2.2.1 General overview

Human speech is normally produced by pushing air through a set of cavities and articulators, as conveniently described by Kenneth Pike, for example. The human speech apparatus includes five cavities: oral, nasal, pharyngeal, pulmonic and esophageal (p. 85). The cavities can be either active or passive, depending on the sound in question (p. 86), but what is required is that some part acts as the initiator of the sound. In humans, the initiator used most commonly is the lungs, but the larynx, lips, tongue, and esophagus can also be used as the initiator(s). Most often the initiator is located inside the air chamber, e.g. the way the lungs are located beneath the oral and nasal cavities (p. 88).

However, sounds can be produced in other ways, and while most human speech sounds are made on a pulmonic egressive airstream, there are also sounds that do not require an air stream, e.g. gnashing or grinding of teeth (Abercrombie Reference Abercrombie1967: 24) or by moving the jaw from side to side, clashing or scraping the teeth together (Catford Reference Catford1977: 63), or simply opening and closing the lips (Pike Reference Pike1943: 103). Such sounds are called percussives (Pike Reference Pike1943: 103; Catford Reference Catford1977: 63). The ultimate goal of sound production is the creation of vibration in a transmitting medium (air in this case).Footnote ²

In human speech, Ladefoged & Maddieson (Reference Ladefoged and Maddieson1996: 77) summarize the situation thus:

Pulmonic egressive speech can be referred to as ‘modified breathing’ (Abercrombie Reference Abercrombie1967: 25; Ball & Rahilly Reference Ball and Rahilly1999: 20). Rosner & Pickering (Reference Rosner and Pickering1994: 2) and Clark, Yallop & Fletcher (Reference Clark, Yallop and Fletcher1995: 17) observe that no known language seems to use ingressive airflow as a distinctive feature. According to Davenport & Hannahs (Reference Davenport and Hannahs1998), among the six possible types of airstream mechanism, two are not found: ingressive pulmonic and velaric egressive, but, they say, ‘it is unclear why this is so’ (p. 9). More recently, Bloomer, Griffiths & Merrison (Reference Bloomer, Griffiths and Merrison2005: 234) observe that while ‘[m]ost speech sounds . . . are made using a pulmonic egressive airstream mechanism . . . [i]t is . . . possible to reverse the direction of airflow so that the air moves from outside the body down towards the lungs’.

The six possible air stream combinations are shown in table 1. As can be seen in this table, glottalic ingressive airstream is used in the production of implosives, and velaric ingressive airstream is used in the production of clicks.Footnote ³ The latter can also be found in English in the production of such non-speech sounds as tut-tut and tsk-tsk (Jespersen Reference Jespersen1922: 419; Mitchell Reference Mitchell1957: 22; Gimson Reference Gimson1994: 33; Abercrombie Reference Abercrombie1967: 31; Catford Reference Catford1977: 70, 72; Jackson Reference Jackson1980: 3; Roach Reference Roach1992: 20; Brinton Reference Brinton2000: 45; McMahon Reference McMahon2002: 25). However, Cruttenden (Reference Cruttenden2001: 32) points out that the ingressive pulmonic airstream may occur in English as ‘a common way of expressing surprise or pain’. He also mentions an ‘ingressive interdental fricative for hurt’ (Reference Cruttenden1986: 180), while Jackson (Reference Jackson1980: 3) mentions that ingressive lung air is sometimes used in ‘moments of tension’.

Table 1

Summary of airstream mechanisms (see also Ladefoged Reference Ladefoged1971: 23).

* John Ohala (p.c.) points out that certain kinds of non-speech acts can be executed this way, e.g. ‘squirting out a mouthful of water’.

Ingressive ‘clicks’, used for paralinguistic purposes, are also noted in French. Thus, Havet (Reference Havet1875) points out that ingressive t is used to express doubt; and ingressive palatal t expresses surprise, but can also be used to call horses; ingressive u and ou are kisses (normal and ‘gros’, respectively), while ingressive f expresses surprise and une satisfaction de gourmet (Havet Reference Havet1875: 221; see also Passy Reference Passy1895: 114; Reference Passy1912: 112).Footnote ⁴ Jespersen (Reference Jespersen1922: 419) mentions that an ‘inbreathed voiceless l’ is used to express delight in eating and drinking, something with which Passy (Reference Passy1895: 114; Reference Passy1912: 112) agrees; and Nyrop (Reference Nyrop1924) notes that inspiratory [f] signals doubt, hesitation, indifference, pain as well as ‘satisfaction du gourmet’ (p. 3).

Beach (Reference Beach1938: 74) points out that

Further, Vendryes (Reference Vendryes1950: 39) mentions that in French

Finally, an exhausitive account of paralinguistic click usage is provided by Gil (Reference Gil, Haspelmath, Dryer, Gil and Comrie2005), who – besides repeating after Jespersen (Reference Jespersen1922) that tut-tut and tsk-tsk occur in English – lists a variety of languages from all over the world where paralinguistic click sounds are used, both for affective meanings and to denote ‘yes’ or ‘no’ (e.g. in Hebrew and San'ami Arabic). Gil also provides typological distribution maps.

While most speech is produced on a pulmonic egressive airstream, the alternatives are used for other purposes. However, as Pike (Reference Pike1943: 89) pointed out, more sounds can be produced on lung air than with any of the other mechanisms. Concerning ingressive phonation, he has the following to say (pp. 32–33; footnotes removed):

Although these sounds do not constitute speech, he provides symbols for the annotation of ingressive sounds (Pike Reference Pike1943: 90) and even a (separate) symbol for the annotation of ingressive airstream (Pike Reference Pike1947: 6).

Why does speech so overwhelmingly make use of only one of these possibilities? Catford (Reference Catford1977: 67) provides an answer:

Catford (Reference Catford1977) also points out that pulmonic pressure proper differs between egressive and ingressive production. While speech normally lies in the range of 1 to 12 cm H₂O, the pulmonic initiator can generate up to 180 cm H₂O egressively (Reference Catford1977: 66). Ingressively, however, ‘the absolute maximum inspiratory pressure that can be generated in the vocal tract is somewhat lower than maximum expiratory pressure, being about 100 cm H₂O’ (p. 68).Footnote ⁷ He summarizes this comparison by concluding that (p. 68)

Catford (Reference Catford1988: 57) restates this by simply saying that the ‘shape of the glottis is such that it vibrates well under pressure from below, but not at all well under pressure from above’.

Not only is pulmonic ingressive phonation in general less sonorous and harsher-sounding than its egressive counterpart but it is also less suited to the production of certain specific sounds. As Abercrombie (Reference Abercrombie1967: 25) and Catford (Reference Catford1977: 68) pointed out, the English word yes is only produced ingressively on the ye part, the final s being egressive.Footnote ⁹ This is attributable to physiological-acoustic reasons. Catford (Reference Catford1988: 20–22) addresses this issue in some detail:Footnote ¹⁰

Ohala (Reference Ohala, MacNeilage and Kreidler1983: 192) provides an account of the differences between egressive and ingressive sound production in general, and describes the overall differences in sub- and supra-glottal pressure between the two initiation processes; he also discusses the shape of the vocal folds. Ohala points out that certain sounds, notably sibilants such as [s,ʃ], do not easily lend themselves to ingressive production. However, he also mentions that he can ‘phonate in falsetto voice about as well ingressively as egressively’ (p. 192).Footnote ¹² This is attributable to the fact that in falsetto production, ‘the vocal cords are considerably thinned and thus have a more symmetric coronal profile’ (p. 192).Footnote ¹³

Although there are different ways in which the human speech apparatus can be used to produce sounds, the overwhelming majority of sounds are produced on a pulmonic egressive airstream. The reason for this is attributable to phenomena such as breath control, the shape of the vocal cords, sub- and super-glottal pressure. We have also seen that even when speech is in fact produced on a pulmonic ingressive airstream, not all sounds can be produced, at least not ones with similar phonetic-acoustic properties.

2.2.2 Detailed description

This section gives an in-depth description of the physiological and acoustic detail of inspiratory vocalization.

Moore & Van Leden (Reference Moore and Von Leden1958) provided a detailed account of how the vibratory pattern of the inspiratory voice differs from that which occurs during exhalation based on high-speech cinematography of the larynx, and described the transition from exhalation to inhalation as a five-step process (p. 220):

1. 1. Vocal folds stop vibrating during the moment exhalation ceases

2. 2. During this transition the motionless vocal folds are partially abducted

3. 3. The arytenoid cartilages are abducted and remain so throughout inspiratory phonation

4. 4. Vocal folds begin to vibrate with a small lateral displacement at the edges

5. 5. This motion is followed by a sudden medial sweep of the folds that accomplishes the vibratory closure of the glottis

When these authors looked at a single vibratory cycle, they observed a ‘marked difference from the vibrations of expiratory phonation’ (p. 220), including ‘an unusually long opening phase and a very brief closing phase’, and also that it ‘is evident that all points on the glottal margins do not move at the same time, at the same rate, or the same degree’ (p. 221).

Hartlieb, Luchsinger & Pfister (Reference Hartlieb, Luchsinger and Pfister1960) compared expiratory and inspiratory phonation at low, middle, and high frequencies, and observed among other things that the inspiratory voice had a shorter closed phase.

Orlikoff, Baken & Kraus (Reference Orlikoff, Baken and Kraus1997) investigated F0, electroglottographic (EGG) and airflow patterns in sixteen men and women who alternated between expiratory and inspiratory phonation. They observed that F0 increased by 5.1 semitones (on average) during inspiratory phonation, that inspiration showed a more symmetrical pattern of vocal fold contact, that both jitter and EGG amplitude perturbation were significantly higher during inspiration, and that the airflow rate was on average 48.5% higher during inspiration. They also observed that both inspiratory pulse and falsetto could be produced by at least some of the subjects, ‘indicating some control over the mode and frequency of vocal fold vibration when driven by an ingressive airflow’ (p. 1838).

Hoole, Munhall & Mooshammer (Reference Hoole, Munhall and Mooshammer1998) investigated how tongue movements are affected by changes from egressive to ingressive airstreams. They observed ‘consistent reductions in forward tongue movement during consonant closure in the ingressive conditions’ (p. 131). They also observed that ‘respiratory activity for the ingressive tokens is rather vigorous compared to the egressive tokens’ (p. 142), thus confirming the observation made by Orlikoff et al. (Reference Orlikoff, Baken and Kraus1997) that airflow rate is higher during inspiration. (See also Hoole Reference Hoole, Hahn, Schneider and Hahn1999.)

Kelly & Fisher (Reference Kelly and Fischer1999) studied laryngeal adduction, F0 and intensity in egressive and ingressive phonation. They observed a decrease in vocal fold contact and an increase in F0 during inspiration, but no consistent differences with regard to intensity. They also pointed out that all subjects produced ingressive speech with a ‘breathy quality’ (p. 395), and that ingressive phonation is used ‘for artistic purposes such as jazz singing’ (p. 400).

Robb et al. (Reference Robb, Chen, Gilbert and Lerman2001) – who adhered to the Lehmann (Reference Lehmann1965) terminology of ‘reverse phonation’ – studied vowel articulation produced on egressive and ingressive airstreams. They repeated the observations mentioned above that F0 is higher in ingressive speech, but also found that F1 and F2 values varied (in both directions) as a function of the particular vowel being produced.

2.4.1 Inspiratory phonation in various pathological conditions

In a paper on spastic dysphonia, Critchley (Reference Critchley1939) discusses previous publications on dysarthria where the term inspiratory phonation was used to describe the condition in question. The characteristics of this particular voice condition are said to be voiced sounds ‘emitted in a peculiarly harsh and constrained fashion, the pitch being at times unduly low and at others excessively high’ (p. 96). However, the speech condition thus described is, in fact, not ingressive speech, but ‘[a]lthough the whole of the speech takes place during expiration, the sound-effect is unnatural and suggests the pictoresque though incorrect title of “inspiratory speech”’ (p. 96). However, Critchley does not entirely rule out the possibility that some such speech may indeed be produced on an inspiratory airstream, although he makes the (erroneous) observation that ‘only in whispered speech is it possible to speak upon an inspiratory phase’ (p. 102).Footnote ¹⁴ Another early source is Curry (Reference Curry1940), who mentions that ‘inspiratory voice may come also as a disorder or pathology of normal voice’ (p. 2). Even though the title of Critchley's paper includes the phrase ‘inspiratory phonation’, it seems likely that it did not refer to the use of pulmonic ingressive phonation, at least not in any important way.

