Acoustic analysis of singleton and geminate affricates in Italian

Simone Faluschi and Maria Gabriella Di Benedetto

INFOCOM Department, University of Rome 'La Sapienza'

Via Eudossiana, 18, 00184 Rome, Italy

E-mail: simone.fal@tiscalinet.it

gaby@acts.ing.uniroma1.it

Abstract

The aim of this project was to carry out a study based on the analysis of gemination of affricate consonants in Italian. Disyllabic words such as VCV (vowel-consonant-vowel) or VCCV (vowel-consonant-consonant-vowel) containing the four affricate Italian consonants [ʧ, ʤ, ʦ, ʣ], in their singleton and geminate version co-articulated with the three Italian vowels [a, i, u], were the subject of this study. The parameters that were considered are those time-related, frequency-related and energy-related ones. Next, a statistical analysis on the values obtained was carried out in order to establish the meaningfulness of the average values obtained. Those measures that were considered were taken at different sampling points of the utterance, so that it was possible to observe how they changed during the utterance sequence. Results show that pre-consonant vowel and consonant durations significantly change when passing from singleton to geminate utterance. Frequency and energy parameters do not show great differences like the durations of vocal and consonant although they present significant differences between singleton and geminate utterances. Results in time domain are in agreement with previous studies about Italian consonants (stops, liquids, fricatives and nasals) but it is not so for frequency and energy measurements.

Introduction

In Italian, there are minimal pairs formed by words which can be guessed only by the gemination of a consonant. Words belonging to these minimal pairs are graphically distinguished by writing twice the consonant that doubles (for example: micia (pussy-cat) and miccia (fuse)). A problem still unsolved is about which vocal signal parameters change where moving from singleton to geminate utterance and how they can affect perception. To give an answer to this problem, concerning Italian language, several studies - belonging to the GEMMA project in progress at the INFOCOM Department at University 'La Sapienza' in Rome - have been performed on other consonant, such as stops, liquids, fricatives and nasals. Results on stop consonants were reported in Rossetti (1993, 1994) and Esposito and Di Benedetto (1999). Liquids were analyzed in Argiolas et al. (1995) while the analysis of fricatives was reported in Giovanardi (1998) and Giovanardi and Di Benedetto (1998). Finally results on nasal consonants were reported in Mattei (1999) and Mattei and Di Benedetto (2000). Main results confirm those obtained in previous studies about the gemination of Italian consonants. They are a significant increase of consonant duration and an as much reduction of the pre-consonant vowel, that are confirmed by previous studies about perception of stop (Esposito and Di Benedetto, 1999) and nasal (Mattei and Di Benedetto, 2000) consonants.

As already pointed out in (Mattei and Di Benedetto, 2000) gemination has been analysed in other languages as well: (Shrotriya, 1995), (Blumstein et al., 1998), (Rochet and Rochet, 1995), (Cohn et al., 1999), (Abramson, 1999) (Local and Simpson, 1999), (Arvaniti, 1999), (Louali and Maddieson, 1999). This studies, by the way, do not concentrate particularly on affricate consonants. The only one analysing this class of consonants was (Abramson, 1999) although in a pre-test perceptive analysis results presented a too high percentage of mistakes so that he was compelled to discard that class.

1 Speech materials and measurements

In Italian, several disyllabic words form minimal pairs which can be distinguished on the sole basis of gemination of one consonant. Native speakers exhibit a natural attitude in producing disyllabic words of minimal pairs identified by the presence or absence of gemination of one consonant. The above consideration led to the creation of the GEMMA project database formed by a set of vowel-consonant-vowel disyllabic words (the singleton case) and vowel-consonant-consonant-vowel disyllabic words (the geminate case) which would serve as the basis for studying gemination in all possible geminated consonantal forms of Italian. The words in the database included the entire set of those Italian consonants which appear in singleton and geminated forms. Affricate consonants were studied when co-articulated with the three Italian vowels that are at the extreme positions in the phonetical diagram [a, i, u] and represent a subset of Italian vowels [a, e, ɛ, i, o, ɔ, u]. Words were pronunced by six pronunciation defectless and dialectal inflexionless native speakers. In order to keep under total control all the parameters used, words were pronounced alone instead of in whole sentences, so that those elements, such as intonation or internal stresses, that could have influenced the utterances were eliminated. Every word was pronounced three times by each speaker. Our procedure in building up the databese of utterances was motivated by the strong need of having data with well-controlled parameters, as a starting point for future more realistic settings. The analyzed words in the present study were therefore 3 for each affricate consonant (which are [ʧ, ʤ, ʦ, ʣ] and their geminate version) and 6 for each speaker in three repetition, leading to a total of 3x4x2x6x3=432 utterances (216 singletons and 216 geminates).

