Foreign Accent Syndrome Andforce of Articulation

Foreign Accent Syndrome andforce of articulation

INTRODUCTION

Some neurological syndromes, such as the Foreign Accent Syndrome, appeal so much to the imagination that newspapers report on the typical symptoms. These symptoms are often exaggerated or misunderstood in the non-medical press. OnSeptember 3rd, 2013,the British newspaper ‘Daily Mail’ published an article entitled: ‘The woman from Plymouth who woke up with a Chinese accent’. In this article, a 38-year old woman was described who could not speak for a while after a migraine insult. Later on, her speech developed into a ‘Chinese accent’.The phenomenon that someone after brain damage speaks with a strange accent is called Foreign Accent Syndrome (FAS).

FAS is an intriguing disorder, which was first described almost a century ago (Marie, 1907), yet it continues to be poorly understood. The FAS speaker’s inability to make phonetic and phonemic contrasts of his native language results in a perceived foreign accent in listeners (Gurd, Besel, Bladon, &Bamford,1988; Takayama, Sugishita, Kido, Ogawa, Akiguchi, 1993).The number of cases published is restricted, around one hundred until now. This suggests that the incidence is low. However, it may be that FAS largely escaped the attention of the research community. This can be due to two reasons: First, FAS often occurs as a transient stage in recovery or deterioration. Relatively few people haveapersisting FAS. Next, in cases with a persisting form, the accent is in general one of the only symptoms: intelligibility in daily living is not or hardly affected. As such,speakers with FAS tend not to be referred to speech-language therapists, and other physicians appear to assume there are no communication concerns. However, many reports allude to marked psychosocial consequences of living with FAS (Miller,2010).

The current study will describe a case study of a Dutch male with FAS, AA. We will try to show that the foreign accent of AA is due to the fact that he speaks with much more force of articulation than normal speakers of Dutch. Due to this force of articulation, characteristics of different languages become apparent.

Characteristics of FAS

The distinct speech characteristics of FAS described in the literature mainly concern segmental and prosodic deficits, although not necessarily both levels are affected. Among the segmental characteristics, vowels are more vulnerable than consonants. The vowel anomalies have shown a variety of patterns, including vowel tensing, and thus lack of vowel reduction in unstressed syllables (Van Lancker, Bogen, Canter, 1983; Whitaker, 1982; Blumstein,Alexander, Ryalls, Katz,Dworetzky,1987; Ingram,McCormack Kennedy,1992), schwas becoming more like full vowels, and monophthongisation of diphthongs, and a tendency towards fronting and raising (Whitaker, 1982; Gurd et al., 1988;Dankovičová, Gurd, Marshall,2001;DankovičováHunt, 2011). Other acoustic analyses, however, demonstrated an overall reduction in the acoustic vowel space, due to a restricted F1 range (Ingram et al., 1992; Moonis et al., 1996). With respect to vowel length, lengthening (Van Lancker et al., 1983; Graff-Radford, Cooper, Colsher,Damasio,1986, ArdilaRoselli, Ardila,1988; Perkins, Ryalls, Carson,Whiteside, 2010) as well as shortening have been reported (Pick, 1919). Nevertheless, acoustic data of some individuals with FASshowed normal vowel durations, although there is an overall greater variability in vowel formants (Blumstein et al., 1987). Alterations of consonants are also mentioned in the literature, including change of place of articulation and manner, andvoicing errors (Whitaker, 1982; Ardila et al.,1988; Gurd et al., 1988; Ingram et al.,1992).

In terms of prosodic characteristics, the feature most frequently reported in FAS cases is a change in prosody, specifically a tendency towards ‘syllable-timing’ in languages where ‘stress-timing’ isexpected (e.g. Blumstein et al.,1987; Gurd, Coleman, Costello,Marshall, 2001; Scott Clegg, RudgeBurgess,2006; Haley, Roth, Helm-Estabrooks, Thiessen, 2010). The perception of a ‘syllable-timed’ rhythm can be due to a number of characteristics: unusual equal syllable durations (isochronicity), for example, due to lack of vowel reduction (e.g. Whitaker, 1982) and the use of epenthetic vowels (Ardilaet al.,1988; Ingram et al., 1992).A recent study ofKuschmann and Lowit (2012) investigated intonation patterns to identify the characteristics that were compromised in FAS to their ability to signal new and given information (information status) within sentences. Results showed that speakers with FAS frequently place pitch accents on given information instead of de-accenting these elements.