Stewart (Reference Stewart1968: 153) mentions that ‘pulmonic ingressive fricatives’ were observed in deaf children.

Frint & Kelemen (Reference Frint and Kelemen1969) describe a case of inspiratory phonation with a psychogenic origin (which lasted for months), while Nwokah et al. (Reference Nwokah, Hsu, Davies and Fogel1999) describe inspiratory phonation as one of several ‘pathological’ deviations from normal laughter.Footnote ¹⁵

Coombes (Reference Coombes, Yule and Rutter1987: 358) mentions that some ‘cerebral-palsied children . . . sometimes phonate on an ingressive air stream’, and ingressive speech as part of the speech in speakers with cerebral palsy is also mentioned in Hewlett & Beck (Reference Hewlett and Beck2006: 29). Finally, O'Regan & Brown (Reference O'Regan, Brown, Cockerill and Carroll-Few2001: 60) point out that children with hyperkinetic bulbar palsy sometimes phonate on an ingressive airstream.

2.4.2 Inspiratory phonation as a part of stuttering

Early sources to mention ingressive phonation as part of stuttered speech are Viëtor (n.d.^{a, b}: 4; Reference Viëtor1884: 3), Zwaardemaker (Reference Zwaardemaker1909: 7) and Eijkman (Reference Eijkman1937: 12/1955: 11). More recently, Van Riper (Reference Van Riper1982: 129) describes sudden inhalations as one of the secondary behaviors among some people who stutter:

Also, Ball et al. (Reference Ball, Code, Rahilly and Hazlett1994: 79) report that one of their subjects exhibited ingressive speech:

It has been shown several times that people who stutter have reaction times longer than people who do not stutter on a number of tasks, be they speech-related or non-speech related (e.g. Dinnan, McGuiness & Perrin Reference Dinnan, McGuiness and Perrin1970; Cross & Luper Reference Cross and Luper1979, Reference Cross and Luper1983; Cross, Shadden & Luper Reference Cross, Shadden and Luper1979; Venkatagiri Reference Venkatagiri1981). A question that presents itself is whether inspiratory phonation is performed differently by people who stutter and people who do not stutter, and this issue has been addressed in a couple of studies.

Reich, Till & Goldsmith (Reference Reich, Till and Goldsmith1981) compared laryngeal and manual reaction times in adults that stuttered and adults that did not stutter, and found that people who stuttered performed significantly slower than people who did not stutter only on speech phonation tasks, while no reaction time differences were found on non-speech phonation tasks, which included inspiratory phonation. Till et al. (Reference Till, Reich, Dickey and Seiber1983) performed a similar study on adults that stuttered and adults that did not stutter, and obtained the same results, i.e. while people who stuttered had slower reaction times on speech phonation tasks, no such differences were found with regard to inspiratory phonation. This could, they argue, be taken as suggestive evidence that inspiratory phonation is not a normal mode of producing speech.

2.4.3 Psychiatric observations

Key (Reference Key1975a) mentions that ‘vocal behavior of mentally disturbed patients sometimes shows more use of ingressive air movement during speech’ and that ‘[p]sychiatrists have noted that a disturbed patient may speak for long periods of time with only ingressive air’ (Reference Key1975a: 69–70). However, there is no reference to any corroborating psychiatric work.

2.4.4 Ingressive speech as a tool and treatment

Ingressive vocalization has also been used as a tool within the medical discipline, both when examining patients and as a treatment for patients with some type of speech disorder. Already as early as in the 19th century, Merkel (Reference Merkel1866: 278–280) pointed out that inspiratory speech was useful in the treatment of certain speech disorders.

Powers, Holtz & Ogura (Reference Powers, Holtz and Ogura1964) used inspiratory phonation to examine the larynx and pharynx for medical reasons. They observed (p. 14) that inspiratory phonation was ‘of considerable value’ in this respect and that during inspiratory phonation, the ‘ventricle is widely dilated and the enlargement of the true cord is easily seen’ and that the ‘configuration of the medial aspect of the hypopharynx is easily identified’.

In a follow-up of the Powers et al. (Reference Powers, Holtz and Ogura1964) study, Lehmann (Reference Lehmann1965) pointed out that ‘[d]uring reverse phonation the main change is distention of the ventricles . . . Other changes are less striking’ (p. 216), and concluded that reverse phonation ‘is useful in determining the extent of vocal cord tumors, in differentiating exophytic from invasive tumors, and in deciding upon appropriate therapy’ (p. 221).

Réthi (Reference Réthi1965) used inspiratory–expiratory phonation to treat stuttering and dysphonia. Kollár (Reference Kollár1973) reported successfully treating hundreds of patients diagnosed with recurrent palsy, ventricular voice, vox fistulosa and other disorders by having them phonate on inspiratory airstreams. Kollár (Reference Kollár1973) also pointed out that some of his patients produced much better-sounding speech on ingressive airstream than they did on an egressive airstream. Inspiration can also be used constructively by people who stutter, as observed by Van Riper (Reference Van Riper1982: 130):

Harrison et al. (Reference Harrison1992: 375) report on a patient who uses inspiratory speech deliberately as a communicative tool:

The authors report that the patient is able to speak for prolonged stretches of time in this manner, with only a need to clear his throat with a cough or two every five or ten minutes. All in all, the patient has found a good means of keeping up his communicative abilities, and ‘has found IS more comfortable than ES’ (p. 381). (See Colton, Casper & Leonard (Reference Colton, Casper and Leonard2006: 354–355) for a recent discussion of Harrison et al. Reference Harrison1992.)

Reich & Till (Reference Reich and Till1983) examined reaction times in ten women with spasmodic dysphonia compared to ten normal-speaking women. They found no performance differences on the inspiratory phonation task and concluded that spasmodic dysphonic patients only exhibit longer reaction times on ‘speech-like or motorically complex’ tasks (p. 16), implying that inspiratory phonation is not ‘speech-like’.

Rammage, Morrison & Nichol (Reference Rammage, Morrison and Nichol2000: 167) simply observe in their chapter on voice therapy that ‘ingressive airflow techniques can be trained’.

There are other sources within voice therapy that mention inspiratory phonation as either a tool or a treatment. For example, Schulman (Reference Schulman and Stemple1993) indicates that this is still an acknowledged method within speech pathology.Footnote ¹⁶ For recent accounts of the use of ingressive phonation as a therapeutic tool, see Maryn, De Bodt & Van Cauwenberge (Reference Maryn, De Bodt and Van Cauwenberge2003), Finger & Cielo (Reference Finger and Cielo2007) and Shiba, Isono & Nakazawa (Reference Shiba, Isono and Nakazawa2007).

2.5.1 Formant production

Miller et al. (Reference Miller, Sulter, Schutte and Wolf1997) asked a singer to produce the vowels /i/, /u/, /a/ and /ɛ/ in three different ways: singing, with a vocal fry and on ingressive airstream. Using magnetic resonance imaging (MRI) and linear predictive coding (LPC) they compared the acoustic and spatial dimension of the vocal tract, and observed (pp. 8–9) that

Dedouch et al. (Reference Dedouch, Horáček, Vampola, Švec, Kršek, Havlik and Zolotarev2002) used ingressive phonation to estimate formant frequencies for the five short vowels in Czech. One of the authors, Jan Švec (p.c.), reports that ingressive speech was used simply because it made the estimation of the formant frequencies simpler, and that ingressive speech does not normally occur in Czech. Fuchs & Perrier (Reference Fuchs and Perrier2005) also used ingressive phonation when studying kinematic behaviour in speech production.

2.5.2 F0 declination and expanding rib cages

It has often been observed that there is an F0 declination in phrases and sentences, and a number of hypotheses have been proposed to explain this phenomenon. Dunn & Ohala (Reference Dunn and Ohala2000) hypothesized that F0 declination is a feature of planned speech, used as a means to help the listener, and set out to reassess the hypothesis that F0 declination is automatic and caused by the collapsing rib cage, which, during speech, creates an inward air flow. Dunn & Ohala asked ten speakers to produce a total of 260 scripted utterances, a half on an egressive airstream, and a half on an ingressive airstream. They found that 75% of the egressive utterances exhibited F0 declination, while 50% of the ingressive utterances showed F0 declination. The most important conclusion, however, was not the quantitative difference between the egressive and the ingressive utterances, but the fact that despite the expanding rib cage, half the ingressive utterances were still produced with F0 declination. A similar experiment was carried out by Ohala, Dunn & Sprouse (Reference Ohala, Dunn and Sprouse2004), and the same results were obtained. The authors take this as partial support for their claim that F0 declination is not automatic, but purposely imposed.

3.2.1 General observations and sundry animals

Segond (Reference Segond1848:14–16) mentions that dogs, foxes, cats, horses, donkeys and birds (e.g. chickens) employ ingressive phonation. Milne-Edwards (Reference Milne-Edwards1876–1877) – who seems to have read Segond – mentions that both the ass and the horse (pp. 600, 601) exploit ingressive airstream in their vocalizations, while Waller (Reference Waller1891: 149/1896: 153) points out that ‘animals such as the cat, pig, and ox, vocalise with inspiration’. Horses are also mentioned by Pressman & Kelemen (Reference Pressman and Kelemen1955: 547).

In his classic work, The Mechanism of the Larynx, Negus (Reference Negus1929: 366–367) includes a paragraph on ‘inspiratory phonation’:

Further, according to Negus ‘the Bush Dog appears to make sounds in this manner . . . A Horse and an Ass neigh partly in inspiration’ (Negus Reference Negus1929: 367, referring to Milne-Edwards Reference Milne-Edwards1876–1877). Negus also pointed out that ‘Cats purr continuously, both when blowing out and taking in the breath, with a partly closed glottis’ (p. 367), and this is also mentioned by Darwin (Reference Darwin1872: 129):

That cats purr on inhalation is a widespread observation, and is encountered both in the scientific literature and in popular accounts; early sources include Zwaardemaker & Eijkman (Reference Zwaardemaker and Eijkman1928), Eijkman (Reference Eijkman1937/1955), Moelk (Reference Moelk1944) and Hussel (Reference Hussel1949). Zwaardemaker & Eijkman (Reference Zwaardemaker and Eijkman1928: 14) and Eijkman (Reference Eijkman1937: 12/1955: 11) confirm also that the cat miaows ingressively, and Moelk (Reference Moelk1944) notes also that ingressive phonation in the cat is employed in a wide variety of sounds besides purring, with a variety of communicative functions.

Remmers & Gautier (Reference Remmers and Gautier1972) conducted an electro-myographic study of feline purring. More recent works include Bradshaw & Cameron-Beaumont (Reference Bradshaw, Cameron-Beaumont, Turner and Bateson2000: 71), Sunquist & Sunquist (Reference Sunquist and Sunquist2002), who also point out that purrs ‘are louder and rougher-sounding during inspiration than during expiration’ (p. 422), and Fisher (Reference Fisher2005: 41) tells us that purring occurs ‘both while the cat is breathing in and while it is breathing out’.Footnote ¹⁹ Sunquist & Sunquist (Reference Sunquist and Sunquist2002: 423) mention also the existence of an ingressive ‘vocalized gasp’ which is present in the leopard's sawing (i.e. roaring sequence).