All the utterances were produced and recorded at the Speech Laboratory of the INFOCOM Department at the University of Rome 'La Sapienza' (Italy). Equipement used for this experiment is of a superior quality and the recording took place into a sound-treated room. The operator who look after this recording procedure was an acoustic trained subject. When there was an evident pronunciation mistake or if it was judged unnatural, the speaker was compelled to repeat it again. Three were the recording sessions that were done, each corresponding to one of the three repetitions.

The set of words analysed in the present study is reported in Table I. Geminate utterances are indicated by a double grapheme of the first consonant letter.

	ʧ		ʤ		ʦ		ʣ
a	aʧa	atʧa	aʤa	adʤa	aʦa	atʦa	aʣa	adʣa
i	iʧi	itʧi	iʤi	idʤi	iʦi	itʦi	iʣi	idʣi
u	uʧu	utʧu	uʤu	udʤu	uʦu	utʦu	uʣu	udʣu
TABLE I The complete set of words analyzed

Once recorded, the utterances were digitized using UNICE software by VECSYS (Vecsys, 1989) through which it has been possible to elaborate the signal. It was filtered at 5 kHz (enough to study the main spectral properties of the vocal signal), sampled at 10 kHz while every sample was represented by using 16 bits. The resulting signal was stored into PC to procede into next analysis performed by UNICE, such as spectrograms, DFT (Discrete Fourier Transform) and LPC (Linear Predictive Coding) spectra. For these elaborations a pre-emphasizing filter (with α=0.95) on the signal and a Hamming window of 256 samples were used.

The GEMMA project analyses the measurement of a set group of standard parameters (Giovanardi and Di Benedetto, 1998) and (Mattei and Di Benedetto, 1999). Being the affricate consonants very particular in having an occlusive and a fricative phase it has been decided to modify the standard parameters in order to study more correctly the utterances. The following are the parameters used for this study:

1. duration of the pre-consonant vowel, indicated as V1d. The vowel onset was identified by the appearance of a glottal pulse followed by other regular glottal pulses. In those cases in which a glottal excitation was visible before regular vowel voicing, the vowel onset was taken as the beginning of regular vowel voicing, and the initial glottal excitation was discarded. Vowel offset was identified, by examination of both the waveform and the spectrogram

2. duration of the occlusive part of consonant, indicated below as C1d

3. duration of the fricative part of consonant, indicated below as C2d

4. duration of all consonant, indicated below as Cd (= C1d + C2d)

5. duration of the vowel following the consonant, indicated as V2d. The V2 onset was identified, by visual inspection of both the waveform and the spectrogram. The V2 offset was identified as the temporal sampling point where the glottal pulse disappeared

6. duration of the whole utterance, indicated as Utd

7. total energy of V1. X_i is the sample i, t₁ and t₂ are the temporal sampling points of vowel onset and vowel offset, respectively

8. average power of V1

9. total energy of occlusive phase of C, indicated below as E_totC1 and computed as for V1, with t₁ and t₂ that corresponding to V1 offset (C1 onset) and C2 onset (C1 offset), respectively

10. average power of occlusive phase of C, indicated below as P_mC1 and computed as for the average power of V1

11. total energy of fricative phase of C, indicated below as E_totC2 and computed as for V1, with t₁ and t₂ that corresponding to C1 offset (C2 onset) and C2 offset, respectively

12. average power of fricative phase of C, indicated below as P_mC2 and computed as for the average power of V1

13. total energy of the whole consonant, indicated below as E_totC and computed like usual