Characteristics often resemble the characteristics of Apraxia of Speech (AoS), a motor speech disorder resulting from brain damage and different authors see FAS as a kind of compensation strategy in relation to an underlying (mild) form of AoS (MariënVerhoeven, 2007, Whiteside Varley, 1998; Moen, 2000; Miller, Lowit, & O’Sullivan, 2006; Fridrikssonet al., 2005; Kanjee, Watter, Sévigny,Humphreys, 2010; Roy, Macoir, Martel-Sauvageau, Boudreault, 2012). According to Varley, Whiteside, Hammil, and Cooper (2006) FAS and AoS are due to the same underlying deficit, but differ with respect to the possibility of a patient to compensate or control their speech problems. Characteristics of FAS sometimes also resemble characteristics of dysarthria. This holds, for example for the syllable isochronicity seen in FAS speakers, but also in dysarthric speakers (Kent Rosenbek, 1982). However, in dysarthric speakers articulatory deficits are caused by an impairment of the articulatory musculature. These deficits remain constant under all speech conditions, which is not seen in FAS.

Of course, listeners’ perception and familiarity with other languages plays an important role in perceiving a foreign accent. Results from studies on perceiving a foreign accent in healthy speakers (Flege, 1988, Di Dio, Schulz,Gurd, 2006) show that listeners can detect foreignness in milliseconds. However, a feature that is salient for one listener may not be for another (SouthwoodFlege, 1999) and furthermorethese salient features are interpreted through the listener’s experiential and attitudinal filters (Miller et al., 2006).Recently, Verhoeven, De Pauw, Pettinato, Hirson, Van Borsel,and Mariën(2013) even showed that the foreign impression of the listeners is different from the impression they get from real foreign accents. This means that we assume that the foreign accent in FAS must not be seen as the deficit itself. Rather the speech of a speaker with FAS is altered due to a brain deficit in such a way, that it no longer sounds like the original language of the speaker.We assume that the different aspects of the perception of FAS can be combined to one underlying characteristic, namely increased force of articulation. [1]

We will argue that listeners perceive a foreign accent in our FAS speaker because of a larger amount of fortition (force of articulation) in his speech. Synchronically as well as diachronically the term ‘fortition’ is used for segment changes, such as approximants changing into fricatives; fricatives into plosives and voiced obstruents into voiceless obstruents. Changes in the opposite direction are called ‘lenition’ (Lass, 1984). Fortition implies strengthening of the overall force of the sound (Crystal, 2008).[2]An increase in fortition will result in clear segmentation, lack of assimilation and lack of vowel reduction and this is exactly what has been reported for FAS. Laver (1994) also assumes that fortis sounds are characterized by a high general speech tension.

Our attempt to explain FAS is comparable to former attempts of Miller et al. (2006)Dankovicová and Hunt (2011) and Perkins and Ryalls (2013) who tried to explain the perceived foreignness of FAS speakers on the basis of vowel, consonant cluster and stress pattern changes playing an important role in making their accent sounding as foreign. Part of these characteristics will be used in our model as well, but we will try to extend it to characteristics of fortition in general.

We expect that the characteristics of fortis speech in FAS show that the direction in perception of an accent will go from a language with relatively more lenition, like Dutch, into a language with relatively less lenition, like German.Because this larger amount of force of articulation manifests itself in different aspects of AA’s speech, we assume that we are able to account for the variation in perceived foreign accents.One could assume that perception of an accent always moves towardsa language spoken by a larger community. However, Moonis, Svearer,Blumstein, Kurowski, Licho, Kramer, and Drachmon (1996), Coelho andRobb (2001), Hall, Anderson, Filley, Newcombe, and Hughes (2003), Blumsteinet al. (1987) all give examples of FAS in which an American FASspeaker is perceived as speaking with a French accent, although English is spoken by a larger number of people.

We will also try to show that it is not just the matter of one single aspect that could explain the outcomes, like for example stress-timed languages perceived as syllable-timed languages. For AA some speakers reported a French accent, which could be explained in this respect, but others reported a German accent, where Dutch and German are both stress-timed languages. Therefore, we will try to explain the perceived accents on the basis of a combination of phonological and phoneticaspects of the speech of our FAS speaker.We will try to show this, by investigatingthe effects of speech rate, stress patterns, vowel reduction, voice onset time, and assimilationin the speech of AA. Before we describe the characteristics of AA, we first will provide an introduction into somephonological aspects that are typical for the Dutch language.

