What's new, pussycat? On talking to babies and animals
What's new, pussycat? On talking to babies and animals
复制标题
DOI：
10.1126/science.1069587
复制
发表时间：
2002-05-24
期刊：
影响因子：
56.9
通讯作者：
Vollmer-Conna, U
中科院分区：
文献类型：
--
作者：
Burnham, D;Kitamura, C;Vollmer-Conna, U
与婴儿交谈时，成年人总是使用特殊的语音记录，其特点是基频（音调）升高、语调轮廓夸张和情感丰富 (1, 2)。研究还发现，母亲在对婴儿说话时元音发音过多，但在与其他成年人交谈时却不会（3）。这种现象无处不在，发生在各种语言中——英语、俄语、瑞典语和日语 (3, 4)——并且被认为可以通过放大母语元音的语音特征来促进婴儿的语言发展 (3)。然而，这种言语风格的普遍存在意味着几乎不可能获得其作为语言教学工具的直接证据。显然，我们不能要求护理人员不要对婴儿使用儿语，因为它似乎是自动引发的。因此，由于语音输入的性质无法改变，我们决定从另一个角度来解决这个问题——通过实验操纵接收者的性质。之前已经注意到宠物与婴儿的语音惊人的相似性（5），尽管尚未尝试对音调或情感语音成分进行客观比较。针对宠物和婴儿的语音之间的这种相似性是否意味着当我们与宠物交谈时也会发生元音过度发音？我们是否（也许是无意识地）试图教我们的动物如何说话或至少理解我们的语言？或者，元音过度发音可能只是我们对婴儿和宠物使用高度情绪化语言的副产品。为了解决这个问题，我们在三个领域对 12 位母亲与其婴儿、宠物和另一位成人的言语进行了客观比较：(i) 音调，这是基频的心理相关性；(ii) 情感，通过低通滤波的言语评级来衡量，可以听到语调和节奏，但无法理解单词； (iii) 元音过度发音，通过绘制“角”元音/i/、/u/和/a/的第一和第二共振峰（F1和F2）值并比较所得元音三角形来客观化（3）。这些母亲的语音样本都是以澳大利亚英语为母语的单语人士，是用便携式专业随身听（索尼）录制的，母亲们在家中随身携带了领夹麦克风。为了获得必要的角元音信息，我们要求母亲们与每个接受者进行 10 到 15 分钟的自然互动，并玩耍并说出三个提供的玩具：一只“绵羊”、一只“鞋子”和一只“鲨鱼”。母亲们在自己的时间内分别录制与 6 个月大的婴儿、宠物猫或狗以及另一位成人交谈的内容（参见补充文本和表 S1）。对音高、情感和元音三角形进行方差分析，以测试婴儿、宠物和成人定向言语的差异（在与五只猫和七只狗的言语中没有发现差异）。对于音调（图 1A），婴儿和宠物的言语在统计上相当 [F (1, 11) 0.03，P < 0.05]，但对婴儿的言语音调 [F (1, 11) 6.58，P < 0.05] 和宠物的言语音调 [F (1, 11) 36.52，P < 0.001] 显着高于对成人的言语音调。对五个等级的低通滤波语音（参见音频 S1、S2 和 S3）的评级进行了因子分析，并得出了由此产生的影响因子的分数。婴儿的情感比针对宠物的言语更大[F (1, 11) 10.76, P 0.01]，但婴儿的情感[F (1, 11) 94.34, P 0.001]和宠物的言语[F (1, 11) 54.44, P 0.001]都高于成人的言语（图1B）。婴儿、宠物和成人语音的元音绘制在图 1C 的 F1-F2 空间中。母亲的元音三角形区域在婴儿导向中
When talking to babies, adults invariably use a special speech register characterized by elevated fundamental frequency (pitch), exaggerated intonation contours, and high affect (1, 2). It has also been found that mothers hyperarticulate vowels when addressing their infants but not when speaking to other adults (3). This phenomenon is ubiquitous, occurring across various languages—English, Russian, Swedish, and Japanese (3, 4)—and is thought to facilitate infants’ linguistic development by amplifying the phonetic characteristics of native language vowels (3). However, the very ubiquity of this speech style means that it is practically impossible to obtain direct evidence of its function as a language-teaching device; clearly, we cannot ask caregivers not to use baby-talk with infants, as it appears to be elicited automatically. So, as the nature of the speech input cannot be changed, we decided to approach this issue from another angle—by experimentally manipulating the nature of the recipients. The uncanny similarity of pet-to infant-directed speech has been noted previously (5), although no objective comparison of either pitch or affective speech components has been attempted. Does this similarity between pet-and infantdirected speech imply that vowel hyperarticulation also occurs when we talk to our pets? Are we (perhaps unconsciously) trying to teach our animals how to speak or at least understand our language? Or maybe vowel hyperarticulation is simply a by-product of the highly emotional speech we use to both our infants and pets. To resolve this issue, we made objective comparisons of 12 mothers’ speech to their infant, their pet, and another adult in three domains:(i) pitch, which is the psychological correlate of fundamental frequency;(ii) affect, which is measured by ratings of low-pass–filtered speech, in which the intonation and rhythm can be heard but the words cannot be understood; and (iii) vowel hyperarticulation, which is objectified by plotting first and second formant (F1 and F2) values of the “corner” vowels,/i/,/u/, and/a/, and comparing the resultant vowel triangles (3). Speech samples of the mothers, all monolingual native speakers of Australian English, were recorded on a portable Professional Walkman (Sony) with lapel microphone left with the mothers in their homes. To obtain the requisite corner vowel information, we asked mothers to play with and name three provided toys, a “sheep,” a “shoe,” and a “shark,” in naturalistic 10-to 15-min interactions with each recipient. Mothers made separate recordings in their own time talking to their 6-month-old infant, to their pet cat or dog, and to another adult (see supplementary text and table S1).Analyses of variance were conducted for pitch, affect, and vowel triangles to test differences in infant-, pet-, and adult-directed speech (no differences were found in speech to the five cats and seven dogs). For pitch (Fig. 1A), infantand pet-directed speech was statistically equivalent [F (1, 11) 0.03, P 0.05], but pitch in speech to both infants [F (1, 11) 6.58, P 0.05] and pets [F (1, 11) 36.52, P 0.001] was significantly higher than pitch in speech to adults. Ratings of low-pass–filtered speech (see Audio S1, S2, and S3) on five scales were factor analyzed, and scores from the resultant affect factor were derived. Affect was greater in infant-than in pet-directed speech [F (1, 11) 10.76, P 0.01], but affect in both infant-[F (1, 11) 94.34, P 0.001] and pet-directed speech [F (1, 11) 54.44, P 0.001] was higher than in adult-directed speech (Fig. 1B). Vowels for infant-, pet-, and adult-directed speech are plotted in F1-F2 space in Fig. 1C. Mothers’ vowel triangle areas in infant-directed