14. average power of the whole consonant, indicated below as P_mC

15. instantaneous energy at V1 center, indicated as E_iV1, computed in a window of 256 samples centered on V1

16. instantaneous energy at V1-C1 transition, indicated as E_iV1-C1: The window of 256 samples is centered on V1 offset

17. instantaneous energy at C1 center, indicated as E_iC1 and computed as E_iV1

18. instantaneous energy at C1-C2 transition, indicated as E_iC1-C2: The window of 256 samples is centered on C1 offset (C2 onset)

19. instantaneous energy at C2 center, indicated as E_iC2 and computed as E_iV1

20. instantaneous energy at C2 offset, indicated as E_iC2off, computed right before the first 256 samples of V2 onset

21. F0, A0, F1, A1, F2, A2, F3, A3, at V1 center (where F1,F2 and F3 are the fornants and A1, A2 and A3 their amplitudes)

22. F0, A0, F1, A1, F2, A2, F3, A3, at V1 offset

23. F0, A0, F1, A1, F2, A2, F3, A3, at the transition from V1 to C

24. F0 and A0 at the onset of voiced consonants

25. F0 and A0 at the centre of occlusive phase of voiced consonants

26. F0 and A0 at the centre of fricative phase of voiced consonants

27. F0 and A0 at the offset of voiced consonants

28. F0, A0, F1, A1, F2, A2, F3, A3, at V2 onset

29. F0, A0, F1, A1, F2, A2, F3, A3, at V2 centre

FIGURE 1 Sampling points selected for the computation of the acoustic parameters and indication of the frames where frequency parameters were estimated.

2 Results of acoustic analysis

In this paragraph there are time, frequency and energy domains results of the study. To understand wether existing differences between values obtained for singleton and geminate utterances are statistically meaningfulness, these statistic tests were used: Mono and Multivariate Anova, the maximum a-posteriori classification test, the maximum likelihood classification test and the Spearman Rank Correlation Coefficient r_s. We separately show the results in time and frequency domain.

2.1 Results in the time domain

The durations of the phonemes are showed in Table II. It contains the average values of V1d, C1d, C2d, V2d and Utd for each of the four consonants co-articulated with each of the three vowels and the corresponding standard deviations. As we can see in Table II those parameters which present relevant differences between singleton and geminate utterances are V1d, C1d and C2d (remember that C1 and C2 together form the whole consonant). We notice a general tendency to shorten the first vowel duration and to lenghten the consonant duration (both in occlusive and fricative phase); on the contrary, there are no relevant variations in the second vowel duration. These statements are supported by a statistic analysis based on these data, showed on Table III. Those values that modify significantly (p<0.05) from singleton to geminate utterance, are highlighted through bold characters. They indicate that a strong significancy is found for V1d, C1d and C2d, except for [ʦ] articulated with [i]. Utd too, grows significantly in rather all the geminate utterances, although not so strongly as those presented above. Moreover, the second consonant duration does not vary significantly between singletons and geminates, except for [ʤ] articulated with [a].