3.2.2 Birds

As was observed by Negus (Reference Negus1929), songbirds sustain tones for very long stretches of time and one of the reasons that this is possible is because they produce sounds on both inspiration and expiration. This is addressed by Hunter (Reference Hunter1980: 101), who, in a study on the Eurasian nightjar (Caprimulgus europaeus), observes that

Contrary to Negus (Reference Negus1929), Hunter (Reference Hunter1980: 102) states that it is ‘widely accepted that avian vocalizations occur only during exhalation’, but he also points out that ‘this idea is largely based on a limited amount of work on dead birds’ (p. 102). He concludes ‘that nightjars may sustain long vocalizations by singing one part of their song, the short trill, during inhalation’ (p. 103).

Gaunt & Gaunt (Reference Gaunt and Gaunt1977: 15) observe that the chicken (Gallus gallus) ‘may reverse the flow of air by inhaling . . . may be accompanied by a distinct, sometimes quite loud, wheezing sound’. Hunter (Reference Hunter1980) follows Gaunt & Gaunt noting that the turtle dove (Streptorella turtur) produce parts of their vocalizations on inspiration, and he cites Kneutgen (1969),Footnote ²⁰ who hypothesized that the white-rumped shama (Copsychus malabaricus) probably vocalized during inhalation.

In a more recent study on the zebra finch (Taeniopygia guttata), Goller & Daley (Reference Goller and Daley2001: 2301) report that ‘some male zebra finches also sing particular syllables of their motif during inspiration’. Interestingly, it seems that this feature of male zebra finch singing is learned behavior, copied from ‘tutors’; Goller & Daley (Reference Goller and Daley2001: 2304) speculate that given the ‘growing evidence that songbird females exhibit a preference for complex songs’, perhaps inspiratory syllables are especially attractive.Footnote ²¹ Unlike the inspiratory vocalizations made by the nightjar, which were made during long stretches of song, the zebra finches produce their inspiratory vocalizations on short songs. From an acoustical point of view, inspiratory syllables are ‘acoustically distinct from all other zebra finch song syllables’ (Goller & Daley Reference Goller and Daley2001: 2301), having a high fundamental frequency. However, when Leadbeater, Goller & Riebel (Reference Leadbeater, Goller and Riebel2005) investigated whether ingressive sound production was more attractive to female zebra finches, they concluded (p. 909) that although ‘songs containing inspiratory phonation clearly differ both in performance aspects (syllable rate and motif length) and in structure from those that are wholly expiratory’ and that ‘inspiratory syllables are indicative of enhanced motor control . . . [s]ong motifs containing inspiratory phonation were not systematically preferred . . . implying that this unusual trait does not impact significantly upon song attractiveness in an intersexual context’.

Finally, Wild, Goller & Suthers (Reference Wild, Goller and Suthers1998) studied muscle activity during song in the zebra finch and the cowbird (Moluthrus ather), and concluded that both expiratory and inspiratory muscles are active during vocalization in both birds. However, it is not completely clear in their article whether actual singing took place during inhalation, as distinguished from their observation that inspiratory muscles were active during song.

3.2.3 Frogs

Turning to amphibians, Weber (Reference Weber1974) studied the bioacoustics of two types of painted frog, the Discoglossus pictus and the Discoglossus sardus. He found that several types of calls exploited inspiratory phonation, notably mating calls and release calls.

Schneider (Reference Schneider, Frisch, Ryan, Wilczynski, Hetherington and Walkowiak1988: 543) observes that ‘in male Bombina (Discoglossidae) the inspiratory airstream produces the calls, although the closely related genus Alytes vocalizes in the usual way, by expiration’.

Wilczynski, McClelland & Rand (Reference Wilczynski, McClelland and Rand1993: 428) mention that ‘H. microcephala usually produced its secondary notes as doublets, or “biphasic notes”’ (sonogram on p. 429), while Cameron & Cogger (Reference Cameron and Cogger1992: 18) refer to the ‘low soft biphasic notes’ of the Litoria genimaculata.

3.2.4 Monkeys and apes

Turning to our closest relatives, monkeys and apes, several sources mention ingressive call production. Tenaza (Reference Tenaza1989: 200) discusses langurs (Simias concolor), endemic to the Mentawai islands off the west coast of Sumatra:

Richman (Reference Richman1976: 719–720) mentions that gelada monkeys ‘produce long strings of alternations of egressive and ingressive phonations’ (see also Richman Reference Richman1978: 571, 573; Reference Richman1993: 722), and includes spectrograms of ingressive phonation (Reference Richman1976: 720–722; Reference Richman1978: 574, 576).

Itani (Reference Itani1963: 15) mentions that the ‘wild Japanese monkey’ produces ‘sounds uttered in inhalation’ and that in one group studied ‘are included both sounds uttered by the exhaling breath and by the inhaling breath’ (p. 19).

That various species of gibbons exploit ingressive call production is mentioned in a range of papers by Thomas Geissmann and colleagues (Geissmann Reference Geissmann1984: 219, 222, 231; Reference Geissmann1991: 360; Reference Geissmann and Wallin2000: 113 et passim; Geissmann & Dallmann Reference Geissmann and Dallmann2002: 25; Geissmann & Nijman Reference Geissmann and Nijman2006: 5, 9):

Interestingly, the ingressive parts of the calls are also produced ‘approximately one fifth higher’ than the egressive parts (Geissmann Reference Geissmann1984: 219), as has been pointed out for human sound production by e.g. Robb et al. (Reference Robb, Chen, Gilbert and Lerman2001) (see section 2.2.2 above). Spectrograms of ingressive sound production can be found in Geissmann (Reference Geissmann1984: 27; Reference Geissmann and Wallin2000: 114).

Orangutans are also known to produce ingressive sounds. Already in 1922, Hornaday mentioned that under excitement ‘the orang can produce a roar by inhalation’ (Hornaday Reference Hornaday1922: 61). Dixson (Reference Dixson1998: 49) mentions that the orangutan's long call ‘consists of a series of loud moaning vocalizations, interspersed with “bubbling sounds” produced by inhalation’. Geissmann (Reference Geissmann and Wallin2000: 113, 115), citing a number of works, also mentions ingressive sound production in the orangutan (Pongo pygmaeus) calls. A sonogram of an orangutan ‘inhalation sound’ is given in Davila Ross & Geissmann (Reference Davila Ross and Geissmann2007: 313), who also mention (p. 314) that orangutans ‘often purr during inhalation’.

Turning to chimpanzees, an early source that mentions ingressive calls is Marler (Reference Marler1969), which refers to grunts that ‘alternate rapidly with “panting” sounds during inhalation’ (p. 97). The paper also includes (p. 96) spectrograms of inspiratory pant grunts. Goodall (Reference Goodall1986) reports several cases of inspiratory vocalization in the sound production of chimpanzees, notably laughter (p. 130) and general ‘pant-hoots’, which include a variety of subclasses such as ‘inquiring pant-hoots’, ‘arrival pant-hoots’, and ‘roar pant-hoots’ (p. 134; see also Marshall, Wrangham & Arcadi Reference Marshall, Wrangham and Arcadi1999: 826). More recently, Notman & Rendall (Reference Notman and Randall2005: 180, 187) confirm inhalation in the ‘build-up phase’ of chimpanzee (Pan troglodytes) pant-hoot, which ‘consists of a series of shorter exhalation/inhalation, or “panting” elements’ (p. 180).Footnote ²² Berntson et al. (Reference Berntson, Boysen, Bauer and Torello1989: 777) mention ‘ingressive voiceless noise ranging from 1 to 6 kHz with four to six periodic clicklike pulses between 6.5 and 7.5 kHz’ as one the traits of chimpanzee laughter. Arcadi (Reference Arcadi1996) mentions that the build-up phase of chimpanzee pant hoots are ‘voiced on both inhalation and exhalation’ (p. 163), and that ‘the inhalations between the concluding elements were loudly voiced’ (p. 169), and he provides a spectrogram that shows ingressive phonation (p. 170). Marler (Reference Marler and Wallin2000: 42) provides a spectrogram of a chimpanzee pant hoot, and the caption text points out that sound ‘is produced on both inhalation and exhalation’. Similarly, Falk (Reference Falk and Wallin2000: 210) mentions that chimpanzee calls are ‘voiced on both inhalation and exhalation’. Geissmann (Reference Geissmann and Wallin2000: 114) provides a spectrogram of a chimpanzee pant hoot that includes ingressive stretches.

Geissmann (Reference Geissmann and Wallin2000: 116–117) points out that biphasic notes have been reported for many different species of monkeys and apes, including Cercocebus, Lophocebus, Macaca silenus, Papio, Presbytis potenziani:thomasi, Simias concolor, Trachypithecus, Cercocebus galeritus, Trachypithecus johnii, and Semnopithecus entellus, but, interestingly, have not been observed in the gorilla (p. 115–116).

Finally, primate researcher Roger Fouts mentions that ‘chimpanzee laughter occurs on the inhale and exhale, while ours is primarily done on our exhales’ (quoted in Johnson Reference Johnson2007: 61).

Summing up, it is clear that ingressive sound production is very common in the calls produced by monkeys and apes, however intriguingly ‘absent in gorillas’ (Geissmann Reference Geissmann and Wallin2000: 116).Footnote ²³

3.3.1 Infant cry

Infant cry is partly characterized by inspiratory phonation. Already Zwaardemaker (Reference Sipma1909: 7) mentions that small children produce ‘sons inspiratoires’ [inspirated sounds]. Negus (Reference Negus1929), citing Darwin (Reference Darwin1872), pointed out that ‘the sobbing in infants is produced “at the moment when the inspiration conquers the resistance of the glottis, and the air rushes into the chest”’ (Negus Reference Negus1929: 367), and Trask (Reference Trask1999: 9) also notes that ingressive airstream is found in ‘child sobbing’. Bosma (Reference Bosma and Jacob1964: 105) mentions that ‘commonly the laryngeal vocalization is continued longer than expiration, so that the vocalization is carried over into the succeeding inspiration’. Other early sources which mention ingressive phonation as a normal part of infant cry are Truby & Lind (Reference Truby, Lind and Lind1965) and Bosma, Truby & Lind (Reference Truby, Lind and Lind1965). Bosma et al. (Reference Bosma, Truby, Lind and Lind1965) made the same observation, and pointed out that the ‘pitch of the ingressive phonation is almost invariably higher than the pitch of the associated egressive phonation’ (p. 72), an observation echoed by McMahon (Reference McMahon2002: 25) in a discussion of voice disguise. Truby & Lind (Reference Truby, Lind and Lind1965: 35) also mention that

Golub & Corwin (Reference Golub and Corwin1982) and Corwin & Golub (Reference Corwin, Golub, Milunsky, Friedman and Gluck1985) – who listed inspiratory phonation as one of four acoustic types of phonatory behavior that infants exhibit during crying, viz. phonation, hyperphonation, dysphonation and inspiratory phonation (Corwin & Golub Reference Corwin, Golub, Milunsky, Friedman and Gluck1985: 166). Grau, Robb & Cacace (Reference Grau, Robb and Cacace1995) – focus on the least researched of these four types (i.e. inspiratory phonation) and discuss its anatomical and physiological bases. They observe that ‘F0 of inspiratory cries was found to be significantly higher in comparison to pre-inspiratory and post-inspiratory cries’ (p. 378), for which they offer a range of alternative explanations, mainly physiological. They also point out that their study was based on newborns and that one might predict a decrease in inspiratory cry phonation beginning at 4–6 months of age. Stark (Reference Stark, Fletcher and Garman1986: 167) mentions that vegetative sounds are equally likely to be ingressive or egressive, while Papoušek, Papoušek & Koester (Reference Papoušek, Papoušek, Koester, Izard and Read1986: 111) list ingressive phonation as one of the characteristics of ‘maximal displeasure’ cry. Other sources that include mentions or analyses of inspiratory phonation in infant cry (from various perspectives) are Drachmann (Reference Drachmann, Brettschneider and Lehmann1980), Oller & Eilers (Reference Oller, Eilers, Papousek, Jürgens and Papoušek1992: 182), Protopapas (Reference Protopapas1993), Robb & Goberman (Reference Robb and Goberman1997), Robb, Goberman & Cacace (Reference Robb, Goberman and Cacace1997), Gierut & Champion (Reference Gierut and Champion2000), Leonard (Reference Leonard2000, Reference Leonard, Fletcher and MacWhinney1997), Koopmans-van Beinum, Clement & van der Dikkenberg-Pot (Reference Koopmans-van Beinum, Clement and van den Dikkenberg-Pot2001), Lederman (Reference Lederman2002), Solé (Reference Solé2002), Branco, Behlau & Rehder (Reference Branco, Behlau and Rehder2005), Branco et al. (Reference Branco, Fekete, Rugolo and Rehder2007) and Ertmer, Young & Nathani (Reference Ertmer, Young and Nathani2007).Footnote ²⁵ Finally, Nathani, Ertmer & Stark (Reference Nathani, Ertmer and Stark2006) mention ingressive phonation as part of level four (out of five levels) in infant speech development (Reference Nathani, Ertmer and Stark2006: 367–368).