Phonological aspects of Dutch

Stress pattern and vowel duration

Dutch, like German and English,has been described as a stress-timed language in which stressed syllables occur at even intervals (Rietveld Van Heuven, 2001). In these languages, stressed syllables are allegedly longer than unstressed syllables.Vowel reduction takes place in unstressed syllables. Abercrombie (1967), following Pike (1945), claims that in languages like French, Spanish and Arabicall syllables have approximately the same duration. These languages are called syllable-timed languages. Vowel reduction is less common in these languages.

Although Abercrombie’s controversial distinction of languages in stress-timed and syllable-timed languages has been challenged, for instance by Roach (1982) and Liberman (2008), it led to differences in prosodic organization of these languages. In Dutch, syllables are grouped into bounded trochaic feet in which the syllable in the weak non-initial position is often reduced to schwa. In a syllable-timed language like French, on the other hand, syllables are grouped in unbounded feet with final stress. All syllables have full vowels. This is for example illustrated in example 1. In (1a) the Dutch version of the word “onomatopee” is given, in (1b) the French one .

(1)Feet parameter

a. Bounded feet (Dutch): (o no)∑(ma to) ∑(pee)∑[ʔonəma:təpe:]

b.Unbounded feet (French): (o no ma to pee) ∑[ʔo:no:ma:to:pe:]

Voice onset time

Many language systems exhibit a contrast between so-called voiced plosives (/b,d,ɡ/) and voiceless plosives (/p,t,k/). The phonetic differences between these categories can be measured as differences in Voice Onset Time (VOT), i.e. the time when vocalcord vibration starts following the release burst of the plosive. However, the phonetic realization of this contrast in plosives is not the same for all languages. Dutch realises initial prevocalic plosives such as /b/ and /d/ with prevoicing: vocal cord vibration starts before the release burst of the plosive(Rietveld & Van Heuven, 2001, Jansen, 2004). In other words, VOT has a negative value. Other language systems, such as German, do not exhibit this prevoicing.[3] The opposition /b/:/p/ in these languages is accomplished by means of a longer VOT in the voiceless counterpart, which is perceived as aspirated. Jansen (2004) distinguishes the two manners of categorization as an opposition between negative and short lag positive VOT (Dutch), on the one hand, and as an opposition between short lag and long lag VOT (German, English), on the other hand. He defines short lag as a VOT between 0-36 msec. and long lag as a VOT larger than 36 msec. Thus, voiceless plosives in Dutch, such as /p/ and /t/, are realised with a short lag positive VOT, whereas voiceless plosives in German are realised with a long lag positive VOT. The longer the VOT, the more powerful the realization.Therefore, German exhibits more force of articulation in the realisation of these plosives than Dutch.

Assimilation

As mentioned above, force of articulation causes more segmentation and, therefore, less assimilation and co-articulation. Grijzenhout (2000) claims that the German word beweisbar(”provable”) is preferably realised as [bəʋɑisbaːr] with [sb], whereas the Dutch cognate bewijsbaaris realised as [bəʋɛizbaːr], with [zb]. Both German and Dutch colloquial speech exhibit optionally regressive (and progressive) assimilation of voice in obstruent clusters,but only in Dutch the assimilated forms are preferred. Both languages also exhibit regressive place assimilation of nasals before obstruents. Again, the realization of a word like [ʔɪŋkoːpən] for inkopen(”purchase”) is preferred to [ʔɪnkoːpən] in Dutch, whereas in German [ʔɛinkaufən] is preferred to [ʔɛiŋkaufən].We see that German exhibits less assimilation than Dutch (Grijzenhout, 2000)and, therefore, we conclude that German shows more force of articulation than Dutch with respect to the concatenation of segments.