	V1d	C1d	C2d	V2d	Utd		V1d	C1d	C2d	V2d	Utd		V1d	C1d	C2d	V2d	Utd
aʧa	160.0	73.1	100.9	112.3	446.3	iʧi	137.4	64.0	122.4	104.6	428.4	uʧu	163.6	66.0	103.7	131.7	465.0
(StD)	27.6	34.7	20.5	19.6	43.8	(StD)	20.8	29.2	16.2	17.9	29.8	(StD)	27.4	37.9	24.0	23.7	32.0
atʧa	113.2	137.8	128.7	107.5	487.2	itʧi	99.3	122.8	158.4	110.7	491.3	utʧu	110.9	151.1	123.0	125.0	509.9
(StD)	19.2	13.9	28.1	12.2	29.3	(StD)	17.9	20.4	26.1	21.0	37.5	(StD)	25.4	39.4	24.7	22.4	51.7
aʤa	169.0	92.0	49.1	142.3	452.3	iʤi	166.7	95.9	52.6	141.6	456.8	uʤu	173.5	85.7	44.1	146.1	449.5
(StD)	20.6	18.9	13.6	26.1	47.4	(StD)	28.3	17.5	15.7	30.6	53.4	(StD)	32.1	21.1	16.5	26.5	45.0
adʤa	127.3	156.1	61.5	125.9	470.9	idʤi	111.7	162.1	74.1	129.4	477.3	udʤu	120.2	154.0	61.3	137.3	472.8
(StD)	16.0	17.7	11.0	15.9	42.2	(StD)	21.3	28.2	25.5	30.6	56.6	(StD)	21.6	21.3	20.8	29.9	67.7
aʦa	121.3	89.6	129.8	109.9	450.6	iʦi	106.7	84.4	149.6	109.7	450.4	uʦu	133.2	73.3	140.7	115.3	462.5
(StD)	23.3	11.0	34.0	23.1	37.0	(StD)	25.9	20.2	31.3	18.1	32.2	(StD)	30.6	26.9	22.4	16.3	41.1
atʦa	106.0	112.2	167.0	117.4	502.6	itʦi	94.5	114.0	171.0	123.2	502.7	utʦu	103.8	96.3	178.8	115.1	493.9
(StD)	18.7	18.8	22.0	20.6	43.5	(StD)	17.9	31.4	34.7	22.8	48.0	(StD)	21.9	20.4	19.4	15.8	40.4
aʣa	163.4	89.9	78.6	139.7	471.7	iʣi	148.4	85.9	90.9	148.1	473.4	uʣu	150.8	81.6	80.9	139.7	453.0
(StD)	24.7	13.5	19.3	18.9	42.9	(StD)	37.5	16.5	21.6	20.7	35.7	(StD)	23.7	18.8	18.1	23.8	44.8
adʣa	127.8	139.8	102.3	136.3	506.2	idʣi	104.7	136.5	120.2	139.7	501.1	udʣu	117.7	116.8	112.3	136.4	483.1
(StD)	24.5	35.3	23.0	29.0	57.4	(StD)	23.9	36.4	38.1	19.0	53.0	(StD)	17.1	26.9	29.4	20.0	43.1
TABLE II Average values (and standard deviations) of V1d, C1d, C2d, V2d and Utd, over all repetitions and speakers. All values are in msecs.

		A					I					U
		V1d	C1d	C2d	V2d	Utd	V1d	C1d	C2d	V2d	Utd	V1d	C1d	C2d	V2d	Utd
ʧ	F ratio	34.89	53.90	53.90	0.78	10.83	34.69	49.22	24.66	0.88	31.05	35.75	43.56	5.62	0.78	9.82
	p value	0.0000	0.0000	0.0018	0.3842	0.0023	0.0000	0.0000	0.0000	0.3554	0.0000	0.0000	0.0000	0.0235	0.3845	0.0036
ʤ	F ratio	46.04	110.50	9.15	5.14	1.54	43.36	71.55	9.23	1.43	1.25	34.23	93.40	7.55	0.88	1.48
	p value	0.0000	0.0000	0.0047	0.0298	0.2231	0.0000	0.0000	0.0046	0.2394	0.2722	0.0000	0.0000	0.0096	0.3546	0.2328
ʦ	F ratio	4.73	19.22	15.15	1.06	14.91	2.69	11.29	3.80	3.86	14.75	10.96	8.32	29.77	0.00	5.36
	p value	0.0367	0.0001	0.0004	0.3094	0.0005	0.1102	0.0019	0.0595	0.0576	0.0005	0.0022	0.0068	0.0000	0.9679	0.0268
ʣ	F ratio	18.84	31.39	11.15	0.17	4.19	17.39	28.90	8.04	1.62	3.39	23.24	20.71	14.83	0.20	4.22
	p value	0.0001	0.0000	0.0020	0.6785	0.0485	0.0002	0.0000	0.0077	0.2120	0.0745	0.0000	0.0001	0.0005	0.6587	0.0477
TABLE III Results of the Anova test performed on V1d, C1d, C2d, V2d and Utd. F-values are listed. The null hypothesis can be rejected at the p level of significance indicated on the table. Bold characters indicate significantly different values.