Inspiratory phonation in infants can also be due to pathological conditions or acquisition problems in speech development. Ingram & Terselic (Reference Ingram and Terselic1983) and Ingram & Ingram (Reference Ingram and Ingram2001: 275) discuss a child that produced ingressive instead of egressive [s], but who dropped this phonological behavior after intervention. Gillis, Schauwers & Govaerts (Reference Gillis, Schauwers, Govaerts, Schauwers, Govaerts and Gillis2002: 26) discuss infants with cochlear implants, and point out that for ‘vocal development the distinction between cry and non-cry sounds [in infants] is crucial. The latter sounds are egressive such as normal speech sounds, while the former are ingressive as well as egressive’. Chin (Reference Chin2003: 857–858) mentions an infant with a cochlear implant who produced the ingressive nasals [m^↓] and [n^↓] in free variation with their egressive counterparts. (For a recent and succinct review of ingressive, or ‘non-pulmonic-egressive’, speech in clinical data, see Ball & Müller Reference Ball and Müller2007.)

3.3.2 Adult sounds

Several non-speech sounds are produced on an ingressive pulmonic airstream in adults. Already Segond (Reference Segond1848: 3–4) mentioned that ingressive voice was used in laughter, crying and hiccoughs, as well as in speaking. Bell (Reference Bell1867: 64) mentions that the ‘lips closed and opened with accompanying inspiration’ constitutes an ‘interjection of pain’. Trall (Reference Trall1875: 24) mentions sighing, groans, smacking and clicking, while Kitchen (Reference Kitchen1885: 56) lists yawning, hiccoughs, sobbing, laughing, crying and chuckling'. Kitchen (Reference Kitchen1885: 69) also mentions that ‘especially in women, the effect of a sudden fright . . . usually causes the emission of a scream . . . produced by the inspiratory current’. Passy (Reference Passy1895: 114; Reference Passy1912: 112) mentions that ingressive [f] signals pain, while ingressive [l] signals pleasure. Other early sources that mention that ingressive speech is possible include Turner (Reference Turner1857: 99), who mentions that ‘sighing consists of a long and forcibly drawn inspiration’. ‘Sighing’ is also mentioned by Pressman & Kelemen (Reference Pressman and Kelemen1955: 547). Milne-Edwards (Reference Milne-Edwards1876–1877: 490), von Meyer (Reference Meyer1884) and Negus (Reference Negus1929: 367) list ‘Hiccough, Gaping, and Stammering’ as inspiratory ‘disturbances’ (von Meyer Reference Meyer1884: 174). Voegelin & Voegelin (Reference Voegelin and Voegelin1959: 11) mention that ‘inspiratorial vowels’ are used to produce ‘folkloristic and emotional exclamations, and to even longer stretches of speech in situations of excitement’.

Moore & Von Leden (Reference Moore and Von Leden1958: 220) observe that ‘inspiratory vocalization can be produced by anyone with a normal larynx, and it is heard frequently in laughter, crying, rapid emotional speech, and children's vocal play’.Footnote ²⁶ Abercrombie (Reference Abercrombie1967: 24) informs us that yawning and snoring might occur on ingressive airstream – also mentioned by Viëtor (n.d.^{a, b}: 4; Reference Viëtor1884: 3) and Key (Reference Key1975a: 69) – as can sobs (Abercrombie Reference Abercrombie1967: 25; also Viëtor, n.d.^{a, b}: 4; Reference Viëtor1884: 3). We also saw earlier that laughing, crying and rapid counting can be produced ingressively (Collins & Mees Reference Collins and Mees1981: 26; Crystal Reference Crystal1987: 125; Rogers Reference Rogers2000: 255; MacMahon Reference MacMahon, Aarts and McMahon2006: 363), and that ingressive production is also encountered when the speaker expresses pain (Klinghardt Reference Klinghardt1914: 30; Bhaskararao 1976: 14; Crystal Reference Crystal1997: 197; Anderson Reference Anderson1998: 124) or surprise (Crystal Reference Crystal1997: 197), or when simply wanting to be silly (Roach Reference Roach1992: 54). That ingressive phonation is part of laughter is also mentioned by Richman (Reference Richman1976: 720).

Peterson & Harary (Reference Peterson, Harary and Jakobson1980: 146) mention that both vowels and consonants can be produced ingressively, and Johnson (Reference Johnson1998: 78) mentions that ingressive phonation occurs in English and German when speakers try ‘to talk while out of breath, crying, or laughing’; Collins & Mees (Reference Collins and Mees1981: 26) mention ‘sobbing, or out of breath’ as well as when ‘counting quickly’.Footnote ²⁷ Hewlett & Beck (Reference Hewlett and Beck2006: 29) note that ingressive phonation occurs in ‘running out of breath, or speaking while laughing’, and Otis (Reference Otis, Mathew and Sant'Ambrogio1988: 530) also notes laughter, as well as ‘whooping cough’.

Orlikoff et al. (Reference Orlikoff, Baken and Kraus1997: 1838) observe that ‘inhaled breath . . . is, in fact, not an uncommon form of phonation, usually manifest as a brief audible gasp in vocal expressions of surprise, fear, or joy’, but also that it ‘is not surprising that anecdotal evidence points to phonation on inspired air as generally difficult, uncomfortable, and prone to elicit involuntary coughing’.

Having listed these non-linguistic and paralinguistic uses of ingressive phonation, let us turn to speech proper. After all, as Smalley (Reference Smalley1963: 490) mentions,

3.3.3 Ventriloquists and (other) shamans

Negus (Reference Negus1929: 367) mentions that ‘ventriloquial voice’ can be produced ingressively. It is possible that this information was obtained from Milne-Edwards (Reference Milne-Edwards1876–1877: 490), who also remarked that ingressive airstream is employed by ventriloquists: ‘Beaucoup des effets acoustiques produits par la ventriloquie sont obtenus ainsi’ [Many of the acoustic effects produced by ventriloquism are obtained thus [ingressively]].

The word ‘ventriloquism’ comes from Latin ventriloquus, literally meaning ‘belly speaker’, which in turn is derived from the Greek word engastromythos (Connor Reference Connor2000: 50); it was long believed that ventriloquists in fact spoke through their bellies. Lipton (Reference Lipton1983) and Findley (Reference Findley1998) point out that ventriloquism is a very old phenomenon and that throughout history has mainly been used by shamans and priests to make people believe that either statues or effigies, or even the dead, were speaking.Footnote ²⁸ For example, Lipton (Reference Lipton1983: 604) mentions that ventriloquists ‘believe that the miracles of early oracles such as that of Delphi were probably utterances of priests referring ventriloquial sounds to statues’. Findley (Reference Findley1998: 191–192) notes that Inuit shamans ‘made extensive use of ventriloquism’ (see section 4.1 below, on throat games).

Another idea that was proposed to explain ‘belly speaking’ was that ventriloquists achieved the effect by speaking ingressively. The first person to suggest that ventriloquists spoke on an ingressive airstream was most likely van Helmont (Reference Helmont1657: 22), but it is Amman (Reference Amman1694, Reference Amman and Rieber1873) that this theory is most commonly attributed to.Footnote ^29, Footnote ³⁰ Amman, whose main interest was in helping deaf people learn to speak, also claimed to have heard ingressive voice used, by both deaf speakers and ventriloquists (Reference Amman1694: 90–91; Reference Amman and Rieber1873: 131).

The first person to question the van Helmont/Amman ‘ingressive theory’ was, in all likelihood, von Haller (Reference Haller1761: 434), but well into the 19th century many, if not most, researchers held that at least part or most of the ventriloquists's speech was produced ingressively. Thus, people like Nol[l]et (1754–1765: 470), Arnott (Reference Arnott1831: 515) and Liskovius (Reference Liskovius1846: 50) supported the ingressive speech explanation of ventriloquism, while others, like Richerand (Reference Richerand1837: 300) rebutted it.Footnote ³¹ It is of interest to note, however, that the ingressive theory was believed to be true as late as in Pressman & Kelemen (Reference Pressman and Kelemen1955: 547).

Müller (Reference Müller1843: 704) writes thus:

Silliman (Reference Silliman1861 284) mentions that ventriloquism ‘is supposed by many investigators to consist chiefly in the use of inspiratory sounds; this is true only to a certain extent’, and Merkel (Reference Merkel1866: 279–280) mentions that it was previously believed that ventriloquists made use of ingressive phonation, but that this method ‘in neuerer Zeit, nur selten von den Bauchrednern verwendet würde’ [in more recent times only rarely is used by ventriloquists]; see also Merkel (Reference Merkel1857: 640).

The history of various theories of how ventriloquists produce their voices is perhaps most succinctly summarized by Kodman (Reference Kodman1955: 1066):

Recent sources, e.g. Lipton (Reference Lipton1983) or Findley (Reference Findley1998) do not mention ingressive speech as part of the production of ventriloquial voice. Rather, the focus is on speaking without moving the lips or jaws, or to divert attention towards a puppet or similar device, and is ‘based on acceptance of an illusion’ (Lipton Reference Lipton1983: 602).Footnote ³²

However, modern ventriloquists confirm Merkel's (Reference Merkel1857, Reference Merkel1866) claim that certain effects in ventriloquism are indeed produced ingressively. Stephen Knowles (p.c.; http://www.stephenknowles.us/) reports (with ample demonstration) that he uses ingressive phonation to achieve certain effects, like infant cry, dog barks, pig squeals or other animal sounds (including donkey and horse sounds). Swedish ventriloquist Fredrik Eklöv (p.c.) reveals that he personally never speaks on inhalation airstream since it is hard on the vocal cords, but that he knows that other ventriloquists do it, and that many books mention it. Both Knowles and Eklöv believe that ingressive production probably was more common in earlier times when the ventriloquists' goal was to make their audience believe that the voices either came from statues, the dead, or another dimension, since the ingressive voice has an ‘other-worldly’ quality to it.Footnote ³³ It is not hard to imagine that a prophecy from the Pythea at Delphi (the oracle) given in ingressive voice would have sounded more impressive than the same prophecy given in ‘normal’ voice.

That audiences were indeed religiously affected by ingressive phonation is also pointed out by Voltaire (1734: 21–22; 1880: 90), who, in his Troisieme lettre sur les Quakers (Voltaire 1734: 17–25; 1880: 88–94), mentions that the Quaker George Fox impressed his proselytes by speaking on ingressive airstream:

It is not absolutely clear whether ‘inspiration’ here refers to phonetic or religious inspiration, of course. However, Segond (Reference Segond1848: 7) and Milne-Edwards (Reference Milne-Edwards1876–1877: 490) both take it for granted that Voltaire is referring to ingressive voice; the following is from Segond (Reference Segond1848: 7):

Perhaps the strongest indication that this interpretation is in fact the correct one is that Voltaire refers to the Pythea at Delphi, at the time regarded to be a ventriloquist, and consequently an ingressive speaker.

