C)
By emphasis we mean the sounds that are pharyngeals or pharyngealized.The pharyngeals are /χ/ and /ʁ/ and the pharyngealized sounds are /sˁ, tˁ, ðˁ, đ/.Labials and dorsals are not considered by coronals for emphatic assimilation in all the four Yemeni dialects under investigation as illustrated by (36)  In TD, we have observed that the sound /sˁ/ in words like /sˁaʁiir/ is pronounced as /z/ [zaʁiir] by uneducated informants of this dialect.However, educated informants of TD pronounce /sˁ/ as it is pronounced in standard Arabic.We have also observed that in TD, our uneducated informants produce /tˁ/ as a depharyngealized [t].However, their educated

Flourishing
the western areas beginning from Hajja until Mocha.
and completely assimilates with the velar sound.
Assimilation of /n/ to pharyngeals

Table 1 .
Consonant phonemes of all Ds compared to Standard Arabic (SA)

Table 5 .
Assimilation of /al/ to coronalsBased upon the production of the YA informants, we can formulate the following rule:

Table 7 .
Assimilation of /al/ to coronals & non-coronals in Saber Mount (23)/n/ does not assimilate to /b/ in TD and THD dialects if the /n/ sound is preceded by a consonant; however, if /n/ is preceded by a vowel, it assimilates to /b/ (cf 16).In SD, a vowel is inserted between the two stop sounds.In HD, /n/ assimilates with /b/, when a vowel is inserted before /n/ as illustrated by (14) below: Although /n/ assimilates with the uvular /q/, it does not assimilate with the pharyngeals /ħ/ or /ʕ/ due to the fact that the pharyngeals cannot be nasalized.This is shown in (16) below:Finally, non-nasal coronal sonorants /l, r/ do not assimilate to a following non-coronal, as shown in(23)below: / ʔ ant ʁariib/ [ ʔ ant ʁariib] 'you are strange' When coronal fricatives are followed by non-coronals, they are not considered for place assimilation, as shown in (22) below: (22) / θ alaaθ banaat/ [ θ alaaθ banaat] 'three girls' /naas kaθiir/ [naas kaθiir] 'a lot of people' (28)in non-sonorants, the non-stridents /θ/ and /ð/ do not assimilate to the stridents /tˁ/, /ʃ/ /s/, /z/ and /sˁ/ in all the four dialects.Similarly, stridents /ʃ/ /s/, /z/ and /sˁ/ do not assimilate to non-stridents /θ/ and /ð/ as illustrated in(28)below: However, we have observed that the above process is not applicable to TD and THD as the affricate /ʤ/ is not part of the phonemic system in these two dialects as shown in the following examples: TD & THD:/mahallooʃ ɡanbi/ [mahallooʃ ɡanbi] 'He is not beside me'TD & THD:/maʔindanaaʃ ɡamal/ [maʔindanaaʃ ɡamal] 'there is no camel' Within non-stridents, the alveolar /t/ assimilates to the dental /θ/ in all the four dialects as illustrated in ( (35) produced by our informants have shown that the alveolar /t/ and /d/ assimilate to the pharyngealized /ðˁ/ but not the vice versa as shown in(34)below: Labials and dorsals are not targeted by coronals for voicing assimilation in all dialects.5.2.2.3 non-coronals & non-coronals (NC & NC)When two juxtaposed non-coronal sounds are different only in voicing, they are not considered for voicing assimilation.However, when they are different in any other features, voicing assimilation is allowed.The non-application of voicing assimilation is illustrated by examples from SD and HD.In SD, a vowel epenthesis blocks voicing assimilation; in HD, /q/ is pronounced as a heavy lenis /G/, which blocks assimilation to /k/.Voicing assimilation is shown in TD & THD as shown in(35)below: :/ ʕ ilaaʤ ʃaamil/ [ ʕ ilaaʃ ʃaamil] 'complete treatment'We have also observed that within non-strident obstruents, all segments are triggers and targets of total assimilation (place, voice, and emphasis) in all the four dialects except for HD as shown in (33) below.In SD, the /d/ is usually