A Spearman Rank correlation test was carried out in order to verify if the relation between phonemes durations, in particular those between V1d, C1d and C2d, was based on gemination. These were the results:

· when the two singleton and geminate groups are put together we obtain these results for r_s: V1d vs C1d r_s= - 0.471; V1d vs C2d r_s = - 0.474

· when the two utterance groups are kept separately, then these are the value obtained for r_s: V1d vs C1d: r_s = - 0.271 for singletons and r_s not significative for geminates; V1d vs C2d: r_s = - 0.432 for singletons and r_s = - 0.321 for geminates.

Therefore, the correlation between V1d and C1d can be attributed to the presence of gemination. The same can be said for the correlation between V1d and C2d, although not with the same strength.

On the basis of the results achieved studying time-domain quantities, it was decided to use the parameters V1d, C1d, C2d, Cd, Cd/Utd, Cd/V1d, C1d/V1d and C2d/V1d to try an utterance classification based on time parameters using the Maximum Likelihood Criterion (Dillon W.R. and Goldstein M., 1984). All this, with connected error percentages, can be seen in Table IV. Applying the MLC criterion on all the utterances, we can find these error percentages: on Cd/Utd we have 19.9%, on Cd/V1d 16.7%, on C1d/V1d 13.9%, on C2d/V1d 30.3%, on V1d 23.4%, on C1d 16.9%, on C2d 36.3%, on Cd 17.6%. The best result is obtained using parameter C1d/V1d but, if we consider specific consonant classes, the result can be even much better (for example, with parameter C1d/V1d there are no mistake for [ʤ] and with parameter Cd there is only a mistake percentage of 4.63% for [ʧ]).

	Cd/Utd			Cd/V1d			C1d/V1d			C2d/V1d
	EPP	Errors	Err. %	EPP	Errors	Err. %	EPP	Errors	Err. %	EPP	Errors	Err. %
Overall	0.43	86	19.91	1.6	72	16.67	0.76	60	13.89	0.78	131	30.32
Male	0.42	40	18.52	1.68	30	13.89	0.69	21	9.72	0.78	66	30.56
Female	0.43	45	20.83	1.55	40	18.52	0.85	36	16.67	0.78	65	30.09
[ʧ]	0.44	15	13.89	1.6	9	8.33	0.65	9	8.33	0.87	20	18.52
[ʤ]	0.37	6	5.56	1.26	3	2.78	0.91	0	0.00	0.36	17	15.74
[ʦ]	0.49	31	28.70	1.9	26	24.07	0.77	29	26.85	1.41	28	25.93
[ʣ]	0.42	15	13.89	1.45	14	12.96	0.76	13	12.04	0.75	18	16.67
[a]	0.43	23	15.97	1.29	19	13.19	0.74	15	10.42	0.78	38	26.39
[i]	0.48	32	22.22	1.75	25	17.36	0.92	22	15.28	1.13	41	28.47
[u]	0.44	27	18.75	1.57	21	14.58	0.68	18	12.50	0.87	41	28.47
	V1d			C1d			C2d			Cd
	EPP	Errors	Err. %	EPP	Errors	Err. %	EPP	Errors	Err. %	EPP	Errors	Err. %
Overall	136	101	23.38	102	73	16.90	129	157	36.34	208	76	17.59
Male	135	47	21.76	93	26	12.04	146	82	37.96	215	36	16.67
Female	135	54	25.00	112	41	18.98	129	67	31.02	208	37	17.13
[ʧ]	137	20	18.52	95	10	9.26	139	30	27.78	224	5	4.63
[ʤ]	136	15	13.89	125	5	4.63	67	34	31.48	166	8	7.41
[ʦ]	124	34	31.48	91	27	25.00	157	28	25.93	244	19	17.59
[ʣ]	153	28	25.93	100	17	15.74	108	28	25.93	201	11	10.19
[a]	145	32	22.22	104	17	11.81	107	53	36.81	198	23	15.97
[i]	122	33	22.92	102	24	16.67	139	48	33.33	235	28	19.44
[u]	135	27	18.75	94	24	16.67	129	54	37.50	207	24	16.67
TABLE IV Results of the unidimensional Maximum Likelihood Criterion used for classifying singleton vs. geminate.