4.1.1 The katajjaq of the Inuit (Canada)

According to Charron (Reference Charron1978: 245),

Katajjaq (katajjait in plural) can be seen as some form of recreation and is commonly performed to celebrate a successful hunt or to greet visitors from other villages. According to Charron (Reference Charron1978: 246), anyone can play, ‘but in actual practice, it seems that the performance is mostly an activity of women and some children’. A katajjaq is normally performed by two persons sitting very close to each other and facing each other. Sometimes the acoustics are enhanced through resonators such as oil drum lids, cooking pots or bread pans. On the performance of katajjait, Charron (Reference Charron1978: 246–247) states:

And on the linguistic material used (p. 247):

Interesting as it is in and by itself, our specific interest in this work is the fact that katajjaq has as one of its features inhaled sounds, i.e. ‘performers are not merely breathing in and out; rather, their inspiration and expiration is realized at specific intonation levels. . . . Instead of using breathing only as punctuation or as a means for filling the lungs and activating the vocal cords, the Inuit also use it melodically’ (Charron Reference Charron1978: 251). Or, more succinctly, a ‘discernible, regular feature is the use of sound generated both by inspiration and expiration of the airstream’ (p. 252). Further mention of inspirated pitches in katajjait are found in Cavanagh (Reference Cavanagh1976: 45), Beaudry (Reference Beaudry1978: 269), Nattiez (Reference Nattiez, Falck and Rice1982: 134; Reference Nattiez1983b: 457, 460) and Bagemihl (Reference Bagemihl1988: 2, 4–5, 7).

On how to classify katajjait, Bagemihl (Reference Bagemihl1988: 2) states that the katajjaq ‘can and should be considered a primarily linguistic rather than musical system’. However, he continues (p. 7):

4.1.2 The rekkukara of the Ainu (Japan)

Throat games are also found in Japan, among the Ainu people. Their particular form of throat games is called rekkukara, and Collaer (Reference Collaer1956: 203), in his work on sixteen Japanese folksongs recorded between 1947 and 1951, includes a transcription of rekkukara (recorded in 1951):

Nattiez (Reference Nattiez1983a: 33) compared the katajjaq and the rekkukara, and stated that the two forms ‘can be considered as two manifestations of the same circumpolar music culture’. He also pointed out (p. 34) that the rekkukara has become extinct, since the ‘last interpreter of rekkukaras died in 1976’. The most striking similarity between the katajjaq and the rekkukara, according to Nattiez (p. 36), is that they ‘both systematically use a combination alternating between voiced and unvoiced sounds, inhaled and exhaled’.

4.3.1 Tsou in Taiwan

Fuller (Reference Fuller1990) studied the speech of a Taiwanese speaker from the southern Taiwan village of Punguu, where the language Tsou is spoken. The subject's speech included two pulmonic ingressive allophonic fricatives, viz. [f↓] and [h↓], both of which appeared before stops. Fuller observed that these sounds had never been described in the literature on Tsou. He also pointed out (p. 11) that whenever a pulmonic ingressive is produced, ‘there is a relatively long time period between the pronunciation of the ingressive phone and the onset of the following egressive phone’.

Fuller's claim is contested by Ladefoged & Zeitoun (Reference Ladefoged and Zeitoun1993). They report from a visit to Punguu, where they had 14 subjects record all the words (one of which was not known to any of their speakers) listed in Fuller (Reference Fuller1990). The speakers included the parents of Fuller's subject. For both /f/ and /h/, Ladefoged & Zeitoun observed only egressive productions in all speakers, and consequently concluded that Fuller's speaker must be special, and not representative of the language Tsou. Ladefoged & Zeitoun conclude (p. 14) that ‘it is absolutely clear that these sounds do not occur in the speech of other Tsou speakers’.

If Ladefoged & Zeitoun are correct, then Fuller (Reference Fuller1990) has only shown that it is possible to have ingressive allophones as part of a speaker's idiosyncratic speech behavior. Which, of course, is interesting in itself.Footnote ³⁹

4.3.2 Damin (Lardil) in Australia

As we have seen, there is strong evidence to discard the claim that there are ingressive phones in Tsou. There was, however, a language where an ingressive phoneme did exist – the ritual language Damin used in Australia.

Hale (Reference Hale and Sebeok1973; see also Hale Reference Hale, Grenoble and Whaley1998: 205) described the language Lardil, spoken on Mornington Island in the Gulf of Carpentaria, south of New Guinea, in Australia. Although Lardil is of interest in its own right, what we will focus on here is a variety of Lardil – or ‘auxiliary form of speech’, to use Hale's (Reference Hale and Sebeok1973: 442) wording – called Damin.

When Lardil men underwent subincision they also learned a special secret language, Damin. The vocabulary of Damin was small, perhaps around 250 words, and it was said that it could be learned in a single session. The basic grammar was identical to Lardil, but with regard to phonology Damin differed not only from Lardil but also from any other known language. Damin included several sounds not found anywhere else in Australia, such as nasalized clicks /m!/, /nh!/, an apico-alveolar /n!/, and an ejective bilabial stop /p'/.Footnote ⁴⁰ However, most interesting of all, an ingressive lateral fricative /L/ was also included, not found in any other known language (Hale Reference Hale and Sebeok1973: 443). Catford (Reference Catford1977: 65) commented thus:

Dixon (Reference Dixon1980: 218), in his work on Australian languages, reached a similar conclusion:

The last fluent speaker of Damin died more than a decade ago (Waugh Reference Waugh1995), so this language can no longer be studied.Footnote ⁴¹

4.4.1 Latvia

Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 56) report the use of ingressive ‘yes equivalents’ in Latvian.

4.4.2 Estonia

Kalevi Wiik (p.c.) asserts that ingressive speech is used in Estonian. Diana Krull of Stockholm University confirms (p.c.) that it is as common as it is in Swedish (i.e. very common). Ingressive speech in Estonian is also mentioned in Shorrocks (Reference Shorrocks and Tristram2003: 377, 387), including the word jah ‘yeah’. Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 56) also report the use of ingressive ‘yes equivalents’ in Estonian.

4.4.3 Russia, Armenia and Georgia

Rousselot (Reference Rousselot1924: 489) mentions that ingressive speech has been demonstrated in Russian, and among the people in Armenia (p. 489) and Georgia (p. 491). Liston (Reference Liston1971: 481 et passim) mentions that interjections in Russian are often produced with ingressive features, including ‘[h], [w], [ψ]’ [sic], and others. Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 56) also report the use of ingressive ‘yes equivalents’ in Russian spoken in the St. Petersburg area.

4.4.4 Finland

Peters (Reference Peters1981: 4) mentions that Finnish ‘has a bilabial pulmonic’ ingressive. Wiik (Reference Wiik1981: 14) mentions that both the word juu ‘yes’ in Finnish and short phrases, especially at the end of utterances, are produced on ingressive airstream. Wiik suggests that this habit is more common in female speech, and perhaps also when the speaker is in a ‘tired state of mind’.Footnote ⁴⁶ Hakulinen (Reference Hakulinen and Ivars1993: 54–55) lists the following items joo, mm, nii ‘yes’, on ‘to be 3sg’, ei ‘no’, just and jaaha ‘oh yes, allright’. Abondolo (Reference Abondolo1998: 9), referring to Laver (Reference Laver1994: 170), mentions that ‘[l]ong stretches of Finnish . . . are often spoken on an ingressive pulmonic airstream’. Ursin (Reference Ursin2002: 17) mentions that ‘ingressive air flow to the lungs is utilized’ for speech production in Finnish. Shorrocks (Reference Shorrocks and Tristram2003: 377) points out that ingressive speech in Finnish is ‘relatively common’ and ‘seems to be associated with informal speech, but is construed as being standard rather than dialectal’. Lennes Mietta (University of Helsinki) confirms (p.c.) the items above, and adds instances like ai kauhee and herrajumala ‘oh my God!’. Finally, Forsblom-Nyberg (Reference Forsblom-Nyberg1995) reports ingressive speech in Finland Swedish.

4.4.5 Sweden

Elert (Reference Elert1989: 84) observes that the word ja ‘yes’ is sometimes produced on inspiratory airstream in Swedish, and that this is especially common among women. He also mentions that this is very striking for foreigners.

Key (Reference Key1975a: 69) mentions that ‘Scandinavians, especially the women, sometimes say the word ja ‘yes’ with the breath drawn in- -often unvoiced’, and again (p. 150) that ‘the women but not the men, express agreement by articulating ‘ja’ with air drawn in’. Peters (Reference Peters1981: 1) also acknowledges that Swedish makes use of paralinguistic ingressive airstream, noting that there is ‘at least one’ variant of ingressive in Swedish (p. 4).

Extensive research on Swedish dialog has been carried out by Jens Allwood and colleagues at Göteborg University. Since their aim is primarily interactional, it comes as no surprise that they include references to Swedish ingressive feedback signallers in several of their works. For example, Allwood (Reference Allwood1981: 33) points out that

Allwood (Reference Allwood, Linell, Adelswärd, Nilsson and Pettersson1988: 97) lists several other items that are produced ingressively, all variants of either ‘yes’ (ja, m, a, ɑɑː, ha, ɑː, ʲɑ, jo, j, ho, o) or ‘no’ (nej, nä). Some of these words/sounds occur both in voiced and unvoiced versions. Further mentions and lists of ingressive feedback are given in Allwood (Reference Allwood1981: 33; Reference Allwood and Allwood1985: 33; Reference Allwood, Taylor, Néel and Bouwhuis2000: 19; Reference Allwood2001: 33); Allwood & Strömqvist (Reference Allwood, Strömqvist, Strömqvist and Strömqvist1988: 88); Allwood & Björnberg (Reference Allwood and Björnberg1999: 7); Nivre, Ahlsén & Allwood (Reference Nivre, Allwood and Ahlsén1999); Allwood et al. (Reference Allwood, Björnberg, Grönqvist, Ahlsén and Ottesjö2000, Reference Allwood, Ahlsén, Björnberg, Nivre and Allwood2001a, Reference Allwood, Grönqvist, Ahlsén, Gunnarsson, Kuppevelt and Smith2001b: 3).

Ohala (Reference Ohala, Hardcastle and Marchal1990: 23) observes that ja ‘yes’ and nej ‘no’ are sometimes produced ingressively in ‘Scandinavian’ (which is likely to include Danish). Laver (Reference Laver, Aronoff and Rees-Miller2000: 162) mentions the paralinguistic use of ingressive speech ‘in Scandinavian cultures’. Landqvist (Reference Landqvist2001) includes descriptions of inspiratory ja ‘yes’ and mm ‘yes’, and points out that the most common place for it to occur is as the third ‘yes’-turn, to close a topic. Landqvist (Reference Landqvist1999) also includes the version hh ‘yes’. Lindahl (Reference Lindahl2001: 6, 18, 24, 48) lists ingressive ja and mm (both meaning ‘yes’) but erroneously claims (p. 24) that they are always voiceless.

Eklund (Reference Eklund2002), in a comparison between human–human and human–machine dialogs, observed the items ja ‘yes’, mm ‘yes’, a ‘yeah’, nej ‘no’, jorå ‘affirmative yes’, jadå ‘indeed’ and bra ‘good’. He also observed that ingressive ja constituted about 10% of all ja, which made it as frequent as egressive nej ‘no’, med ‘with’ or inte ‘not’ in his data, corresponding to the 33rd most common item (token) overall. Shorrocks (Reference Shorrocks and Tristram2003: 378) mentions ingressive ja in Swedish and observes that it is ‘emphatic’.

Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 52, 54 et passim) list å herre gud det var ju läckert ‘oh my God it's fantastic’, så sprang ‘then ran’, ja, jo, mm ‘yes’, nej, nä ‘no’.Footnote ⁴⁸ Gunnel Melchers (p.c.) also reports that some female students say their names (both first names and surnames) ingressively when they register at the Department of English at Stockholm University.Footnote ⁴⁹

Salö (Reference Salö2007) focusses on the use of ingressive jo (‘yes’ variantFootnote ⁵⁰), often referred to as ‘Norrland vacuum cleaner’,Footnote ⁵¹ given its alledged frequency in Norrland (the northern part of Sweden). Salö analyzed 46 instances of jo, interviewed local radio stations throughout Sweden and concluded that the frequency of jo actually increases the further north in Sweden one goes, confirming popular perceptions. It must be borne in mind, however, that interviews are not a totally reliable data source, and that the conclusion might be based on popular perceptions rather than confirming them.