Phonological aspects of Dutch and FAS speech

Only if we relate our phonetic analyses with the differences in phonological organization and preferences between languages, can weaccount for the variation in perceived accents in Dutch FAS speech. FAS speech is characterised by a lack of lenition processes, such as vowel reduction and assimilation. For example, lack of assimilation can be observed in the way AA realises a sentence such as ik ben alleen (”I am alone”) (obtained from an interview with AA). In this case, we observe a lack of regressive assimilation of voice: the /k/ of ik is expected to be realised as [ɡ] under influence of the right-hand neighbour /b/ in ben. In the upper graphs of Figure 1, a spectrogram including the pitch contour as well as an oscillogram of a control speaker (see Methods section) are shown. The vertical lines in the oscillogramindicate pulses. At these positions the vocal cords vibrate. The first two words are realized as [ʔɪɡbɛn]. As can be observed, the entire fragment is voiced. In the lower part of Figure 1, the spectrogram and an oscillogramof the realisation of the same two words by AA are depicted. Again the vertical lines in the oscillogramindicate pulses, the voiced parts of the fragment. When the graphs are compared, one can see the absence of assimilation in the fragment of AA, which may explain the perception of a German accent.

Figure 1 about here

METHOD

Participants

AA is a right-handed, monolingual Dutch male. At the age of 59, he presented with a temporally expressive aphasia and a right-sided hemi-hyperesthesia. The symptoms disappeared after an hour. Medical examination revealed a stenosis of the internal coronary artery. Seventeen days later the patient underwent surgery to removethe carotid stenosis, which was without complications. However, when AA awoke from the anesthesia, he noticed reduced muscular strength in his right leg and a complete paresis of the right arm. On the first day after surgery, the patient could not speak at all, later his speech improved. He was diagnosed with a stroke. When he spoke again,the employees of the hospital wondered from which country he came, as they heard him speaking with an unknown accent.

Neurological findings

Neurological examination showed a mild right-sided facial paresis, an almost complete paresis of the right arm graded as MRC 2, and a mild paresis of the right leg, graded as MRC 4. During neurological examination no signs of aphasia were present, only prosodic changes in his speech.

Magnetic resonance imaging (MRI) revealed an ischemic infarction in the left middle cerebral artery territory and tissue loss in the right occipital lobe consistent with a small earlier infarction. Areas involved were the left posterior supramarginalgyrus and the pre and post central gyrus. MRI data are shown in figure 2. Several months after strokea second MRI revealed gliosisin the affected regions.

Figure 2 about here

Neurolinguistic findings

A neurolinguistic analysis revealed that AA’s speech rate was reduced and his speech wassometimes halting. Speech and language assessment was donewith subtests of the Aachen Aphasia Test (AAT: Graetz, De Bleser,Willmes, 1992), and severalsubtests of the Dutch PALPA(Bastiaanse, Bosje,Visch-Brink,1995). The Boston Naming Test (Kaplan, Goodglass,Weintraub, 1983) was administered to assess word finding. Scores are given in Table 1.

Table 1 about here

The scores on all tests were within normal limits. AA had no word-finding problems, used no semantic paraphasias and spoke grammatically correct. Repetition of non-words with one or two syllables was good, but repetition of non-words with more than two syllables resulted in halteringspeech and slips of the tongue, followed by self-correction. Thishalteringwas less severe in repetitions of words. In spontaneous speech, prosodic changes included unusual patterns of timing and intonation. Reduced speech rate and less stress variation were noted for polysyllabic words. The intonation of some of his sentences was inadequate. In conclusion, no evidence of aphasia was found, but there were clear symptoms of FAS in AA’s speech. A diadochokinesis task, mentioned in the test for dysarthria and Apraxia of Speech (DYVA; Dharmaperwira-Prins,1998) was administered and this task was difficult to perform for AA, especially with regard to alternating diadochokinesis (/pataka/).Unfortunately, the DYVA is not a standardised or validated test. Although a valid test for Apraxia of Speech (AoS) was not available when AA was tested, clinical judgement gave the impression that AoS could not be ruled out. However, assessment of the RadboudDysartria Test (Knuijt,Kalf 2007) didnot reveal symptoms of dysarthria.

Control data

All data of AA’s speech that were used for the analyses were compared with the data of a matched control speaker: a 59-year old, monolingual Dutch speaker without a history of neurological illness and without speech and language problems. He lived for his entire life in the same area as AA and speaks with the same regional accent.

Materials and analysis

Qualitative analysis

33 first-year students of Linguistics and 19 second-year students of Speech Therapy with no or basics knowledge in speech and language deficits were asked to listen to a fragment of one minute of spontaneous speech of AA. This fragment did not contain reference to the diagnosis. The students were asked whether they assumed that the speaker had an accent and if so, what accent they heard. They were told that it was possible to decide that there was no specific accent, a local accent, a dialect or an accent from other languages. In case of an accent they had to write down what accent they heard.