2.2 Results in the frequency domain

In Table V there are the average frequency parameters values and their standard deviations obtained by working out the average of all utterances. All measured values and other interesting average results can be found in (Faluschi, 2000). On the basis of a statistical analysis about a multivariated ANOVA, there are not statistically meaningful differences between taken measurements except for slight changes in the fundamental frequency (F0) when it is measured in very specific frames. F0 is about 14 Hz and 12 Hz higher in the geminate form (+9% and +8%) in the V1 offset and V1 to C transition frames respectively. No significant variations were observed in F1, F2 and F3. Formant amplitudes A1, A2 and A3 are significantly higher in their geminate form (1-3 dB) in V1 center, V1 offset, V1 to C transition and V2 onset frames. Performing a Maximum Likelihood Criterion based on the statistically significative frequency parameters, we are led to about 64% of correct classifications, an average too low.

2.3 Results in the energy domain

Table VI reports the data about the energy-based parameters. An Anova mulivariate test was performed to determine if statistically significative differences between averages exist. Bold characters indicate significantly different values. A general tendency is to pronounce geminate words with more emphasis than singleton ones. This is confirmed also by the previous analysis were there is a higher amplitude of formants in geminate utterances. Performing a Maximum Likelihood Criterion based on the energy parameters, leads to a poor average 61.7% of correct classification. In this case too, the average percentage of mistakes is too high.

		V1 CENTER
		F0		A0		F1		A1		F2		A2			F3		A3
Singleton		160		13		550		34		1581		32			2895		26
(StD)		44		6		310		9		777		8			466		11
Geminate		169		13		552		36		1588		34			2870		26
(StD)		43		7		303		8		769		8			548		11
		V1 OFFSET
		F0		A0		F1		A1		F2		A2			F3		A3
Singleton		151		13		487		29		1684		27			2877		23
(StD)		44		5		242		8		679		9			469		9
Geminate		165		14		506		32		1694		30			2792		25
(StD)		43		7		256		7		683		8			656		10
		V1 TO C TRANSITION
		F0		A0		F1		A1		F2		A2			F3		A3
Singleton		148		12		448		26		1716		22			2864		21
(StD)		43		5		201		8		638		8			469		7
Geminate		160		13		466		27		1728		25			2778		22
(StD)		44		5		220		7		651		8			648		9
		C1 ONSET / C1 CENTER / C2 CENTER / C2 OFFSET
		F0		A0		F0		A0		F0		A0			F0		A0
Singleton		139		11		126		9		112		9			120		10
(StD)		36		5		32		5		23		4			27		5
Geminate		70		5		62		4		40		3			48		4
(StD)		78		7		69		6		58		6			64		7
		V2 ONSET
		F0		A0		F1		A1		F2		A2			F3		A3
Singleton		134		10		413		28		1724		25			2758		22
(StD)		32		5		167		6		508		7			409		7
Geminate		138		11		415		29		1743		27			2689		23
(StD)		33		5		162		7		488		7			614		8
		V2 CENTER
		F0		A0		F1		A1		F2		A2			F3		A3
Singleton		130		10		476		30		1617		27			2798		21
(StD)		31		5		258		6		653		7			421		8
Geminate		133		10		480		31		1626		28			2724		21
(StD)		32		5		250		7		644		7			627		9
TABLE V Frequency-based parameters. Mean values and Standard Deviation with respect of all the repetitions, speakers, vowels and consonants. Frequencies are in Hz, amplitudes in dB.
			EtotV1		PmV1		EtotC1		PmC1		EtotC2		PmC2	EtotC		PmC
	Singleton		93.7		62.1		75.9		47.1		78.4		49.1	81.4		59.1
	(StD)		6.0		5.5		5.9		6.0		4.8		5.3	4.1		4.5
	Geminate		94.1		63.7		76.9		45.8		80.9		50.5	83.4		59.4
	(StD)		5.7		5.2		6.0		5.9		5.1		5.5	4.5		4.7
					EiV1cent		EiV1-C1		EiC1cent		EiC1-C2		EiC2cent	EiC2offs
	Singleton				86.8		79.4		67.3		68.6		73.1	72.5
	(StD)				5.9		5.1		9.0		6.6		6.2	5.1
	Geminate				88.8		80.9		63.2		68.2		74.3	74.5
	(StD)				5.3		4.7		9.5		6.2		6.6	5.5
	TABLE VI Energy-based parameters. Mean values and Standard Deviation with respect of all the repetions, speakers, vowels and consonants. All values are in dB. Bold characters indicate significantly different values.