4.4.6 Norway

The oldest source found that mentions ingressive speech in Norwegian is Broch & Selmer (Reference Broch and Selmer1938: 8). It includes the following passage:

According to Peters (Reference Peters1981: 1), Norwegian makes use of paralinguistic ingressive airstream. He notes that there are ‘three, perhaps four’ ingressive variants in Norwegian (p. 4) and that some Norwegians use ingressives when speaking English to him (p. 84), and reports people using ingressives when speaking to animals, viz. a parrot, a dog, a mouse and a rabbit (p. 93). As previously mentioned (for Swedish), Ohala (Reference Ohala, Hardcastle and Marchal1990: 23) observed that ja ‘yes’ and nej ‘no’ are sometimes produced ingressively in ‘Scandinavian’.

Laver (Reference Laver1994: 169; see also Laver Reference Laver, Aronoff and Rees-Miller2000: 162) observes that

Finally, Kobayashi (Reference Kobayashi2001) devotes her entire M.A. dissertation to the use of ingressive ja ‘yes’, jo ‘affirmative yes’ and nei ‘no’ in Norwegian.

4.4.7 Denmark

Jespersen (Reference Jespersen1926: 124–125) mentions inspiratory phonation in Danish:

Almost exactly the same passage appears, in Danish, in Jespersen (Reference Jespersen1966: 78), without the annotation symbol, however. Jespersen also mentions there that ingressive sounds are used to express happiness, especially over good food and drinks.

Fischer-Jørgensen (Reference Fischer-Jørgensen1962: 20) provides the following information:

Peters (Reference Peters1981: 1) mentions that Danish makes use of paralinguistic ingressive airstream, of which there are two variants (p. 4). As was previously mentioned (for Swedish and Norwegian), Ohala (Reference Ohala, Hardcastle and Marchal1990: 23) observes that ja ‘yes’ and nej ‘no’ are sometimes produced ingressively in ‘Scandinavian’.

Stølen (Reference Stølen, Bucholtz, Liang, Sutton and Hines1994: 668) mentions that the ‘particle Ja’ is ‘a typical listener signal of assent . . . pronounced with an ingressive pulmonic airstream mechanism’. She also mentions that it is commonly used in telephone conversations, ‘as a listener's way of compensating for the absence of visual cues’ (p. 671). In a second study, Stølen (Reference Stølen1995) returns to the use of the ingressive affirmative particle ja in informal conversation. She concludes (pp. 222–223) that ‘significantly more ingressive particles were used by women than by men in the role of the listener and by female than male speakers in utterances in which the particle was used in an inner-directed way’.

Grønnum (Reference Grønnum1998: 8) reports the following:

Finally, Laver (Reference Laver1994: 169) mentions that an ingressive ja is used to express sympathy, agreement or commiseration (as in Norwegian), while an ingressive nei can also be used to express sympathy (see also Laver Reference Laver, Aronoff and Rees-Miller2000: 162).

4.4.8 Germany

Winteler (Reference Winteler1876: 5) mentions two different usages of inspiratory speech in German, voice disguise and the affirmative particle:

Sievers (Reference Sievers1885: 23) mentions that, in German, the particles ja and juch are sometimes produced ingressively, and that more rarely also the particle so (which is then pronounced ho (cf. section 2.2.1). Interestingly, he also mentions that whistling can be produced ingressively. Sacleux (Reference Sacleux1905: 35) states that ya in German is sometimes produced ingressively, and Passy (Reference Passy1912: 112) mentions ja ‘yes’ too, without specifying which country/language the ja is from, however. Forchhammer (Reference Forchhammer1924: 142, 156) mentions that words like ‘najn’ nein/‘no’ and ‘javōl’ jawohl/‘yes’ are sometimes produced ingressively, and he proposes a phonetic character to indicate ingressive production (see appendix B).

Pitschmann (Reference Pitschmann1987) studied the use of ingressive airstream in German and the Scandinavian languages. Although his study focused on Icelandic, he mentions that both ja ‘yes’ and nein ‘no’ are produced ingressively by German speakers. Pätzold & Simpson (Reference Pätzold and Simpson1995: 512) state that ‘German ja produced on a pulmonic ingressive airstream is doing both linguistic and interactional work; counting out loud can be carried on while inhaling’. Finally, Hall (Reference Hall2003: 12, footnote) repeats the observation mentioned above that a ‘brief ingressive “Ja!” can sometimes be heard in German’.

4.4.9 Switzerland

Dieth (Reference Dieth1950) lists three occasions where one might encounter inspiratory speech, despite its inferior phonetic-acoustic characteristics: when counting or adding; when speaking in affection, for example, no!, nee ‘why not’, f ‘how stupid!’ and ja; and in voice disguise, the so-called Fensterle, where village boys courted their loved ones and wanted to keep their own identities secret.

4.4.10 Austria

Von Kempelen (Reference Kempelen1791: 103–104) mentions that ingressive speech occurs both when people speak ahead of themselves and in Catholic prayer:

4.4.11 The Netherlands

Perhaps the longest lists of ‘words’ produced on ingressive airstream that appear in the literature are found in Sipma (Reference Sipma1913/1966, Reference Sipma1949), who described speech on inhalation in Frisian. In Sipma (Reference Sipma1913/1966: 18–19), the following is found:

4.4.12 Belgium

Peters (Reference Peters1981: 46) mentions that French oui ‘yes’ occurs ingressively in Belgium although no mention of ingressive speech in Flemish has been found in any of the sources consulted.

4.4.13 France

Already in the 19th century, Segond (Reference Segond1848: 3) pointed out that speaking on ingressive airstream is possible, and as was previously noted, both Havet (Reference Havet1875) and Sacleux (Reference Sacleux1905) listed several ingressive sounds used for paralinguistic purposes in French. Havet (Reference Havet1875: 221) suggested that ingressive sounds be annotated using mirror-turned and upside-down letters, an idea that he attributed to Ballu (‘Suivant M. Ballu’) and that Sacleux (Reference Sacleux1905: 44) adopted later, too.

Passy (Reference Passy1895: 114; Reference Passy1912: 112) and Sacleux (Reference Sacleux1905: 35) mention that the word oui ‘yes’ is sometimes produced ingressively. Rousselot (Reference Rousselot1924: 488) mentions that oui and non ‘no’, as well as a few other sounds, mainly used to address animals, are sometimes produced ingressively in French and that he had encountered ingressive speech in Breton (p. 489), although in Breton the phenomenon involves single phones rather than full words (pp. 492–494). This passage from Rousselot is widely quoted, in e.g. Sacleux (Reference Sacleux1905: 44), Stopa (Reference Stopa1935: 15) and Vendryes (Reference Vendryes1950: 39). Vendryes (Reference Vendryes1950: 39, quoted here from Vendryes Reference Vendryes2003: 33) reports that

Peters (Reference Peters1981: 46) mentions a number of reports of ingressive oui in French, both first and second hand. That oui occasionally is produced ingressively in French is also mentioned in Fitch (Reference Fitch1994: 11).

However, the most authoritative source on ingressive speech in French is Léon (Reference Léon1993: 52):

Thus, Léon (Reference Léon1993) mentions sobs, impatience and strong emotion as causes, and also that women are the most frequent users of ingressive oui. The use of an ingressive oui in French is also mentioned by Ohala (Reference Ohala, MacNeilage and Kreidler1983: 192; Reference Ohala, Hardcastle and Marchal1990: 23), Sampson (Reference Sampson1999: 30), Collins & Mees (Reference Collins and Mees2003: 26) and Shorrocks (Reference Shorrocks and Tristram2003: 378).

4.4.14 England

A translation of Viëtor (Reference Viëtor1884), Rip[p]man (Reference Rip[p]man1910/1918: 2) is an early source which mentions the occurrence of ingressive speech in English:

Later, Ripman (e.g. Reference Ripman1933?: 33; Reference Ripman1947?: 33) mentions that English speakers ‘very occasionally, in conversation, say a word, such as yes, while taking in breath’. Sacleux (Reference Sacleux1905: 35) mentions that yes is sometimes produced ingressively in English. Passy (Reference Passy1912: 112) also mentions yes, and observes that while the ye part is ingressive, the s is ‘normal’ (i.e. egressive).

Although Jespersen (Reference Jespersen1926) mainly treats Danish, he cites (p. 125) two sources attesting the use of ingressive phonation in English:

Hockett (Reference Hockett1955) notes that speech can be produced on intake of air, although the larynx makes such sounds harder to produce, especially voiced sounds that come out hoarse. On English, he says (p. 26):

Abercrombie (Reference Abercrombie1967: 27), Catford (Reference Catford1977: 68) and Laver (Reference Laver1994: 169) mention that yes is occasionally produced ingressively by English speakers, with the final [s] egressive. Catford (Reference Catford1977: 68) also mentions that no is sometimes produced ingressively. Malmberg (Reference Malmberg1968: 313) mentions that ‘in English no may be said entirely ingressively, and yes may be uttered with pulmonic ingressive ye but pulmonic egressive [s]’. Wells & Coulson (Reference Wells and Coulson1971: 2) mention that yes ‘may be said ingressively sometimes’, and Peters (Reference Peters1981: 1) states carefully that British English ‘possibly’ makes use of paralinguistic ingressive airstream.

Crystal (Reference Crystal1987: 125) also says that yes and no are sometimes produced ingressively, as is also the case with laughing, crying and counting rapidly. He also tells us (p. 125) that ‘ingressive speech is of poor quality, muffled, and croaky, and many people find it unpleasant to listen to’. Shorrocks (Reference Shorrocks and Tristram2003: 388) reports ingressive use of yeah and no as ‘confirmed’ for locations in Cumbria, Durham, Teesside and East Yorkshire. Kreidler (Reference Kreidler2004: 18) only observes that it ‘is possible’ to produce yes on an ingressive airstream, and Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 56) mention that the words aye and neh occur ingressively in Cumbria, in the village of Patterdale.

4.4.15 Wales

Peters (Reference Peters1981: 1) mentions that Welsh English makes use of paralinguistic ingressive airstream. He mentions (p. 45) that a ‘JA type ingressive’ was collected from a female speaker of Welsh English in 1979. Morris Jones (Reference Morris Jones1999: 85) notes that in Welsh spoken in North Wales ‘responsives and nonresponsives’ can sometimes be ‘delivered with ingressive air mechanism’.

4.4.16 Ireland and Northern Ireland

Peters (Reference Peters1981: 1) mentions that Irish English makes use of paralinguistic ingressive airstream and points out (p. 7) that the negative form no can also occur ingressively. Shorrocks (Reference Shorrocks and Tristram2003: 389) reports that

Lamb (Reference Lamb2003: 97) also reports having heard it in Ireland.Footnote ⁶²

4.4.17 Scotland, Faroe, Orkney and Shetland Islands

Johnson (Reference Johnson1962: 49) provides an account of the ‘remarkable native of Shetland, Laurence Williamson’, a self-trained folklorist who studied, among other things, the phonetics of the Shetland dialect. Johnson gives the following description of a festivity which features inspiratory speech (pp. 56–57; cited later in Marwick Reference Marwick1975: 116):

The tradition of bands of people disguising themselves, going from home to home and (more or less) scaring people is found in the entire North Atlantic region. Gunnell (Reference Gunnell2001, Reference Gunnell1995) treats the Shetland tradition of grøleks, and the Faroese grýlur, who used ‘grunting’ to hide their identities, and (‘more recently’) also ‘reversed speech’, i.e. speaking on inhalation airstream (Gunnell Reference Gunnell2001: 43; this is also described in Marwick Reference Marwick1975: 91, 116). Gunnell (Reference Gunnell2001: 34) reports that this tradition is also encountered in Sweden, western Norway and northern Iceland. Whether or not the different versions of identity disguise all included ingressive speech is not mentioned. Gunnell (Reference Gunnell2001: 44) offers a hypothesis that there is a link between these variants and the southern Scottish mumming tradition, the ‘Galoshins’, which then also links it with Newfoundland. Moreover, ‘reverse speech’ (Gunnell Reference Gunnell1995: 117, 164, 169) was also used both in the julebukk tradition and among skeklers in the Shetland, the Orkneys, as well as in Scandinavia.

J. K. F. Anthony (cited in Gould & Okamura Reference Gould, Okamura and Wyke1974: 363) mentioned that ‘women between 50 and 60 years of age or so sometimes speak . . . on inhaling, as a normal thing’, in the Edinburgh area.

That ingressives are still used in Scottish English and Gaelic is shown by Thom (Reference Thom2005), whose M.A. dissertation possibly constitutes the most exhaustive survey of ingressive speech ever carried out. She collected various forms of data – including video tapes (of TV shows) and recordings in sound archives – but obtained most of her results from a nationwide survey that yielded 121 responses from all over Scotland. Her consultants provided her with the longest list of ingressive items found in the literature (alongside the work by Pieter Sipma and Jens Allwood and his colleagues cited in this article), and includes aye ‘yes’, ken, weel, uh huh, wow, no, wait, eh, yin for one, away you go, here, ouch, feech, what, yep, well, ooh, right, och, ach, saise, weesh, siadh, and many others (Thom Reference Thom2005: 39–40 et passim). Thom observes that ingressive speech is most common in Aberdeenshire, Morayshire, the Highlands, Orkney, Shetland Islands and the Western Islands. Somewhat surprisingly, she reports that 35% of her consultants thought that ingressive speech is more common in men, as opposed to 9% who thought it is more common in women. Most people thought that ingressive speech is more common among the elderly and several consultants thought it was a trait of ‘agricultural workers’ rather than fishermen (pp. 31–33). Many consultants also associated ingressive speech with Gaelic (p. 34).

Lamb (Reference Lamb2003: 96–97) writes the following on Gaelic:

Lamb also points out (p. 97) that the phrase uell uell ‘well well’ ‘should be performed ingressively for full effect’.

Thom (Reference Thom2005) mentions that ingressive speech is also encountered in Faroese, where short phrases can be produced ingressively.Footnote ⁶⁴ Examples of what can be heard are ja ‘yes’, nei ‘no’ and veit ikki ‘I don't know’. Helgason (Reference Helgason2002) recorded several instances of ingressive speech in Faroese, including entire phrases like eh eg veit ikki ‘eh, I don't know’, making it similar to Icelandic in that entire phrases might be produced ingressively.Footnote ⁶⁵

4.4.18 Iceland

While Peters (Reference Peters1981: 4) mentions that there is ‘at least one’ ingressive particle in Icelandic, Pitschmann (Reference Pitschmann1987) studied the use of ingressive airstream in German and the Scandinavian languages. He observes (p. 154) that Icelandic has an outstanding position:

He also observed that not only do Icelandic men and women utter entire phrasesFootnote ⁶⁶ on an ingressive airstream when they speak Icelandic, they also use it to a large degree when speaking English. Pitschmann (Reference Pitschmann1987: 157) provides the examples please, thank you, you're welcome, that poor guy, I was so mad, etc., all produced (in English) ingressively by Icelanders.

As an example of an Icelandic phrase, Pitschmann (Reference Pitschmann1987: 158) mentions ég veit það ekki ‘I don't know’. However, since certain phones are more suited to ingressive articulation than other phones, the dental fricatives /þ/ and /ð/ are not pronounced ingressively, but either dropped or changed into plosives.

4.4.19 Greenland

Cranz (Reference Cranz1765) mentions ingressive speech on Greenland already in the middle of the 18th century. More recent reports of ingressive speech in Greenland have not been found, however.

4.4.20 Canada

Canada is a place where ingressive speech (English) is widely reported. In the words of Kirwin (Reference Kirwin1971: 24),

Mumming (or mummering)Footnote ⁶⁷ is related to the previously mentioned skeklers and grýlurs, and is also linked to more contemporary traditions such as the Philadelphia Mummer's Parade, the New Orleans Mardi Gras, the Northern English Sword Dance, the German tradition of ‘belsnickles’ in, for example, Nova Scotia, Pennsylvania and Virginia (Halpert Reference Halpert, Halpert and Story1969: 39). Christmas mumming includes groups making house visits, while attempting complete disguise, including voice disguise, most often realized as phonation on ingressive airstream.Footnote ⁶⁸ Mumming occasionally degenerates into ‘boisterous and violent behavior’ which resulted in official bans of the tradition (Halpert Reference Halpert, Halpert and Story1969: 51).

Newfoundland Christmas mumming is well described and there are several accounts of it in the literature.Footnote ⁶⁹ Williams (Reference Williams, Halpert and Story1969) described janneying in a small fishing community in Trinity Bay, where ‘groups of two or three men will make the rounds of the houses of their friends and neighbours’ (p. 209). To hide their identities, janneys always used janney-talk (p. 211):

Firestone (Reference Firestone, Halpert and Story1969: 66) described mummers, or ‘janneys’, who during the twelve days of Christmas go from house to house while donning disguises. He reports that

Szwed (Reference Szwed, Halpert and Story1969) treats mumming in a small village in the Gulf of St. Lawrence, and informs us that not only is speech ‘reversed’ in mumming, but also dress (p. 110):

Chiaramonte (Reference Chiaramonte, Halpert and Story1969: 82–83) reports that

Faris (Reference Faris, Halpert and Story1969: 130–131) reports:

Sider (Reference Sider1976: 102, 115) notes that while the first mention of mumming dates back to 1819, it had disappeared in the mid-1970s, and he discusses ‘the recent rapid decline of Christmas Mumming’ (see also Sider Reference Sider2003: 155).

Paddock (Reference Paddock1981: 35) describes the Carbonear dialect of Newfoundland in detail, and devotes some space to ingressive speech and phonation:

He also discusses (pp. 35–36) the ingressive speech encountered in more phonetic detail than most other linguists do:

Paddock finishes his account of ingressive speech with a very interesting and important observation (p. 36), which highlights the interactional aspects of ingressive speech:

In his Dictionary of Prince Edward Island English, Pratt (Reference Pratt1988: 177) mentions that ‘the word yes indicated by forming the y consonant with the mouth and sucking in the breath rather than expelling it . . . also found in New Brunswick speech’. Steinbergs (Reference Steinbergs1993) studied the use of the paralinguistic sympathetic ingressive affirmative in the St. John's area in Newfoundland, and recorded uses of yes, yeah and no in males and females. She confirms, on the basis of a quantitative analysis, the oft-repeated claim that women are more frequent users of ingressive articulation (p. 9). In his survey of Canadian English, Woods (Reference Woods1999: 318) lists 18 ‘ingressive speech occurrences’ divided between one male and four female ‘ingressive speakers’. Lamb (Reference Lamb2003: 97) reports hearing ingressive speech in Cape Breton and much of Newfoundland.

Poyatos (Reference Poyatos2002: 148–149) – discussing narrative literature, cinema, etc. – mentions that ingressive ‘air intake’ is a means to indicate ‘fear, surprise, indignation, physical pain, etc.’ and provides the examples ‘Oh God’, ‘Bill, Bill’, ‘A Gentleman’ and ‘Ah-h-h’. Although these examples are most likely not authentic, it is clear that Poyatos think they could be since ingressive speech is ‘a most typical feature in, for instance, the Canadian province of New Brunswick’ (p. 149). Shorrocks (Reference Shorrocks and Tristram2003) sets out to trace the origin of ingressive speech in Canada, and studies recordings made on the West Coast, the Great Northern Peninsula and at Trinity Bay, Newfoundland. He observes items such as yeah, no, oh and ah. He also observes that ingressive yeah ‘seems to be used very frequently in telephone conversations’ (p. 386) and reports that consultants hypothesize that it is more common among older people (p. 387). Shorrocks does not rule out the possibility of multiple sources for the use of ingressives in Newfoundland English, although it seems likely that a major contributor is the influx of Irish-English settlers. Clarke & Melchers (Reference Clarke, Melchers, Filppula, Klemola, Palander and Penttilä2005: 59 et passim) mention that the items yeah, mm and no are encountered in Newfoundland.

4.4.21 The United States

Starting with American English, in the Linguistic Atlas of New England (LANE, Kurath et al. Reference Kurath, Hanley, Bloch, Lowman and Hansen1943), inspirated versions of words like yes, yep, ya, eh ya, why and similar items are reported from several locations in Massachusetts (Kurath et al. Reference Kurath, Hanley, Bloch, Lowman and Hansen1943, maps 588 and 589).Footnote ⁷⁰ Without specifying a geographical area, Fries (Reference Fries1952: 52) mentions that oh can be produced on ‘ingressive air’ in English.

Peters (Reference Peters1981) studied the paralinguistic affirmative in English and the Scandinavian languages. Although his main focus is on speakers from Vinalhaven, Maine, one of his hypotheses is that ingressive forms were brought to Maine and other parts of the Northeastern United States by Scandinavian immigrants. He also reports evidence of ingressive speech in Massachusetts (e.g. Kurath et al. Reference Kurath, Hanley, Bloch, Lowman and Hansen1943), where the forms why and why yes occur (Peters Reference Peters1981: 17). Peters also devotes some space (pp. 22–24) to a discussion of whether there is evidence of ingressive forms in Middle English, which has been argued, but concludes that there is no evidence for such a hypothesis. An interesting feature of Peters's study is that he had collected much data from films, radio shows and TV shows. Peters concludes (p. 195) that he is ‘not able to find any further evidence that use of ingressive had been influenced by the arrival of Scandinavian immigrants’.

That ingressive speech still exists in Maine is confirmed by Aimee Baker (p.c., 2005) who reports having heard it ‘around’ in Maine. Hockett (Reference Hockett1955: 26) mentions that the Maidu (Indians) have two consonants that in combination with the following vowel are always produced ingressively:

Key (Reference Key1975a: 69) mentions that the ‘Maidu Indians use a long inspirated high vowel sound [iː] for an exclamation’. Napoli (Reference Napoli1996: 30) also mentions the Maidu Indians when discussing ‘ingressive pulmonic air pressure’.

A very interesting case is described in Hill & Zepeda (Reference Hill, Zepeda and Jonz1999). They discuss the use of pulmonic ingressive speech in the Southern Arizona language Tohono O'odham. In this language, ‘women use PIAS [pulmonic ingressive airstream] in their everyday conversational speech to construct an intimate interactional quality’ (p. 16). The authors observe that whisper has exactly the same distribution in women's speech, and seemingly plays the same role. While some women use ingressive speech more than whisper, most women use both types. In Tohono O'odham, ingressive speech is obviously a female characteristic (p. 17):

Hill & Zepeda (Reference Hill, Zepeda and Jonz1999: 21–22) found no dialect differences in the use of ingressive speech, despite huge differences in e.g. speech rate in the Tohono O'odham language. Nor did they find any age-related variation. The examples they provide include the following phrases, produced on ingressive airstream:

Hill & Zepeda provide an in-depth analysis of the occurrence of ingressive speech in Tohono O'odham, including discourse, grammar and gender perspectives, and also mention that ingressive voicing in women's speech does not possess the ‘croaking quality’ that Catford (Reference Catford1977) mentions for ‘inverse speech’ (Hill & Zepeda Reference Hill, Zepeda and Jonz1999: 38).

4.4.22 Mexico

Nigel Ward (p.c.) reports the existence of ‘ingressive backchannels’ in Mexican Spanish as spoken by a female speaker in her twenties from the Chihuahua state in Northern Mexico. The items in question, transcribed as ‘ih!’, are not words proper but are exclamations used to express ‘surprise/empathy/concern’. The data are described in Acosta-Reyes (Reference Acosta-Reyes2004).