3 Discussion and conclusion

First of all, we are obliged to introduce a foreword. Affricate consonants have peculiar and very distinctive characteristics that made necessary a split of the consonant itself into two parts: the first is named C1 and it indicates the stop phase while the second, named C2, indicates the fricative phase. This distinction had not been necessary in the previous studies about other Italian consonants and this is why it will not always be possible to compare the results of this study with those of the previous ones.

In time domain there is evidence for a strong connection between gemination and phonemes duration. In particular, as it was for previous studies about gemination of Italian consonants, it was evident, moving from singleton to geminate utterances, an increase of consonant duration (both in the C1 stop phase and C2 fricative phase) and a decrease of pre-consonant vowel duration, while there are not significant variations in second vowel (V2) duration. Even the total duration of the utterance (Utd) moves up in geminate form, although not in such a significant way such as in V1, C1 and C2. This leads to suppose that a compensation effect (even if incomplete) takes place between V1 and C durations. This compensation effect is the element that tends not to make the phonetic rhythm change too much.

In the frequency domain, F0 increases passing from singleton to geminate only in two frames, an average quantity of about 13 Hz that is perceptively relevant. To vary are also the amplitudes of the three measured formants, A1, A2 and A3 in every analysis frame, except for V2 center. These three are higher of 2 dB average in geminate utterances compared to singleton ones.

As far as the energy and power of pronunciations are concerned, we see the tendency to give more emphasis to geminates instead of singletons.

For affricates it was found that the average difference between singles and geminates in terms of V1d is 38 ms (»-25% for geminates), in C1d is 51 ms (»+62%for geminates), in C2d is 27 ms (»+28% for geminates) while considering the whole consonant Cd=C1d+C2d the difference is 78ms (»+44% for geminates). For fricatives (Giovanardi M., 1998) the average V1d difference was 49 ms (»-28% for geminates) and in Cd 98 ms (»+73% for geminates). As regards stops (Esposito A., Di Benedetto M.G., 1999), V1d difference was 43 ms (»-26% for geminates) and 92 ms for the stop closure duration (+101% for geminates). Finally, for nasals (Mattei M., Di Benedetto M.G., 1999) the average V1d difference was 59 ms (»-32% for geminates) and in Cd 121 ms (»+134% for geminates).

The straightforward comparisons that can be done with previous studies are between the C1 occlusive phase of the affricate consonants and the stop consonants (Esposito A., Di Benedetto M.G., 1999) and also between the C2 fricative phase of the affricate consonants and the fricative ones (Giovanardi M., 1998).

There is evidence to assert that differences between singleton and geminate durations are less sharp and clear than in other Italian consonants.

The singleton/geminate classification based on the Maximum Likelihood Criterion using the Cd and Cd/V1d parameters leads to the following results, belonging to previous studies:

1) 0.47% of errors in nasals for both parameters

2) 12 % of errors in fricatives for both parameters

3) 4% and 8% of errors in stops for Cd and Cd/V1d, respectively

In Table IV, it is possible to see that mistake percentages for affricates are worse compared to those obtained for other classes of consonants.

Finally, an additional similarity between nasals, fricatives, and stops was found; the Spearman Rank correlation coefficient between Cd and V1d was for all these consonants equal to values from -0.71 to -0.78, while for affricates this value is lower.

As a general conclusion, the most relevant outcomes of the present work can be summarized as follows.

The general tendency of shortening the pre-consonant vowel and of lengthening the consonant in geminate utterance, observed on stops, fricatives and nasals in previous studies, is confirmed for affricates.