4.4.23 Colombia

Flórez (Reference Flórez1957: 162–163) includes the following passage on affirmation adverbs in Colombian Spanish:

4.4.24 Argentina

Vidal de Battini (Reference Vidal de Battini1949: 25), in her book on Argentine Spanish rural speech of San Luis, includes this intriguing passage:

Nardi (Reference Nardi1958, Reference Nardi1960) picks up on this thread and points out that ingressive speech ‘es frecuente en el habla rural de una gran parte de nuestro pais’ [is frequent in the rural speech in a great part of our country] (Reference Nardi1958: 153). He also points out that it occurs in dialog and that it is predominantly observed in the speech of women and children (p. 154). Apart from sí ‘yes’ (already mentioned by Vidal de Battini Reference Vidal de Battini1949), Nardi (Reference Nardi1960: 223–224) lists several other items, including the following:Footnote ⁷³

It is clear from these, and the other, examples given by Nardi that ingressive phonation is not restricted to but a small number of fixed words or phrases, but is used frequently whenever the discourse situation calls for its use. Indeed, Nardi identifies seven different conversational functions of ingressive speech – very similar to what has been observed for e.g. the Nordic languages – including exclamations, courtesy phrases, and other circumstances under which it is observed, such as the final words or phrases and in whispers.

Nardi (Reference Nardi1960: 226) discusses possible origins of this phenomemon, and posits that it is of Spanish origin, or that it is indigenous to Latin American Spanish, or that its origins are to be found in the substrate languages of South America. He notes that he had found no sources that mention the occurrence of ingressive phonation in Spain, but that ingressive phonation is observed in the Indian language Mapuche in the Argentinian province Neuquén (El Malleo, Auca Pan, Ruca Choroy), possibly hinting at substrate influences.

Nardi also points out that ingressive phonation is known to exist in European cultures but is less known in (Argentine) Spanish. Contrary to its most common use in Europe (i.e. single words), entire phrases like ‘un trabajo matador’ are produced ingressively in Argentine Spanish.

4.4.25 Chile

Navarro Tomás (Reference Navarro Tomás1945: 56) mentions that in Chilean Spanish, an ‘enfático ya’ [emphatic ‘yes, indeed, sure’] is sometimes observed. As was noted earlier, Vidal de Battini (Reference Vidal de Battini1949: 25) mentions that the ingressive speech characteristic of certain parts of Argentina is also found in Chile, obviously implying Spanish.

4.4.26 Turkey

There is some evidence that ingressive speech might occur in Turkish. Laura DilleyFootnote ⁷⁴ reports (p.c.) that a Turkish consultant confirms pulmonic ingressives ‘in short utterances perhaps as a kind of paralinguistic marker’.

4.4.27 Cyprus

Catford (Reference Catford1988: 31) reports that

4.4.28 Malta

Borg & Azzopardi-Alexander (Reference Borg and Azzopardi-Alexander1997: 299) report that in Maltese

4.4.29 India

In a paper on prosodic functions in Marathi (spoken in southern and western India, in the provinces of Maharashtra, Goa, Daman and Diu, and Dadra Nagar Haveli), Kelkar (Reference Kelkar1991/1992: 293) notes that as a part of what is called ‘marginal phonology’ or ‘speech gestures’, two ingressive sounds are found, ‘[ s pulmonic ingressive], [ ḁ pulmonic ingressive]’. Although the author is not explicit as to the exact function of these sounds, it seems clear that their use is paralinguistic.

4.4.30 China

Key (Reference Key1975a: 69) mentions, without any specification, that ‘Chinese’ includes a sound ‘[Ss!] with the air sucked in’.

4.4.31 Mongolia

Several people with first-hand experience from Mongolia confirm that the words tiim ‘yes’ and ugui ‘no’ are frequently produced on ingressive air in Khalka Mongolian. Thus, Matthew Patenaude (Baptist Bible Translators Institute, Bowie, Texas) mentions (p.c.) that it is ‘a very common thing to hear’. He also points out that it is more common in the countryside, although urban dwellers also do it. Amy LaCross (Linguistics Department, University of Arizona, Tucson) agrees (p.c.) that ingressive phonation of tiim and ugui is the ‘everyday, informal way of doing it’. She also points out that the negative suffix particles are sometimes produced ingressively. Joakim Enwall (Department of Linguistics and Philology, Uppsala University, Sweden) says (p.c.) that the [m] in tiim is frequently dropped, and that the [i] turns into a fricative, resulting in an ingressive [txː], while ugui is produced (ingressively) as [qxu].

Finally, Carsten Naeher (Department of General Linguistics, Institute for Communications (IfK), Bonn) mentions (p.c.) that his Mongolian teacher (in Germany), a native speaker of Khorchin Mongolian, produced ingressive speech ‘quite regularly’, and ‘by no means confined to the words ‘yes’ or ‘no’’. Interestingly, Naeher also points out that his teacher spoke ingressively in German (cf. section 4.4.18 above and Pitschmann Reference Pitschmann1987: 157) as well.

4.4.32 Japan

The Japanese ‘hiss’ is mentioned in several sources. An ‘alveolar continuant, with inspiration’ is mentioned by Trager (Reference Trager1958: 7; Reference Trager1964: 277), and Critchley (Reference Critchley1939: 102) mentions an honorific, which is a ‘hissing inspiratory sound’. Key (Reference Key1975a: 69) also lists the ‘Japanese hiss, which is produced with air drawn in’.Footnote ⁷⁵ Maynard (Reference Maynard1990: 259) provides the following description of the Japanese hiss:

Mey (Reference Mey1993: 166–167) gives the following account of the Japanese hiss:

4.4.33 The Philippines

Conklin (Reference Conklin1949, Reference Conklin1959) describes different ways to modify normal speech patterns used by the Hanunóo people of the Minduro province in the Philippines. The Hanunóo ‘are pagan jungle farmers living in small hamlets dispersed over a mountainous area of about 800 square kilometers’ (Conklin Reference Conklin1959: 634). Normal linguistic forms can either be modified by rearrangement, substitution, affixation of reduplication, but there are also four ways to modify the manner of articulation or vocalization, one of which is ingressive phonation (Conklin Reference Conklin1959: 632, 633):

In what situations, then, is ingressive speech used? Conklin provides the answer:

4.4.34 Indonesian Province of Papua (Western New Guinea/Irian Jaya)

Bromley (Reference Bromley1961: 47) reports that in the Lower Grand Valley Dani dialect, spoken in the Baliem Valley of what was then Central Netherlands New Guinea ‘[i]ngressive lung air occurs as an extra-systematic airstream most commonly used in attention-assuring forms spoken by a listener periodically throughout the speech of a speaker’. Although Bromley's terminology is different, it is clear that the use of ingressive phonation in Dani is identical to its feedback use in several of the previously mentioned languages.

4.4.35 Papua New Guinea

Chung & Chung (Reference Chung and Chung1993: 5) report from the New Ireland (Papua New Guinea) language Kuot that a ‘[d]eep ingressive breath’ signals ‘That's right’ or ‘It's correct’. This is confirmed, in more detail, in Lindström (Reference Lindström2002: 123; bold in the original):

4.4.36 Vanuatu

Thieberger (Reference Thieberger2004: 79) reports that in the language South Efate, spoken in Central Vanuatu, ingressive voice is used to ‘encode resignation on the part of the speaker event described as if to say “that's the way it is”’. Thieberger provides examples where tefla ‘like that’ and su rowat ‘house of sago leaves’ are produced on ingressive voice.Footnote ⁷⁶

4.4.37 New Zealand

Professor Paul Warren (p.c.) reports that ingressive yeah occurs in New Zealand English, expressing surprise and incredulity.

4.4.38 Australia

Despite its occurrence in New Zealand English, it has not been possible to confirm that ingressive speech exists in Australian English. Most people consulted, including linguists and phoneticians, responded that they would be surprised if it did not exist, but that they cannot confirm or prove its existence. And as we all know, ‘absence of evidence is not evidence of absence’.

The ritual language Damin included an ingressive phoneme, but we also saw that the remarkable phonology of Damin probably could be attributed to it being deliberately invented, most likely with the intention of making the language ‘special’ (cf. section 4.3.2 above).

There is, however, a language where ingressive speech occurs in a pattern similar to Damin. Laughren (Reference Laughren1984: 87) describes speech styles in Warlpiri, of Central Australia:

What we see here is that ingressive speech seems to be a female phenomenon, and although Laughren (Reference Laughren1984) does not explicitly mention what kind of linguistic items are produced ingressively, it would seem to be the case that this might concern longer stretches of speech, not only single words.

Amery (Reference Amery, Cunningham, Ingram and Sumbuk2006: 169), mentions ‘ingressive whispers’ when discussing Australia's indigenous languages, notably Youlŋu and Aṉangu.

4.4.39 Malawi

Sam Mchombo (p.c.) reports that loud inhalation sounds are used as a paralinguistic marker in Chichewa. It functions as an ‘I'm following’ or ‘I'm listening’ marker. As far as Mchombo knows, no words are produced ingressively, though, but it is still intriguing that inhalations serve the same function in Chichewa as they do in the northern European languages. Laurel MacKenzie (p.c.) also reports that one of her Malawi consultants uses inhalations when she speaks English, with the same function.

4.4.40 Guinea, Sierra Leone and Liberia

Childs (Reference Childs1995: 146) mentions that in Kisi, an ‘ingressive pulmonic’ version of έ ‘yes’ is used ‘to show agreement, comparably to the way in which it is used in South Africa, northern Europe, and Canada’. He also points out (p. 1) that around 20.4% of Kisi speakers reside in Sierra Leone and that around 19.3% reside in Liberia (1995: 1).

4.4.41 Togo

Kirk Miller (p.c.) reports that a non-lexical, voiced, ‘ingressive schwa’ is very frequently used in Ewe for backchanneling, very often accompanied ‘by raising of the eyebrows’. He mentions that speakers ‘also use it when speaking French’, on oui ‘yes’, although ‘not frequently’.

4.4.42 Tanzania

Pike (Reference Pike1947: 22, footnote 1) includes the following, intriguing, remark:

Note that Pike uses the phrase ‘for example’ here, implying that the Wazanaki are not the only tribe known to employ ingressive assent.

4.4.43 South Africa

Given its occurrence in Dutch, it is not unreasonable to expect ingressive speech in Afrikaans. Professor Daan Wissing (p.c., 2005) confirms this:

In later correspondence, Wissing reports that his students and colleagues also agree that ingressive speech is used in Afrikaans.

Childs (Reference Childs1995: 146), when discussing Kisi (see section 4.4.40 above), mentions that an ‘“ingressive pulmonic” is also used to show agreement . . . comparably to the way in which it is used in South Africa, northern Europe, and Canada’, however without specifying in what language in South Africa it occurs.

4.4.44 ‘Arabia’

Without specifying either country or region, Sacleux (Reference Sacleux1905: 35) states that the interjection nam in ‘arabe’ [Arabic] is sometimes produced ingressively.

4.4.45 ‘Pays des Hottentots’

Finally, Rousselot (Reference Rousselot1924: 489) asserts that the ‘terrain classique’ for ingressive words is ‘le pays des Hottentots’:

Rousselot does not specify exactly what country/countries he has in mind. According to Alexandre & Binet (Reference Alexandre and Binet1958), the Pahouins was a collective term for a group of tribes (Fang, Boulou, Beti) who (at the time of publishing, 1958) inhabited ‘Cameroun’, ‘Guinée Espagnole’, ‘Gabon’ and ‘Moyen-Congo’ (see Alexandre & Binet Reference Alexandre and Binet1958: 10–11 and final fold-out map on page 153). The authors do not mention ingressive speech in their chapter on language (ch. III: 18–25).

Pither Medjo (p.c., by way of Lolke van der Veen) confirms that ingressive phonation is sometimes used to say ‘oui’ in Fang, Ewondo and Bulu, but never to say ‘no’.

Sacleux (Reference Sacleux1905: 35) mentions that the interjections e, ye, o and ndo are produced ingressively in ‘les langues africaines’ [the African languages].