In previous studies, significant variations in the energies of utterances have not been observed, except for the total energy of the consonant in nasals that is stronger in the geminates and it is due, as far as the author is concerned, to a longer duration of the consonant while the tendency for the affricates is to emphasise the geminate utterance (even if differences are limited to very few dB).

In conclusion, we want to point out that there are studies in progress concerning synthetised utterances of affricate consonants obtained using the articulatory synthetiser HLsyn. Moreover, the project will pilot an experiment of perceptive analysis in order to support the idea that, as there is evidence in this study, the duration of phonemes is the most important factor to distinguish a singleton utterance from a geminate one. In order to illustrate this theory we insert below two link-buttons: the first connects to the singleton synthetised utterance of the [ʧ] articulated with the [a] vowel, while the second connects to the corresponding geminate synthetised utterance.

References

Abramson A. S. (1999) "Fundamental frequency as a cue to word-initial consonant length: Pattani Malay" ICPhS99 San Francisco pp 591-594.

Argiolas F., Macrì F., Di Benedetto M.G. "Acoustic analysis of Italian [r] and [l]," Journal of the Acoustical Society of America 97, no. 5, pt.2, pp.3418, 1995.

Arvaniti A. (1999) "Effects of speaking rate on the timing of single and geminate sonorants" ICPhS99 San Francisco pp 599-602.

Blumstein S.E., Pickett E., Burton M. (1998). “Effects of speaking rate on Singleton/Geminate consonant contrast in Italian”, unpublished manuscript.

Canepari L. (1979) "Introduzione alla fonetica" Piccola Biblioteca Einaudi.

Cohn A.C., Ham W.H., Podesva R. J.(1999) "The phonetic realization of singleton-geminate contrasts in three languages of Indonesia" ICPhS99 San Francisco pp 587-590.

Esposito A., Di Benedetto M.G (1999), "Acoustic and Perceptual Study of Gemination in Italian Stops", Journal of the Acoustical Society of America.

Faluschi S. (2000) "Analisi Acustica e Sintesi delle consonanti affricate singole e geminate in Italiano" Degree Thesis in Electronics Engineering, Rome University "La Sapienza".

Giovanardi M. (1998) "Analisi Acustica e Sintesi delle consonanti fricative singole e geminate in Italiano" Degree Thesis in Electronics Engineering, Rome University "La Sapienza".

Giovanardi M. and Di Benedetto M.G. (1998)." Acoustic analysis of singleton and geminate fricatives in Italian" European student journal of language and speech.

Local J.and Simpson A.P. (1999) "Phonetic implementation of geminates in Malayalam nouns" ICPhS99 San Francisco pp 592-595.

Louali N.and Maddieson I. (1999) "Phonological contrast and phonetic realization: the case of Berber stops" ICPhS99 San Francisco pp 603-606.

Mattei M. and Di Benedetto M.G. (2000) "Acoustic analysis of singleton and geminate nasals in Italian" European student journal of language and speech.

Mattei M. (1999) "Analisi Acustica delle consonanti nasali singole e geminate in Italiano" Degree Thesis in Electronics Engineering, Rome University "La Sapienza".

Muljacic Z. (1972). "Fonologia della lingua Italiana", (Il Mulino, Bologna).

Rochet,L.B., and Rochet,A.P. (1995) “The perception of the single-geminate consonant contrast by native speakers of Italian and Anglophones” in proceedings of ICPhS95, edited by K.Elenius and P.Branderud,Vol 3 (Arne Strombergs, Stockholm ) pp. 616-619.

Rossetti R. (1994) "Gemination of Italian stops", Journal of the Acoustical Society of America, 95, 2pSP25, pp.2874.

Vecsys (1989). The Unice User Manual (Vecsys - Chemin du Chene rond - 91570 Bièvres, France).

Shrotriya N., Siva Sarma A.S., Verma R., Agrawal S.S. (1995). “Acoustic and perceptual characteristics of geminate Hindi stop consonants”, in Proceedings of ICPhS95, edited by K.Elenius and P.Branderud, 4, (Arne Strombergs Grafiska Stockolm), pp.132-135.

Vecsys (1989). The Unice User Manual (Vecsys - Chemin du Chene rond - 91570 Bièvres, France).